Interventional Radiology
Original Research
Imaging and Percutaneous Treatment of Secondarily Infected Hepatic Infarctions
Audio Available | 
Share
OBJECTIVE. The objective of our study was to describe the imaging features and success rate of percutaneously treated infected hepatic infarctions.
MATERIALS AND METHODS. Three hundred ninety-two patients had percutaneous liver abscess aspiration and drainage or aspiration and intraoperative débridement at our institution between 1990 and 2003. One hundred fifty-one of these patients underwent CT at least 2 days before the drainage procedure and immediately before the procedure. Retrospective review of the imaging and medical records identified 13 patients with microbiologically documented liver abscesses who had liver lesions consistent with hepatic infarction on the baseline CT.
RESULTS. Twenty-one hepatic infarctions in 13 patients were documented on baseline CT, 15 of which became secondarily infected. Ten of 15 patients with infected infarctions had undergone either hepatic transplantation or the Whipple procedure. Although the left lobe was slightly more commonly infarcted than the right lobe (54% vs 46%, respectively), right lobe infarctions were more commonly superinfected than left lobe infarctions (61% vs 39%); however, neither of these distinctions was statistically significant. Twelve of 13 patients underwent percutaneous drainage. The duration of catheter drainage was significantly longer in patients in whom catheter drainage was complicated by biliary communication than those without biliary communication (61 vs 19 days, respectively). Eleven of 12 patients (92%) responded to drainage such that they survived to discharge from the hospital.
CONCLUSION. Patients with hepatic infarctions are at risk for secondary infection, particularly those patients having undergone surgery involving the porta hepatis. Percutaneous abscess drainage can be performed safely with excellent technical and clinical outcomes in this complex patient population.
Keywords: abscess drainage, hepatic infarction, hepatic transplantation, polymicrobial infections, superinfection, Whipple procedure
| Introduction |   |
|---|
The increased use of CT in the evaluation of the postoperative patient has led to the increased recognition of disease processes once considered rare [1, 2]. One such disease process is hepatic infarction [3]. Although the liver parenchyma is partially protected from infarction because of its dual blood supply, several reports have documented the increased detection and the CT characteristics of these lesions [4, 5]. Hepatic infarctions appear as well-defined, hypodense, wedge-shaped, and geographic lesions extending from the hepatic hilum to the hepatic capsule [6–10]. Infarctions result from hepatic hypo perfusion with or without associated vascular occlusion, either hepatic arterial or portal venous. The patient population who develop hepatic infarction has not been clearly delineated, but an association of hepatic infarction with surgery involving the porta hepatis, including hepatic transplantation and pancreaticoduodenectomy (Whipple procedure), is suspected [11, 12].
The complicated postsurgical patient who develops hepatic infarction may be susceptible to additional complications, including infection. Secondary infection of infarcted parenchymal tissue may result in liquefaction and abscess formation [13]. The imaging appearance of hepatic abscesses on CT has been well described as a lesion with a thin enhancing rim with hypoattenuating nonenhancing internal contents [14–17]. Both portal venous and hepatic venous thrombosis may be associated with hepatic abscess formation [18]. Once parti ally liquefied, infected hepatic infarctions may be suitable for percutaneous abscess drainage [19–22]. Multiple studies have documented the safety and efficacy of percutaneous abscess drainage, although complications including postprocedure sepsis do occur [23].
Percutaneous drainage has long been the accepted primary therapy of parenchymal organ abscesses, including those of the liver, but no reports published to date describe the clinical background, imaging appearance, or outcome of patients with percutaneously drained infected hepatic infarctions. The purpose of this investigation was to describe the patient population, imaging characteristics, and success rate of percutaneously drained hepatic infarctions that have become abscesses as the result of secondary infection.
| Materials and Methods | |
|---|
Before initiation of this research activity, appropriate institutional review board (IRB) approval was obtained for data collection limited to medical records, including radiologic images. The IRB waived patient informed consent for this retrospective study, which was conducted in compliance with HIPAA.
For the purposes of this investigation, the baseline CT scan was defined as a scan that was obtained at least 2 days before the aspiration and drainage procedure or aspiration and intraoperative débridement (range, 2–53 days), whereas the preprocedure CT scan was defined as the scan obtained within 24 hours of the procedure. A search of the interventional radiology procedures database yielded 392 patients who had undergone percutaneous liver needle aspiration and catheter drainage or needle aspiration and intraoperative débridement from January 1, 1990, to December 31, 2003, at our institution. Review of the radiology records of these 392 patients identified 151 patients with both an immediate preprocedure contrastenhanced CT scan within 24 hours of the procedure and an additional baseline CT at least 2 days before the procedure. Retrospective review of the imaging and medical records identified 13 patients with microbiologically documented liver abscesses who had liver lesions consistent with hepatic infarction on the baseline CT.
Hepatic infarctions were defined on the basis of their appearance on the initial CT scans, as described in the literature [6–9]. These imaging features include at least one of the following five features: wedge-shaped, hypodense, geographic in contour, well-defined border, and extension from the perihilar region to the hepatic capsule. Changes in the appearance of the hepatic infarctions between the two CT examinations were assessed and scored as described later in this article. Hepatic lesions smaller than 2 cm were excluded from analysis.
The baseline CT scans in all 13 patients were retrospectively reviewed to determine the size, number, location, segmental involvement, and radiographic appearance of the hepatic infarctions by consensus of two staff radiologists with experience in gastrointestinal and interventional radiology. If hepatic infarction involved any portion of a segment, the segment was scored as involved. The preprocedure CT scans of all patients were also evaluated retrospectively to assess for a change in the appearance of infarction, including rounding of the lesion, development of an enhancing wall, and intralesional formation of gas. Changes in the size of the lesions when compared with baseline and segmental involvement were also tabulated on the preprocedure scan.
Statistical analysis was also performed. Because individual patients may have had more than one infarct and infarcts may involve multiple segments, a chi-square analysis that accounted for this clustering could not be performed because of the limited sample size. Acknowledging these limit ations, we still performed a chi-square analysis to assess statistical significance.
Due to the duration of this review (1990–2003), several generations of imaging equipment were used. The CT scanners used ranged from singledetector nonhelical scanners (three patients) to 4- to 16-MDCT scanners (10 patients). All of the scanners used (HiLight Advantage and LightSpeed) were manufactured by GE Healthcare. Of the 13 patients, 11 patients underwent at least one IV contrast-enhanced examination (baseline scanning, preprocedure scanning, or both) before percutaneous abscess aspiration and catheter drainage or aspiration and débridement were performed. IV contrast material was not used in two hepatic transplant patients due to the preference of the referring transplant surgery service.
For the one patient who received IV contrast material and was imaged using single-detector nonhelical CT, 100–150 mL of iodinated contrast material was injected at 2 mL/s using a scanning delay of 40 seconds and 5-mm contiguous slices. For the 10 patients who were scanned with helical CT, 100–150 mL of nonionic contrast material was injected at 2.5 mL/s with a scanning delay of 70 seconds using 5-mm contiguous slices. If IV contrast material was administered, delayed scans through the liver were usually obtained because that is the general policy of our institution's CT technologists in the setting of hepatic findings. The iodinated contrast medium used was one of the following three agents: iohexol (Omnipaque, GE Healthcare), iopromide (Ultravist, Bayer HealthCare), or iopamidol (Isovue, Bracco Diagnostics).
Percutaneous procedures were performed by one of 10 fellowship-trained interventional radiologists working with a fellow training in either abdominal or interventional radiology. All patients underwent initial needle aspiration with either an 18or a 20-gauge needle. Twelve of the 13 patients underwent percutaneous abscess drainage at the same time as the initial needle aspiration. In the remaining patient (patient 12 in Table 1), the results of the Gram stain of the aspirate were reviewed with the transplant surgeon, who chose immediate surgical débridement rather than percutaneous drainage. The interventional radiologist determined the approach, number of catheters inserted, and catheter size, ranging from 8- to 14-French, depending on the nature of the aspirated material, location in the liver, and number of loculations. Standard interventional radiology methods were implemented using a tandem-trocar technique under either sonography or CT guidance.
The interventional treatment success was divided into technical and therapeutic components. The procedure was considered a technical success if the catheter was placed in the appropriate location and material drained through the catheter. Therapeutic success was defined as a positive patient response to percutaneous abscess drainage with improved clinical condition at discharge from hospital. For each patient, the total number of catheters used, the presence of biliary communication with the abscess cavity, and the duration of catheter drainage were tabulated. Biliary communication was documented either by direct catheter injection of the abscess catheter under fluoroscopy or by visual inspection of both the volume and bilious appearance of the drainage contents.
The patients' medical records were reviewed to determine the cause for the initial hepatic infarction and long-term clinical outcome. Laboratory, clinical, and microbiologic data, including species of infecting organism, were collected and tabulated.
Statistical analysis, including the chi-square test and one-sided Wilcoxon's rank sum test, was performed with the aid of SAS software (version 9.1, SAS Institute).
| Results | |
|---|
Of the 13 patients included in the study population, seven (54%) were men and six (46%) were women, with ages ranging from 29 to 76 years old (Table 1). The average duration from baseline to preprocedure CT ranged from 2 to 53 days, with a mean of 20 days between the two CT examinations. At CT immediately before the procedure, eight patients (62%) had a fever of > 100.5° F (38.1° C) and a WBC count that ranged from 4,500 to 41,600 cm3, with 10 of 13 patients (77%) having a WBC count of > 11,000 cm3.
Five patients (38%) with infected infarctions had undergone a Whipple procedure, five patients (38%) had undergone hepatic transplantation, one patient (8%) had severe pancreatitis secondary to unresectable distal extrahepatic cholangiocellular carcinoma, one patient (8%) had recently undergone chemo therapeutic hepatic arterial embolization, and one patient (8%) had undergone radiofrequency ablation of liver metastasis of unknown primary. In six patients (46%), the hepatic artery was occluded, which was determined either by lack of hepatic arterial enhancement on contrast-enhanced CT (four patients) or by concurrent hepatic arteriography (two patients). Pneumobilia, identified by linear branching patterns of air within the nondependent liver, was present in six patients (46%).
Overall, 21 infarctions involving 35 individual segments were identified in the 13 patients. On the baseline CT of the 13 patients, eight (62%) had one infarction, two (15%) had two infarctions, and three (23%) had three infarctions. The average size of the infarctions was 7.5 cm in maximum axial diameter (range, 3–17 cm), and no infarction had internal foci of gas on the baseline CT. Eleven of 13 patients (85%) had lesions that were peripherally rather than centrally located in the liver. The left lobe of the liver tended to be more commonly infarcted than the right lobe, with 54% (19/35) versus 46% (16/35) involvement, respectively; however, a chi-square analysis showed that this difference was not statistically significant.
 View larger version (169K) | Fig. 1A —72-year-old man (patient 5 in Table 1) who underwent Whipple procedure for pancreatic carcinoma and presented with hepatic infarction that shows interval rounding of secondarily infected infarction when compared with baseline infarction. Contrast-enhanced CT scan shows subtle peripheral wedge-shaped hypodensity (arrow) in segment VI of liver; this finding is consistent with hepatic infarction. |
 View larger version (173K) | Fig. 1B —72-year-old man (patient 5 in Table 1) who underwent Whipple procedure for pancreatic carcinoma and presented with hepatic infarction that shows interval rounding of secondarily infected infarction when compared with baseline infarction. Contrast-enhanced CT scan obtained 53 days after A shows interval rounding (arrow) of segment VI lesion with rim enhancement; these findings are consistent with infected infarction. Purulent fluid from subsequent percutaneous abscess drainage grew α-hemolytic Streptococcus organisms. |
The 21 infarctions involved a total of 35 individual segments, comprising approximately 34% (35/104 segments) of the segmental anatomy of the liver or an average of 2.7 segments per patient. Segments II and III were the most common segments to be involved, with 54% (7/13) of the patients having at least partial infarction of these segments. Segments II and III were also more likely to be infarcted in the postoperative patient subgroup, with 70% (7/10) having infarction versus 0% (0/3) of the nonsurgical patients.
Of the 21 infarctions in our study population of 13 patients, 15 infarctions (71%) became infected; abscess formation was doc umented by microbiologic aspiration and culture. In total, 12 of 13 patients (92%) developed one infected infarction and the remaining patient developed three infected infarctions. The average size of the infected infarctions was 7.0 cm in diameter (range, 3.3–11.8 cm). The median size of the infected infarctions was 6.6 cm, which was slightly increased from the baseline infarction size of 6.2 cm, although this difference was not statistically significant. In approximately 92% (12/13) of the patients, interval rounding of the infarction, defined as a more rounded configuration of the superinfected infarction on preprocedure CT, was seen when compared with the more geographic appearance of the infarction at baseline CT (Fig. 1A, 1B).
Interval development of intralesional foci of gas was seen in approximately 38% (5/13) of the patients with infected infarctions (Fig. 2A, 2B, 2C). The internal gas formation appeared focally localized with a mottled appearance, an air–fluid level, or both. The localized pattern of gas within the secondarily infected infarctions was different from the linear branching pattern of pneumobilia. Only one of the six patients who had underlying pneumobilia developed internal gas within the infected infarction; this finding suggests that internal gas within these lesions was a result of gas-forming organisms rather than of communication with the biliary system.
The 15 infected infarctions involved a total of 23 individual hepatic segments of the 13 patients, comprising approximately 22% (23/104 segments) of the segmental anatomy of the liver. Although the trend was for the left lobe of the liver to be more commonly infarcted, infarcts in the right lobe trended to be more often secondarily infected; however, this difference was not statistically significant at chi-square test analysis. In total, 61% (14/23) of the infected segments involved the right lobe (segments V–VIII) and 39% (9/23) of the infected segments involved the left lobe (segments I–IV). Hepatic segments V and VI were the most commonly superinfected hepatic segments, with seven patients (54%) developing superinfection of one or both of these segments.
 View larger version (160K) | Fig. 2A —29-year-old woman (patient 4 in Table 1) who underwent hepatic transplantation for metastatic carcinoid. Images show evolution of appearance of hepatic infarction from baseline examination to secondary infection with abscess formation and intralesional gas. Contrast-enhanced CT scan shows peripheral wedge-shaped lesion in segments II and III (black arrow), which is consistent with hepatic infarction. Linear high-attenuation area (white arrow) anterior to main portal vein represents biliary stent. |
 View larger version (170K) | Fig. 2B —29-year-old woman (patient 4 in Table 1) who underwent hepatic transplantation for metastatic carcinoid. Images show evolution of appearance of hepatic infarction from baseline examination to secondary infection with abscess formation and intralesional gas. Contrast-enhanced CT scan obtained 15 days after A shows interval rounding, intralesional gas formation (black arrow), and rim enhancement; these findings are consistent with infection with abscess formation. Linear high-attenuation area (white arrow) anterior to main portal vein represents biliary stent. |
 View larger version (147K) | Fig. 2C —29-year-old woman (patient 4 in Table 1) who underwent hepatic transplantation for metastatic carcinoid. Images show evolution of appearance of hepatic infarction from baseline examination to secondary infection with abscess formation and intralesional gas. Unenhanced CT scan obtained after procedure shows drainage catheter (arrow) in place. Purulent fluid from percutaneous abscess drainage grew Enterococcus organisms. |
Of the 12 patients who underwent percutaneous abscess drainage, 10 had one catheter placed, and two patients had three catheters placed for a total of 16 catheters inserted. Catheter placements were completed during a single visit to the interventional radiology suite in all patients except one who returned to the radiology suite for placement of an additional catheter due to incomplete drainage by the first catheter of a large 10.1-cm infected hepatic infarction. All catheter placements were technically successful with each catheter placed in the intended location and yielding drainage material.
Of the 12 patients who underwent percutaneous abscess drainage, the average duration of catheter placement was 43 days with a range of 3–187 days. Biliary communication was documented in seven of 12 patients (58%) either by direct catheter injection or by visual inspection of the catheter drainage contents. In the seven patients with biliary communication, the average duration of catheter drainage was 61 days; in the five patients without biliary communication, the average catheter duration was 19 days. Using a one-sided Wilcoxon's rank sum test, the duration of catheter drainage was statistically significant (p = 0.029).
Twelve of 13 patients (92%) underwent percutaneous abscess drainage. The infarction in the remaining patient was cultured and débrided intraoperatively. This latter patient ultimately underwent transplantation with a second orthotopic liver 16 days after surgical débridement because of complications of hepatic artery thrombosis and was discharged from the hospital 30 days after transplantation. Of the 12 patients who underwent percutaneous therapy, 11 (92%) responded to percutaneous abscess drainage with their clinical condition improved enough to allow hospital discharge. The one patient (8%) who did not respond to percutaneous abscess drainage underwent technically successful catheter placement for attempted palliation; this patient died 4 days later from widespread hepatic metastatic disease. One of the 11 pa tients who responded to percutaneous abscess drainage ultimately underwent orthotopic liver retransplantation 60 days after the percutaneous abscess drainage procedure because of persistent hepatic artery thrombosis.
There were 15 culture-positive infected infar ctions in the 12 patients who underwent needle aspiration and catheter drainage. Seven of 13 patients (54%) had polymicrobial abscesses, with an abscess from one patient yielding four separate species. Enterococcus species were isolated from six of the 13 patients (46%) patients, coagulase-negative Staphylococcus species were isolated from three pa tients (23%), Klebsiella pneumoniae were isolated from three patients (23%), and Candida species were isolated from two patients (15%). Other organisms that were isolated in one patient each included Staphylococcus aureus, Enterobacter cloacae, Escherichia coli, α-hemolytic Streptococcus organisms, non hemolytic Streptococcus organisms, Clostri dium perfringens, Corynebacterium species, and a fungus, Torulopsis glabrata.
| Discussion | |
|---|
Hepatic infarction is an entity that was once considered rare but is now better recognized because of the increased use of CT. Despite the dual blood supply to the liver parenchyma, hepatic infarctions occur and can become secondarily infected. In this series of patients, 77% of patients had recently undergone major surgery involving the porta hepatis, including hepatic transplantation or the Whipple procedure. Given the complexity of surgery in this subgroup of patients, compromised vascular supply to the porta hepatis likely explains the hepatic infarctions seen in this patient population. Hepatic infarctions after surgery involving the porta hepatis are well established, particularly after liver transplantation [8], although this complication has been described with other procedures including the Whipple procedure [5].
The secondary infection of liver infarctions is very rare given that we observed only 13 cases over a 13-year period at our hospital during which we performed 392 percutaneous liver abscess aspiration and drainage procedures. In patients already at increased risk of infection, we propose that abnormal liver tissue provides a nidus for secondary infection. In our series, the left lobe tended to be more commonly infarcted than the right lobe of the liver (54% vs 46%, respectively), with the lateral segments II and III most commonly involved. Conversely, infarctions of the right lobe tended to be more commonly secondarily infected than infarctions of the left lobe (61% vs 39%, respectively), with the inferior segments V and VI most commonly secondarily infected. This predisposition for infarction of the left lobe of the liver was most pronounced among the 10 surgical patients (70% vs 0%). We speculate that in this subgroup of postsurgical patients, the predisposition for infarction of the left lobe may reflect the surgical approach to these procedures involving the porta hepatis with greater retraction and potential ischemic time to the left lobe of the liver. The propensity for infarctions of the right lobe to become secondarily infected may reflect an increased degree of biliary reflux into the right lobe of the liver in this postoperative population lacking sphincter of Oddi competence.
The characteristic CT appearance of acute hepatic infarctions has been described as wedge-shaped and clearly demarcated regions of nonperfusion within the liver parenchyma that extend to the liver capsule [6–9]. Once the infarcted tissue becomes infected, the classic appearance of these lesions changes to approach the typical appearance of abscesses of other types. Rounding of the hepatic infarction on preprocedure CT when compared with baseline CT and development of internal foci of gas within the infarction were imaging features seen in 92% and 38% of patients in our series, respectively. Our series also showed slight interval enlargement of the median lesion size from 6.2 to 6.6 cm, although this difference was not statistically significant.
Although individual hepatic infarctions resemble typical abscesses once they become infected, this patient population appears more complex than the typical hepatic abscess patient. Infected hepatic infarctions often are complicated by communication with the biliary system, as seen in 58% of patients who had catheters placed in this investigation. An increased rate of biliary communication was associated with a longer duration of catheter drainage, with a mean of 61 days in patients with biliary communication versus 19 days in those patients without biliary communication, a statistically significant difference. Second, this patient population more often had polymicrobial infections, with 54% of patients having more than one organism isolated microbiologically. Finally, given that 77% of patients had recently undergone complex surgery, including liver transplantation or the Whipple procedure, these patients typically require more intensive and complicated medical care because of their recent major surgery and comorbidities.
One potential limitation to this investigation is the lack of a pathologically established gold standard of bland infarction in the patient population before superinfection. Although the imaging appearance of hepatic infarctions overlaps with the appearance of inflammatory lesions, the diagnosis of hepatic infarction on baseline CT can be made confidently because of the significant change in lesion appearance between the baseline and preprocedure CT scans, which were spaced by a mean interval of 20 days in our series. Another potential limitation to this investigation is the variation in the CT equipment used over the course of the study period from 1990 to 2003. Three different generations of CT scanners were used, varying from nonhelical scanners to 16-MDCT scanners. Of the 13 patients, three were imaged before 1997 with single-detector nonhelical CT and thus the image quality and sensitivity were likely decreased when compared with later-generation scanners.
Despite the complexity of care of this patient population, percutaneous abscess drainage can be performed safely in this patient population with excellent technical and longterm clinical outcomes. In the 12 patients who underwent percutaneous therapy, 11 (92%) responded to percutaneous abscess drainage such that they survived to discharge from the hospital. Our institutional preference for management of liver abscesses has been and continues to be placement of a percutaneous drainage catheter when safe and technically feasible. Although some have advocated simple aspiration with culture-guided antibiotic therapy [24], our experience with these patients is that they did not improve on antibiotic therapy alone and that portions of the abscess may liquefy with time, thereby justifying placement of an indwelling catheter.
In summary, infected hepatic infarctions are an uncommon but important subset of liver abscesses among patients presenting for percutaneous intervention. Infected hepatic infarctions tend to occur in patients who have undergone complex surgery that compromises blood flow to the porta hepatis, including the Whipple procedure or hepatic transplantation. In this series, although the left lobe of the liver was more commonly infarcted, the right lobe was more commonly secondarily infected; however, this distinction was not statistically significant. In those patients with infarctions complicated by biliary communication, the duration of catheter drainage was significantly longer than in those in whom biliary communication did not occur (61 vs 19 days, respectively). Despite the increased complexity of this patient population, including biliary communication and polymicrobial abscesses, percutaneous abscess drainage can be performed safely with excellent technical and long-term clinical outcomes.
Address correspondence to P. R. Mueller ([email protected]).
We thank Elkan Halpern for help with statistical analysis for this study.
| References | |
|---|
