Summary of Evidence

Confocal laser endomicroscopy has been proposed for use in conjunction with colonoscopy to evaluate colorectal lesions, for surveillance of Barrett’s esophagus, and for evaluation of other suspicious lesions (e.g. lung, bladder, gastric, celiac disease). CLE is unique in that it is designed to characterize the cellular structure of lesions immediately. Confocal laser endomicroscopy can thus potentially be used to make a diagnosis of polyp histology, particularly in association with screening or surveillance colonoscopy, which could allow for small hyperplastic lesions to be overlooked rather than removed and sent for histologic evaluation. Using CLE would reduce risks associated with biopsy and reduce the number of biopsies and histologic evaluations. Limitations of CLE systems include a limited viewing area and depth of view. Another issue is the standardization of systems for classifying lesions view with these devices. The available evidence is insufficient to determine whether there is an improvement in health outcomes with the use of CLE. 

Rationale

Colorectal/gastrointestinal lesions

For colorectal lesions, several systematic reviews have compared the diagnostic accuracy of confocal laser microscopy (CLE) with a reference standard. Su et al (2013) reviewed studies on the efficacy of CLE for discriminating colorectal neoplasms from non-neoplasms. To be included in the review, studies had to use histologic biopsy as the reference standard, and the pathologist and endoscopist had to be blinded to each other’s findings. Selected studies also had to use a standardized CLE classification system. Patients had to be at increased risk of colorectal cancer (CC) due to personal or family history, have previously identified polyps, and/or have inflammatory bowel disease. Two reviewers independently assessed the quality of individual studies using the modified Quality Assessment of Diagnostic Accuracy Studies tool, and studies considered at high risk of bias were excluded from further consideration. Fifteen studies (n = 719 adults) were selected. All were single-center trials, and 2 were available only as abstracts. In all studies, suspicious lesions were first identified by conventional white-light endoscopy with or without chromoendoscopy and then further examined by CLE. Meta-analysis of the 15 studies found an overall sensitivity for CLE of 94% (95% confidence interval [CI], 88% to 97%) and a specificity of 95% (95% CI, 89% to 97%) compared with histology. Six studies included patients at increased risk of CC who were undergoing surveillance endoscopy; 5 studies included patients with colorectal polyps and 4 studies included patients with inflammatory bowel disease. In a predefined subgroup analysis by indication for screening, the pooled sensitivity and specificity for surveillance studies were 94% (95% CI, 90% to 97%) and 98% (95% CI, 97% to 99%), respectively. For patients presenting with colorectal polyps, the pooled sensitivity of CLE was 91% (95% CI, 87% to 94%) and the specificity was 85% (95% CI, 78% to 90%). For patients with inflammatory bowel disease, the pooled sensitivity was 83% (95% CI, 70% to 92%) and the specificity was 90% (95% CI, 87% to 93%). In other predefined subgroup analyses, the summary sensitivity and specificity were significantly higher (p<.001) in studies of endoscopy-based CLE (97% and 99%, respectively) than in studies of probe-based CLE (87% and 82%, respectively). In addition, the summary sensitivity and specificity were significantly higher (p<.01) with real-time CLE in which the macroscopic endoscopy findings were known (96% and 97%, respectively) than in blinded CLE in which recorded confocal images were subsequently analyzed without knowledge of macroscopic endoscopy findings (85% and 82%, respectively). 

A systematic review by Dong et al (2013) included studies that compared the diagnostic accuracy of CLE with conventional endoscopy. Reviewers did not explicitly state that the reference standard was a histologic biopsy, but this was the implied reference standard. Six studies were included in a meta-analysis. All were prospective, and at least 5 included blinded interpretation of CLE findings (in 1 study, it was unclear whether the interpretation was blinded). In a pooled analysis of data from all 6 studies, the sensitivity was 81% (95% CI, 77% to 85%) and the specificity was 88% (95% CI, 85% to 90%). Reviewers also conducted a subgroup analysis by type of CLE used. When findings from the 2 studies on endoscopy-based CLE were pooled, the sensitivity was 82% (95% CI, 69% to 91%) and the specificity was 94% (95% CI, 91% to 96%). Two studies may not have been sufficient to obtain a reliable estimate of diagnostic accuracy. When findings from the 4 studies on probe-based endoscopy were pooled, the sensitivity was 81% (95% CI, 76% to 85%) and the specificity was 75% (95% CI, 69% to 81%).

A meta-analysis by Wanders et al (2013) searched for studies that reported on the diagnostic accuracy of several new technologies used to differentiate between colorectal neoplasms and non-neoplasms. To be selected, studies had to use the technology to differentiate between non-neoplastic and neoplastic lesions and to use histopathology as the reference standard. Blinding was not an inclusion criterion. Eleven eligible studies identified included an analysis of CLE. Meta-analysis yielded an estimated sensitivity of 93.3% (95% CI, 88.4% to 96.2%) and a specificity of 89.9% (95% CI, 81.8% to 94.6%). Meta-analysis limited to the 5 studies that used endoscopy-based CLE found a sensitivity of 94.8% (95% CI, 90.6% to 98.92%) and a specificity of 94.4% (95% CI, 90.7% to 99.2%). When findings of the 6 probe-based CLE studies were pooled, the sensitivity was 91.5% (95% CI, 86.0% to 97.0%) and specificity was 80.9% (95% CI, 69.4% to 92.4%).

A study by Xie et al (2011) in China included 116 consecutive patients who had polyps found during CLE (1 patient was excluded from the analysis). All patients had an indication for colonoscopy (19 were undergoing surveillance after polypectomy, 2 had a family history of CC, 3 had inflammatory bowel disease, 91 were seeking a diagnosis). All patients first underwent white-light colonoscopy. Endoscopy-based CLE was used on the first polyp identified during withdrawal of the endoscope (ie, 1 polyp per patient was analyzed). Real-time diagnosis of the polyp was performed based on criteria used at the study center (adapted from the Mainz classification system). The polyps were biopsied or removed, and the histopathologic diagnosis was determined. Real-time CLE diagnosis correctly identified 109 (95%) of 115 adenomas or hyperplastic polyps. Four adenomas were misdiagnosed by CLE as hyperplastic polyps (2 were tubulous adenomas, 2 were tubulovillous adenomas), and 2 hyperplastic polyps were misdiagnosed as adenomas. The overall sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of CLE diagnosis were 93.9% (95% CI, 85.4% to 97.6%), 95.9% (95% CI, 86.2% to 98.9%), 96.9% (95% CI, 89% to 99%), and 94.8% (95% CI, 89.1% to 97.6%), respectively. For polyps less than 10 mm in size, CLE diagnosis had a sensitivity of 90.3% and a specificity of 95.7%; for polyps 10 mm or larger, sensitivity was 97.1% and specificity was 100%.

Buchner et al (2010) published findings on 75 patients who had a total of 119 polyps. Patients were eligible for participation if they were undergoing surveillance or screening colonoscopy or undergoing evaluation of known or suspected polyps identified by other imaging modalities or endoscopic resection of larger flat colorectal neoplasia. White-light colonoscopy was used as the primary screening method. When a suspicious lesion was identified, it was evaluated by virtual chromoendoscopy and a probe-based CLE system. After the imaging techniques, the appropriate intervention (i.e., polypectomy, biopsy, endoscopic mucosal resection) was performed, and all resected specimens underwent histopathologic analysis by a pathologist blinded to CLE information. Confocal images of the 119 polyps were evaluated after all procedures were completed; the evaluator was blinded to the histology diagnosis and the endoscopic appearance of the lesion. Diagnosis of confocal images used modified Mainz criteria; polyps were classified as benign or neoplastic. According to histopathologic analysis, there were 38 hyperplastic polyps and 81 neoplastic lesions. The use of CLE correctly identified 74 of 81 neoplastic polyps (sensitivity, 91%; 95% CI, 83% to 96%). In addition, CLE correctly identified 29 of 38 hyperplastic polyps (specificity, 76%; 95% CI, 60% to 89%). In contrast, virtual chromoendoscopy correctly identified 62 neoplastic polyps (sensitivity, 77%; 95% CI, 66% to 85%) and 27 hyperplastic polyps (specificity, 71%; 95% CI, 54% to 85%).

Another study from the same academic medical center as Buchner et al (2010) was published by Shadid et al (2012). The study compared 2 methods of analyzing CLE images: real-time diagnosis and blinded review of video images after endoscopy (known as "offline" diagnosis). The study included 74 patients with 154 colorectal lesions. Eligibility criteria were similar to the Buchner et al (2010) study (previously discussed)¾selected patients were undergoing surveillance or screening colonoscopy. Patients had a white-light colonoscopy, and identified polyps were also evaluated with virtual chromoendoscopy and probe-based CLE. At the examination, an endoscopist made a real-time diagnosis based on CLE images. Based on that diagnosis, the patient underwent polypectomy, biopsy, or endoscopic mucosal resection, and histopathologic analysis was done on the specimens. Images from CLE were deidentified and reviewed offline by the same endoscopist at least 1 month later. In the second review, the endoscopist was blinded to the endoscopic and histopathologic diagnosis. Of the 154 polyps, 74 were found by histopathologic analysis to be non-neoplastic, and 80 were neoplastic (63 tubular adenomas, 12 tubulovillous adenomas, 3 mixed hyperplastic-adenoma polyps, 2 adenocarcinomas). Overall, there was no statistically significant difference in the diagnostic accuracy between real-time CLE diagnosis and blinded offline CLE diagnosis (ie, CIs overlapped). The sensitivity, specificity, PPV, and NPV for real-time CLE diagnosis were 81%, 76%, 87%, and 79%, respectively. For offline diagnosis, these values were 88%, 77%, 81%, and 85%, respectively. For larger polyps, there was a nonsignificant trend in favor of better diagnostic accuracy with real-time compared with offline CLE. However, in the subgroup of 107 smaller polyps (<10 mm in size), the accuracy of real-time CLE was significantly less than offline CLE. For smaller polyps, the sensitivity, specificity, PPV, and NPV of real-time CLE were 71%, 83%, 78%, and 78%, respectively; for offline CLE, they were 86%, 78%, 76%, and 87%, respectively

Ypsilantis et al (2015) published a systematic review that included retrospective and prospective studies reporting the diagnostic accuracy of CLE for the detection of residual disease after endoscopic mucosal resection of gastrointestinal lesions. After examining full-text articles, 3 studies (1 RCT, 2 prospective, nonrandomized comparative studies) met the eligibility criteria. Studies included patients with BE, gastric neoplasia, and colorectal neoplasia. There was significant heterogeneity among studies. In a per-lesion meta-analysis, pooled sensitivity of CLE for detecting neoplasia was 91% (95% CI, 83% to 96%) and pooled specificity was 69% (95% CI, 61% to 76%). Based on the small number of studies and heterogeneity among studies, reviewers concluded that the evidence on the utility of CLE in assessing the adequacy of endoscopic mucosal resection was weak.

Barrett’s Esophagus

DeMeester et al (2022) published a meta-analysis of prospective studies and RCTs evaluating the diagnostic accuracy of probe-based CLE as an adjunct to random four-quadrant biopsies in patients with BE. A total of 9 studies (N=688) were included. Results for CLE were reported in comparison to histopathological results (highest grade diagnosis detected by standard white light endoscopy targeted or random four-quadrant biopsies or from resection histopathological analysis) as the diagnostic reference. The following results were obtained for CLE for the diagnosis of high-grade dysplasia (HGD) or esophageal adenocarcinoma: pooled sensitivity, 96% (95% CI, 65% to 100%); pooled specificity, 93% (95% CI, 71% to 99%); pooled PPV, 69% (95% CI, 49% to 84%); pooled NPV, 98% (95% CI, 93% to 100%). The relative increase in neoplasia detection using CLE compared with the Seattle protocol randomized biopsies was 243% (95% CI, 122% to 482%); the absolute increase was 5% (95% CI, 1% to 9%). Dysplasia prevalence with Seattle protocol randomized biopsies was 4% (95% CI, 1% to 11%), and with CLE was 9% (95% CI, 2% to 29%). Investigators concluded that the addition of probed-based CLE-guided biopsies provides a significantly higher diagnostic yield for dysplasia and cancer and reduces sampling error compared with random four-quadrant biopsies alone.

Vithayathil et al (2022) conducted a randomized crossover trial of standard high-resolution white-light Seattle protocol endoscopy or autofluorescence imaging-guided probe-based CLE in patients referred for surveillance of nondysplastic BE or flat dysplasia at 2 high-volume tertiary centers in the United Kingdom. A total of 154 patients were recruited, of whom 8 were excluded based on presence of clear macroscopic lesions consistent with BE-related neoplasia upon first endoscopy. An additional patient was excluded due to a protocol breach (use of chromoendoscopy) and 11 patients withdrew consent. A total of 134 patients completed both arms of the study, with crossover occurring after a 6- to 12-week interval. Endoscopists were blinded to the endoscopy and histology results of the pretrial endoscopy and other study arm. In the per-lesion analysis, optical diagnosis by CLE had a sensitivity and specificity for high-grade dysplasia (HGD)/intramucosal cancer (IMC) of 69.2% and 73.2%, respectively. In the per-patient analysis, there was no difference in the sensitivity of CLE for dysplasia compared with Seattle protocol for HGD/IMC (76.5% for both; p=1.00) or all grades of dysplasia (74.3% vs. 80.0%, respectively; p=.48). The specificity of CLE was 60.7% for HGD and 66.7% for all grades of dysplasia. Use of a 3-biomarker panel consisting of 1 or more of optical dysplasia on CLE, aberrant p53 on immunohistochemistry, and/or aneuploidy on flow cytometry was associated with a per-patient sensitivity and specificity of 94.1% and 49.6% for HGD and 91.4% and 56.6% for all grades of dysplasia, respectively. The authors concluded that CLE has similar diagnostic accuracy for dysplasia compared with standard Seattle protocol endoscopy. In addition, the use of molecular biomarkers can further improve diagnostic accuracy. Several study limitations were noted: (1) it cannot be excluded that prior biopsy sites may have appeared as irregularities on second endoscopy due to the crossover study design, (2) sensitivity for detecting dysplasia was inconsistent across endoscopists, and (3) results may not be generalizable to general practice centers.

Canto et al (2014) reported on a single-blind, multicenter trial conducted at academic centers with experienced endoscopists. It included consecutive patients undergoing endoscopy for routine BE surveillance or for suspected or known neoplasia. Patients were randomized to high-definition white-light endoscopy with random biopsy (n=98) or white-light endoscopy with endoscopy-based CLE and targeted biopsy (n=94). In the white-light endoscopy-only group, 4-quadrant random biopsies were taken every 1 to 2 cm over the entire length of the BE for patients undergoing surveillance and every 1 cm for patients with suspected neoplasia. In the CLE group, biopsy specimens were obtained only when there was CLE evidence of neoplasia. Final pathologic diagnosis was the reference standard. A per-patient analysis of diagnostic accuracy for diagnosing BE-related neoplasia found a sensitivity of 40% with white-light endoscopy only and 95% with white-light endoscopy plus CLE. Specificity was 98% with white-light endoscopy only and 92% with white-light endoscopy plus CLE. When the analysis was done on a per-biopsy specimen basis and when CLE was added, sensitivity was substantially higher, and specificity was slightly lower. The median number of biopsies per patient was significantly higher in the white-light endoscopy group (4 biopsies) compared with the CLE group (2 biopsies; p<.001). The investigators analyzed the number of cases in which CLE resulted in a different diagnosis. Thirty-two (34%) of 94 patients in the white-light plus CLE group had a correct change in dysplasia grade after CLE compared with initial endoscopic findings. Six (19%) of the 32 patients had lesions, and the remaining 26 did not. In 21 of the 26 patients without lesions, CLE changed the plan from biopsy to no biopsy. The remaining 62 (65%) of 94 patients in the white-light endoscopy plus CLE group had concordant diagnoses with both techniques. Because the trial was conducted at academic centers and used endoscopy-based CLE, findings may not be generalizable to other clinical settings or to probe-based CLE.

Richardson et al (2019) conducted a prospective study at 8 centers in the United States to compare probe-based CLE to conventional histology using the Seattle Protocol (random 4-quadrant biopsy) to identify intestinal metaplasia among 172 patients undergoing screening or surveillance endoscopy for BE. Endoscopists recruited for the study were early users of CLE with less than 2 years of experience and no formal pathology training. All patients underwent a standardized endoscopy with white light and narrow band imaging evaluation, identification of landmarks, and recording of columnar lined esophagus visualized according to the Prague classification. Patients then received fluorescein followed by optical biopsy; images were interpreted both in real time and immediately following the procedure. After CLE images were acquired, esophageal biopsies were taken via the Seattle Protocol. Endoscopists were able to identify intestinal metaplasia among 99 patients (57.6%) using CLE compared to 46 patients (27%) using the Seattle Protocol (p<.0001). Dysplasia was identified in 6 patients using CLE compared to 2 patients using the Seattle Protocol (both of which were also identified via CLE). Confocal laser endomicroscopy also identified significantly more patients with intestinal metaplasia compared to the Seattle Protocol among those with visible columnar lined esophagus (75 vs. 31 patients, respectively; p<.0001), but not among those without columnar lined esophagus (24 vs. 15 patients; p=.067). Identification of intestinal metaplasia was not found to be significantly different when comparing CLE to expert review.

Guidelines

In 2022, the American College of Gastroenterology (AGA) published a clinical practice update expert review on new technology for surveillance and screening in Barrett’s Esophagus. The article makes the following best practice advice statement relevant to screening and surveillance for BE:

• "Screening and surveillance endoscopic examination should be performed using high-definition white light endoscopy and virtual chromoendoscopy, with endoscopists spending adequate time inspecting the Barrett’s segment."

None of the best practice advice statements mentioned CLE. While the article did summarize data in support of innovative screening technologies such as CLE, the panelists noted that: "the use of these techniques was not required for a high-quality exam and the data to date did not support its routine use." However, the panelists also noted that "these technologies were promising and carried potential benefits in select cases and currently might be best utilized in expert centers."

In 2019, The American Society for Gastrointestinal Endoscopy (ASGE) published a guideline on screening and surveillance of Barrett esophagus (BE) which recommends against routine use of confocal laser endomicroscopy compared with white-light endoscopy with Seattle protocol biopsy sampling in patients with BE undergoing surveillance. An older guideline from the Society (2012) on the role of endoscopy in BE and other premalignant conditions of the esophagus stated the following: “Adjuncts to white-light endoscopy used to improve the sensitivity for the detection of BE and dysplastic BE include chromoendoscopy, electrical enhanced imaging, magnification, and confocal endoscopy.

In 2006, with reaffirmation in 2011, ASGE published guidelines on the role of endoscopy in the surveillance of premalignant conditions of the upper gastrointestinal (GI) tract. Regarding the use of confocal endoscopy as an adjunct to white-light endoscopy, the guidelines stated that this technique is “still in development.”

In 2015, ASGE published guidelines on the role of endoscopy in benign pancreatic disease that stated ‘confocal endomicroscopy is an emerging technology that may prove useful for the evaluation of indeterminate pancreatic strictures." Similarly, in the ASGE's 2016 guidelines on the role of endoscopy in the diagnosis and treatment of cystic pancreatic neoplasms, they acknowledged that CLE was an emerging technique for pancreatic lesion evaluation, but made no formal recommendations regarding its use.

Reference List

1. Buchner AM, Shahid MW, Heckman MG, et al. Comparison of probe-based confocal laser endomicroscopy with virtual chromoendoscopy for classification of colon polyps. Gastroenterology. Mar 2010; 138(3): 834-842.
2. Canto MI, Anandasabapathy S, Brugge W, et al. In vivo endomicroscopy improves detection of Barrett's esophagus-related neoplasia: a multicenter international randomized controlled trial (with video). Gastrointest Endosc. Feb 2014; 79(2): 211-221.
3. Chandrasekhara V, Chathadi KV, Acosta RD, et al. The role of endoscopy in benign pancreatic disease. Gastrointest Endosc. Aug 2015; 82(2): 203-214.
4. DeMeester S, Wang K, Ayub K, et al. High-definition probe-based confocal laser endomicroscopy review and meta-analysis for neoplasia detection in Barrett's esophagus. Techniques and Innovations in Gastrointestinal Endoscopy. 2022;24(4):340-350.
5. Dong YY, Li YQ, Yu YB, et al. Meta-analysis of confocal laser endomicroscopy for the detection of colorectal neoplasia. Colorectal Dis. Sep 2013; 15(9): e488-495.
6. Evans JA, Early DS, Fukami N, et al. The role of endoscopy in Barrett's esophagus and other premalignant conditions of the esophagus. Gastrointest Endosc. Dec 2012; 76(6): 1087-1094.
7. Hirota WK, Zuckerman MJ, Adler DG, et al. ASGE guideline: the role of endoscopy in the surveillance of premalignant conditions of the upper GI tract. Gastrointest Endosc. Apr 2006; 63(4): 570-580.
8. Muthusamy VR, Chandrasekhara V, Acosta RD, et al. The role of endoscopy in the diagnosis and treatment of cystic pancreatic neoplasms. Gastrointest Endosc. Jul 2016; 84(1): 1-9.
9. Muthusamy VR, Wani S, Gyawali CP, et al. AGA Clinical Practice Update on New Technology and Innovation for Surveillance and Screening in Barrett's Esophagus: Expert Review. Clin Gastroenterol Hepatol. Dec 2022; 20(12): 2696-2706.e1.
10. Qumseya B, Sultan S, Bain P, et al. ASGE guideline on screening and surveillance of Barrett's esophagus. Gastrointest Endosc. Sep 2019; 90(3): 335-359.e2.
11. Richardson C, Colavita P, Dunst C, et al. Real-time diagnosis of Barrett's esophagus: a prospective, multicenter study comparing confocal laser endomicroscopy with conventional histology for the identification of intestinal metaplasia in new users. Surg Endosc. May 2019; 33(5): 1585-1591.
12. Shahid MW, Buchner AM, Raimondo M, et al. Accuracy of real-time vs. blinded offline diagnosis of neoplastic colorectal polyps using probe-based confocal laser endomicroscopy: a pilot study. Endoscopy. Apr 2012; 44(4): 343-348.
13. Su P, Liu Y, Lin S, et al. Efficacy of confocal laser endomicroscopy for discriminating colorectal neoplasms from non-neoplasms: a systematic review and meta-analysis. Colorectal Dis. Jan 2013; 15(1): e1-12.
14. Vithayathil M, Modolell I, Ortiz-Fernandez-Sordo J, et al. Image-Enhanced Endoscopy and Molecular Biomarkers Vs Seattle Protocol to Diagnose Dysplasia in Barrett's Esophagus. Clin Gastroenterol Hepatol. Nov 2022; 20(11): 2514-2523.e3.
15. Wanders LK, East JE, Uitentuis SE, et al. Diagnostic performance of narrowed spectrum endoscopy, autofluorescence imaging, and confocal laser endomicroscopy for optical diagnosis of colonic polyps: a meta-analysis. Lancet Oncol. Dec 2013; 14(13): 1337-1347.
16. Xie XJ, Li CQ, Zuo XL, et al. Differentiation of colonic polyps by confocal laser endomicroscopy. Endoscopy. Feb 2011; 43(2): 87-93.
17. Ypsilantis E, Pissas D, Papagrigoriadis S, et al. Use of confocal laser endomicroscopy to assess the adequacy of endoscopic treatment of gastrointestinal neoplasia: a systematic review and meta-analysis. Surg Laparosc Endosc Percutan Tech. Feb 2015; 25(1): 1-5.