Cesare Hassan is a gastroenterologist working in an Endoscopic Centre mainly directed towards colorectal cancer prevention. His main research fields are the appropriateness of colonoscopy in the open-access setting, and the prevention and management of colorectal polyps. He is currently involved in international trials focused on colorectal neoplasia. He has also performed research on the application of new radiological techniques in inflammatory bowel diseases. He is the author of more than 50 papers.

Cesare Hassan and Angelo Zullo
Gastroenterology and Digestive Endoscopy, ‘Nuovo Regina Margherita’ Hospital, Rome, Italy

Address for correspondence:
Cesare Hassan, MD
Ospedale Nuovo Regina Margherita
U.O. di Gastroenterologia ed Endoscopia Digestiva
Via Morosini 30, 00153, Rome, Italy
Tel: +39 (0)6 6830 8650
Fax: +39 (0)6 6830 8660
Email: gastroroma@virgilio.it

Abstract
Colorectal cancer (CRC) is the second-leading cause for cancer-related death worldwide. Colonoscopy is a safe procedure which can identify colorectal polyps and early cancers with the highest accuracy, and
efficaciously remove them. Indirect evidence, based on large randomised and observational studies, has clearly shown colonoscopy to reduce mortality from CRC. Colonoscopy appears to be better accepted by the patients, and is more cost-effective when compared with virtual colonoscopy. The detection of flat lesions in a high proportion of patients, even in Western populations, has largely increased after the pioneering of ‘magnifying’ and ‘high-resolution’ endoscopy. The endoscopic identification of such lesions could constitute a crucial advantage over virtual techniques.

Introduction
Colorectal cancer (CRC) is the second-leading cause of cancer-related death worldwide. Indeed, its incidence is rising, even in countries in which it was previously regarded as substantially low, such as the Eastern populations.1 Screening for colorectal cancer is based on two main factors: CRC carcinogenesis involves a precancerous polypoid stage that can be stopped by polypectomy (Figures 1a and 1b); and patients diagnosed with early lesions have a proven better outcome in comparison to those with lesions at an advanced stage. Indeed, the adenoma-carcinoma sequence is a single, indivisible continuum that begins as a mild epithelial dysplastic polyp and can progress through dysplasia of increasing severity to a locally aggressive carcinoma. Its subsequent invasion through the lymphatic and haematic pathways determines the patient’s prognosis. Colonoscopy is the only technique which can both identify colorectal polyps and early cancers with the highest accuracy, and efficaciously remove them with polypectomy/mucosectomy procedures at the same index examination. Undeniably, these benefits have led to this examination becoming the most suitable screening tool for CRC.

Figure 1. Conventional colonoscopy image (a) and resected specimen in a case of pedunculated colonic polyp (b).

Colonoscopy for CRC screening
The ultimate aim of a screening procedure is to save lives, thereby reducing the death rate from cancer. Although indirect, there is evidence that colonoscopy achieves this aim. Large, randomised clinical trials have shown a significant reduction of mortality of about 33% using the faecal occult blood test (FOBT), mainly due to the use of colonoscopy in patients with a positive result.²

Moreover, the widely documented effectiveness of flexible sigmoidoscopy in reducing CRC mortality³ can be reasonably extended to colonoscopy. In addition, a large observational study found a 75–90% lower CRC incidence in patients who had undergone colonoscopy compared to historical controls.⁴ Further evidence from mathematical models and analysis of cost-effectiveness clearly supports the colonoscopy technique as a cost-effective screening tool for CRC.⁵

A safe procedure for general population screening Some criticism has, however, been directed to colonoscopy. The common criticism is that it is too invasive and is potentially dangerous for use in screening healthy people. Recent evidence undeniably shows colonoscopy to be a safe procedure in screening average-risk patients. Indeed, the major complication rate from the diagnostic procedure was as low as 0.1% and no higher than 0.3% for polypectomy.⁶

The second usual criticism of colonoscopy is that it is poorly accepted by an asymptomatic subject. However, substantial evidence clearly shows that it is the bowel preparation – and not the examination itself – that is associated with adverse events. Indeed, with the examination conducted under sedation, colonoscopy is more acceptable to patients than a noninvasive procedure, such as virtual colonoscopy.⁷

Thirdly, to screen all the 50–60-year-old general population is, according to some pessimists, impossible due to the limited resources of endoscopic centres. However, it has been estimated that reducing the number of inappropriate examinations, avoiding flexible sigmoidoscopy screening, and improving the organisation of the endoscopic centres in the USA could allow an increase of more than 2.6 million colonoscopies per year, covering the need for screening the general population.⁸

CT colonography
CT colonography, also known as ‘virtual colonoscopy’, has been recently proposed as a screening technique. In the earlier reports – mainly based on unselected populations – virtual colonoscopy was found to be disappointing overall. Indeed, the specificity for lesions larger than 1 cm was reported to be low, with a false positive rate of higher than 15% in many work-ups.⁹ Analysis of cost-effectiveness did not find CT colonography to be competitive with respect to colonoscopy due to the high number of colonoscopies that would have to be performed for either a true- or a false-positive result.¹⁰ Moreover, many small polyps would have been missed, so that a negative examination could not be felt to be as reassuring as a negative colonoscopy with regard to the possibility of future lesions.

According to a recent study11, some of the previous limitations of CT colonography have been overcome with a primary three-dimensional approach (Figures 2a and 2b). In this study, in which only asymptomatic subjects were screened, an acceptable accuracy for lesions = 6 mm in diameter was shown with a specificity of about 80%.11 However, some considerations need to be put forward. Firstly, almost half of this low-risk population had some polyps, and half of these polyps were more than 6 mm in diameter. Therefore, even with a high accuracy, 25–30% of the patients would need a colonoscopic procedure, and the same proportion of patients will be left with the emotive reaction of having a ‘non-removed’, potentially advanced, diminutive polyp. If we assumed radiologists were able to reduce the high cost of virtual colonoscopy to that of diagnostic colonoscopy, even this would result in a prohibitive 30% rise in the overall cost of CRC screening. Secondly, such promising results need to be confirmed on a wider scale and in different settings. Thirdly, it has been shown that significantly more patients complained of discomfort after CT colonography than after endoscopy. Moreover, after a positive virtual colonoscopy when endoscopy is not feasible on the same day – as this is easily foreseeable in the clinical practice – how does the radiologist explain to the patient that they have to return on another day for a second bowel preparation for polypectomy? It is clear that population screening with CT colonography would involve a complex relationship between radiology and endoscopy that would complicate and not ease the organisation of such a huge workload. According to these observations, the overall impression is that CT colonography currently seems to be able to attain what colonoscopy achieved two decades ago – that is a high accuracy in detecting colorectal polyps and carcinoma – but without resulting in either a clear better acceptance from the patients or in a reduction of screening costs.

Figure 2. (a) Endoluminal, ‘endoscopic-like’ view, showing irregular stricture; (b) “virtual” double contrast enema.

Outlook and conclusions
At the beginning we stated that one of the main postulates of CRC screening is to detect lesions of the polypoid route of the adenoma-carcinoma sequence. However, it is now clear that not all the colorectal cancers come from polyps, since a de novo pathogenesis has been reported and claimed to be responsible for at least 15% of CRC.12 In recent years, endoscopy has made major steps forward in this field with the introduction of so called ‘magnifying’ and ‘high-resolution’ endoscopy.

Such powerful technological improvements have allowed the detection of flat lesions in a high rate of patients, even in Western populations in which they were previously scarcely reported.13,14 Such lesions are associated with a high prevalence of high-grade dysplasia or early cancer, especially in the right colon, and they are presumably responsible for the remaining 10–15% risk of CRC in already-screened subjects. If the endoscopy approach also proves to detect flat lesions with a high accuracy, we could experience a distinct improvement in CRC screening, and colonoscopy would emerge to be by far the most accurate procedure in this field.

Key Learning Colonoscopy, for CRC screening: • Identifies colorectal polyps and early cancer with high accuracy • Enables removal of early lesions by polypectomy at the index examination • Reduces CRC mortality • Is a cost-effective screening tool • Is safe • Is more acceptable to patients compared to virtual colonoscopy • Using magnifying and high-resolution endoscopy, enables detection of flat lesions