Westheim, Birgitte H.*; Østensen, Anniken B.†; Aagenæs, Ingegerd‡; Sanengen, Truls†; Almaas, Runar*
Percutaneous liver biopsy is commonly used as a diagnostic tool and for assessment and staging of liver disease (1). It plays an essential role in the therapeutic management of liver-transplanted children presenting with biochemical or clinical abnormalities indicating allograft dysfunction.
Complications such as hemorrhage, bile leak, arteriobiliary fistula, sedation-related complications, abdominal pain, pneumothorax, hemothorax, hemoperitoneum, and death have been reported (2–7). Ultrasound-guided liver biopsies are associated with fewer complications and a lower number of inadequate samples compared to blind biopsies, and the mortality rate has decreased (8–10). A higher complication rate has been reported in children compared with adults (7). Although the rate of major complications after ultrasound-guided percutaneous liver biopsies in children is reported to be low, 0% to 4.6% (6,7,11–14), the overall incidences of complications depend on the definition used and range from 5% to 18% (6,7,13,15).
Liver-transplanted children are commonly treated with antiplatelet medication such as acetylsalicylic acid (ASA) to prevent thrombosis of the hepatic artery. Treatment with antiplatelet medication is considered a relative contraindication to percutaneous liver biopsy. The indication for doing a liver biopsy often arises rapidly and an antiplatelet-free window before intervention is not possible. Thus, the clinician must choose between performing a liver biopsy while the patient is taking ASA or to make a judgment without liver biopsy as suggested by Sornsakrin et al (13). Whether ASA increases the bleeding risk associated with percutaneous liver biopsy in children has not, however, been systematically investigated. Furthermore, little data exist regarding timing of discontinuation of ASA and bleeding prevention with platelet substitution.
In our department, the surveillance protocol in the last 10 years has included a postbiopsy ultrasound control the day after biopsy to detect any previously undiscovered bleeding or other complications. To our knowledge, the result of a routine control ultrasound the day after percutaneous liver biopsy has not previously been reported in children.
Our aim was to investigate whether treatment with antiplatelet medication and other potential risk factors increase the likelihood of bleeding complications in association with percutaneous liver biopsy in children and assess the value of performing a routine postbiopsy ultrasound the day after the procedure.
The study protocol was approved by the institutional review board at Oslo University Hospital. We retrospectively analyzed all of the percutaneous liver biopsy procedures performed on children in the Department of Pediatrics, Rikshospitalet, Oslo University Hospital, from January 2000 to March 2011. All of the biopsy procedures were reviewed individually. Data were collected from patient medical charts, laboratory results, imaging studies, and anesthesia charts and included patient demographics (age, weight); clinical indication for biopsy; diagnosis at time of review; ultrasound findings before, during, and after biopsy; number of passes; and use of antiplatelet (ASA, nonsteroidal anti-inflammatory drugs) and anticoagulant (warfarin, low-molecular-weight heparin [LMWH]) medication.
Baseline values of hemoglobin (Hb), international normalized ratio (INR), platelet count, and postbiopsy values of Hb were recorded. Recorded sonographic images of 40 procedures were re-evaluated by an experienced pediatric radiologist (I.A.) to confirm the original description.
Biopsy Technique and Surveillance Protocol
All of the biopsies were ultrasound guided and performed by a pediatric interventional radiologist. Coagulation abnormalities were corrected with fresh frozen plasma, prothrombin complex concentrate, or thrombocyte transfusion if necessary. Prophylactic antibiotics were not administered routinely. All of the patients were given general anesthesia and a local subcutaneous administration of Marcain 2.5 mg/mL. A Bard Magnum core biopsy system (Bard Biopsy Systems, Tempe, AZ) was used with either Angiotech (Angiotech Pharmaceuticals, Vancouver, Canada) Ultracore Biopsy Needle 18-G, 100 mm or Bard Magnum disposable core tissue biopsy needle 18-G, 160 mm, with 15 or 22 mm core length.
The surveillance protocol included bed rest for the first hours after the procedure and close clinical and biochemical monitoring of the patients until the next day. Control of Hb was routinely performed at 8:00 PM the day of the biopsy and 8:00 AM the next day. All of the patients were kept in the hospital overnight for routine ultrasound check the next consecutive day.
Outcome and Definitions
Bleeding complications were the major outcome of the study, but other complications were also registered. Bleeding was recorded in patients presenting any signs of postbiopsy bleeding, including unexplained Hb drop (at 8:00 AM) and sonographical indications of bleeding. Bleeding was classified as the following:
1. Major (hemorrhage requiring surgery, Hb drop [>2 g/dL] due to hemorrhage requiring blood transfusion)
2. Minor (sonographically detected minor bleeding in the abdominal cavity or subcapsular hematoma not requiring therapy, unexplained Hb drop [>2 g/dL])
Complications were classified as the following:
1. Major (major bleeding, other complications requiring surgery, intensive care, or medical therapy with prolonged hospital stay)
2. Minor (minor bleeding, abdominal pain requiring oral analgesics, minor anesthetic complications, other minor complications not causing persisting consequences or extending the length of hospital stay)
Findings not classified as complications were sonographic evidence of minimal and localized bleeding at the site of biopsy (n = 16). Two patients had a drop in Hb >2 g/dL at 8:00 AM. A thorough sonographic examination demonstrated no bleeding at the first evaluation or at re-evaluation of recorded images. Furthermore, both patients had other plausible reasons for Hb drop such as hemodilution (1 patient received a high volume of fluid) and a minor bleeding associated with the establishment of a central line. There was no clinical suspicion of bleeding at the time of the procedure, in the follow-up, or at the time of review.
Fifteen patients experienced sedation-related complications. Six had inadequate oxygen saturation immediately after sedation, but all responded to supplemental oxygen. Two patients experienced laryngeal spasms, 1 of whom had an associated bronchoscopy. Three patients were transiently circulatory unstable and responded to either atropin or a fluid bolus. Three patients developed fever without signs of septicemia. One patient developed a rash that was successfully treated with dexchlorpheniramine.
The effect of risk factors on complication rate was studied using the Fisher exact test. Data are given as odds ratio with 95% confidence interval. Univariate regression was also used. Independent factors with a P < 0.10 were included in a multiple regression model. The factors were tested for multicollinearity. Comparison of drop in Hb between children younger than 3 months and older than 1 year was done with an unpaired Student t test. A 2-tailed P < 0.05 was considered statistically significant. Calculations were performed with GraphPad Instat version 03.10 for Windows (GraphPad Software Inc, San Diego, CA).
A total of 275 percutaneous liver biopsies were performed on 190 children. A total of 178 of the biopsies were performed on native livers. A total of 25 of the biopsies were taken from a focal liver lesion/tumor. The mean age was 5.8 ± 5.2 years (range 0.06–17.9). The mean weight was 22 kg (range 2.1–75). A total of 9% of the procedures (n = 26) were performed in children younger than 3 months and 25% (n = 68) of the procedures were performed in children younger than 12 months. A total of 83 of the patients were girls (44%) and 107 were boys (56%). Clinical indications for the biopsies are presented in Figure 1, and primary diagnoses are presented in Figure 2.
Mean Hb level before biopsy was 11.0 ± 1.7 g/dL (n = 275). Mean Hb controlled at 8:00 PM (usually 5–6 hours after biopsy) was 10.7 ± 1.7 g/dL (n = 272) and at 8:00 AM the day after the procedure 10.8 ± 1.7 g/dL (n = 269). A >2.0-g/dL loss in Hb level was recorded in 1.5% of the biopsies (at 8:00 AM). Changes in Hb and complications related to age at procedure are presented in Table 1. Major and minor complications are presented in Table 2.
Mortality rate was 0%. Major bleeding occurred in 3 patients (1.1%), and the total rate of major complications was 1.5% (n = 4). One patient (10 kg), suspected of having an acute rejection after liver transplantation, experienced life-threatening hemorrhage localized to the biopsy site and developed a large (2×6 cm) intrahepatic hematoma. Hb dropped from 11.6 to 5.3 g/dL. A total of 100 IE/kg of LMWH had been administered the day before. The abdominal bleeding required surgical evacuation twice and the patient received several transfusions. He did not have any persisting consequences.
The second patient (25 kg), admitted due to liver failure of unknown origin, received prothrombin complex concentrate immediately before biopsy. Approximately 12 hours later, the patient's condition deteriorated and abdominal cavity hemorrhage was detected and evacuated surgically.
A patient (46 kg) with acute liver failure received a transfusion of platelets, fresh frozen plasma, prothrombin complex concentrate, and erythrocytes before liver biopsy. Twelve hours after the procedure, the patient experienced variceal bleeding and developed encephalopathy. He was stabilized with sclerotherapy and supportive therapy. Whether transfusion before the procedure or hemodynamic changes during and after the procedure could have triggered the variceal bleeding or if the timing was incidental was undetermined.
The fourth patient (17 kg) underwent 8 biopsies from a liver lesion and subsequently experienced abdominal pain, tachycardia, and metabolic acidosis. Ultrasound demonstrated fluid in the abdominal cavity; however, there was no significant drop in Hb (from 8.4–7.8 g/dL). He was transferred to the intensive care unit for blood transfusion and hemodynamic monitoring. The patient experienced no further complications and was transferred back to the ward within hours.
The overall minor complication rate was 25%, including minor sedation-related complications (described in Methods), abdominal pain, and minor bleeding. A total of 28 incidents of minor bleeding (10%), including fluid in the abdominal cavity or subcapsular hematoma detected with ultrasound, were recorded. None of these required any intervention. Patients with minor bleeding had a more pronounced (P = 0.0007) drop in Hb (0.78 ± 0.91) than those without bleeding (0.09 ± 1.00).
Abdominal pain requiring analgesic treatment was registered after 12% of the procedures (n = 32), and 8 of these procedures were associated with bleeding of some kind. An extra ultrasound control was performed after 21 procedures due to abdominal pain or vital changes indicative of blood loss. These controls revealed 11 cases of bleeding. Six of these bleeding incidents had not been visualized during the biopsy procedure. One was a major bleeding incident, described above.
Routine Postbiopsy Ultrasound
Routine control ultrasound was routinely performed the day after the biopsy and 266 (96.7%) radiological descriptions were available. Bleeding was visualized in 7.1% (19/266) of these controls. The bleeding had not been detected clinically or on previous ultrasound imaging in 2.6% (n = 7). Free abdominal fluid was detected in 4 and subcapsular hematoma in 3. None of these minor bleeding incidents required any intervention, changed therapy, or altered length of hospital stay.
Risk Factors for Complications
Focal Space Occupying Lesion/Tumor
A total of 25 of 275 biopsies were taken from a focal lesion such as hepatoblastoma (n = 6), hepatocellular carcinoma (n = 5), and other focal lesions (n = 14). Most procedures included additional biopsies of surrounding liver parenchyma, and samples were obtained with a 22-G fine needle in addition to the routinely used 18-G needle. Mean number of passes was 4.9 (range 1–8). One major bleeding incident, 5 minor bleeding incidents, and 5 other minor complications occurred. The total bleeding rate was significantly higher than in the rest of the population (odds ratio 2.84 [1.04–7.78]; P = 0.047) (Table 3). There was no difference in Hb drop between patients with focal lesions (0.21 ± 0.88) and those without (0.16 ± 1.05; P = 0.82).
The platelet count was <70 × 109 cells/L before 14 procedures. Platelet concentrates were transfused in 11, 4 received prothrombin complex concentrates, and 2 received plasma. There were 3 minor complications, but no bleeding was recorded. INR was ≥1.7 before 18 procedures. All of the patients with INR ≥1.7 received platelets (n = 9), fresh frozen plasma (n = 7), and/or prothrombin complex concentrate (n = 12) before procedure. One patient had a subcapsular hematoma not requiring any therapy, and 4 other minor complications occurred.
Acute Liver Failure
Of 275 biopsies, 12 were taken from patients with acute liver failure. One major and 1 minor bleeding and 3 other complications, including 1 major bleeding, occurred. Patients with acute liver failure had a significantly increased odds ratio for major complications: 26.1 (3.3–204.7; P = 0.01).
ASA treatment (3 mg/kg body weight, maximum 75 mg) had been given the last 5 days before 55 of the procedures. Five had also received LMWH the day before the procedure. Preprocedural treatment was given as platelets (n = 17), fresh frozen plasma (n = 4), and/or desmopressin and tranexamic acid (n = 1).
Six patients had minor bleeding incidents such as subcapsular hematoma or modest amounts of free abdominal fluid not requiring any therapy. No major bleeding occurred. Patients treated with ASA did not have significantly increased odds ratio for all bleeding 0.96 (0.37–2.26) or minor bleeding 1.10 (0.42–8.87) (Table 3). Mean drop in Hb (0.11 ± 0.82) was not significantly different from the mean decrease in the patients without ASA (0.18 ± 1.09) (P = 0.66).
Three of the 17 who had received platelets before the procedure and 3 of the 38 children who did not receive platelets before biopsy had minor bleeding (odds ratio 2.50 [0.45–13.9]; P = 0.36).
In 18 biopsies treatment with LMWH had been given the day before biopsy. One major and 3 minor bleeding and 2 other minor complications occurred. The odds ratio for all bleeding of 2.43 (0.75–7.93) was not significantly different from the rest of the population (P = 0.13). The mean decrease in Hb (0.54 ± 0.69) was not significantly different from those without LMWH (0.14 ± 1.05) (P = 0.11).
A total of 97 liver biopsies were performed in liver-transplanted children. A total of 50 of the 97 biopsies were performed within 4 months after liver transplantation. The most frequent indication for doing liver biopsy on transplanted livers was suspected rejection, followed by suspected posttransplantation lymphoproliferative disease or de novo autoimmune hepatitis (Figure 1). There were 7 minor hemorrhages and 10 other minor complications. Transplanted children did not have an increased risk for bleeding complications compared with the rest of the population. The odds ratio for all of the complications was significantly lower in the transplanted patients compared with the rest of the population 0.52 (0.29–0.96); P = 0.04.
Univariate regression with “all bleeding” (both major and minor bleedings) as the dependent factor demonstrated P < 0.1 for LMWH (P = 0.0997) and focal lesion (P = 0.035) (Table 4). These 2 were included in a multiple regression model (P = 0.02; R2 = 0.028). Although focal lesion (P = 0.024) made a significant contribution, LMWH (P = 0.066) was not significant.
Univariate regression with “major bleeding” as the dependent factor resulted in P < 0.10 for LMWH (P = 0.0498) and acute liver failure (P = 0.022). These 2 were included in a multiple regression model (P = 0.0050; R2 = 0.038). Both LMWH (P = 0.034) and acute liver failure (P = 0.009) made significant contributions.
Univariate regression with “minor bleeding” as the dependent factor resulted in P < 0.10 just for focal lesion (P = 0.089) and no multiple regression model was made.
We found that treatment with ASA 5 days before ultrasound-guided percutaneous liver biopsy in children did not increase the rate of bleeding complications. Routine postbiopsy ultrasound on the consecutive day detected previously unrecognized minor bleeding in 2.6% of the biopsy controls, but none of these required intervention and the findings did not change therapy or the length of hospital stay. The routine postbiopsy ultrasound increases the detection rate of bleeding and thus strengthens the finding that ASA does not increase the bleeding rate.
This is of practical importance because many patients who undergo liver transplants receive low-dose ASA to prevent thrombosis of the hepatic artery. The concern that treatment with ASA may increase the procedure-related bleeding risk may influence the decision whether to perform a liver biopsy or to treat on clinical suspicion. Our finding is in line with the report of Atwell et al (16), who did not find an increased risk of bleeding in 3195 patients taking aspirin within 10 days before percutaneous biopsies (including liver, kidney, lung, and pancreas biopsies). There are several differences in these studies in population (mean age was 5.8 and 56 years, respectively), type of biopsies studied, and methodology, but the conclusions are parallel.
The incidents of minor bleeding in the ASA-treated patients who received fresh platelets before biopsy was slightly higher than those in the patient group who had not. The number of patients treated with platelets was, however, too low to conclude whether pretreatment with fresh platelets is beneficial or adverse in low-dose ASA-treated children and remains to be investigated.
In contrast to the lack of bleeding risk in ASA-treated patients, LMWH treatment the day before the procedure significantly increased the risk of bleeding complications. Heparin administration on the day of biopsy has been demonstrated to be an independent risk factor for bleeding after liver biopsy in an adult population (17). Thus, extra care and surveillance should be provided to patients recently treated with LMWH when a percutaneous liver biopsy is indicated.
The risk of bleeding is greatest within the first 2 hours after the procedure (18), but delayed bleeding up to 18 days has been reported (19). Terjung et al (17) found that 70% of the major and 34% of the minor bleeding incidents were detected later than 24 hours in an adult population. Because that study did not routinely include postbiopsy ultrasound, the timing of the bleeding was unknown. In a prospective study of 40 adult patients, 23% had ultrasound-detected hematomas after percutaneous liver biopsy after 24 hours (20). In that study, a follow-up ultrasound at day 7 did not find new hematomas in a randomly selected subgroup (n = 10). In children, no previous studies have evaluated the value of doing postbiopsy ultrasound routinely. In the present study, routine postbiopsy ultrasound discovered minor bleeding in 7.1% of the cases, and in 2.6%, postbiopsy ultrasound revealed unsuspected bleeding, but none of these required intervention. The routine postbiopsy ultrasound was in addition to the frequent use of extra ultrasounds in the first hours after the procedure performed on clinical suspicion of bleeding. Postbiopsy ultrasound did not change therapy, length of hospital stay, or handling of the patients. The results of the present study question the value of doing postbiopsy routinely.
Although the value of doing postbiopsy ultrasound is debatable, more subclinical bleeding otherwise unrecognized is detected and affects the complication rate. In contrast to the higher rate of recorded minor complications, the incidence of major complications in the present study (1.5%) is low compared with other studies (2.4%–4.6%) (6,7). The recorded complication rate depends not only on the quality of the procedure but also on the definition of minor complications and how closely the patients are monitored and examined after the procedure.
The recorded rate of bleeding complications is affected by the performance or not of routine postbiopsy ultrasound and whether symptoms such as abdominal pain and Hb drop are regularly followed up with an extra ultrasound.
The frequency of complications also varies with the population studied. It has been proposed that infants are at higher risk for bleeding complications than older children (7,15). We found a significantly more pronounced drop in Hb in infants younger than 3 months compared with older children between 1 and 18 years old. Intravenous fluid is administered more frequently and during a prolonged period of time in the youngest population, and thus hemodilution may account for a part of the difference; however, children younger than 3 months had a higher frequency of minor bleeding compared with children older than 1 year.
All of the major complications in the present study were associated with risk factors such as acute liver failure, biopsy from a focal lesion, or LMWH the day before biopsy. Terjung et al (17), in an (mostly) adult population, found that 7 of 10 of the clinically overt bleeding incidents occurred in patients with an increased a priori bleeding risk, and they found an increased risk of bleeding after liver biopsy in patients with acute liver failure. In line with this, we found that acute liver failure was a significant risk factor for major complication.
Biopsies taken from a focal space-occupying lesion are at particular risk, with a 2.8 odds ratio for bleeding. Furthermore, focal lesion was identified as an independent risk factor for postbiopsy bleeding. Cohen et al (2) reported a 5-fold increase in transfusion rate in patients with cancer or bone marrow transplantation. One of the factors for the increased risk may be the increased number of passes.
Liver-transplanted children had a significantly reduced frequency of complications after percutaneous liver biopsy compared with the rest of the population. This may seem contradictory to the a priori risk with frequent use of medication of LMWH and ASA; however, the present study did not indicate an increased risk associated with ASA. Moreover, the liver-transplanted group included fewer patients with other risk factors such as focal lesions and acute liver failure. Our incidence of 1% of major complications in liver-transplanted children is comparable to the 1.7% of Sornsakrin et al (13).
Ultrasound-guided percutaneous liver biopsy in children has a low incidence of major complications. Minor bleeding is frequent and is detected both on extra ultrasounds indicated by symptoms and on routine postbiopsy ultrasound the day after the procedure. Routine postbiopsy ultrasound the day after the procedure did not reveal any unexpected major bleeding requiring therapy that had not been detected earlier with on-demand ultrasound. The results of the present study challenge the value of performing postbiopsy routinely on patients with normal a priori risk for bleeding; although the majority of these minor complications required no or minimal therapy, it is important to be aware that small bleeding occurs frequently and the avoidance of major bleeding requires awareness of risk factors.