CT colonography without cathartic preparation: positive predictive value and patient experience in clinical practice
- First Online:
- Cite this article as:
- Zueco Zueco, C., Sobrido Sampedro, C., Corroto, J.D. et al. Eur Radiol (2012) 22: 1195. doi:10.1007/s00330-011-2367-0
- 278 Downloads
To determine the positive predictive value (PPV) for polyps ≥6 mm detected at CT colonography (CTC) performed without cathartic preparation, with low-dose iodine faecal tagging regimen and to evaluate patient experience.
1920 average-risk patients underwent CTC without cathartic preparation. Faecal tagging was performed by diatrizoate meglumine and diatrizoate sodium at a total dose of 60 ml (22.2 g of iodine).The standard interpretation method was primary 3D with 2D problem solving. We calculated per-patient and per-polyp PPV in relation to size and morphology. All colonic segments were evaluated for image quality (faecal tagging, amount of liquid and solid residual faeces and luminal distension). Patients completed a questionnaire before and after CTC to assess preparation and examination experience.
Per-polyp PPV for detected lesions of ≥6 mm, 6–9 mm, ≥10 mm and ≥30 mm were 94.3%, 93.1%, 94.7% and 98%, respectively. Per-polyp PPV, according to lesion morphology, was 94.6%, 97.3% and 85.1% for sessile, pedunculated and flat polyps, respectively. Per-patient PPV was 92.8%. Preparation without frank cathartics was reported to cause minimal discomfort by 78.9% of patients.
CTC without cathartic preparation and low-dose iodine faecal tagging may yield high PPVs for lesions ≥6 mm and is well accepted by patients.
• Computed tomographic colonography (CTC) without cathartic preparation is well accepted by patients
• Cathartic-free faecal tagging CTC yields high positive predictive values
• CTC without cathartic preparation could improve uptake of colorectal cancer screening
KeywordsCT-colonography Faecal tagging Patient acceptance Colorectal cancer Bowel preparation colorectal polyp Adenoma
Despite being a preventable neoplasia, colorectal cancer is the second leading cause of cancer death in the developed world. Early detection and removal of the precursor lesion significantly reduces the incidence and mortality associated with this neoplasia . CT colonography (CTC) is increasingly accepted by both patients and professionals , owing to its non-invasiveness and high sensitivity for polyp and neoplasia detection. This sensitivity is comparable to that obtained using optic colonoscopy (OC) [3, 4]. However, the need for thorough colonic cleansing by means of cathartics still remains a major barrier limiting patient acceptance [5, 6, 7]. Tagging of faecal matter has been a substantial improvement in this line, as the need for exhaustive colonic cleansing is relaxed, while allowing good discrimination between polyps and faeces [3, 7, 8, 9].
Iodine or barium solutions have been used for faecal tagging. In some reports, they are used with no additional colonic preparation [8, 10, 11], while in other studies they were associated with low residue diets [12, 13, 14, 15], stool softening or reduced catharsis with a variety of doses and administration patterns [16, 17, 18, 19]. Some researchers conclude that lowering the dose, or even abolishing the use of cathartics, combined with faecal tagging, increases patient acceptance of CTC [14, 15, 18], achieving, in some instances, a sensitivity for polyp detection comparable to that of conventional cathartic preparation. . There has been no general agreement as to which tagging agent to use or on quantity and timing of contrast medium administration, so far. However, the prospect of replacing conventional preparation with cathartics drives research to find a new method combining diagnostic reliability, ease of preparation and patient acceptance. Barium agents have the advantage of not producing diarrhoea. Nevertheless they suffer from the important drawback of primarily tagging the solid stools rather than the liquid components . High-osmolarity iodine contrast medium, if used at low doses, softens the stools and provides a more homogeneous mix, with limited collateral effects. This reduction of side effects, especially diarrhoea, can lead to an improvement in the experience of CTC both for diagnosis and screening purposes. The essential factor is to achieve good quality colon cleansing and tagging of residual stool, which guarantees accuracy in the diagnosis, permitting its use in a real clinical setting. Performance measures which can be obtained in routine clinical CTC practice include the test-positive rate, the false-positive rate (FPR), and the positive predictive value (PPV) as only CTC with positive results made OC .
To our knowledge, no previous study, with a large series of patients, has evaluated the PPV and FPR for lesions ≥6 mm detected at CTC without laxatives and faecal tagging using low iodine doses (22.2 g).
The objective of our study was to retrospectively assess the PPV and FPR as performance measures for lesions ≥6 mm detected at CTC performed in clinical practice without cathartic preparation and faecal tagging with low-dose iodine and evaluate clinical experience.
Materials and methods
The present retrospective study was approved by the Institutional Ethics Committee for Clinical Research with waiver of informed consent.
Our retrospective study group was composed of 1920 patients (1191 women and 729 men; mean age 64.4) referred for CTC by physicians in our hospital over a 36-month interval. This group included 1600 asymptomatic average-risk adults between 50 and 79 years old who underwent CTC screening, and 320 adults between 29 and 90 years old who presented minor abdominal symptoms (changes in intestinal function, abdominal discomfort) and underwent CTC because of a relative contraindication to cathartic preparation or rejection OC. Patients with rectal bleeding or hematochezia within previous 12 months, colo-rectal cancer or polyp surveillance, family history of colo-rectal cancer, polyposis syndrome, inflammatory bowel disease or hereditary non-polyposis colo-rectal syndrome were excluded.
All patients received the same preparation, consisting of a two-day low-fibre diet, avoiding intake of all foods high in fibre, including fruits, vegetables, whole-grain bread and whole-grain cereals (no specific meal kit was used) and a liquid diet on the day before CTC examination, consisting of a enteral liquid diet (1500 ml) (Iso Source standard, Nestlè, Frankfurt, Germany, supplied by the hospital’s pharmacy) and clear liquids.
For faecal tagging, an oral iodinate contrast agent (diatrizoate meglumine and diatrizoate sodium) with an iodine concentration of 370 mg/ml (Gastrografin; Schering, Berlin, Germany, supplied by the hospital pharmacy), was administered at a total dose of 60 ml (22.2 g of iodine) over 42 h according to the following pattern:
First day, low fibre diet; second day, low fibre diet and a dose of Gastrografin (7.5 ml) diluted in 250 ml of water at each of the three main meals starting at lunch time (lunch, snack and dinner); third day, enteral liquid diet (300 ml) and a dose of Gastrografin (7.5 ml) diluted in 250 ml of water at each of the five main meals (breakfast, lunch, dinner, mid-morning and mid-afternoon snacks). At least, 2 L of water per day must be drunk. No full cathartic preparation was used before CTC. Trained nurses, dedicated to CTC, thoroughly instructed patients on this preparation (typical instruction time was ≤ 5 min).
Images were made with 16-slice CT (SOMATOM Sensation, Siemens Medical Solutions, Erlangen, Germany). The imaging parameters include a 16 × 0.75-mm collimation, a slice thickness of 1 mm, reconstruction increment 0.7 mm and 0.5-s rotation time. A low-dose protocol was used with a tube current of 25 mAs and a kilovoltage of 100 kV. In all patients supine and prone acquisitions were performed. Colonic distension for supine imaging was begun in left lateral decubitus. Then patients were instructed to roll into the supine position until completion. Approximately 30–40 puffs of room air were carefully insufflated using a manual balloon via a flexible rectal catheter. Colonic distension was assessed on the scout view and additional insufflation was performed if there was insufficient colonic distension. All CTC examinations were performed under direct radiologist supervision to ensure optimal image quality. No spasmolytic agents were used. The typical in-room time for CTC was 11 min.
The first 700 patients completed two self-administered questionnaires (following prior consent from the Hospital’s Clinical Research Ethics Committee), one, before CTC, covering: a) assessment of the preparation discomfort on a four-point scale (1: minimum, 2: mild, 3: moderate, 4: severe), b) presence of diarrhoea, number of stools, abdominal discomfort or other side effects. In the second questionnaire, after CTC, patients assessed the distress associated with the examination on a four-point scale (1: minimum, 2: mild, 3: moderate, 4: severe).
Imaging processing and interpretation were performed on a Wizard workstation (Siemens Medical Solutions, Erlangen, Germany) with specialised software: Syngo-colonography CT2006G-W, by one of the two radiologists with prior experience in reading more than 300 colonoscopically verified CTCs. Our standard interpretation method was primary 3D fly-through with 2D problem solving . Electronic cleansing software was not available. Lesions were measured with an electronic calliper at the multiplanar reformation (MPR) setting, which showed the maximal diameter of the colonic lesion detected. The colon was divided into six segments for lesion location (caecum, ascending, transverse, descending, sigmoid, rectum). The location, morphology and size were noted for every polyp to be at least 6 mm in diameter. Polyp size was categorised as small (for 6- to 9-mm lesions) or large (for all lesions measuring 10 mm or more). Polyp morphology was classified as sessile, pedunculate or flat (lesions protruding less than 3 mm from the mucosa) . Masses were defined as ≥3 cm (30 mm) and described as polyps, carpet lesions, annular or hemi-circumferential.
Consistency of faecal residue (solid, liquid, solid/liquid).
Total amount of liquid residue per segment (five-point scale) assessed in axial sections in supine decubitus: as the largest air fluid level relative to the maximum antero-posterior diameter in the same segment: 1: > 75%, 2: 50–75%, 3: 25–50%, 4: <25%, 5: no residual fluid . For segments with several different levels, only the largest was considered.
Residual stool, graded as 1: greater than 75% of the lumen filled with stool, 2: 50–75% of the lumen filled with stool, 3: 25–50% of the lumen filled with stool, 4: less than 25% of the lumen filled with stool, 5, no stool.
Quality of faecal tagging (visual five-point scale): 1: completely untagged, 2: poorly tagged, 3: average tagging, 4: well tagged, 5, excellent tagging.
- e)Distension (five-point scale): 1: collapsed, 2: poor, 3: adequate, 4: good, 5: excellent (Fig. 1).
Fibre-optic colonography was performed on patients with positive findings for polyps or masses 3–92 days after positive CTC. Lesion measurement at OC was performed using visual comparison against an open forceps or another endoscopically inserted device.
The records and reports from both examinations were retrospectively reviewed from June to December 2010 for the evaluation of CTC–OC agreement on lesions ≥ 6 mm. This data collection was performed by two radiologists not involved in the readings. Findings were categorised as matched lesions OC-CTC if their size measurements were within a 50% margin of error and if they were located in the same or in an adjacent segment [3, 24].
Pathological analysis classified adenomas as tubular, tubullo-villous, villous, serrated, high-grade dysplasia or adenocarcinoma. Neoplastic lesions were defined as adenoma or adenocarcinoma. Advanced adenomas were defined as tubular adenomas (≥ 10 mm diameter) or adenomas of any size with more than 25% villous component or high-grade dysplasia . Advanced neoplasia included both advanced adenomas and adenocarcinomas. Invasive carcinoma was defined as a malignant extension past the muscularis mucosae.
A follow-up of clinical histories and histological registries of all 1920 patients was performed in order to identify subsequent clinical events or diagnosis related to polyps or colonic neoplasias.
Matched lesions were considered as CTC true-positive (TP). Lesions were categorised as CTC false-positive (FP) if detected on CTC but without matching OC, and as CTC false-negative (FN) when polyps were detected on OC but not on CTC. Subjects with at least one TP lesion per size category were classified as CTC–TP cases; those with no TP and one or more FP polyps were classified as FP cases. The positive predictive values (PPV) of CTC findings were calculated per polyp in relation to size and morphology and per patient. The Statistical Package for Social Sciences, version 17.0 (SPSS, Chicago, IL, USA) was used for data recording and analysis.
Descriptive statistics were computed for quantitative (mean and standard deviation) and qualitative variables (frequency, percentage and confidence intervals). The Student t-test was used to analyse differences between means and the Chi-squared test for the association between qualitative variables. PPV and their confidence intervals were used for comparison of the two examination procedures. Effects were considered statistically significant when P < 0.05.
Histopathology of neoplastic polyps and masses detected at CT colonography according to lesion size
Neoplastic polyps 6–9 mm [n = 124]
Neoplastic polyps ≥ 1 cm [n = 109]
Masses ≥ 3 cm [n = 46]
Positive predictive value (PPV) and false-positive rate (FPR) for polyps ≥6 mm detected at CT colonography
Positive predictive value (PPV)
False positive rate (FPR)
All polyps(≥6 mm)
According to lesion size
Small (6–9 mm)
Mass (≥3 cm)
According to lesion morphology
A total of 20 false-positive ≥ 6-mm polyps were found in 14 patients obtaining a per-patient FPR of 5.9% (14/236). Only 1 false-positive lesion >30 mm, due to endometriosic infiltration, was found (Table 2). OC identified 13 individual polyps ≥6 mm that were not detected in CTC examinations in 10 out of 236 patients (4.2%) undergoing complete OC examination. All these polyps were small polyps (6–9 mm). In each of these patients additional polyps of similar or larger size were found at prospective CTC, showing concordant OC findings.
Three patients showed mild vasovagal reactions, which were all resolved in the radiology service in 30–45 min. No allergic reactions occurred.
Qualitative analysis for faecal residue, tagging and distension across the six different colonic segments
4.77 ± 0.64
4.75 ± 0.67
4.72 ± 0.69
4.70 ± 0.72
4.69 ± 0.77
4.78 ± 0.62
4.73 ± 0.64
4.41 ± 0.58
4.64 ± 0.52
4.48 ± 0.64
4.23 ± 0.76
3.71 ± 1.02
4.11 ± 0.86
4.26 ± 0.55
4.97 ± 0.21
4.97 ± 0.22
4.95 ± 0.28
4.85 ± 0.60
4,63 ± 1.00
4,43 ± 1.23
4.80 ± 0.45
4.71 ± 0.76
4.88 ± 0.37
4.91 ± 0.29
4.80 ± 0.61
4.58 ± 0.88
4.14 ± 1.15
4.66 ± 0.49
4.99 ± 0.14
4.98 ± 0.18
4.92 ± 0.47
4.78 ± 0.68
4,57 ± 0.82
4,97 ± 0.24
4.87 ± 0.30
The CTC examination was minimally unpleasant for 55.0% (385/700) of patients surveyed and severely unpleasant for 6% (42/700 Fig. 5).
In our study, comparison with other diagnostic approaches was based on indirect quality measures such as the test positive rate, false-positive rate and positive predictive value, as, in clinical practice, only positive CTC findings are referred for OC for polypectomy or for biopsy [20, 26, 27]. A high PPV is essential for CTC to be considered as an efficient, non-invasive technique, since it protects against unnecessary duplicities in screening or diagnostic tests. The high overall per-patient PPV, 92.8% at the 6-mm size threshold, reflects a very good agreement in positive findings between CTC and subsequent OC. This result is in line with recent previous studies in clinical practice with conventional preparation with laxatives showing a 90–92% PPV [20, 27] and substantially outperforms a large published multi-centre CTC screening trial . The low per patient FPR (5.9%), also in line with these studies [20, 27, 28], makes our CTC screening approach highly cost-effective. It could be argued that in a clinical practice set-up, where false negative rates cannot be assessed, high PPV could result from only truly relevant lesions being detected. However, out of 236 patients who underwent OC after positive CTC findings, only in 10 (4.2%) were diagnosed additional polyps not previously detected by CTC. Of the 33 FP and FN polyps, 5 were due to differences in localisation and 6 to differences in size which broke the matching rules, while 22 were undetected in CTC or OC. Besides, the follow-up of clinical histories and cancer registries of all patients did not reveal any clinical events related to polyps or neoplasia. Out of 167 patients who had an OC or a new CTC since the initial CTC was performed (6 to 42 months earlier), only 2 subjects had two small polyps (6–9 mm) detected which were either not present or undetected in the initial CTC. If they were really false-negatives, would increase our FN rate by only 1.2%. No large polyps or colonic neoplasia were found.
Our CTC positive rate for lesions ≥6 mm, and the prevalence of advanced neoplasia (8.7%), were slightly higher than in others screening trials [4, 20, 28]. This could be due to the higher mean age in our series, since the prevalence of advanced neoplasia increases from 5.7% in patients aged 50–59 to 13% in patients aged 70–75 . However, this can be taken as an indication that relative performance of CTC is not lessened when no cathartics are used in patient preparation. The relative high rate of cancer in this population could also be attributed to the same factor, since 56% of these lesions were found in patients over 79 years.
Ionic iodinated agents such as diatrizoate meglumine and diatrizoate sodium are hypertonic and exert a mild osmotic laxative effect because of an increased colonic fluid load, resulting in liquid faeces with homogeneous tagging and, therefore, allowing for better detection of lesions. Moreover, when the patient shifts from supine to prone decubitus, or between oblique positions, the residual liquid or semi-liquid faecal material is readily redistributed, offering a larger area of “clean” mucosa for the study as a whole. On the contrary, solid faeces tend to form an irregular layer that adheres to the mucosa. In our study, 86.2% of patients had liquid faeces, which we believe, in agreement with Zalis et al. makes interpretation easier  and reduces reading time (< 10 min in our study).
Since residual stool is generally recognised as a major source of potential error at CTC, the scarce faecal residue, either liquid or solid, along with the excellent labelling achieved with the iodine faecal tagging preparation in our study are remarkable results, especially so given the small volume of iodine applied with regard to that used in other studies. The fact that patients took diatrizoate meglumine and diatrizoate sodium in small doses over 42 h was probably essential for the high level of efficiency in tagging. Although recent studies [32, 33, 34, 35, 36] report good results when iodinated faecal tagging agents are administered over 24 h, in our opinion, the need for adequate tagging in all segments should prevail over the benefits of a reduced period of iodine administration, since the latter is not among the main sources of distress for patients. Moreover, side effects of the iodinated agent (diarrhoea, abdominal pain, nausea and vomiting) may be reduced when administered at lower doses, even if this is over a longer period of time. Few studies have been conducted with iodine tagging as the sole preparation [10, 12, 15, 19, 32, 33, 34, 35, 36, 37], each of these used different doses of iodine. Iannaccone et al.  administered 200 ml of diatrizoate meglumine and diatrizoate sodium (total of 74 g iodine over 48 h), achieving faecal tagging judged as excellent in 98.5% of patients and episodes of diarrhoea or abdominal discomfort reported in 10.3% of patients. Other researchers applied smaller quantities (60 g of iodine) over 24–48 h [10, 19, 34], but, when side effects were assessed, a large percentage of patients was affected, particularly by diarrhoea. In our study, the amount of iodine used (60 ml of diatrizoate meglumine and diatrizoate sodium, containing 22.2 g of iodine) was substantially lower than in any previous study, except those of Keeling et al.  who used 15 g of iodine in frail elderly patients with a limited objective of ruling out gross pathological conditions and Liedenbaum et al.  who compared the effects of different doses of iodine (45 g–22.5 g). In our study, diarrhoea, with a maximum of three stools per day and graded as mild by all except 12 patients, was declared by only 7.4% (52/700) of patients. This is probably one of the key reasons why the preparation experience in our study was graded as excellent, since 94% of patients found it minimally or mildly unpleasant and non-interfering with their common daily activity, while only 5.9% considered it moderately to severely upsetting. This remarkable result indicates that this protocol enables high quality tagging and faecal cleansing with almost no significant diarrhoea and represents an encouraging outcome in CTC acceptance for screening purposes. Contrary to previous publications in which laxative cleansing before CTC is the most uncomfortable part of the whole procedure [5, 6, 18, 38], in our study the percentage of patients who considered themselves minimally disturbed by the preparation was significantly higher (78%) than those who were minimally disturbed by the examination (55%, Fig. 5). However, 89.6% of patients graded the examination as minimally-mildly unpleasant. Given this good acceptance, the excellent colonic distension achieved (4.87 mean distension) and the absence of complications, the possibility of modifying our colonic distension system, with room air, was not considered. Both techniques, manual room air insuflation and automated CO2 delivery for CTC are safe techniques, both cause an acceptable level of discomfort during and after the examination, and they produce reliable colonic distension . On similar grounds, we avoided a generalised use of spasmolytics, a more controversial practice as i.v. drug administration lengthens examination time, increases the possibility of additional side effects and adds a new source of discomfort [6, 39].
Several limitations of our study deserve specific discussion. First of all, those stemming from being performed with an observational purpose, thus preventing the estimation of direct performance indexes such as sensitivity or specificity. Nonetheless, the good surrogate performance measures (92.8% CTC-OC concordance rate and 5.9% per patient FPR) suggest that there was no significant underdiagnosis of patients with colonic lesions and that relevant lesions have been adequately detected and removed, since the prevalence of advanced neoplasia obtained in our study is in line with other screening trials.
Secondly, questionnaires about patient experience of preparation and examination were carried out, once only, immediately after the procedures, but not repeated at a later stage as in some other studies [18, 32, 33]. This could have altered patient perception, although we would expect that our conditions would, if they had any effect, bias our results towards poorer acceptance. Thirdly: The CTC reading was performed by one of two experienced radiologists in our service, as double readings would not be feasible in our clinical practice set-up.
The length of the preparation protocol could be an additional concern. Like most studies performed with a tagging-only bowel preparation for CTC [7, 12, 15, 19, 31, 32, 33], ours used a low-fibre diet with clear instructions for the patients. Most studies prescribe this diet for 2 or 3 days before the CTC examination. Low-fibre diets reduce residual bowel content and improve subjective tagging quality of residual faeces . They are varied enough to barely cause any burden on patients, as evidenced by Liedenbaum el al who found no significant differences in acceptance, with respect to degree of burden, between patients who followed the restricted diet and those who did not . On the other hand, the enteral liquid diet used on the day before CTC covers all dietary requirements and prevents hunger sensation. The only drawback would be that, as it is furnished by the hospital pharmacy, it adds to the overall cost of the examination, albeit to very limited amount. In this respect, the substitution of this diet for conventional liquid foods is currently under consideration. With regard to the amount of tagging agent (8 × 7.5 ml over 42 h in our scheme), we find it preferable to administer it in a larger number of small doses provided all segments are adequately clean and labelled and side effects are reduced. Therefore, the assayed regime provides an optimal balance between tolerance and safety, the two key factors for increasing CTC acceptance.
In conclusion, our study shows that CTC without cathartic bowel preparation and iodinated agents for faecal tagging can obtain high PPV values and a low rate of false-positive results for ≥ 6-mm polyps comparable to those obtained with conventional preparation with laxatives. Furthermore, because of a good patients’ experience, this method could really improve the acceptance of CTC for colorectal cancer screening.
We thank C. P. Cadórniga for critical review of this manuscript and A. Salgado for his help with the data analysis.