Health & Medical stomach,intestine & Digestive disease

CT Colonography: Accuracy, Acceptance, and Safety

CT Colonography: Accuracy, Acceptance, and Safety

Accuracy in Screening


A meta-analysis published in 2011 showed that CTC has an estimated per-patient sensitivity of 80% for adenomas ≥6 mm and of 88% for advanced neoplasia ≥10 mm (including both advanced adenomas and invasive cancer) based on the aggregated results of five CTC screening studies (figures 1–3). This meta-analysis predominantly included data on average-risk participants (n=4086, <1% high risk), defined as individuals who were asymptomatic, had no family history of CRC and no personal history of polyps, CRC or IBD. Three of the five screening studies reported per-patient sensitivities for advanced neoplasia ≥6 mm ranging from 84% to 93% (data only provided in De Haan et al). One further study published after this review showed similar results. No CRCs were missed by CTC in these six screening trials. The high sensitivity of CTC for the detection of CRC was confirmed by a meta-analysis including studies on both average and high-risk participants, showing a sensitivity of 96% for CRC. Colonoscopy, on the other hand, has a per-patient sensitivity of 92% for adenomas ≥6 mm, of 88% for advanced neoplasia ≥10 mm and of 95%–97% for CRC. These data indicate that CTC and colonoscopy have a comparable sensitivity for CRC and CRC precursor lesions of 10 mm and larger.



(Enlarge Image)



Figure 1.



Large pedunculated advanced adenoma (tubulovillous adenoma with high-grade dysplasia) in an asymptomatic 54-year-old man undergoing routine CT colonography (CTC) screening. (A) Three dimensional (3D) colon map shows the precise location of the polyp in the sigmoid colon (red dot with blue arrow), which is provided to the gastroenterologist to guide polypectomy. (B and C) Transverse two dimensional (2D) images with polyp (B) and soft tissue (C) windowing shows the lobulated soft tissue lesion in the sigmoid colon (arrow). Note the coating of the lesion surface with oral contrast, which can aid in detection at CTC. (D) 3D endoluminal CTC image better depicts the thin stalk of this pedunculated lesion, which measured 2.8 cm. (E) Image from same-day optical colonoscopy for polypectomy. The mucus cap seen at optical colonosccopy corresponds to the oral contrast coating seen at CTC. Tubulovillous adenoma with high-grade dysplasia was found at pathology.







(Enlarge Image)



Figure 2.



Early asymptomatic cancer detected in a 51-year-old man. (A) Three dimensional (3D) endoluminal image that simulates the retroflexed anorectal view at optical colonoscopy shows a relatively flat 2 cm lesion (arrowheads) adjacent to the rectal catheter. (B and C) Transverse two dimensional (B) and endoluminal 3D (C) images suggest a shallow central depression to this rectal lesion (arrowhead). (D) Endoscopic image confirms the presence and morphology of the 2 cm lesion, which contained a focus of invasive adenocarcinoma at biopsy. (E) Transrectal ultrasound suggests a superficial T1 lesion (calipers), which was confirmed at pathology after resection (T1N0M0). The patient is doing well 4 years later, without evidence of recurrence.







(Enlarge Image)



Figure 3.



Large right-sided serrated lesion (traditional serrated adenoma with high-grade dysplasia) found at screening CT colonography (CTC) in a symptomatic 56-year-old man. (A and B) Supine two dimensional (2D) (A) and three dimensional (3D) (B) CTC images show a relatively flat-appearing 3.5 cm lesion (arrowheads) near the hepatic flexure. (C and D) Prone 2D (C) and 3D (D) CTC images show that the lesion is not sessile but rather drapes broadly into the lumen. Note the unusual palisading appearance of contrast extending into crevices of this lobulated lesion. (E) Image from same-day optical colonoscopy shows a lobulated appearance. The mucus coating again corresponds to the oral contrast seen at CTC.




Diminutive Polyps


The main goal of CTC screening is to detect advanced adenomas, as the majority of CRCs develop from these. The majority of lesions smaller than 6 mm, the so-called diminutive lesions, are hyperplastic or tubular adenomas of little or no clinical significance. The chance of these lesions being malignant or containing high-grade dysplasia at the time of detection is estimated to be much <1%. Referring CTC-detected diminutive lesions to polypectomy is not a clinically efficacious or cost-effective strategy. Reporting of these small lesions at CTC is, therefore, not recommended.

Flat Polyps


Although the majority of intracolonic lesions has a sessile or polypoid morphology, a considerable minority is flat. A US study including data of 5107 screening subjects showed that 13% of 954 detected polyps were flat, but also showed that this type of lesions were less likely to have advanced histology compared with polypoid lesions of similar size. This is an important finding, as lesions with a flat morphology are more difficult to visualise during endoscopy as well as on CTC. Most studies on the sensitivity of CTC for flat lesions are characterised by small sample sizes. A Korean study comparing CTC with colonoscopy, reported that CTC detected 12 (67%) of 18 flat advanced neoplastic lesions. These findings confirmed a previous Korean study, which observed that the sensitivity of CTC for detection of flat polyps was <50%. Lesions had to have a minimum height of 2 mm and a diameter of 7 mm before CTC could detect them. Once flat lesions reach 3 cm in diameter, they may be termed 'carpet lesions' and are more readily detectable at CTC. One useful feature for detecting flat lesions is the propensity for oral contrast to coat the lesional surface, which increases the conspicuity at CTC.

Flat lesions can be divided in the more common superficially elevated type, and the more rare depressed type. It has been shown that depressed lesions are more important than other flat lesions, as they more often contain high-grade dysplasia or invasive carcinoma beyond the submucosal layer than other flat lesions. These results were in line with the results of a Japanese study showing that invasive cancer rates in lesions between 6 and 10 mm were 0.18% in flat elevated lesions, while 43.2% of depressed lesions in the same size category contained invasive cancer. In a screening cohort of 1233 subjects, no depressed flat lesions were detected, suggesting that the incidence of relevant flat lesions is low in screening cohorts. This was confirmed by a study published in 2008, showing a prevalence for depressed colorectal neoplasia of 0.18% based on the colonoscopy screening results of 8372 average-risk subjects.

Reading Strategy


CTC can be evaluated on 2D images with or without 3D problem-solving or by using primary 3D evaluation with 2D problem-solving (figures 1–4). The primary 2D read strategy is faster than primary 3D read and requires no special expertise on advanced 3D interpretation techniques. However, a large screening study showed that when less experienced readers use 3D as primary evaluation technique, their sensitivity for intracolonic findings is higher compared with their sensitivity with primary 2D read. Another large screening study from Johnson et al showed no significant difference between primary 2D and 3D reading strategies. At this point, there is still conflicting evidence regarding the possible difference of sensitivity using primary 2D or 3D read. The best strategy is probably to use both 2D and 3D for lesions detection, as they are likely complementary.



(Enlarge Image)



Figure 4.



Extracolonic malignancy (primary peritoneal high-grade papillary serous carcinoma) developing rapidly in a 55-year-old woman undergoing surveillance CT colonography (CTC) for a small 6 mm polyp detected at initial CTC screening two years earlier. (A and B) Three dimensional (3D) endoluminal image (A) from the initial CTC screening study shows a small 6 mm polyp (blue region) in the sigmoid colon, which is shown as marked by the automatic computer-aided detection (CAD) algorithm. Two dimensional (2D) transverse CTC image (B) through the upper abdomen shows excellent colonic distention and no extracolonic abnormalities. (C and D) 2D CTC image (C) from polyp surveillance study performed 2 years later shows interval development of omental caking (arrowheads) and serosal thickening of the transverse colon by the primary peritoneal tumour (arrow). 3D endoluminal view (D) from the vantage point of the blue arrow in C shows extrinsic luminal narrowing with preservation of the mucosal surface. The 6 mm sigmoid polyp (not shown) was stable in size and no other intrinsic colonic abnormalities were seen. Such detection of unsuspected extracolonic cancer at CTC is a more common occurrence than colorectal cancer among asymptomatic adults.




Computer Aided Detection


The main objective of the use of computer aided detection (CAD) is to identify lesions that have been missed by the observer (figure 4). Regge et al have shown in a prospective study with low-risk individuals that double reading with CAD, as second reader significantly improves the sensitivity for 6–9 mm polyps compared with unassisted interpretation by one radiologist (from 65% to 77%) at the expense of a longer reporting time, indicating that CAD is very useful as second reader. CAD might also be considered as a first-reader technique or as stand-alone. However, most studies on evaluating these strategies in a screening population used retrospective study designs. Further, medicolegal issues will be a hurdle for primary CAD reads.

Experience of CTC Observers


The yield of CTC, as with any imaging technique, is observer-dependent. A retrospective analysis of routinely collected data in the English Bowel Cancer Screening Program showed that the detection rate and positive predictive value of CTC are significantly higher at centres with experienced radiologists (>1000 examinations) and at centres with more than 175 cases per radiologist per year. This finding is in line with the results of a Dutch study in which six physicians and three radiographers were trained by means of a structured training programme. The average sensitivity for detection of lesions 6 mm or larger increased significantly from 76% in the first set of 50 CTC examinations to 91% in the fourth. The estimated number of CTC examinations for a sufficient sensitivity was 164, suggesting that sufficient training requires at least 164 training cases. However, among radiologists experienced with CTC, Pooler et al showed a much more uniform performance among radiologists for polyp detection compared with published variation among gastroenterologists at colonoscopy.

Leave a reply