High-definition endoscopy with digital chromoendoscopy for histologic prediction of distal colorectal polyps

Distal diminutive polyps (polyps ≤ 5 mm) are frequently observed in daily clinical routine and can be found in more than 50 % of the screening population [1, 2]. However, diminutive polyps have a very low prevalence of advanced histologic features [3, 4] and their cancer prevalence ranges between 0 and 0.08 % [5, 6]. Therefore, histology of diminutive polyps is mostly used to guide post-polypectomy surveillance. Since conventional white-light endoscopy cannot reliably distinguish histological features of colorectal polyps, it is standard practice to remove all polyps for histopathological evaluation. However, this routine pathologic evaluation of all resected diminutive polyps results in considerable costs to the health care system for the management of a group of lesions which eventually have only limited clinical significance. The annual up-front cost savings in the US of forgoing the pathologic assessment would exceed one billion dollars per year [7].

Recently, the American Society for Gastrointestinal Endoscopy (ASGE) proposed two new paradigms for the management of diminutive polyps in the so called PIVI-statement [8]. One paradigm describes endoscopic resection of colorectal polyps without submitting them for pathological assessment (“resect and discard”) while the other paradigm describes to leave diminutive hyperplastic rectosigmoid polyps in place without resection [8]. A prerequisite for both approaches is that an accurate real-time endoscopic assessment of the colorectal polyp histology could be made [8].

Within the recent years, dye-less chromoendoscopy techniques (DLC) are rapidly evolving allowing characterization of the tissue in real-time without the requirement of traditional dye-spraying. DLC includes optical (i.e. NBI, CBI) and digital (i.e. i-scan, FICE, SPIES) chromoendoscopy [9‐12].

Recently, it was shown that NBI is reliable for in vivo prediction of distal polyp histology with a negative predictive value of ≥ 90 % [13‐16]. Nonetheless, data on the prediction of distal polyp histology with digital chromoendoscopy are scarce to date. Here, we prospectively assessed the potential of digital chromoendoscopy for real-time in vivo prediction of distal diminutive polyp histology according to the requirements of the ASGE PIVI statement.

Here, we have shown that digital chromoendoscopy can reliable predict histology of distal diminutive polyps in real time according to the recommendations of the ASGE PIVI statement.

Distal diminutive colorectal polyps are common in the general screening population. Thus, an accurate endoscopic in vivo identification of polyps that are hyperplastic and therefore do not possess a risk of developing colorectal cancer is of major importance to reduce costs and risks associated with their redundant removal. Based on these considerations, the ASGE has proposed the PIVI statement on the real-time assessment of colorectal diminutive polyps [8] in which it is recommended that the technology used to guide the decision to leave a suspected hyperplastic polyp in the rectosigmoid in place should provide at least 90 % negative prediction for adenomatous histology. Further, in order for diminutive colorectal polyps to be resected and discarded without pathologic examination, a new technology should provide a ≥ 90 % agreement with histopathologic assessment in assigning post-polypectomy intervals.

Within this study we thus set off to prospectively evaluate digital chromoendoscopy for the real-time in vivo assessment of distal diminutive colorectal polyps and specifically questioned whether digital chromoendoscopy is sufficiently accurate (i) to allow distal hyperplastic polyps to be left in place without resection or (ii) to be resected and discarded without pathological assessment. We found that digital chromoendoscopy is indeed able to differentiate adenomatous from non-adenomatous histology in real time with a high overall sensitivity and specificity. Importantly, digital chromoendoscopy exhibited a negative prediction for ruling out adenomas of 98 % in high confidence predictions and predicted endoscopic surveillance intervals with ≥ 90 % agreement compared to histology based US and European surveillance guidelines, thereby exceeding by far the thresholds recommended by the ASGE for leaving suspected hyperplastic polyps in place and for resecting and discarding diminutive polyps without pathologic assessment [8]. Of note, these values included diminutive polyps ≤ 5 mm of the rectum, sigma and descending colon. When focusing on polyps of the rectum and sigmoid colon only, digital chromoendoscopy exhibited an equally high diagnostic performance for both, the negative prediction of adenomatous histology and the prediction of surveillance intervals.

The results of this study are consistent with previous studies on optical chromoendoscopy which have shown that NBI can predict polyp histology in real-time with high accuracy [14, 16, 24, 25]. Further, the available data illustrate that NBI indeed holds the potential to facilitate the management of distal diminutive polyps as it could accurately exclude adenomatous lesions with a sufficient negative prediction and also allowed for prediction of post polypectomy surveillance intervals in real time [13, 15, 26].

Recently, it has been demonstrated that digital chromoendoscopy is superior compared to standard white-light colonoscopy for the detection of colorectal neoplasms [27], and a prospective cohort study demonstrated that NBI and i-scan exhibit a similar efficiency for the histological prediction of diminutive polyps [22]. Further, analyzing polyps throughout the whole colon, it was shown that digital chromoendoscopy could predict histology in diminutive colorectal polyps with high accuracy and could also predict surveillance intervals [21]. However, in this report high definition white-light endoscopy exhibited the same high negative prediction for adenomatous histology as digital chromoendoscopy [21]. As already discussed by the authors, the results of this study might exhibit a certain bias, as the study followed a strict protocol of cleaning all polyps with water, simethicone, and N-acetylcysteine prior to the assessment of polyps [21].

Within this study, we included polyps of the distal colorectum and assessed them in real time under clinical conditions without pre-staining of the mucosa.

Importantly, our results show that the diagnostic performances of assessing histology are similar when all distal diminutive polyps were analyzed (including descending colon) or when the analysis was limited to diminutive polyps in the rectosigmoid only. These findings have important implications: Firstly, consistent with the results of the HiScope study [21], they provide further evidence that digital chromoendoscopy is indeed capable to translate the “leaving distal hyperplastic polyps in place” and “resect and discard” strategies into clinical practice, and secondly, they also allow a first insight that the “leaving in place” paradigm might not only be limited to polyps in the rectosigmoid, but could also be extended to the descending colon. However, the later aspect needs to be verified in a separate and large study cohort specifically addressing this question.

Although it has recently been suggested that SSAs may show changes in the microvasculature [28‐30], their endoscopic identification and differentiation is challenging. Previous studies have shown that almost one third of all SSAs are misdiagnosed as hyperplastic polyps on NBI [31]. Since SSAs can progress to cancer through the serrated neoplasia pathway and can lead to sporadic microsatellite instability with a high colon cancer rate [32‐34], their accurate endoscopic identification is of high clinical relevance. Although limited in number, all 4 SSAs included in the current study were correctly predicted be to adenomatous lesions by digital chromoendoscopy. Based on this, larger studies that assess whether digital chromoendoscopy can accurately predict SSA histology are highly warranted.

The strengths of this study are its prospective design and the assessment of a question that is highly relevant to daily endoscopic routine, and to which the answers are likely to influence and change daily practice. As a limitation, this study was conducted as a single center study and endoscopies were performed by a single endoscopist and it is anticipated that the interpretation of digital chromoendoscopy findings are also subject to inter-individual variations. Thus, studies in which less-experienced and community physicians perform digital chromoendoscopy to predict histology of diminutive polyps will be needed before this approach becomes widely clinically accepted. Nevertheless, digital chromoendoscopy is, similar to NBI, a simple “push-of-a-button” technology and data on NBI illustrate that the appearance of adenomatous polyps can be learned within 15 to 20 min with moderate inter- and substantial intraobserver agreement [35‐37]. Similarly, endoscopists with varying levels of experience can accurately predict polyp histology using digital chromoendoscopy after a single training session [38, 39] and digital chromoendoscopy exhibits comparable accuracies for the detection of colonic lesions by non-expert and expert endoscopists [40].

Further, we did not include a separate control arm assessing whether high-definition endoscopy without the concomitant utilization of digital chromoendoscopy can adequately predict polyp histology. However, it has been shown recently that the adenoma detection rate does not differ between procedures utilizing high-definition or standard-definition endoscopes [41]. Further, it is known that high definition endoscopy in combination with digital chromoendoscopy is superior compared to white-light colonoscopy for the detection of colorectal neoplasms [27]. Based on these existing data, we thus aimed to address the remaining question whether high definition endoscopy in combination with digital chromoendoscopy is accurate to predict real time polyp histology, a question which can be well addressed with the design of our study. Nevertheless, it seems clear that in the future multi-center and multi-observer studies including experts and non-expert endoscopists should be performed to assess the inter-observer agreement and accuracy for predicting histology and also to assess the learning curve for predicting histology in diminutive colonic polyps using digital chromoendoscopic modalities. Further, it is important to demonstrate in the future that digital chromoendoscopy is as precise as optical chromoendoscopy and conventional dye based endoscopy.

In summary, we have shown that digital chromoendoscopy allows an accurate real time prediction of the histology of diminutive polyps in the distal colon and by far exceed the thresholds recommended by the ASGE for a new technology to leave diminutive polyps in place or to resect and discard them without pathological assessment. Based on our results, further studies on the prediction of polyp histology with digital chromoendoscopy should be performed with the inclusion of less experienced physicians and polyps of the whole colon.