Jpn J Radiol DOI 10.1007/s11604-017-0613-3
ORIGINAL ARTICLE
Peritoneal fluid of low CT Hounsfield units as a screening criterion for traumatic bowel perforation Yon‑Cheong Wong1 · Li‑Jen Wang1 · Cheng‑Hsien Wu1 · Huan‑Wu Chen1 · Being‑Chuan Lin2 · Yu‑Pao Hsu2
Received: 5 September 2016 / Accepted: 4 January 2017 © Japan Radiological Society 2017
Abstract Purpose To investigate whether peritoneal fluid of low CT Hounsfield units is an important screening criterion for traumatic bowel perforation. Materials and methods We performed a retrospective study on two cohorts of blunt trauma patients who had peritoneal fluid. Intravenous and oral contrast was used for the first cohort (61 patients) as opposed to intravenous contrast only for the second cohort (60 patients). We compared the CT Hounsfield units of peritoneal fluid with bowel perforation. The optimal cutoff value of CT Hounsfield units was determined, and its diagnostic values for bowel perforation were calculated. Results The mean CT Hounsfield units (HU) of peritoneal fluid with bowel perforation were significantly lower (30.3 ± 9.0 versus 44.1 ± 13.6 HU, p = 0.008) in the second cohort. The optimal cutoff value was 43 HU, and its sensitivity, specificity, accuracy and positive likelihood ratio were 100.0, 69.2, 73.3% and 3.3, respectively, for bowel perforation. Comparisons of CT HUs of peritoneal fluid with bowel perforation in the first cohort that used additional oral contrast for CT did not show statistically significant differences.
* Yon‑Cheong Wong
[email protected] 1
Emergency and Critical Care Radiology, Department of Medical Imaging and Intervention, Chang Gung Memorial Hospital, Chang Gung University, 5 Fu‑Hsing Street, Gueishan, Taoyuan 33333, Taiwan
2
Division of Trauma and Emergency Surgery, Department of Surgery, Chang Gung Memorial Hospital, Chang Gung University, Gueishan, Taoyuan, Taiwan
Conclusion Peritoneal fluid of low CT HU is a sensitive and important CT screening criterion for traumatic bowel perforation. Keywords CT · Bowel perforation · Peritoneal fluid · Hounsfield unit · Diagnostic values
Introduction Blunt bowel and mesenteric injuries occur in less than 5% of the blunt abdominal trauma patients [1–3]. Among them, patients with life-threatening mesenteric hemorrhage or transmural traumatic bowel perforation (TBP) warrant immediate laparotomy to avoid mortality or severe peritonitis [4]. Many specific or suggestive CT findings of TBP have been reported [2, 4–10]. However, identification of most of these CT features requires expertise of radiologists or experienced trauma surgeons [5, 7]. Less experienced CT readers could easily overlook subtle abnormalities such as a bowel wall defect, bowel wall contusion, small extraluminal air pockets and mesenteric fat stranding [5, 7]. In contrast, there is less discrepancy in identifying peritoneal fluid on CT scans among readers from different training backgrounds such as radiologists, surgeons or residents [11]. Peritoneal fluid is not only sensitive for TBP, but its absence can also exclude TBP because of its high negative predictive value [4, 7]. However, the major drawback of peritoneal fluid is its low specificity for TBP; therefore, it has been deemphasized [4, 6, 7, 12]. Therefore, the aim of this study was to analyze whether we could improve its specificity by measuring the CT Hounsfield units (CTHU). Our hypothesis is that spillage of bowel fluid resulting from TBP is generally low in CTHU compared to hemoperitoneum caused by
13
Jpn J Radiol
Fig. 1 Flow diagram for patient inclusion and final number of patients with traumatic bowel perforation
solid organ injuries [4, 13]. Consequently, we compared the CTHU of peritoneal fluid in patients with blunt abdominal trauma and investigated its diagnostic value as a screening criterion for TBP. We also analyzed whether or not oral contrast medium would affect the results because it could potentially alter the density of bowel spillage.
Materials and methods We performed a retrospective study on prospectively collected data of peritoneal fluid CTHU in blunt abdominal trauma patients who were admitted to our institution. This study was approved by our institutional review board, and informed consent was waived. Data were retrieved from our two previous research projects performed at two different periods, respectively. A total of 106 patients were admitted during period 1 (16 months); 109 patients were admitted during period 2 (12 months). At period 1, oral contrast medium (2% diluted iodinated contrast medium) was routinely prescribed to patients 60 min before CT examination for gastrointestinal opacification. In contrast,
13
at period 2, none of the patients were given oral contrast medium before CT examination. Among all patients (periods 1 and 2), those who had CT evidence of peritoneal fluid in at least one of the three peritoneal spaces (right paracolic gutter, left paracolic gutter and pelvic cavity) were included for this study. Patients who had no peritoneal fluid or only minimal fluid (visualized in ≤two contiguous slices) in the rectovesical pouch were excluded. We finally included 61 (57.5%) patients from period 1 and 60 (55.0%) patients from period 2 to form the first and second cohorts, respectively, for the following analyses (Fig. 1). All CT examinations were performed with a multidetector CT scanner (LightSpeed QX/i Scanner, General Electric Medical Systems, Milwaukee, WI). The routine anatomic coverage was from the lower lung to inferior pubic rami. All CT examinations were obtained with 120 kVp and automatic tube current modulation. A uniphasic intravenous injection of 120 ml Omnipaque 350 (Iohexol, 350 mgI/ml, GE Healthcare, Ireland) was given at 3 ml per second. Portal venous phase images obtained at 70 s after initiation of intravenous contrast medium injection were reconstructed
Jpn J Radiol
Fig. 2 A 55-year-old male with traumatic bowel perforation with peritoneal fluid of 27 average HU. a CT scan at the aortoiliac bifurcation level shows mesenteric fat stranding (arrowhead), diffuse bowel wall thickening and peritoneal fluid. A cursor (o) is placed in the fluid collected in the paracolic gutter. It records 27 HU on the right and 24 HU on the left. b CT scan at the pelvic level shows a cursor (o) in the pelvic fluid of 30 HU
Fig. 3 A 39-year-old male of traumatic bowel perforation with peritoneal fluid of 42 averaged HU. a CT scan at the renal vein level shows mesenteric root fat stranding (arrowhead) and peritoneal fluid. The cursor (o) in the paracolic gutter fluid records 34 HU on the right and 31 HU on the left. b CT scan at the pelvic level shows fluid collection with layers. The cursor (o) is placed in the higher attenuating layer, and it measures 61 HU
to 5-mm thickness and 5-mm interval for viewing on a picture archiving and communication system. The CT scanner was calibrated daily with a phantom according to our institutional guidelines. All CTHUs of peritoneal fluid were prospectively measured and recorded at the time of CT examination. We used an oval-shaped region of interest (ROI) and manually placed it in the peritoneal fluid to measure the CTHU in three peritoneal spaces (Fig. 2a, b). The size of the ROI was adjusted according to the anatomic space. If the fluid collection was heterogeneous in density, we placed the ROI in different densities and calculated the averaged CTHU for analysis (Fig. 3a, b). All CT examinations with heterogeneous fluid collection and extraluminal air were recorded. Comparisons of TBP with mean CTHU, heterogeneous fluid collection and extraluminal air were performed. All medical charts were retrospectively reviewed for demographics, surgical records, discharged diagnosis,
follow-up records of outpatient visits and patient outcome by one investigator who did not know the peritoneal fluid CTHU. Diagnosis of TBP was established if the surgical record confirmed the bowel perforation. Those patients who did not undergo surgery were followed up at outpatient clinics for 3 months. They were cleared from bowel perforation if they did not develop bowel-related peritonitis during the follow-up period. All categorical variables were compared by chi-square test or Fisher’s exact test if one of the cells had an expected count of less than five. The means of continuous variables were compared by independent t test. The area under the receiver-operating characteristic curve (AUC) was used to determine the optimal cutoff value of peritoneal fluid CTHU for TBP using the maximal Youden’s index value. The diagnostic values of the cutoff CTHU and extraluminal air were then calculated. A two-sided p value of <0.05 was considered statistically significant.
13
Jpn J Radiol
Table 1 Comparison of age, CT characteristics and outcome of patients between the first and second cohorts
Characteristics
First cohort (n = 61)
Second cohort (n = 60)
p value
Age (years) CTHU of peritoneal fluid Heterogeneous fluid collection Extraluminal air Bowel perforation
33.1 ± 13.4 38.3 ± 13.0 31 (50.8%) 8 (13.1%) 7 (11.5%)
33.3 ± 14.2 42.2 ± 13.9 27 (45.0%) 7 (11.7%) 8 (13.3%)
0.952* 0.112* 0.522* 0.809* 0.757*
Death
5 (8.2%)
4 (6.7%)
1.000#
* Independent t test for continuous data and chi-square for categorical data #
Fisher’s exact test
Table 2 Comparisons between patients with and without traumatic bowel perforation in the first and second cohorts Characteristic
First cohort
Second cohort
Bowel perforation (n = 7)
No perforation (n = 54)
CTHU of peritoneal fluid
32.7 ± 13.0
39.1 ± 12.9
0.224* 30.3 ± 9.0
44.1 ± 13.6
0.008*
Heterogeneous fluid collection
2 (28.6%)
29 (53.7%)
0.255# 1 (12.5%)
26 (50.0%)
0.063#
Extraluminal air
6 (85.7%)
2 (3.7%)
p value
Bowel perforation (n = 8)
<0.001# 5 (62.5%)
No perforation (n = 52)
2 (3.8%)
p value
<0.001#
CTHU CT Hounsfield units * Independent t test #
Fisher’s exact test
Results A total of 52 patients (32 from first cohort, 20 second cohort) underwent surgery. Fifteen patients (15/121, 12.4%) were surgically proven to have TBP (7 from the first cohort, 8 from the second cohort). None of the nonoperated patients developed delayed peritonitis during hospitalization or the period of outpatient clinic followup. Overall, nine patients (9/121, 7.4%) died. Of them, seven died of massive bleeding, one of colon injury and 1 of small bowel injury. As tabulated in Table 1, the first and second cohorts did not differ in age, mean CTHU of peritoneal fluid, frequencies of heterogeneous fluid collection, extraluminal air, TBP and death. Table 2 summarizes the comparisons of TBP with mean CTHU, heterogeneous fluid collection and extraluminal air. In the second cohort, the mean CTHU of peritoneal fluid was significantly lower in patients with TBP than in patients without TBP (30.3 ± 9.0 versus 44.1 ± 13.6 HU, p = 0.008). Likewise, the AUC of peritoneal fluid CTHU for TBP (0.821, p = 0.004) in the second cohort was significant, and the optimal cutoff value of CTHU for TBP was 43 HU. The calculated sensitivity, specificity, accuracy and positive likelihood ratio of peritoneal fluid ≤43 HU for TBP were 100.0, 69.2, 73.3% and 3.3 respectively.
13
In contrast, although the mean CTHU of peritoneal fluid of the first cohort was lower in patients with TBP than in patients without TBP (32.7 ± 13.0 versus 39.1 ± 12.9 HU, p = 0.224), the difference was not statistically significant. Thirty-one patients of the first cohort and 27 patients of the second cohort had heterogeneous peritoneal fluid collection. Comparison showed heterogeneity of peritoneal fluid was not associated with TBP in either cohort (Table 2). Although frequencies of extraluminal air were low in both cohorts (Table 1), they were significantly associated with TBP (Table 2). The calculated sensitivity and specificity of extraluminal air and peritoneal fluid ≤43 HU for TBP were tabulated in Table 3. Peritoneal fluid ≤43 HU was more sensitive than extraluminal air for TBP in the second cohort.
Discussion Although peritoneal fluid and extraluminal air are both sensitive CT features of TBP, peritoneal fluid is more easily identified by CT readers regardless of their specialty backgrounds and levels of training [11]. In contrast, detection of a small volume of extraluminal air not only requires meticulous searching on every CT scan slice, but also the chance of detecting it also depends on the time lapse between
Jpn J Radiol Table 3 The calculated sensitivities and specificities of extraluminal air and peritoneal fluid ≤43 HU for traumatic bowel perforation
Characteristics
Sensitivity (%)
Specificity (%)
Extraluminal air (first cohort) Extraluminal air (second cohort) Peritoneal fluid ≤43 HU (first cohort)
85.7 (95% CI; 59.8–111.6) 62.5 (95% CI; 29.0–96.0) 71.4 (95% CI; 38.0–104.9)
96.3 (95% CI; 91.3–101.3) 96.2 (95% CI; 90.9–101.4) 42.6 (95% CI; 29.4–55.8)
100 (95% CI; 100.0–100.0)
69.2 (95% CI; 56.7–81.8)
Peritoneal fluid ≤43 HU (second cohort)
HU Hounsfield units, CI confidence interval
injury and CT examination [4, 12, 14]. Consequently, peritoneal fluid may potentially serve as an early sign of TBP even before extraluminal air is detected [12]. Peritoneal fluid is not only a sensitive CT feature, but it also has a high negative predictive value for TBP [4, 7]. However, because of its low specificity and low positive predictive value, peritoneal fluid has never been considered a sole diagnostic criterion for TBP [4, 6, 7, 12]. In this study, we could improve the specificity of peritoneal fluid for TBP by measuring its CTHU. Although there are opposing views on the radiodensity of peritoneal fluid in abdominal injuries [4, 13, 15], our results showed that peritoneal fluid of low CTHU is significantly associated with TBP. This can be attributed to the fact that bowel spillage from TBP is usually low in radiodensity. By using peritoneal fluid ≤43 HU as a CT criterion, we could achieve a high accuracy (73.3%) and positive likelihood ratio (3.3) for TBP without compromising its sensitivity (100%). The purpose of oral contrast medium is to facilitate the opacification of the gastrointestinal tract. Theoretically, the spillage of oral contrast medium from the perforated bowel is the most specific CT feature of TBP. However, its sensitivity as the sole criterion for diagnosing TBP is extremely low [16]. Other studies have also found that CT with oral contrast medium does not substantially improve the diagnostic value of multidetector CT for blunt abdominal trauma [9, 17–19], but rather can delay the CT examination for other life-threatening injuries. Oral contrast medium would potentially elevate the CTHU of bowel content. However, the degree of CTHU elevation is variable because of the dilution effect of oral contrast medium when it passes through the long gastrointestinal tract. Unlike the results in the second cohort in which low attenuating fluid ≤43 HU is significantly associated with TBP, ingestion of oral contrast medium before CT examination in the first cohort could negatively affect the diagnostic value of peritoneal fluid CTHU. Our results also showed that surgeries were more frequently required for patients with peritoneal fluid ≤43 HU than for those patients with peritoneal fluid >43 HU. However, two TBP patients died of severe sepsis and multiple organ failure because of delayed diagnosis and treatment. Fakhry et al. documented in their study that timely
treatment for TBP can improve the prognosis and any delay in the surgical treatment, even as short as 8 h, would be sufficient to result in morbidity and mortality, and the rates may increase as the delay time is prolonged [20]. False positivity of low-attenuating peritoneal fluid for TBP could occur in patients with intraperitoneal rupture of the urinary bladder as a result of extravasation of unopacified urine [21]. The CTHU of extravasated urine varies from patient to patient and depends on the homogeneity of urine opacification. It can be affected by the time lapse between intravenous contrast medium injection and CT examination, renal excretory function and ureter patency. Nonetheless, as opposed to TBP, patients with bladder rupture would have hematuria on arrival. In this study, we excluded patients with minimal fluid in the rectovesical pouch because of inconsistency in placing the ROI as well as inconsistency in measuring CTHU within the small fluid collection. Minimal fluid could occur in a patient with hypotensive shock [15, 16, 21–27]. Other associated CT features of hypotensive shock include a flat inferior vena cava, peripancreatic tissue edema, an abnormal increase of pancreatic enhancement and bowel mucosal enhancement [27]. Presence of these features could help us distinguish hypotension complex from TBP. The retrospective nature and small sample size are the main limitations of this study. Measurement of the CTHU is a simple task if the peritoneal fluid density is homogeneous. However, if the fluid has layers of different densities, then the measurements could have variable values. In this study, we recorded the average CTHU to avoid discrepancies.
Conclusion Peritoneal fluid of low CTHU (≤43 HU) is significantly associated with bowel perforation. It is a sensitive and important CT screening criterion for TBP in patients with blunt abdominal trauma. Acknowledgements This study was partly supported by research grants from the National Science Council, Taiwan (NSC 93-2314-B-182A-093 and NSC 95-2314-B-182A-198).
13
Compliance with ethical standards Conflict of interest Yon-Cheong Wong received research grants from the National Science Council, Taiwan (less than 2,000,000 yen). The other authors declare that they have no conflict of interest. IRB approval This study was approved by the IRB of our institution (Re: 101-4636B).
References 1. Watts DD, Fakhry SM. Incidence of hollow viscus injury in blunt trauma: an analysis from 275,557 trauma admissions from the East multi-institutional trial. J Trauma. 2003;54:289–94. 2. Brofman N, Atri M, Hanson JM, Grinblat L, Chughtai T, Brenneman F. Evaluation of bowel and mesenteric blunt trauma with multidetector CT. Radiographics. 2006;26:1119–31. 3. LeBedis CA, Anderson SW, Soto JA. CT imaging of blunt traumatic bowel and mesenteric injuries. Radiol Clin North Am. 2012;50:123–36. 4. Wu CH, Wang LJ, Wong YC, et al. Contrast-enhanced multiphasic computed tomography for identifying life-threatening mesenteric hemorrhage and transmural bowel injuries. J Trauma. 2011;71:543–8. 5. Butela ST, Federle MP, Chang PJ, et al. Performance of CT in detection of bowel injury. AJR Am J Roentgenol. 2001;176:129–35. 6. Ekeh AP, Saxe J, Walusimbi M, et al. Diagnosis of blunt intestinal and mesenteric injury in the era of multidetector CT technology—are results better? J Trauma. 2008;65:354–9. 7. Atri M, Hanson JM, Grinblat L, Brofman N, Chughtai T, Tomlinson G. Surgically important bowel and/or mesenteric injury in blunt trauma: accuracy of multidetector CT for evaluation. Radiology. 2008;249:524–33. 8. Peters E, LoSasso B, Foley J, Rodarte A, Duthie S, Senac MO Jr. Blunt bowel and mesenteric injuries in children: do nonspecific computed tomography findings reliably identify these injuries? Pediatr Crit Care Med. 2006;7:551–6. 9. Stuhlfaut JW, Soto JA, Lucey BC, et al. Blunt abdominal trauma: performance of CT without oral contrast material. Radiology. 2004;233:689–94. 10. Killeen KL, Shanmuganathan K, Poletti PA, Cooper C, Mirvis SE. Helical computed tomography of bowel and mesenteric injuries. J Trauma. 2001;51:26–36. 11. Yu J, Fulcher AS, Wang DB, et al. Frequency and importance of small amount of isolated pelvic free fluid detected with multidetector CT in male patients with blunt trauma. Radiology. 2010;256:799–805. 12. Saku M, Yoshimitsu K, Murakami J, et al. Small bowel perforation resulting from blunt abdominal trauma: interval change of radiological characteristics. Radiat Med. 2006;24:358–64.
13
Jpn J Radiol 13. Levine CD, Patel UJ, Silverman PM, Wachsberg RH. Low attenuation of acute traumatic hemoperitoneum on CT scans. AJR Am J Roentgenol. 1996;166:1089–93. 14. Kim HC, Shin HC, Park SJ, et al. Traumatic bowel perforation: analysis of CT findings according to the perforation site and the elapsed time since accident. Clin Imaging. 2004;28:334–9. 15. Drasin TE, Anderson SW, Asandra A, Rhea JT, Soto JA. MDCT evaluation of blunt abdominal trauma: clinical significance of free intraperitoneal fluid in males with absence of identifiable injury. AJR Am J Roentgenol. 2008;191:1821–6. 16. Hulka F, Mullins RJ, Leonardo V, Harrison MW, Silberberg P. Significance of peritoneal fluid as an isolated finding on abdominal computed tomographic scans in pediatric trauma patients. J Trauma. 1998;44:1069–72. 17. Lafferty B. Oral contrast omission in the computed tomographic evaluation of blunt abdominal trauma: a literature review. J Trauma Nurs Off J Soc Trauma Nurses. 2012;19:E1–4. 18. Stuhlfaut JW, Anderson SW, Soto JA. Blunt abdominal trauma: current imaging techniques and CT findings in patients with solid organ, bowel, and mesenteric injury. Semin Ultrasound CT MR. 2007;28:115–29. 19. Allen TL, Mueller MT, Bonk RT, Harker CP, Duffy OH, Stevens MH. Computed tomographic scanning without oral contrast solution for blunt bowel and mesenteric injuries in abdominal trauma. J Trauma. 2004;56:314–22. 20. Fakhry SM, Brownstein M, Watts DD, Baker CC, Oller D. Relatively short diagnostic delays (<8 hours) produce morbidity and mortality in blunt small bowel injury: an analysis of time to operative intervention in 198 patients from a multicenter experience. J Trauma. 2000;48:408–14 (discussion 414–405). 21. Cunningham MA, Tyroch AH, Kaups KL, Davis JW. Does free fluid on abdominal computed tomographic scan after blunt trauma require laparotomy? J Trauma. 1998;44:599–602 (discussion 603). 22. Levine CD, Patel UJ, Wachsberg RH, Simmons MZ, Baker SR, Cho KC. CT in patients with blunt abdominal trauma: clinical significance of intraperitoneal fluid detected on a scan with otherwise normal findings. AJR Am J Roentgenol. 1995;164:1381–5. 23. Brasel KJ, Olson CJ, Stafford RE, Johnson TJ. Incidence and significance of free fluid on abdominal computed tomographic scan in blunt trauma. J Trauma. 1998;44:889–92. 24. Eanniello VC 2nd, Gabram SG, Eusebio R, Jacobs LM. Isolated free fluid on abdominal computerized tomographic scan: an indication for surgery in blunt trauma patients? Conn Med. 1994;58:707–10. 25. Venkatesh KR, McQuay N Jr. Outcomes of management in stable children with intra-abdominal free fluid without solid organ injury after blunt abdominal injury. J Trauma. 2007;62:216–20. 26. Ryan MF, Hamilton PA, Sarrazin J, Chu P, Benjaminov O, Lam K. The halo sign and peripancreatic fluid: useful CT signs of hypovolaemic shock complex in adults. Clin Radiol. 2005;60:599–607. 27. Ames JT, Federle MP. CT hypotension complex (shock bowel) is not always due to traumatic hypovolemic shock. AJR Am J Roentgenol. 2009;192:W230–5.