Eur Radiol (2015) 25:669–678 DOI 10.1007/s00330-014-3446-9
Abdominal and pelvic CT: is positive enteric contrast still necessary? Results of a retrospective observational study S. Kammerer & A. J. Höink & J. Wessling & H. Heinzow & R. Koch & C. Schuelke & W. Heindel & B. Buerke
Received: 20 May 2014 / Revised: 5 August 2014 / Accepted: 16 September 2014 / Published online: 15 October 2014 # European Society of Radiology 2014
Abstract Purpose Evaluation of diagnostic accuracy of abdominal CT depending on the type of enteric contrast agent. Methods and materials Multislice CTs of 2,008 patients with different types of oral preparation (positive with barium, n=576; neutral with water, n=716; and no enteric contrast, n=716) were retrospectively evaluated by two radiologists including delineation of intestinal segments and influence on diagnosis and diagnostic reliability exerted by the enteric contrast, using a threepoint scale. Furthermore, diagnostic reliability of the delineation of selected enteric pathologies was noted. CT data were assigned into groups: oncology, inflammation, vascular, pathology, trauma and gastrointestinal pathology. Results Delineation of the bowel was clearly practicable across all segments irrespective of the type of enteric contrast, though a slight impairment was observed without enteric contrast. Although delineation of intestinal pathologies was mostly classified “clearly delimitable” more difficulties occurred without oral contrast (neutral/positive/no contrast, 0.8 %/3.8 %/6.5 %). Compared to examinations without enteric contrast, there was a significant improvement in diagnosis that was even increased regarding the reader’s diagnostic reliability. Positive opacification impaired detection of mucosal enhancement or intestinal bleeding. S. Kammerer (*) : A. J. Höink : J. Wessling : C. Schuelke : W. Heindel : B. Buerke Department of Clinical Radiology, University Hospital Münster, Münster, Germany e-mail: [email protected]
H. Heinzow Department of Gastroenterology and Metabolic Diseases, University Hospital Münster, Münster, Germany R. Koch Institute of Biostatistics and Clinical Research, University Muenster, Münster, Germany
Conclusion Water can replace positive enteric contrast agents in abdominal CTs. However, selected clinical questions require individual enteric contrast preparations. Pathology detection is noticeably impaired without any enteric contrast. Key Points • Neutral oral contrast ensures an equivalent delineation of the bowel. • Neutral contrast ensures a similar detection rate for intestinal pathologies. • Positive enteric contrast should be used in selected questions (suspected fistulas, abscesses). • Detection of pathologies and bowel delineation proved more difficult without oral contrast. • Diagnosis and the reporting physician’s diagnostic reliability are impaired without enteric contrast. Keywords Computed tomography . Abdominal CT . Contrast agent . Intestine . Enteric contrast agent
Introduction Computed tomography (CT) of the abdomen is widely accepted as a means of detecting diseases of the parenchymal organs as well as those affecting the bowel, the mesentery and blood vessels [1–3]. Appropriate examination and patient preparation, including oral contrast administration, are—in addition to other techniques, such as imaging—crucial factors for guaranteeing good-quality diagnostic images [4–6]. Oldergeneration CT systems with fewer slices and a lower spatial resolution compelled radiographers to consider oral contrast agents. They result in fewer misinterpretations in CT imaging, e.g. comparison between lymph nodes and collapsed small bowel [7–10]. Thus, oral administration of positive contrast agents containing barium sulphate or diluted iodine (DI) has become standard practice in abdominal CT, with the exception
of a few, specific indications (e.g. intestinal bleeding, kidney stones). In many countries it is still part of the clinical routine [5, 6, 11–13]. The development of multislice CT (MSCT) has enabled the measurement of isotropic voxels, allowing for high-resolution reconstructions in every spatial direction. Diagnosis of gastrointestinal (GI) pathologies has thus significantly improved and raised the question whether positive oral contrast agent administration is still necessary. In recent studies, the need for positive oral contrast in the era of MSCT has been disproved, though only in certain subgroups. For example, no oral contrast was needed in cases of appendicitis [14–17]. In other studies, the use of neutral enteric contrast agents such as water has been shown to be valuable in special clinical situations, e.g. in oncologic patients , Crohn’s disease or emergency settings [10, 18–21]. Furthermore, the assessment of mucosal hyperenhancement was found to be impaired when using positive oral contrast as a result of very similar densities . Abdominal CT examinations performed with water or without any orally administered contrast agent are expected to be as accurate as CTs using positive oral contrast; the clinical workflow could be optimised and costs further reduced as a result [7, 22]. To the best of our knowledge, no study has methodically proved the impact of different oral contrast agents in abdominal CT on certain diseases and clinical issues. This applies in particular to diseases managed at maximum-care hospitals. Therefore, this retrospective observational study aimed to evaluate the diagnostic accuracy of abdominal CT involving positive, neutral or no orally administered contrast agent with regard to all clinical diagnoses and questions related to CT imaging.
Eur Radiol (2015) 25:669–678
positive oral contrast media for all subsequent patients, unless they had specific clinical issues requiring an individual type of oral contrast preparation. In case of suspected bowel leakage or insufficiency, positive oral contrast was used. According to the clinical indication/task, patients were assigned to groups in order to identify potential differences in the required types of oral preparation. Five groups were chosen to differentiate all clinical indications dealt with in a maximum-care hospital considering the origin of illness: oncology (n=1,359), inflammation (n=225), vascular pathology (n=235), trauma/surgery (n=138) and GI pathology, e.g. bowel ischemia (n=51). Written informed consent for the clinically indicated MSCT images was obtained from every patient prior to the examination; further informed consent for the retrospective study was not necessary, according to our ethics commission. The competent institutional review board approved the study. Patient preparation The oral administration of the contrast agents was standardised. The patients in group A (positive enteric contrast) received 1,000 ml of a liquid contrast suspension containing barium sulphate (Micropaque®, Guerbet, Villepinte, France) and were instructed to drink the whole amount over a period of 60 min prior to the CT. The patients in group B (neutral enteric contrast) received 1,000 ml of mineral water and were instructed to drink it 30 min prior to the examination because of faster transit times from experience. The patients in group C (no enteric contrast) did not receive any oral contrast agent. The distribution of the different types of contrast was balanced: the groups with positive and neutral enteric contrast were of exactly the same size (n=716); the group without enteric contrast was slightly smaller (n=576).
Material and methods Patients MSCT acquisition and patient preparation A total of 2,008 patients [1,225 male (61 %), 783 female (39 %); mean age, 57.7±16.2 years (18–91 years)] were included in this retrospective study. All abdominal CTs performed in our department (including inpatients and outpatients, day and night, without restrictions of clinical indications) between February 2012 and May 2013 and which fulfilled the inclusion criterion were analysed. The inclusion criteria was the availability of an abdominal MSCT with intravenously (i.v.) applied contrast together with orally administered positive or neutral contrast, or no oral contrast agent. In case of insufficient i.v. contrast, CT images were excluded from further analysis. Examinations without any enteric contrast were mostly performed as a result of the patients’ refusal or inability to drink the contrast agent. In September 2012 we generally changed our method of oral opacification in abdominal CTs by using neutral opacification with water instead of
The examinations were performed using a 64-slice CT system (Somatom® Definition, Siemens AG, Medical Solutions, Forchheim, Germany). The i.v. contrast agent (Ultravist®370, Bayer Schering Pharma, Leverkusen, Germany) was applied at a constant injection rate of 3 ml/s. The examination delay was adapted to the anatomical region. Images were obtained at a tube voltage of 120 kV with a collimation of 32 × 0.6 mm, using a special dose-modulation template (CARE dose4D™, Siemens AG, Medical Solutions, Forchheim, Germany) to reduce the radiation exposure [23, 24]. All CT data sets were reconstructed at a slice thickness of 1.5 mm with a reconstruction increment of 0.6 mm. The imaging protocol did not differ from the standardised protocol used in clinical routine .
Eur Radiol (2015) 25:669–678
Data preparation CT data sets were stored in a Picture Archiving and Communication System (PACS), blinded, and sent to a PACS workstation (Centricity™ PACS, GE Healthcare, Fairfield, Connecticut, USA) for blinded evaluation by two experienced radiologists (each of whom had more than 3 years of experience in abdominal imaging).
Data evaluation Two radiologists reviewed and analysed the image data independently and without instructions in interpretation. Reformations corresponded to the standardised definitions in clinical routine. The data sets were presented in a randomised order to avoid memory bias in case the radiologist already dealt with the cases in clinical routine. The following three questions were used as the basis for evaluation: (a) Delineation of the intestine: For outlining of the bowel wall from the surrounding structures, the intestine was divided into nine anatomical segments (duodenum, jejunum, proximal ileum, distal ileum, ascending colon, transverse colon, descending colon, sigmoid colon and rectum) for separate analysis. The delineation characteristics were rated using a threepoint confidence scale: 1, clearly delimitable in all contrast types, positive oral contrast was not necessary to delimitate the structures; 2, partly delimitable, but possible, e.g. by consulting previous images; 3, not delimitable. (b) Detection of selected intestinal pathologies: The images of the bowel were examined for pathological GI findings (i.e. mucosal hyperenhancement, mural stratification, pseudomembranous pattern (as revealed in cases of pseudomembranous colitis), pneumatosis, anastomotic
Fig. 1 Abdominal CT scans with positive, neutral and without enteric contrast. Abdominal CT scans with positive (a), neutral (b) and without any orally applied enteric contrast agent (c). The intestine is opacified by positive contrast agent (a) or by mineral water. The intestinal loops are more distended owing to the orally applied contrast agent (a, b), compared to the examination performed without any enteric contrast (c)
insufficiency/perforation, tumour). The delineation of the pathological intestinal findings was categorised using a three-point confidence scale: 1, clearly delimitable; 2, partly delimitable; 3, not delimitable. (c) Impact of the enteric contrast agent on establishing the diagnosis and its reliability across all examinations and different entities: The impact of the enteric contrast agent on establishing a diagnosis and the readers’ sureness of the diagnosis (diagnostic reliability) across all examinations and with regard to selected diagnostic subgroups (oncology, inflammation, vascular pathology, trauma, GI pathology) was rated by using a three-point scale: 1, improved; 2, no influence; 3, impaired.
Statistical analyses Statistical analysis was performed using SAS (version 9.3 for Windows, SAS Institute Inc., Cary, NC, USA). Inferential statistics were intended to be exploratory (generating hypotheses) instead of confirmatory, and were interpreted accordingly. Hence, p values were interpreted according to Fisher, by assessing the metric weight of evidence against the respective null hypothesis of no effect. P values were considered significant if no greater than 0.05 and highly significant if no greater than 0.01. Standard descriptive statistical analyses were performed. Categorical variables are described as absolute and relative frequencies. Relationships between categorical variables are compared via contingency tables and Fisher’s exact tests. Continuous variables are presented as “mean ± standard deviation” and/or “median (minimum, maximum)” based on their distributional properties. The three-point confidence scales in questions (a)–(c) were compared between the enteric contrasts (no, neutral or positive contrast) using Fisher’s exact tests. Additionally, multinomial logistic regressions were used to obtain pairwise
Eur Radiol (2015) 25:669–678
Fig. 2 Evaluation of the delineation of the intestine with regard to the orally administered contrast as revealed by the three-point scale (1, clearly delimitable; 2, partly delimitable; 3, not delimitable) analysis. Across the entire bowel, delineation was adequate in most cases. However, compared to abdominal CT scans with positive and neutral bowel contrast
(99.1% and 98.4 %, respectively), delineation was—although also quite adequate—statistically noticeably reduced (94.8 %) in scans without any bowel contrast. Consistently, abdominal CT scans without any bowel contrast were evaluated as “partly delimitable” in more cases (5.2 %), compared to CT scans with positive or neutral bowel contrast
comparisons between the enteric contrasts in terms of the influence of the target parameters on diagnosis and on diagnostic reliability. Parameter estimates are reported as odds ratios with 95 % confidence limits.
Results A total of 2,008 CTs were classified as having sufficient quality for evaluation. Responses to questions (a)–(c) did not differ
Table 1 Delineation of bowel segments Rating
538/569 94.6 %
523/574 91.1 %
529/574 92.2 %
539/568 94.9 %
545/569 95.8 %
542/571 94.9 %
544/572 95.1 %
564/570 95.8 %
560/568 98.6 %
706/713 99.0 % 704/712 98.9 % 13/569 5.4 % 7/713 1.0 % 8/172 1.1 % 0 0.0 % 0 0.0 % 0 0.0 %
694/715 97.1 % 706/713 99.0 % 51/574 8.9 % 20/715 2.8 % 7/713 1.0 % 0 0.0 % 1/715 0.1 % 0 0.0 %
699/713 98.0 % 703/712 98.7 % 45/574 7.8 % 14/713 2.0 % 9/712 1.3 % 0 0.0 % 0 0.0 % 0 0.0 %
691/705 98.0 % 697/705 98.9 % 29/568 5.1 % 13/705 1.8 % 8/705 1.1 % 0 0.0 % 1/711 0.2 % 0 0.0 %
695/706 98.4 % 689/697 98.9 % 24/569 4.2 % 11/706 1.6 % 8/697 1.1 % 0 0.0 % 0 0.0 % 0 0.0 %
699/711 98.3 % 698/703 99.3 % 29/571 5.1 % 11/711 1.5 % 5/703 0.7 % 0 0.0 % 1/711 0.2 % 0 0.0 %
698/710 98.3 % 700/702 99.7 % 28/572 4.9 % 12/710 1.7 % 2/702 0.3 % 0 0.0 % 0 0.0 % 0 0.0 %
697/703 99.1 % 690/694 98.4 % 24/570 4.2 % 6/703 0.9 % 4/694 0.6 % 0 0.0 % 0 0.0 % 0 0.0 %
703/709 99.2 % 689/692 99.6 % 8/568 1.4 % 6/709 0.8 % 3/692 0.4 % 0 0.0 % 0 0.0 % 0 0.0 %
None Neutral Positive
None Neutral Positive
Rating of bowel segments delineation with regard to the intestinal contrast (1, clearly delimitable; 2, partly delimitable; 3, not delimitable) showed statistically noticeable differences. In most cases delineation was evaluated as “clearly delimitable”, irrespective of the bowel contrast across all segments. However, compared to examinations with positive and neutral bowel contrast, delineation was reduced (94.8 %) in scans without any bowel contrast. Furthermore, more examinations without any contrast were evaluated as “partly delimitable”. However, irrespective of the intestinal contrast, in almost none of the analysed bowel segments was delineation impossible. Owing to resections of intestinal segments (e.g.Whipple procedure with duodenal resection), the number of analysed segments differs from the number of included patients and the number of anatomical bowel Segments
Eur Radiol (2015) 25:669–678
Fig. 3 Absolute number and distribution of pathological intestinal image findings with regard to the type of enteric contrast. A pathological intestinal image finding was detected in 220 (11.0 %) examinations, mostly in examinations with neutral enteric contrast
significantly between the two readers, and interobserver errors could not be observed. Figure 1 provides examples of CT images with positive, neutral and without any enteric contrast.
vs 97.1–98.2 % for neutral and 98.7–99.7 % for positive enteric contrast. Table 1 illustrates the ranking of the delineation with regard to the enteric contrast and the intestinal segments.
Delineation of the intestine Detection of selected intestinal pathologies The delineation of the bowel was clearly practicable across all bowel segments, irrespective of the type of enteric contrast (Fig. 2). However, although “clearly delimitable” was the outcome in most cases, such a classification was found less frequently in the group without any enteric contrast (94.8 % no contrast vs 98.4 % neutral and 99.1 % positive enteric contrast; p