Eur J Trauma Emerg Surg DOI 10.1007/s00068-013-0346-7
ORIGINAL ARTICLE
Risk factors for liver abscess formation in patients with blunt hepatic injury after non-operative management C.-P. Hsu • S.-Y. Wang • Y.-P. Hsu • H.-W. Chen • B.-C. Lin • S.-C. Kang • K.-C. Yuan • E.-H. Liu • I.-M. Kuo • C.-H. Liao • C.-H. Ouyang • S.-J. Yang
Received: 23 February 2013 / Accepted: 14 October 2013 Ó Springer-Verlag Berlin Heidelberg 2013
Abstract Purpose To identify risk factors for liver abscess formation in patients with blunt hepatic injury who underwent non-operative management (NOM). Methods From January 2004 to October 2008, retrospective data were collected from a single level I trauma center. Clinical data, hospital course, and outcome were all extracted from patient medical records for further analysis. Results A total of 358 patients were enrolled for analysis. There were 13 patients with liver abscess after blunt hepatic injury. Patients with abscess had a significant increase in glutamic oxaloacetic transaminase (GOT, p = 0.006) and glutamic pyruvic transaminase (GPT, p \ 0.0001), and a decrease in arterial blood pH (p = 0.023) compared to patients without abscess in the univariate analyses. In addition, high-grade hepatic injury and transarterial embolization (TAE, p \ 0.001) were also risk factors for liver abscess formation. Five factors (GOT, GPT, pH level in the arterial blood sample, TAE, and highgrade hepatic injury) were included in the multivariate analysis. TAE, high-grade hepatic injury, and GPT level were statistically significant. The odds ratios of TAE and high-grade hepatic injury were 15.41 and 16.08,
C.-P. Hsu S.-Y. Wang Y.-P. Hsu (&) B.-C. Lin S.-C. Kang K.-C. Yuan E.-H. Liu I.-M. Kuo C.-H. Liao C.-H. Ouyang S.-J. Yang Department of Trauma and Emergency Surgery, Chang Gung Memorial Hospital, Chang Gung University, 5, Fu-Hsing Street, Kwei Shan Township, Taoyuan, Taiwan e-mail: [email protected] H.-W. Chen Division of Emergency and Critical Care Radiology, Department of Medical Imaging and Intervention, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan
respectively. A receiver operating characteristic (ROC) analysis was used for GPT, and it suggested cutoff values of 372.5 U/L. A prediction model based on the ROC analysis had 100 % sensitivity and 86.7 % specificity to predict liver abscess formation in patients with two of the three independent risk factors. Conclusions TAE, high-grade hepatic injury, and a high GPT level are independent risk factors for liver abscess formation. Keywords Hepatic injury Liver injury Liver laceration Non-operative management Liver abscess Transarterial embolization
Introduction Hepatic injury occurs in approximately 5 % of all traumarelated hospitalizations [1]. The treatment for hepatic injuries has changed substantially over the last decade, and non-operative management (NOM) has become the current trend of treatment. Most successfully NOM-treated hepatic injuries have spontaneously healed without complication. Croce et al. [1] reported that 15 % of patients demonstrated complete resolution before hospital discharge. The interval between injury and healing is approximately 4 months. However, trauma surgeons occasionally encounter complicated clinical courses during patient recovery. The hepatic-related complication rate in most cases has been reported to range from 0 to 7 % [1–5]. Liver abscess is among these complications, and invasive treatment is sometimes necessary. However, few studies have focused on liver abscess formation in patients with blunt hepatic injuries after NOM. In this study, we attempted to identify
123
C. Hsu et al.
the possible risk factors related to liver abscess formation after blunt hepatic trauma with NOM.
Materials and methods NOM for hepatic injury In our institute, a level I trauma center, we only conducted NOM for patients with hepatic injury when stable hemodynamic status or maintainable hemodynamic status after initial resuscitation was noted (usually 2,000 cc of crystalloid fluid with or without an emergency transfusion of 4 U of whole blood). Initial survey for patients with blunt abdominal trauma included basic laboratory tests (hemogram, hepatic function test, renal function test, etc.). The indications for computed tomography (CT) scanning in our hospital for patients with blunt abdominal trauma include: (1) intra-abdominal fluid accumulation found in bedside echo, (2) elevated hepatic enzyme, either glutamic oxaloacetic transaminase (GOT) or glutamic pyruvic transaminase (GPT) over 200 U/L, or (3) poor image quality at a previous hospital. For every patient with hepatic injury, abdominal CT scans were conducted for all patients technically. The indication for transarterial embolization (TAE) included CT revealing contrast extravasation in the arterial phase or contrast pooling. All of the patients undergoing NOM, with or without TAE, were admitted to the surgical intensive care unit (SICU) for a 72-h observation period, including measurement of the abdominal waist, intraabdominal pressure monitoring, and regular hemoglobin testing. If a new bleeding episode was detected, either repeated TAE or laparotomy for hemostasis was immediately conducted. During the hospitalization period, there was no routine use of antibiotics unless suspicious symptoms or signs of infection were noted.
were excluded because tri-phasic CT scans were not available (all of these patients were transferred patients, and only simple two-phasic CT scans were available). Ninety-five patients who underwent operations were also excluded, including those with initial surgical treatment and those with NOM failure (2 % of patients who underwent NOM initially). All patient CT scans were reviewed by a single radiologist who had majored in trauma-related imaging for years and was blinded to the eventual clinical course of the patients. The hepatic injuries were recorded and graded according to the Hepatic Injury Scale established by the American Association for the Surgery of Trauma (AAST) [6]. The recorded data, including patients’ basic information, laboratory tests, hospital course, and final outcomes, were extracted. The diagnosis of abscess formation was confirmed by the clinical presentation (fever, chillness, leukocytosis, or sepsis) and imaging findings with later microbiological evidence. Statistical analysis Continuous data were analyzed by the Mann–Whitney Utest, while nominal data were analyzed by Fisher’s exact test. All statistically significant factors in the univariate analyses were included in a multivariate logistic regression analysis. A value of p \ 0.05 was considered significant. Independent factors yielded by the logistic regression were then used to depict a receiver operating characteristic (ROC) curve to analyze the sensitivity and specificity of the risk factors for predicting liver abscess formation. All statistical analyses were performed using SPSS version 17.0 software (SPSS Inc., Chicago, IL, USA).
Results
Data collection
General data analysis
This retrospective study was approved by the local institutional review board (IRB) of Chang Gung Memorial Hospital (CGMH). From January 2004 to October 2008 (a 58-month period), data were collected from the CGMH in Linkou, Taiwan. Medical records and radiological studies were reviewed by surgeons and radiologists, respectively. Patients from 18 to 65 years of age who were treated according to the protocol mentioned above were enrolled in this study. Patients who were diagnosed with cirrhosis, diabetes mellitus, or other immune-compromised conditions were excluded. In total, 601 patients with hepatic injury were identified during the study period. Twenty-seven patients with penetrating injuries were excluded. A total of 121 patients
The final number of patients included in the analysis was 358. Among these patients, the median patient age was 27 years. There were 227 males (63.4 %) and 131 females (36.6 %). Ninety-seven patients had an unclear consciousness state indicated by a Glasgow Coma Scale (GCS) score of \15 (97/358). Thirty-eight patients had a systolic blood pressure (SBP) \90 mmHg upon arrival (38/358). There were 306 patients (85.5 %) with grades I–III liver laceration and 52 patients (14.5 %) with grade IV liver laceration. No patients with grade V liver lacerations were included in this report. Thirty-eight patients (10.6 %) underwent TAE for hemostasis. The mean GOT, GPT, total bilirubin (T-Bil), alkaline phosphatase (ALK-P), arterial blood pH, and standard bass excess (SBE) were 374 U/L,
123
Risk factors for liver abscess formation Table 1 Demographics data and outcomes of the patients without and with abscess Factors Age (years)
All patients (n = 358)
Without abscess (n = 345)
With abscess (n = 13)
p-value*
27 (19, 39)
27 (19, 39)
29 (21.5, 39)
n.s.
Male
227 (63.4 %)
216 (62.6 %)
11 (84.6 %)
n.s.
Female
131 (36.6 %)
129 (37.4 %)
2 (15.4 %)
GCS = 15
261 (72.8 %)
251 (72.7 %)
10 (76.9 %)
GCS \ 15
97 (27.2 %)
94 (27.3 %)
3 (23.1 %)
SBP C 90 mmHg
317 (88.5 %)
306 (88.7 %)
11 (84.6 %)
SBP \ 90 mmHg
41 (11.5 %)
39 (11.3 %)
2 (15.4 %)
306 (85.5 %)
303 (87.8 %)
3 (23.1 %)
52 (14.5 %)
42 (12.2 %)
10 (76.9 %)
Gender
Initial GCS n.s.
Initial blood pressure
Grade of liver laceration I–III IV
n.s.
\0.001
TAE Yes
38 (10.6 %)
30 (8.7 %)
8 (61.5 %)
No
320 (89.4 %)
315 (91.3 %)
5 (38.5 %)
\0.001
Laboratory data GOT (U/L)
374 (220, 580)
362 (214.5, 567.5)
728 (344.75, 1,050.5)
0.006
GPT (U/L)
284 (167, 449)
276 (161, 436)
708 (413.5, 1,312.75)
\0.001
0.9 (0.67, 1.3)
0.9 (0.7, 1.2)
1.3 (0.6, 2.7)
n.s.
48 (35, 84)
72 (52, 143)
T-Bil (mg/dL) ALK-P (U/L)
48.5 (35.75, 88.25)
pH in ABG
7.36 (7.31, 7.4)
SBE (mm/L)
-4.2 (-7.4, -1.5)
n.s.
7.36 (7.31, 7.41)
7.29 (7.24, 7.36)
0.023
-4.15 (-6.98, -1.6)
-7.5 (-10.4, -1)
n.s.
Admission length of stay, days
10 (6, 17)
10 (6, 17)
Mortally wounded
12 (3.4 %)
12 (3.5 %)
24 (15.5, 29)
0.002
0 (0 %)
Continuous data are expressed as median (Q3, Q4). Categorized data are expressed as case number (%) GCS Glasgow Coma Scale, SBP systolic blood pressure, TAE transarterial embolization, ICU intensive care unit, GOT glutamic oxaloacetic transaminase, GPT glutamic pyruvic transaminase, T-Bil total bilirubin, ALK-P alkaline phosphatase, ABG arterial blood gas, SBE standard bass excess, ICU intensive care unit * Continuous data were analyzed by using the Mann–Whitney U-test (non-parametric statistical method for small sample sizes). Categorized data were analyzed by using the Fisher’s exact test. p B 0.05 was considered statistically significant
284 U/L, 0.9 mg/dL, 48.5 U/L, 7.36 mm/L, and -4.2 mm/ L, respectively. The median length of hospital stay was 10 days. Thirteen patients (3.6 %) developed a liver abscess. Twelve patients (3.4 %) were mortally wounded. The patient demographics and outcomes are outlined in Table 1. The data of microbiology were heterogenous in our study patients. We summarize the microbiologic data in Table 2. The onset for four patients was during first admission (the same hospitalization for primary trauma). All patients with abscess underwent CT-guided drainage and later recovered well clinically. Univariate analysis Enrolled patients were stratified into two groups: patients with abscess (AB group) and patients without
abscess (NAB group). The AB group included 13 patients. The mean time to develop liver abscess was 18 days (range 8–69 days). All the clinical and laboratory factors were compared between the groups. The results are summarized in Table 1. Higher GOT (p = 0.006) and GPT levels (p \ 0.001), decreased arterial blood pH level (p = 0.023), longer hospital stay, TAE application, and higher grade liver injury were noted in the AB group. With regard to TAE, liver abscesses developed in 1.6 % (5/5 ? 315) of the patients without TAE treatment, while 21.1 % (8/ 8 ? 30) of the patients with TAE treatment developed a liver abscess. Moreover, the patients with low-grade (I– III) liver injury had a lower rate of abscess formation than those with high-grade (IV) injury (1.0 %, 3/306 vs. 19.2 %, 10/52).
123
C. Hsu et al. Table 2 Summary of abscess onset, microbiological profile, and treatment for patients Microbiological results
Number of cases
Gram-positive
4
Species Gram-positive bacilli* Staphylococcus aureus Coagulase-negative staphylococci Viridans streptococci
Gram-negative
2
Escherichia coli Aeromonas hydrophila
Multiorgan infection
7
Escherichia coli Viridans streptococci Morganella morganii Proteus mirabilis Peptostreptococcus species Prevotella intermedia Acinetobacter baumannii Enterococcus faecalis Staphylococcus aureus Enterococcus faecalis Coagulase-negative staphylococci Gram-positive bacilli* Clostridium species Viridans streptococci Enterococcus faecalis
* Gram-positive bacilli: species could not be identified
Fig. 1 ROC curve for GPT cutoff value estimation. GPT glutamic pyruvic transaminase
(p = 0.015), and GPT (p = 0.024) were independent risk factors. The odds ratios of TAE and grade IV liver injury were 15.41 and 16.08, respectively. An ROC curve analysis was performed for GPT, and it yielded a cutoff value of 372.5 U/L (Fig. 1). The sensitivity and specificity of a GPT level C372.5 U/L were 0.900 and 0.668, respectively. Prediction analysis
Table 3 Multivariate analysis for factors predicting liver abscess formation Odds ratio
p-value
GOT
–
n.s.
GPT
1.002
0.024
pH in ABG
–
n.s.
TAE
15.41
0.004
Grade IV liver laceration
16.08
0.015
Binary logistic regression using the forward conditional method was performed NS non-significant, GOT glutamic oxaloacetic transaminase, GPT glutamic pyruvic transaminase, ABG arterial blood gas
Multivariate analysis Five factors with statistical significance in the aforementioned univariate analyses, including GOT, GPT, arterial pH level, TAE, and grade of liver injury, were further analyzed by logistic regression (Table 3). The results revealed that TAE (p = 0.024), grade IV liver injury
123
We attempted to create a prediction model according to the aforementioned three independent factors with an ROC curve analysis. The one-risk-factor group had 100 % sensitivity and 59.7 % specificity. The two-risk-factor group had 100 % sensitivity and 86.7 % specificity.
Discussion NOM for patients with blunt hepatic injury has been the preferred treatment method since the early 1990s [7–9]. In this study, we attempted to identify the possible risk factors for one of the possible complications, abscess formation, after blunt hepatic injury in patients treated with NOM. Three independent factors, GPT over 372.5 U/L, TAE, and high-grade liver injury, were noted in our study. Hepatic enzymes, present in hepatocytes, had been previously documented to indicate hepatic injury and are a sensitive indicator of hepatocyte integrity [10–12]. Severe hepatic injury should be suspected in the presence of high levels of GOT or GPT [13]. The relatively high GPT in
Risk factors for liver abscess formation
patients with abscess formation may simply suggest that more severe injuries are associated with a higher likelihood of abscess formation. TAE is an important procedure for achieving hemostasis in liver trauma. However, the present study revealed that TAE is also an independent risk factor for abscess formation in an injured liver. Few studies are available on the complications after TAE was performed for patients with liver injury. For patients with hepatocellular carcinoma (HCC), transarterial procedures with or without tumor chemotherapy agents is well established, and, therefore, more relevant studies have been published on this topic than for TAE in liver injury. A variety of complications related to transarterial procedures for HCC has been reported. Song et al. [14] published a retrospective review of 2,439 patients with HCC who underwent a total of 6,255 transcatheter oily chemoembolization (TOCE) procedures. Fifteen liver abscesses developed in 14 patients, and the incidence was 0.2 %. The incidence seemed much lower than that reported in our series. The possible hypothesis could be that the portal triad, including the biliary system, portal system, and arterial system, may be damaged after liver trauma, while abscess following transarterial procedures for HCC may be related to only arterial compromise and limited hepatic tissue damage. However, the commonly used diagnostic tools for trauma patients could not clearly determine if the injury was to the bile duct, portal system, or venous system. Further studies to demonstrate the relationship between the biliary, portal, or venous system and post-TAE complications in patients with liver trauma may be needed. Because surgical exploration is usually mandatory for grade V liver injury, no patient with grade V injury was included in our study. Grades I–III and grade IV accounted for 23.1 and 76.9 % of patients with liver abscess in our study, respectively. This indicates that more than threefourths of patients with abscess have a high-grade injury. Moreover, only 1.0 % of patients with grades I–III liver injury had a liver abscess, whereas 19.2 % of the patients with grade IV liver injury developed liver abscess after NOM. This indicates that approximately one-fifth of patients with grade IV liver laceration may develop liver abscess. For high-grade liver injury, Kozar et al. [15] reported 230 patients with grades III–V blunt hepatic injuries. The incidences of complications in the patients with grades III, IV, and V liver lacerations were 1, 21, and 63 %, respectively. Giss et al. [16] reported 185 pediatric blunt hepatic injuries treated with NOM and noted that complications occurred only in patients with grade III or greater injuries. Our findings are compatible with these two studies. We attempted to establish a prediction score for abscess formation. However, due to the small case
number, we could not conduct an adequate validation study of the result. Further external validation is necessary. There were some pitfalls that could not be avoided in this study. First, this was a retrospective study, and all the conditions may not be as well controlled as in prospective observational studies. Some patients who presented without obvious complications during their hospital stay would most likely develop an abscess after discharge, and these patients may have attended another institute for help. As a result, the rate of abscess formation may have been underestimated. In addition, this study emphasized mostly the relationship between liver necrosis and later complication. Other possible links cannot be well assessed, such as patterns of trauma, mechanism of trauma, and associated injuries. The trauma registry system in our institute was established in May 2008. Later comprehensive study for complications after NOM for liver injury may elucidate more than the current retrospective study.
Conclusion According to the present study, transarterial embolization (TAE), grade IV liver laceration, and a high glutamic pyruvictransaminase (GPT) level are risk factors for liver abscess formation in trauma patients who sustain blunt liver injury and are managed non-operatively. Grade IV liver laceration was the most important risk factor, with an odds ratio of 16.08. TAE was the second most important risk factor, with an odds ratio of 15.41, although it is an important treatment for hepatic injury with hemorrhage. A high GPT level was the third most important risk factor, and the cutoff value was 372.5 U/L. The presence of two of these factors may predict a high likelihood of later abscess formation in an injured liver. Conflict of interest On behalf of all authors, the corresponding author states that there is no conflict of interest.
References 1. Croce MA, Fabian TC, Menke PG, Waddle-Smith L, Minard G, Kudsk KA, et al. Nonoperative management of blunt hepatic trauma is the treatment of choice for hemodynamically stable patients. Results of a prospective trial. Ann Surg. 1995; 221(6):744–53; discussion 53–5. 2. Velmahos GC, Toutouzas K, Radin R, Chan L, Rhee P, Tillou A, et al. High success with nonoperative management of blunt hepatic trauma: the liver is a sturdy organ. Arch Surg. 2003;138(5):475–80; discussion 480–1. doi:10.1001/archsurg. 138.5.475. 3. Meredith JW, Young JS, Bowling J, Roboussin D. Nonoperative management of blunt hepatic trauma: the exception or the rule? J Trauma. 1994;36(4):529–34; discussion 534–5.
123
C. Hsu et al. 4. Pachter HL, Knudson MM, Esrig B, Ross S, Hoyt D, Cogbill T, et al. Status of nonoperative management of blunt hepatic injuries in 1995: a multicenter experience with 404 patients. J Trauma. 1996;40(1):31–8. 5. Malhotra AK, Fabian TC, Croce MA, Gavin TJ, Kudsk KA, Minard G, et al. Blunt hepatic injury: a paradigm shift from operative to nonoperative management in the 1990s. Ann Surg. 2000;231(6):804–13. 6. Moore EE, Cogbill TH, Jurkovich GJ, Shackford SR, Malangoni MA, Champion HR. Organ injury scaling: spleen and liver (1994 revision). J Trauma. 1995;38(3):323–4. 7. Farnell MB, Spencer MP, Thompson E, Williams HJ Jr, Mucha P Jr, Ilstrup DM. Nonoperative management of blunt hepatic trauma in adults. Surgery. 1988;104(4):748–56. 8. Federico JA, Horner WR, Clark DE, Isler RJ. Blunt hepatic trauma. Nonoperative management in adults. Arch Surg. 1990;125(7):905–8; discussion 908–9. 9. Pachter HL, Spencer FC, Hofstetter SR, Liang HG, Coppa GF. Significant trends in the treatment of hepatic trauma. Experience with 411 injuries. Ann Surg. 1992;215(5):492–500; discussion 500–2. 10. Stassen NA, Lukan JK, Carrillo EH, Spain DA, Norfleet LA, Miller FB, et al. Examination of the role of abdominal computed tomography in the evaluation of victims of trauma with increased
123
11.
12.
13.
14.
15.
16.
aspartate aminotransferase in the era of focused abdominal sonography for trauma. Surgery. 2002;132(4):642–6; discussion 646–7. Puranik SR, Hayes JS, Long J, Mata M. Liver enzymes as predictors of liver damage due to blunt abdominal trauma in children. South Med J. 2002;95(2):203–6. Lee WC, Kuo LC, Cheng YC, Chen CW, Lin YK, Lin TY, et al. Combination of white blood cell count with liver enzymes in the diagnosis of blunt liver laceration. Am J Emerg Med. 2010; 28(9):1024–9. Karam O, La Scala G, Le Coultre C, Chardot C. Liver function tests in children with blunt abdominal traumas. Eur J Pediatr Surg. 2007;17(5):313–6. doi:10.1055/s-2007-965533. Song SY, Chung JW, Han JK, Lim HG, Koh YH, Park JH, et al. Liver abscess after transcatheter oily chemoembolization for hepatic tumors: incidence, predisposing factors, and clinical outcome. J Vasc Interv Radiol. 2001;12(3):313–20. Kozar RA, Moore JB, Niles SE, Holcomb JB, Moore EE, Cothren CC, et al. Complications of nonoperative management of highgrade blunt hepatic injuries. J Trauma. 2005;59(5):1066–71. Giss SR, Dobrilovic N, Brown RL, Garcia VF. Complications of nonoperative management of pediatric blunt hepatic injury: diagnosis, management, and outcomes. J Trauma. 2006;61(2): 334–9. doi:10.1097/01.ta.0000197605.27190.2c.
ORIGINAL ARTICLE
Risk factors for liver abscess formation in patients with blunt hepatic injury after non-operative management C.-P. Hsu • S.-Y. Wang • Y.-P. Hsu • H.-W. Chen • B.-C. Lin • S.-C. Kang • K.-C. Yuan • E.-H. Liu • I.-M. Kuo • C.-H. Liao • C.-H. Ouyang • S.-J. Yang
Received: 23 February 2013 / Accepted: 14 October 2013 Ó Springer-Verlag Berlin Heidelberg 2013
Abstract Purpose To identify risk factors for liver abscess formation in patients with blunt hepatic injury who underwent non-operative management (NOM). Methods From January 2004 to October 2008, retrospective data were collected from a single level I trauma center. Clinical data, hospital course, and outcome were all extracted from patient medical records for further analysis. Results A total of 358 patients were enrolled for analysis. There were 13 patients with liver abscess after blunt hepatic injury. Patients with abscess had a significant increase in glutamic oxaloacetic transaminase (GOT, p = 0.006) and glutamic pyruvic transaminase (GPT, p \ 0.0001), and a decrease in arterial blood pH (p = 0.023) compared to patients without abscess in the univariate analyses. In addition, high-grade hepatic injury and transarterial embolization (TAE, p \ 0.001) were also risk factors for liver abscess formation. Five factors (GOT, GPT, pH level in the arterial blood sample, TAE, and highgrade hepatic injury) were included in the multivariate analysis. TAE, high-grade hepatic injury, and GPT level were statistically significant. The odds ratios of TAE and high-grade hepatic injury were 15.41 and 16.08,
C.-P. Hsu S.-Y. Wang Y.-P. Hsu (&) B.-C. Lin S.-C. Kang K.-C. Yuan E.-H. Liu I.-M. Kuo C.-H. Liao C.-H. Ouyang S.-J. Yang Department of Trauma and Emergency Surgery, Chang Gung Memorial Hospital, Chang Gung University, 5, Fu-Hsing Street, Kwei Shan Township, Taoyuan, Taiwan e-mail: [email protected] H.-W. Chen Division of Emergency and Critical Care Radiology, Department of Medical Imaging and Intervention, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan
respectively. A receiver operating characteristic (ROC) analysis was used for GPT, and it suggested cutoff values of 372.5 U/L. A prediction model based on the ROC analysis had 100 % sensitivity and 86.7 % specificity to predict liver abscess formation in patients with two of the three independent risk factors. Conclusions TAE, high-grade hepatic injury, and a high GPT level are independent risk factors for liver abscess formation. Keywords Hepatic injury Liver injury Liver laceration Non-operative management Liver abscess Transarterial embolization
Introduction Hepatic injury occurs in approximately 5 % of all traumarelated hospitalizations [1]. The treatment for hepatic injuries has changed substantially over the last decade, and non-operative management (NOM) has become the current trend of treatment. Most successfully NOM-treated hepatic injuries have spontaneously healed without complication. Croce et al. [1] reported that 15 % of patients demonstrated complete resolution before hospital discharge. The interval between injury and healing is approximately 4 months. However, trauma surgeons occasionally encounter complicated clinical courses during patient recovery. The hepatic-related complication rate in most cases has been reported to range from 0 to 7 % [1–5]. Liver abscess is among these complications, and invasive treatment is sometimes necessary. However, few studies have focused on liver abscess formation in patients with blunt hepatic injuries after NOM. In this study, we attempted to identify
123
C. Hsu et al.
the possible risk factors related to liver abscess formation after blunt hepatic trauma with NOM.
Materials and methods NOM for hepatic injury In our institute, a level I trauma center, we only conducted NOM for patients with hepatic injury when stable hemodynamic status or maintainable hemodynamic status after initial resuscitation was noted (usually 2,000 cc of crystalloid fluid with or without an emergency transfusion of 4 U of whole blood). Initial survey for patients with blunt abdominal trauma included basic laboratory tests (hemogram, hepatic function test, renal function test, etc.). The indications for computed tomography (CT) scanning in our hospital for patients with blunt abdominal trauma include: (1) intra-abdominal fluid accumulation found in bedside echo, (2) elevated hepatic enzyme, either glutamic oxaloacetic transaminase (GOT) or glutamic pyruvic transaminase (GPT) over 200 U/L, or (3) poor image quality at a previous hospital. For every patient with hepatic injury, abdominal CT scans were conducted for all patients technically. The indication for transarterial embolization (TAE) included CT revealing contrast extravasation in the arterial phase or contrast pooling. All of the patients undergoing NOM, with or without TAE, were admitted to the surgical intensive care unit (SICU) for a 72-h observation period, including measurement of the abdominal waist, intraabdominal pressure monitoring, and regular hemoglobin testing. If a new bleeding episode was detected, either repeated TAE or laparotomy for hemostasis was immediately conducted. During the hospitalization period, there was no routine use of antibiotics unless suspicious symptoms or signs of infection were noted.
were excluded because tri-phasic CT scans were not available (all of these patients were transferred patients, and only simple two-phasic CT scans were available). Ninety-five patients who underwent operations were also excluded, including those with initial surgical treatment and those with NOM failure (2 % of patients who underwent NOM initially). All patient CT scans were reviewed by a single radiologist who had majored in trauma-related imaging for years and was blinded to the eventual clinical course of the patients. The hepatic injuries were recorded and graded according to the Hepatic Injury Scale established by the American Association for the Surgery of Trauma (AAST) [6]. The recorded data, including patients’ basic information, laboratory tests, hospital course, and final outcomes, were extracted. The diagnosis of abscess formation was confirmed by the clinical presentation (fever, chillness, leukocytosis, or sepsis) and imaging findings with later microbiological evidence. Statistical analysis Continuous data were analyzed by the Mann–Whitney Utest, while nominal data were analyzed by Fisher’s exact test. All statistically significant factors in the univariate analyses were included in a multivariate logistic regression analysis. A value of p \ 0.05 was considered significant. Independent factors yielded by the logistic regression were then used to depict a receiver operating characteristic (ROC) curve to analyze the sensitivity and specificity of the risk factors for predicting liver abscess formation. All statistical analyses were performed using SPSS version 17.0 software (SPSS Inc., Chicago, IL, USA).
Results
Data collection
General data analysis
This retrospective study was approved by the local institutional review board (IRB) of Chang Gung Memorial Hospital (CGMH). From January 2004 to October 2008 (a 58-month period), data were collected from the CGMH in Linkou, Taiwan. Medical records and radiological studies were reviewed by surgeons and radiologists, respectively. Patients from 18 to 65 years of age who were treated according to the protocol mentioned above were enrolled in this study. Patients who were diagnosed with cirrhosis, diabetes mellitus, or other immune-compromised conditions were excluded. In total, 601 patients with hepatic injury were identified during the study period. Twenty-seven patients with penetrating injuries were excluded. A total of 121 patients
The final number of patients included in the analysis was 358. Among these patients, the median patient age was 27 years. There were 227 males (63.4 %) and 131 females (36.6 %). Ninety-seven patients had an unclear consciousness state indicated by a Glasgow Coma Scale (GCS) score of \15 (97/358). Thirty-eight patients had a systolic blood pressure (SBP) \90 mmHg upon arrival (38/358). There were 306 patients (85.5 %) with grades I–III liver laceration and 52 patients (14.5 %) with grade IV liver laceration. No patients with grade V liver lacerations were included in this report. Thirty-eight patients (10.6 %) underwent TAE for hemostasis. The mean GOT, GPT, total bilirubin (T-Bil), alkaline phosphatase (ALK-P), arterial blood pH, and standard bass excess (SBE) were 374 U/L,
123
Risk factors for liver abscess formation Table 1 Demographics data and outcomes of the patients without and with abscess Factors Age (years)
All patients (n = 358)
Without abscess (n = 345)
With abscess (n = 13)
p-value*
27 (19, 39)
27 (19, 39)
29 (21.5, 39)
n.s.
Male
227 (63.4 %)
216 (62.6 %)
11 (84.6 %)
n.s.
Female
131 (36.6 %)
129 (37.4 %)
2 (15.4 %)
GCS = 15
261 (72.8 %)
251 (72.7 %)
10 (76.9 %)
GCS \ 15
97 (27.2 %)
94 (27.3 %)
3 (23.1 %)
SBP C 90 mmHg
317 (88.5 %)
306 (88.7 %)
11 (84.6 %)
SBP \ 90 mmHg
41 (11.5 %)
39 (11.3 %)
2 (15.4 %)
306 (85.5 %)
303 (87.8 %)
3 (23.1 %)
52 (14.5 %)
42 (12.2 %)
10 (76.9 %)
Gender
Initial GCS n.s.
Initial blood pressure
Grade of liver laceration I–III IV
n.s.
\0.001
TAE Yes
38 (10.6 %)
30 (8.7 %)
8 (61.5 %)
No
320 (89.4 %)
315 (91.3 %)
5 (38.5 %)
\0.001
Laboratory data GOT (U/L)
374 (220, 580)
362 (214.5, 567.5)
728 (344.75, 1,050.5)
0.006
GPT (U/L)
284 (167, 449)
276 (161, 436)
708 (413.5, 1,312.75)
\0.001
0.9 (0.67, 1.3)
0.9 (0.7, 1.2)
1.3 (0.6, 2.7)
n.s.
48 (35, 84)
72 (52, 143)
T-Bil (mg/dL) ALK-P (U/L)
48.5 (35.75, 88.25)
pH in ABG
7.36 (7.31, 7.4)
SBE (mm/L)
-4.2 (-7.4, -1.5)
n.s.
7.36 (7.31, 7.41)
7.29 (7.24, 7.36)
0.023
-4.15 (-6.98, -1.6)
-7.5 (-10.4, -1)
n.s.
Admission length of stay, days
10 (6, 17)
10 (6, 17)
Mortally wounded
12 (3.4 %)
12 (3.5 %)
24 (15.5, 29)
0.002
0 (0 %)
Continuous data are expressed as median (Q3, Q4). Categorized data are expressed as case number (%) GCS Glasgow Coma Scale, SBP systolic blood pressure, TAE transarterial embolization, ICU intensive care unit, GOT glutamic oxaloacetic transaminase, GPT glutamic pyruvic transaminase, T-Bil total bilirubin, ALK-P alkaline phosphatase, ABG arterial blood gas, SBE standard bass excess, ICU intensive care unit * Continuous data were analyzed by using the Mann–Whitney U-test (non-parametric statistical method for small sample sizes). Categorized data were analyzed by using the Fisher’s exact test. p B 0.05 was considered statistically significant
284 U/L, 0.9 mg/dL, 48.5 U/L, 7.36 mm/L, and -4.2 mm/ L, respectively. The median length of hospital stay was 10 days. Thirteen patients (3.6 %) developed a liver abscess. Twelve patients (3.4 %) were mortally wounded. The patient demographics and outcomes are outlined in Table 1. The data of microbiology were heterogenous in our study patients. We summarize the microbiologic data in Table 2. The onset for four patients was during first admission (the same hospitalization for primary trauma). All patients with abscess underwent CT-guided drainage and later recovered well clinically. Univariate analysis Enrolled patients were stratified into two groups: patients with abscess (AB group) and patients without
abscess (NAB group). The AB group included 13 patients. The mean time to develop liver abscess was 18 days (range 8–69 days). All the clinical and laboratory factors were compared between the groups. The results are summarized in Table 1. Higher GOT (p = 0.006) and GPT levels (p \ 0.001), decreased arterial blood pH level (p = 0.023), longer hospital stay, TAE application, and higher grade liver injury were noted in the AB group. With regard to TAE, liver abscesses developed in 1.6 % (5/5 ? 315) of the patients without TAE treatment, while 21.1 % (8/ 8 ? 30) of the patients with TAE treatment developed a liver abscess. Moreover, the patients with low-grade (I– III) liver injury had a lower rate of abscess formation than those with high-grade (IV) injury (1.0 %, 3/306 vs. 19.2 %, 10/52).
123
C. Hsu et al. Table 2 Summary of abscess onset, microbiological profile, and treatment for patients Microbiological results
Number of cases
Gram-positive
4
Species Gram-positive bacilli* Staphylococcus aureus Coagulase-negative staphylococci Viridans streptococci
Gram-negative
2
Escherichia coli Aeromonas hydrophila
Multiorgan infection
7
Escherichia coli Viridans streptococci Morganella morganii Proteus mirabilis Peptostreptococcus species Prevotella intermedia Acinetobacter baumannii Enterococcus faecalis Staphylococcus aureus Enterococcus faecalis Coagulase-negative staphylococci Gram-positive bacilli* Clostridium species Viridans streptococci Enterococcus faecalis
* Gram-positive bacilli: species could not be identified
Fig. 1 ROC curve for GPT cutoff value estimation. GPT glutamic pyruvic transaminase
(p = 0.015), and GPT (p = 0.024) were independent risk factors. The odds ratios of TAE and grade IV liver injury were 15.41 and 16.08, respectively. An ROC curve analysis was performed for GPT, and it yielded a cutoff value of 372.5 U/L (Fig. 1). The sensitivity and specificity of a GPT level C372.5 U/L were 0.900 and 0.668, respectively. Prediction analysis
Table 3 Multivariate analysis for factors predicting liver abscess formation Odds ratio
p-value
GOT
–
n.s.
GPT
1.002
0.024
pH in ABG
–
n.s.
TAE
15.41
0.004
Grade IV liver laceration
16.08
0.015
Binary logistic regression using the forward conditional method was performed NS non-significant, GOT glutamic oxaloacetic transaminase, GPT glutamic pyruvic transaminase, ABG arterial blood gas
Multivariate analysis Five factors with statistical significance in the aforementioned univariate analyses, including GOT, GPT, arterial pH level, TAE, and grade of liver injury, were further analyzed by logistic regression (Table 3). The results revealed that TAE (p = 0.024), grade IV liver injury
123
We attempted to create a prediction model according to the aforementioned three independent factors with an ROC curve analysis. The one-risk-factor group had 100 % sensitivity and 59.7 % specificity. The two-risk-factor group had 100 % sensitivity and 86.7 % specificity.
Discussion NOM for patients with blunt hepatic injury has been the preferred treatment method since the early 1990s [7–9]. In this study, we attempted to identify the possible risk factors for one of the possible complications, abscess formation, after blunt hepatic injury in patients treated with NOM. Three independent factors, GPT over 372.5 U/L, TAE, and high-grade liver injury, were noted in our study. Hepatic enzymes, present in hepatocytes, had been previously documented to indicate hepatic injury and are a sensitive indicator of hepatocyte integrity [10–12]. Severe hepatic injury should be suspected in the presence of high levels of GOT or GPT [13]. The relatively high GPT in
Risk factors for liver abscess formation
patients with abscess formation may simply suggest that more severe injuries are associated with a higher likelihood of abscess formation. TAE is an important procedure for achieving hemostasis in liver trauma. However, the present study revealed that TAE is also an independent risk factor for abscess formation in an injured liver. Few studies are available on the complications after TAE was performed for patients with liver injury. For patients with hepatocellular carcinoma (HCC), transarterial procedures with or without tumor chemotherapy agents is well established, and, therefore, more relevant studies have been published on this topic than for TAE in liver injury. A variety of complications related to transarterial procedures for HCC has been reported. Song et al. [14] published a retrospective review of 2,439 patients with HCC who underwent a total of 6,255 transcatheter oily chemoembolization (TOCE) procedures. Fifteen liver abscesses developed in 14 patients, and the incidence was 0.2 %. The incidence seemed much lower than that reported in our series. The possible hypothesis could be that the portal triad, including the biliary system, portal system, and arterial system, may be damaged after liver trauma, while abscess following transarterial procedures for HCC may be related to only arterial compromise and limited hepatic tissue damage. However, the commonly used diagnostic tools for trauma patients could not clearly determine if the injury was to the bile duct, portal system, or venous system. Further studies to demonstrate the relationship between the biliary, portal, or venous system and post-TAE complications in patients with liver trauma may be needed. Because surgical exploration is usually mandatory for grade V liver injury, no patient with grade V injury was included in our study. Grades I–III and grade IV accounted for 23.1 and 76.9 % of patients with liver abscess in our study, respectively. This indicates that more than threefourths of patients with abscess have a high-grade injury. Moreover, only 1.0 % of patients with grades I–III liver injury had a liver abscess, whereas 19.2 % of the patients with grade IV liver injury developed liver abscess after NOM. This indicates that approximately one-fifth of patients with grade IV liver laceration may develop liver abscess. For high-grade liver injury, Kozar et al. [15] reported 230 patients with grades III–V blunt hepatic injuries. The incidences of complications in the patients with grades III, IV, and V liver lacerations were 1, 21, and 63 %, respectively. Giss et al. [16] reported 185 pediatric blunt hepatic injuries treated with NOM and noted that complications occurred only in patients with grade III or greater injuries. Our findings are compatible with these two studies. We attempted to establish a prediction score for abscess formation. However, due to the small case
number, we could not conduct an adequate validation study of the result. Further external validation is necessary. There were some pitfalls that could not be avoided in this study. First, this was a retrospective study, and all the conditions may not be as well controlled as in prospective observational studies. Some patients who presented without obvious complications during their hospital stay would most likely develop an abscess after discharge, and these patients may have attended another institute for help. As a result, the rate of abscess formation may have been underestimated. In addition, this study emphasized mostly the relationship between liver necrosis and later complication. Other possible links cannot be well assessed, such as patterns of trauma, mechanism of trauma, and associated injuries. The trauma registry system in our institute was established in May 2008. Later comprehensive study for complications after NOM for liver injury may elucidate more than the current retrospective study.
Conclusion According to the present study, transarterial embolization (TAE), grade IV liver laceration, and a high glutamic pyruvictransaminase (GPT) level are risk factors for liver abscess formation in trauma patients who sustain blunt liver injury and are managed non-operatively. Grade IV liver laceration was the most important risk factor, with an odds ratio of 16.08. TAE was the second most important risk factor, with an odds ratio of 15.41, although it is an important treatment for hepatic injury with hemorrhage. A high GPT level was the third most important risk factor, and the cutoff value was 372.5 U/L. The presence of two of these factors may predict a high likelihood of later abscess formation in an injured liver. Conflict of interest On behalf of all authors, the corresponding author states that there is no conflict of interest.
References 1. Croce MA, Fabian TC, Menke PG, Waddle-Smith L, Minard G, Kudsk KA, et al. Nonoperative management of blunt hepatic trauma is the treatment of choice for hemodynamically stable patients. Results of a prospective trial. Ann Surg. 1995; 221(6):744–53; discussion 53–5. 2. Velmahos GC, Toutouzas K, Radin R, Chan L, Rhee P, Tillou A, et al. High success with nonoperative management of blunt hepatic trauma: the liver is a sturdy organ. Arch Surg. 2003;138(5):475–80; discussion 480–1. doi:10.1001/archsurg. 138.5.475. 3. Meredith JW, Young JS, Bowling J, Roboussin D. Nonoperative management of blunt hepatic trauma: the exception or the rule? J Trauma. 1994;36(4):529–34; discussion 534–5.
123
C. Hsu et al. 4. Pachter HL, Knudson MM, Esrig B, Ross S, Hoyt D, Cogbill T, et al. Status of nonoperative management of blunt hepatic injuries in 1995: a multicenter experience with 404 patients. J Trauma. 1996;40(1):31–8. 5. Malhotra AK, Fabian TC, Croce MA, Gavin TJ, Kudsk KA, Minard G, et al. Blunt hepatic injury: a paradigm shift from operative to nonoperative management in the 1990s. Ann Surg. 2000;231(6):804–13. 6. Moore EE, Cogbill TH, Jurkovich GJ, Shackford SR, Malangoni MA, Champion HR. Organ injury scaling: spleen and liver (1994 revision). J Trauma. 1995;38(3):323–4. 7. Farnell MB, Spencer MP, Thompson E, Williams HJ Jr, Mucha P Jr, Ilstrup DM. Nonoperative management of blunt hepatic trauma in adults. Surgery. 1988;104(4):748–56. 8. Federico JA, Horner WR, Clark DE, Isler RJ. Blunt hepatic trauma. Nonoperative management in adults. Arch Surg. 1990;125(7):905–8; discussion 908–9. 9. Pachter HL, Spencer FC, Hofstetter SR, Liang HG, Coppa GF. Significant trends in the treatment of hepatic trauma. Experience with 411 injuries. Ann Surg. 1992;215(5):492–500; discussion 500–2. 10. Stassen NA, Lukan JK, Carrillo EH, Spain DA, Norfleet LA, Miller FB, et al. Examination of the role of abdominal computed tomography in the evaluation of victims of trauma with increased
123
11.
12.
13.
14.
15.
16.
aspartate aminotransferase in the era of focused abdominal sonography for trauma. Surgery. 2002;132(4):642–6; discussion 646–7. Puranik SR, Hayes JS, Long J, Mata M. Liver enzymes as predictors of liver damage due to blunt abdominal trauma in children. South Med J. 2002;95(2):203–6. Lee WC, Kuo LC, Cheng YC, Chen CW, Lin YK, Lin TY, et al. Combination of white blood cell count with liver enzymes in the diagnosis of blunt liver laceration. Am J Emerg Med. 2010; 28(9):1024–9. Karam O, La Scala G, Le Coultre C, Chardot C. Liver function tests in children with blunt abdominal traumas. Eur J Pediatr Surg. 2007;17(5):313–6. doi:10.1055/s-2007-965533. Song SY, Chung JW, Han JK, Lim HG, Koh YH, Park JH, et al. Liver abscess after transcatheter oily chemoembolization for hepatic tumors: incidence, predisposing factors, and clinical outcome. J Vasc Interv Radiol. 2001;12(3):313–20. Kozar RA, Moore JB, Niles SE, Holcomb JB, Moore EE, Cothren CC, et al. Complications of nonoperative management of highgrade blunt hepatic injuries. J Trauma. 2005;59(5):1066–71. Giss SR, Dobrilovic N, Brown RL, Garcia VF. Complications of nonoperative management of pediatric blunt hepatic injury: diagnosis, management, and outcomes. J Trauma. 2006;61(2): 334–9. doi:10.1097/01.ta.0000197605.27190.2c.
