European Journal of Radiology 83 (2014) 639–645

Contents lists available at ScienceDirect

European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad

A simple score for predicting mortality in patients with pneumatosis intestinalis Ho-Su Lee a,1 , Young-Whan Cho a,2 , Kyung-Jo Kim a,∗ , Jong Seok Lee b,3 , Seung Soo Lee b,4 , Suk-Kyun Yang a,5 a b

Department of Gastroenterology, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Republic of Korea Department of Radiology and Research Institute of Radiology, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Republic of Korea

a r t i c l e

i n f o

Article history: Received 2 July 2013 Received in revised form 2 January 2014 Accepted 4 January 2014 Keywords: Multidetector computed tomography Prognosis Mortality Pneumatosis intestinalis

a b s t r a c t Background and aim: This study was conducted to identify simple computerized tomography (CT) and clinical predictors of mortality in patients with pneumatosis intestinalis (PI). Thus, the clinical characteristics and outcomes of PI were assessed and the predictors of mortality were identified. Methods: The medical records of 123 patients with PI were reviewed retrospectively. Multivariate logistic regression models were constructed to determine independent predictors of mortality. These data were used to develop a simple score that would predict mortality on the first and seventh day after diagnosis. Results: The median age at diagnosis was 62 (range, 20–91) years. The most common cause of PI was mesenteric vascular ischemia (n = 43, 35.0%). Twenty-nine (23.6%) disease-related deaths occurred during the index admission. Both signs of peritoneal irritation on physical examination and decreased or absent enhancement of the bowel wall were associated with increased mortality. If both factors were absent, the in-hospital mortalities on both the first and seventh days after the diagnosis of PI were less than 5%. However, if both factors were present, the in-hospital mortality was 57% on the first day and 59% on the seventh day. Conclusions: A simple and novel risk score that predicts mortality in patients with PI was proposed. Patients with both peritoneal irritation and decreased or absent enhancement of bowel wall on CT should be observed vigilantly and early intervention should be instituted. © 2014 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Pneumatosis intestinalis (PI) is characterized by the submucosal or subserosal infiltration of gas in the gastrointestinal tract. PI is a clinical sign rather than a disease. Its significance in medical conditions can vary as it can range from being an incidental finding to being a life-threatening condition [1–3].

∗ Corresponding author at: Asan Medical Center, 88 Olympic-ro 43-gil, Songpa-gu, Seoul 138-736, Republic of Korea. Tel.: +82 2 3010 3196; fax: +82 2 3010 8043. E-mail addresses: [email protected] (H.-S. Lee), [email protected] (Y.-W. Cho), [email protected] (K.-J. Kim), [email protected] (J.S. Lee), [email protected] (S.S. Lee), [email protected] (S.-K. Yang). 1 Address: Asan Medical Center, 88 Olympic-ro 43-gil, Songpa-gu, Seoul 138-736, Republic of Korea. Tel.: +82 2 3010 1504. 2 Address: Asan Medical Center, 88 Olympic-ro 43-gil, Songpa-gu, Seoul 138-736, Republic of Korea. Tel.: +82 2 3010 3180. 3 Address: Asan Medical Center, 88 Olympic-ro 43-gil, Songpa-gu, Seoul 138-736, Republic of Korea. Tel.: +82 2 3010 5764. 4 Address: Asan Medical Center, 88 Olympic-ro 43-gil, Songpa-gu, Seoul 138-736, Republic of Korea. Tel.: +82 2 3010 5765. 5 Address: Asan Medical Center, 88 Olympic-ro 43-gil, Songpa-gu, Seoul 138-736, Republic of Korea. Tel.: +82 2 3010 3901. 0720-048X/$ – see front matter © 2014 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejrad.2014.01.003

One complication of PI is the rupture of subserosal gas that results in the presence of free gas within the peritoneal cavity or retropneumoperitoneum; this may require surgical intervention [1,3,4]. Therefore, for a selected population of patients with PI, early surgical intervention may reduce morbidity and mortality. However, some patients with PI will follow a benign course: in such cases, early surgical intervention would be unnecessary and even harmful. Thus, to reduce the morbidity and mortality associated with PI, it is essential that those patients who need surgery and those who should be treated by non-surgical methods are identified in a timely manner. However, the treatment and outcomes of patients with PI have only been examined by small case series studies; moreover, none of these studies have compared the outcomes of surgery and non-surgical methods directly [5–8]. The present study was conducted to determine the clinical characteristics and outcomes of PI and to identify the predictors of mortality. On the basis of the predictors identified in the present study, a simple score that predicts mortality was then developed. 2. Materials and methods This study was approved by the Institutional Review Board of the Asan Medical Center (IRB No 2011-0826). We identified

640

H.-S. Lee et al. / European Journal of Radiology 83 (2014) 639–645

adult patients (20 years or older) with PI on computerized tomography (CT) between January, 2001 and December, 2010 and the medical records of the enrolled patients were reviewed retrospectively. To calculate the follow-up duration, we extracted the last follow-up date of the enrolled patients, whose the last followup date was September, 2011. Also, the data were extracted: age, sex, underlying medical conditions (Charlson comorbidity index) [9], the possible causes of PI, the presence of abdominal pain, and the results of physical examinations (vital signs, abdominal tenderness, and peritoneal irritation signs within 24 h of detecting PI). The age and the laboratory data were extracted and were converted into dichotomous variables by using following criteria: age ≥ 60 years, white blood cell count ≥ 12,000/mm3 , creatinine > 1.5 mg/dL, albumin < 3 g/dL, bicarbonate ≤ 20 mmol/L, and C-reactive protein ≥ 10 mg/dL [2,10]. Hypotension was defined as a mean arterial pressure below 60 mmHg. 2.1. Image interpretation Two board-certified radiologist experts in gastrointestinal imaging (S.S.L. and J.S.L., who had 6 and 10 years of experience, respectively) reviewed all CT images independently. The readers were blinded to the clinical diagnosis of the enrolled patients and the results of other radiological examination other than CT. During each reviewing session, the readers were asked to evaluate the radiological findings in terms of the location and extent of the PI, the pattern of bowel wall enhancement, the presence of extraluminal gas or portal venous gas (PVG), the presence of free peritoneal fluid, and the gas shape of the PI. The following imaging criteria were used: PI location was categorized as small intestine, colon, or small intestine and colon. PI extent was categorized as less than one segment, one to two segments, or more than two segments. One segment was defined as a length equal to the length of the ascending colon. The pattern of bowel wall enhancement was categorized as decreased (or absent) and normal and was determined by comparing the enhancement of the involved bowel wall to the enhancement of the adjacent normal bowel wall. The gas shape of the PI was categorized as round or curvilinear: the gas shape of the PI with cystic was described as round, while that of the PI with linear or cylindrical pattern was described as curvilinear [1,11]. After reviewing the CT findings independently, Cohen’s kappa () values were calculated to determine the degree of interobserver variation between the two board-certified radiologists for each radiological variable. 2.2. Treatment and outcomes The therapeutic modalities were classified as conservative treatment or surgery. Conservative treatment included nasogastric tube decompression, bowel rest, and/or antibiotics. It was given to all patients initially. In all cases, the possible need for surgical treatment was considered during the period of clinical observation. The decision for or against surgery was based on the clinical characteristics or the personal wishes of the patient, and if the patient was considered to be clinically fit enough to undergo resection. The principal endpoint of the study was disease-related death at the index admission, which was defined as the first admission at the time of diagnosis with a PI. Disease-related death was defined as a death that occurred during the course of the index admission, while disease-unrelated death was defined as death occurring at any time after the index admission. 2.3. Follow-up The follow-up data after discharge were obtained by scanning the hospital files of the patients and by contacting the patients or

their relatives by telephone. The follow-up period started on the date of PI diagnosis and ended on the last outpatient clinic visit day. If patients were transferred to another hospital or did not visit the outpatient clinic, the patients were called to check their survival status. 2.4. Statistical analysis The agreement between the image interpretations of the two readers was analyzed by using  statistics. Kappa values below 0.20 were considered to indicate poor agreement, while values of 0.21–0.40 indicated fair agreement, values of 0.41–0.60 indicated moderate agreement, values of 0.61–0.80 indicated good agreement, and values greater than 0.80 indicated excellent agreement [12]. The scoring system that was developed to predict mortality on the first (or seventh) day after the diagnosis of PI was based on a logistic model, as follows. A logistic regression model was developed to predict mortality on the first (or seventh) day by using a bootstrap method. The first step in model development was to evaluate the bivariate relationship between patient characteristics and mortality on the first (or seventh) day. Risk factors that related significantly (P < 0.1) to mortality on the first (or seventh) day in univariate analysis and clinically meaningful factors were chosen as candidate variables of the scoring system. Next, the candidate variables were tested by using a bootstrap resampling procedure in which a logistic regression model with a backward elimination procedure was repeated for each of 1000 bootstrap resamplings. A 50% relative frequency of selection of bootstrap resampling was the criterion for inclusion of predictors in the final logistic model. To evaluate the fit of this model, the C-statistic was used to measure discrimination and the Hosmer–Lemeshow test was adopted to assess calibration. All reported P values are two-sided, and P values < 0.05 were considered to be statistically significant. All data manipulation and statistical analyses were performed by using SAS® Version 9.2 (SAS Institute Inc., Cary, NC). 3. Results 3.1. Patient characteristics Initially, 129 patients with a diagnosis of PI were identified. Of these, four were excluded because they were below the age of 20. Table 1 Charlson comorbidity index weighting of the clinical conditions that were present during the primary diagnosis and distribution of the Charlson comorbidity index in the patients. Comorbid condition

Weight

Number of patients

Myocardial infarct Congestive heart failure Peripheral vascular disease Cerebrovascular disease Dementia Chronic pulmonary disease Connective tissue disease Ulcer disease Mild liver disease Diabetes Hemiplegia Moderate or severe renal disease Diabetes with end-organ damage Any tumor Leukemia Lymphoma Moderate or severe liver disease Metastatic solid tumor Acquired immunodeficiency syndrome

1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 3 6 6

8 33 6 4 1 6 5 6 1 35 0 11 4 25 4 3 13 21 0

H.-S. Lee et al. / European Journal of Radiology 83 (2014) 639–645 Table 2 Possible causes of pneumatosis intestinalis in the patient population. Causes

Number of patients [n = 123, (%)]

Mesenteric vascular disease Idiopathic Bowel obstruction Chemotherapy Adynamic ileus Post-anastomosis Chronic obstructive pulmonary disease Nonspecific enteritis Inflammatory bowel disease Steroid use Endoscopy Kidney transplantation Pseudoobstruction Diverticulum Pancreatitis

43 (35.0) 20 (16.3) 17 (13.8) 13 (10.6) 9 (7.3) 4 (3.3) 3 (2.4) 3 (2.4) 2 (1.6) 2 (1.6) 2 (1.6) 2 (1.6) 1 (0.8) 1 (0.8) 1 (0.8)

Another two were excluded because their medical records were incomplete. Finally, 123 patients (72 male, 58.5%) were enrolled in this study. The median age of the patients was 62 years (range, 20–91). Table 1 presents the distribution of the Charlson comorbidity index of the patients: the Charlson comorbidity index was ≤ 1 in 38 patients (30.9%) and ≥ 2 in the remaining 85 patients (69.1%). The median follow-up duration was 6.47 (range, 0–92.4) months. As shown in Table 2, mesenteric vascular ischemia (43 cases, 35.0%) was the most common cause, followed by bowel obstruction (17 cases, 13.8%). Twenty patients (16.3%) had no significant past medical history and had no related diagnoses. Therefore, these patients were categorized as having idiopathic PI. 3.2. Clinical and laboratory variables Abdominal pain was observed in 87 patients (70.7%) while 14 patients (11.4%) had hypotension. As shown in Table 3, tenderness to palpation and peritoneal irritation signs were observed in 63 (51.2%) and 22 patients (17.9%), respectively. Azotemia and low plasma bicarbonate levels were present in 32 (26.0%) and 43 (35.0%) patients, respectively.

641

81.2%) died from disease-related causes, and the remaining three died from unrelated causes. The remaining five patients for whom surgery was recommended declined to undergo surgery: four of these patients (4/5, 80%) died due to disease-related causes but the remaining patient was discharged from the hospital. The surgical management involved bowel resection in 26 patients for ischemic bowel and exploratory laparotomies in four patients. Bowel resections for obstruction and for perforation were performed in seven and two patients, respectively. 3.5. Mortality Overall, 29 patients (23.6%) died within 1 month. Thus, the cumulative probability of mortality in patients with PI in 1 month was 23.6%. There were no additional disease-related deaths after 1 month. 3.6. Factors associated with death in univariate and multivariate analysis In the univariate analysis of the clinical characteristics, the factors that associated significantly with death were mesenteric ischemia, the presence of hypotension, the presence of tenderness, and peritoneal irritation signs (Table 3). In addition, serum creatinine ≥ 1.5 mg/dL, albumin < 3 g/dL, and bicarbonate < 20 mmol/L associated significantly with increased mortality. Regarding radiological findings, the factors that associated significantly with mortality in the univariate analysis were the presence of PI in both the small bowel and colon, the involvement of more than two segments, decreased or absent bowel wall enhancement, the presence of PVG, the presence of free peritoneal fluid, and a curvilinear shape of the intramural gas. Based on previous studies and the interobserver -values, the following variables were entered into multivariate analysis: the locations involved, the enhancement pattern, the presence of free peritoneal fluid, gas shape, peritoneal irritation signs, and hypotension. Signs of peritoneal irritation (HR: 4.57 95% CI: 1.99–10.39; P = 0.0003), and decreased or absent enhancement on CT (HR: 9.43; 95% CI: 3.18–17.98; P < 0.001) associated independently with increased mortality (Table 4).

3.3. CT variables As shown in Table 3, gas was observed most commonly in the small bowel (in 61 patients, 49.6%). PVG was present in 32 patients (26.0%), and 28 patients (22.8%) had extraluminal gas. Ascites was present in 74 patients (60.2%). The assessments of the readers of the CT variables agreed as follows. The agreement was good for the presence of a wall enhancement pattern in the involved segment ( = 0.754), the number of involved segments ( = 0.709), and the sites involved ( = 0.659). However, the agreement was moderate for the presence of PVG ( = 0.533) and the presence of ascites ( = 0.552), and only fair for the presence of extraluminal gas ( = 0.278). 3.4. Treatment and outcome during the index admission In total, 63 patients (51.2%) were conservatively managed and the remaining 60 patients (48.8%) were recommended to undergo surgery, as shown in Fig. 1. Of the latter, only 39 patients (39/60, 65%) actually underwent surgery at a median of 0 day after the diagnosis of PI (range, 0–55). Of the 39 patients who underwent a surgery, the 22 patients got an operation on the same day of the diagnosis. Eleven of the patients who underwent surgery (11/39, 28.3%) died from disease-related causes. Surgery did not proceed for another 16 patients because their general condition was considered to be too poor for anesthesia. Of these, 13 (13/16,

3.7. Simple criteria predicting mortality on the first or seventh day after the diagnosis of PI A simple scoring system model for predicting mortality in PI is shown in Table 5. A normal bowel wall enhancement pattern on CT was given 0 points, while decreased or absent enhancement was given 2 points. Signs of peritoneal irritation were given 1 point, while the lack of peritoneal irritation signs was given 0 points. The presence of both predictors yielded mortality risks of 57.7% and 59.9% on days 1 and 7 after the diagnosis of PI, respectively. The absence of both predictors yielded mortality risks of 1.8% and 4.0% on days 1 and 7, respectively. The probabilities of mortality on days 1 and 7 relative to the total risk score are shown in Table 6. Figs. 2 and 3 depict representative cases of patients who had neither factor (Fig. 2) or both factors (Fig. 3). The patient with neither factor was discharged without surgery while the patient with both factors died after the operation. 4. Discussion The pathophysiology of PI remains unclear but two main theories, one mechanical and the other bacterial, have been proposed. The mechanical theory postulates that gas enters the wall of the bowel from either the luminal surface through breaks in the mucosa or through the serosal surface by tracking along mesenteric blood

642

H.-S. Lee et al. / European Journal of Radiology 83 (2014) 639–645

Table 3 Analysis of the baseline, clinical, radiologic, and laboratory factors that may predict mortality in patients with pneumatosis intestinalis. Variable

Age ≥ 60 < 60 Gender Male Female Cause Idiopathic Ischemia Obstruction Others Charlson comorbidity index ≥2 ≤1 Tenderness Yes No Peritoneal irritation sign Yes No Hypotension Yes No Location Small bowel + colon Small bowel Colon Segmenta >2 segments 1–2 segment 1.5 mg/dL Yes No Albumin < 3 g/dL Yes No Bicarbonate ≤ 20 mmol/L Yes No C-reactive protein ≥ 10 mg/dL Yes No

Number of patients

Mortality (%)

Univariate Analysis

P value

HR

95% CI

66 57

19 (28.8) 10 (17.5)

1.69

0.79–3.64

0.18

72 51

18 (25.0) 11 (21.6)

1.19

0.56–2.52

0.65

20 43 17 43

2 (10.0) 21 (48.8) 2 (11.8) 4 (9.3)

6.21 1.21 0.96

1.45–26.54 0.17–8.58 0.18–5.24

0.01 0.85 0.96

85 38

20 (23.5) 9 (23.7)

0.97

0.44–2.13

0.94

63 54

23 (36.5) 5 (9.3)

4.73

1.79–12.45

0.002

22 95

12 (54.5) 16 (16.8)

3.96

1.86–8.41

A simple score for predicting mortality in patients with pneumatosis intestinalis.

This study was conducted to identify simple computerized tomography (CT) and clinical predictors of mortality in patients with pneumatosis intestinali...
927KB Sizes 0 Downloads 0 Views