By Marilynn Larkin
NEW YORK (Reuters Health) - 14/3/2019
The accuracy of optical technologies in differentiating cancerous from benign tissue is "insufficient" for most routine clinical applications, including diagnosing pre-cancerous colorectal polyps, researchers say.
"Optical technologies have been used for over 20 years to diagnose cancer or pre-cancer in large bowel polyps during colonoscopy and they have evolved significantly," Sam Mason of Imperial College London, UK, told Reuters Health by email. "This meta-analysis is the largest of its kind, examining five optical technologies used on 33,123 polyps across 102 studies."
"The results demonstrate that the accuracy of optical technologies in diagnosing cancerous and pre-cancerous large bowel polyps during colonoscopy is generally insufficient for routine clinical use," he said. "This is most commonly because adenomas are being misdiagnosed."
"Furthermore, novel chronological analyses have identified that over the past decade, diagnostic accuracies are either plateauing or decreasing as larger and higher quality studies are published," he noted. "This is within the context of a significant publication bias, where the evidence suggests headline accuracies are overestimated through under-representation of small studies demonstrating poorer diagnostic accuracies in the medical literature."
Of the 33,123 colorectal polyps included in the meta-analysis, digital chromoendoscopy differentiated neoplasia (adenoma and adenocarcinoma) from benign polyps with a sensitivity of 92.2% and specificity of 84.0%, regardless of the technology used (narrow-band imaging, Fuji intelligent Chromo Endoscopy, iSCAN) or when looking only at diminutive polyps, according to American Journal of Gastroenterology report, online February 1.
Dye chromoendoscopy had a sensitivity of 92.7% and specificity of 86.6% for all polyp sizes.
Spectral analysis of autofluorescence had a sensitivity of 94.4% and specificity of 50.9%; however, no studies analyzed only diminutive polyps and two of three suitable for meta-analysis had a high risk of bias.
While endomicroscopy had a sensitivity of 93.6% and specificity of 92.5%, no studies assessed only diminutive polyps, and no predictions were presented with high confidence. Computer-aided diagnosis had sensitivity of 88.9% and specificity of 80.4% in a meta-analysis of three studies.
Prediction confidence and endoscopist experience alone did not significantly improve the predictive value of any technology. Digital chromoendoscopy was the only subgroup to demonstrate a negative predictive value for adenoma above 90%, making high confidence predictions of diminutive recto-sigmoid polyps.
"These findings are practice-changing," Mason said, "as they imply that continued incremental improvement to optical technologies is unlikely to meet thresholds of accuracy to make safe therapy decisions during index colonoscopy and that current clinical practice will continue to be based on histological assessment of polyps rather than optical diagnosis."
"Future precision endoscopic technologies are required which will allow clinicians to make decisions at the time of the procedure based on a real-time analysis, preventing unnecessary procedures to remove polyps and much of the histological processing of samples, and resulting in improved safety to patients and cost savings for healthcare providers," he concluded.
Dr. Mark Pochapin, Director of the Division of Gastroenterology and Hepatology at NYU Langone Health in New York City, commented, "the current technology cannot get better than it is to the point that it can substitute for histologic evaluation."
"Further, guidelines depend on both size and histology of a resected polyp to determine surveillance intervals," said Dr. Pochapin, who was not involved in the study. "This article shows that the current technologies are inadequate as a substitute for histology to allow for a recommendation of the timing and interval for the next surveillance procedure."
"A significant positive finding is that digital chromoendoscopy was able to have a high confidence of predicting diminutive recto-sigmoid polyps," he noted. "This is extremely helpful for the endoscopist because a common finding is that of multiple diminutive non-neoplastic polyps in the recto-sigmoid region of the colon."
"By evaluating these small polyps with digital chromoendoscopy and leaving them behind," he said, "the procedure time is reduced, unnecessary polypectomy procedures are avoided, (as are) unnecessary pathology samples sent for histologic review."
"Since almost all colonoscopies have this technology, this positive finding can give confidence in this practice of leaving behind hyperplastic-appearing recto-sigmoid polyps, which is endorsed by some professional societies," he said.
Although optical technologies are not sufficient for histologic assessment, he noted, they are helpful in identifying the margins of a polyp or assessing mucosal tissue more closely to see if a polyp is present.
Am J Gastroenterol 2019.