CLE in the esophagus
Barrett's esophagus (BE) is a disease where clinical trials have most established the yield of CLE. The diagnosis of BE is based on columnar cells and dark ‘goblet’ cells (Fig. 3), suggestive of the presence of intestinal metaplasia. In the presence of dysplasia, CLE shows irregular capillaries and dark irregular epithelial structures.
Figure 3. Confocal laser endomicroscopy image of Barrett's esophagus. Dark spots (arrow) indicate the presence of goblet cells.
Download figure to PowerPoint
Surveillance and treatment are the two major roles of endoscopy in BE. In the surveillance of BE, four-quadrant biopsies are routinely taken according to the Seattle protocol, but this is time-consuming and may cause fibrosis and interfere with future endoscopic treatment such as EMR or ESD. A multicenter randomized controlled trial showed that sensitivity of the diagnosis of high-grade dysplasia or early cancer increased by adding NBI and pCLE: 34.2% by high-definition white-light endoscopy alone (HD-WLE), 45.0% by HD-WLE or NBI and 75.8% by HD-WLE, NBI or pCLE. As the current ASGE guidelines recommend radiofrequency ablation (RFA) treatment for high-grade dysplasia, accurate diagnosis of high-grade dysplasia by CLE enables endoscopic treatment on the spot without waiting for biopsy results.
Two major roles of CLE in the management of BE; that is, detection and decision on the spot, have been confirmed as shown above. Diagnosis of residual lesions after endoscopic treatment of BE is another potential role of CLE, but a multicenter study comparing HD-WLE and HD-WLE with pCLE failed to demonstrate significant differences for the detection of a residual lesion after endoscopic treatment, probably as a result of an underpowered sample size and the unexpectedly higher rate of residual lesions. Among 119 patients who had undergone ablation of BE, decision to re-treat was made based on WLE vs WLE + pCLE. The percentage of patients who had optimal treatment (not overtreated or not undertreated) was only 26% for WLE and 27% for WLE + pCLE. This result again illustrates the difficulty in diagnosing neoplastic change in the background of inflammation (BE per se and ablation). Given the high sensitivity in the diagnosis of a residual lesion after EMR of colon polyps, further investigation with a more sophisticated study design is warranted to evaluate the role of CLE after endoscopic treatment of BE.
Other possibilities of CLE in non-neoplastic conditions are reported in non-erosive reflux disease (NERD) or eosinophilic esophagitis, but these are only preliminary results and the clinical impact of CLE in these conditions are to be determined in a large-scale prospective trial. Targeted biopsies using CLE findings can increase the pathological diagnostic yield in these diseases with diffuse but possibly patchy distribution.
CLE in the colon
There are two entities where CLE has shown its utility in the colon similar to the esophagus: neoplasm and inflammation. In the differential diagnosis of neoplastic vs hyperplastic polyps in the colon, sensitivity of CLE was 91% compared with 77% in virtual chromoendoscopy, whereas specificity was comparable (76% vs 71%). In addition, in the diagnosis of small (<10 mm) colon polyps, pCLE had higher sensitivity than NBI (86% vs 64%, P = 0.008), although specificity of pCLE was lower (78% vs 92%, P = 0.027). From the results of this study, CLE can be used to discard the biopsy specimens of small polyps when CLE or other endoscopic findings are diagnostic for hyperplastic polyps. This ‘resect and discard’ strategy ultimately reduces medical costs, rather than sending the entire specimen to pathological diagnosis, without increasing the risk of colorectal cancer. In addition to the primary diagnosis of colon polyps, the usefulness of CLE was reported in an evaluation of residual tumor after endoscopic treatment of colon polyps. Sensitivity and specificity of CLE were 97% and 77%, compared with 72% and 77% by endoscopy alone, respectively. This study showed that additional endoscopic intervention can be carried out in a single session without waiting for biopsy results when a residual tumor is diagnosed based on CLE. The clinical impact of this application is large because additional biopsies after endoscopic treatment cause further fibrosis at the site of prior endoscopic treatment and make subsequent endoscopic treatment difficult.
The other area where CLE has been investigated is inflammatory disease. The diagnosis of dysplasia or colitic cancer in the background of chronic inflammatory bowel disease is difficult. An initial report demonstrated a high yield of CLE in the diagnosis of intraepithelial neoplasm in ulcerative colitis. However, a subsequent study failed to show an additional yield of CLE. It is likely that these study results can be affected by the degree of inflammation because even a conventional pathological diagnosis is sometimes difficult in the background of severe inflammation. Therefore, a large-scale study with more detailed data should be collected to draw a conclusion.
In addition to the diagnosis of neoplastic change in the background of inflammatory bowel disease, CLE was reported to be potentially useful in predicting relapse of inflammatory bowel disease. This study suggested that local barrier defects indicated by fluorescein efflux through the epithelium can be a predictive marker for relapse. This finding is one of the representatives of functional CLE, which is the advantage of an in vivo diagnostic procedure. Lymphocytic colitis, which sometimes shows patchy distribution, can be diagnosed in vivo at the time of ongoing endoscopy using CLE, and targeted biopsies from the affected area detected by CLE can increase the diagnostic yield of histopathology, compared with random biopsies.
CLE in the stomach
As a result of the low prevalence of gastric cancer in Western countries, there is less evidence in this area compared with the esophagus or colon, but increasing data have been reported, mainly from Asia, where gastric cancer is more prevalent. Gastric cancer often arises from atrophic gastritis caused by Helicobacter pylori infection. In the presence of gastritis, CLE often detects gastric intestinal metaplasia (IM) as seen in BE (Fig. 4). Goblet cells are characteristic of IM on CLE. Sensitivity and specificity of gastric IM were both >90% by the interpretation of experienced endoscopists or in combination with virtual chromoendoscopy. Kiesslich et al. reported that even H. pylori could be detected on CLE. Irregular capillaries and dark irregular, or disorganized, epithelial structures (Fig. 5) are characteristic of dysplasia or cancer, with a sensitivity of 88.9%, a specificity of 99.3% and an accuracy of 98.8% for gastric superficial cancer or high-grade dysplastic lesions. However, the clinical impact of CLE in the management of gastric dysplasia or cancer should be further investigated given the accuracy of the current diagnostic modalities such as NBI with magnification.
Figure 4. Confocal laser endomicroscopy (CLE) image of intestinal metaplasia in the stomach. Goblet cells similar to Barrett's esophagus are seen on CLE.
Download figure to PowerPoint
Figure 5. Confocal laser endomicroscopy (CLE) image of dysplasia in the stomach. CLE shows dark and irregular epithelium.
Download figure to PowerPoint
CLE in the pancreas or biliary tree
Despite the advancement of imaging modalities, the accurate diagnosis of pancreatobiliary diseases without masses is still a problem. In patients with a mass in the pancreatobiliary regions, EUS-FNA is one of the standard diagnostic procedures with high accuracy. However, in patients without masses, the current diagnostic procedure is ERCP. The diagnosis via ERC is based on fluoroscopy and, recently, endoscopy using a video cholangioscope, but the pathological diagnosis via ERCP is limited by the small specimen. Therefore, a novel device has been long awaited for the diagnosis of the indeterminate pancreatobiliary stricture.
A smaller probe (CholangioFlex; Mauna Kea) which goes through an ERCP catheter enables pCLE during ERCP. Although resolution is limited (3.5 μm) compared with those used in the esophagus or colon (1.0 μm), several studies have been conducted to establish the standard criteria for malignancy. CLE findings suggestive of malignancy were epithelial structures, white bands, thick dark bands (>40 μm) or dark clumps. In a large prospective multicenter trial of pCLE during ERCP for pancreatobiliary stricture, pCLE provided a sensitivity of 98%, a specificity of 67% and an accuracy of 81%, compared with a sensitivity of 45%, a specificity of 100% and an accuracy of 75% in pathological examination at index ERCP. Accuracy of the combination of ERCP and pCLE increased up to 90%, compared with 73% by ERCP with tissue acquisition alone (P < 0.001). Although encouraging data were reported, the criteria are somewhat obscure as a result of the lower resolution. Moreover, indeterminate biliary stricture often occurs in the presence of inflammation such as primary sclerosing cholangitis (PSC), which makes the pathological interpretation difficult. Only a small study has been reported regarding the utility of pCLE in the dominant stricture in PSC patients. In 20 dominant strictures, sensitivity and negative predictive value (NPV) were 100%, respectively, but specificity and positive predictive value (PPV) were 66.1% and 22.2%, respectively. Meanwhile, tissue sampling by ERCP showed sensitivity of 0%, specificity of 94.4%, PPV of 0% and NPV of 89.5%. Therefore, more data are necessary to confirm the utility of CLE in the diagnosis of biliary stricture, and improvement of the device is also essential to increase the diagnostic yield.
Another topic of CLE in this field is the diagnosis of pancreatic cysts. Cyst fluid analysis in combination with cytology is routinely done to differentiate mucinous vs non-mucinous cysts, but these diagnostic procedures are far from satisfactory. Cyst fluid carcinoembryonic antigen (CEA) cannot differentiate malignant from benign lesions and cytology is not sensitive due to scant cellularity. Another problem is the heterogeneity of the cyst wall within a pancreatic cyst. As seen in surgical specimens of intraductal papillary mucinous neoplasms (IPMN), even in the presence of dysplasia or cancer, most of the cyst wall is covered by non-dysplastic epithelium. Therefore, multiple sampling, either invasive or non-invasive, is essential for the accurate diagnosis of pancreatic cysts.
A newly developed small CLE, which can be inserted through a 19-gauge EUS-FNA needle, enables needle-based CLE (nCLE; AQ-Flex19; Mauna Kea) under EUS guidance. The first report of nCLE in a porcine model showed possibilities of this in vivo histology procedure in various abdominal organs such as pancreas, lymph node, spleen and liver. Subsequent human clinical trials were focused on the evaluation of the pancreas. The first pilot study which included 16 cysts and two masses showed the feasibility of nCLE through a 19-gauge EUS-FNA needle.
For further evaluation of nCLE for pancreatic cysts, a multinational prospective study (INSPECT trial) was conducted. In the first step, the investigators tried to define nCLE findings in correlation with pathological findings. ‘Papillary projection’ (Fig. 6) or ‘dark ring’ on nCLE was thought to be correlated with villous structures in IPMN. In the second step, the diagnostic yield based on these findings was evaluated. They found that epithelial structures such as papillary projections were 100% specific to mucinous cysts, but its sensitivity was only 57.9%. To increase the diagnostic yield further, we conducted a single-center study, called DETECT study, at University of California, Irvine Medical Center, which evaluated ‘dual through-the-needle imaging’. In this study, prior to nCLE, we carried out cystoscopy through the needle, using a small fiberoptic probe (Spyglass; Boston Scientific, Natick, MA, USA). The procedure of this dual through-the-needle imaging was feasible. In this study, in vivo imaging from multiple areas within a cyst was carried out via fan-like movement of the needle and probe. As a result, both specificity of cystoscopy and nCLE to diagnose mucinous cysts was 100% and sensitivity was 88% with cystoscopy and 75% with nCLE.
Figure 6. Confocal laser endomicroscopy image of a pancreatic cyst. Papillary projection is characteristic of intraductal mucinous papillary neoplasms.
Download figure to PowerPoint
Although nCLE provides novel in vivo imaging inside pancreatic cysts, some hurdles are still to be overcome to bring this into clinical practice. Image quality, such as resolution, is not high enough because of the small size of the probe. In addition, malignant or dysplastic pancreatic cysts cannot be diagnosed based on the current criteria. Lack of pathological diagnosis as the gold standard is also a problem with those studies evaluating the diagnostic yield of pancreatic cysts. Another clinical implication of nCLE for pancreatic cysts is risk stratification of IPMN, which will be possible if subtypes of IPMN (gastric, intestinal, pancreatobiliary or oncocytic) can be diagnosed on nCLE as reported in phenotypes of gastric cancer.[48, 49] Therefore, large-scale prospective studies should be done in patients who will undergo surgical resection to confirm the standard criteria of nCLE for pancreatic cysts. However, differentiation of subtypes or malignant potential of IPMN is impossible so far. The current literature of nCLE for pancreatic cystic neoplasms (PCN) showed only its utility to differentiate mucinous vs non-mucinous cysts, which does not surpass the conventional EUS-FNA procedure by much. As a result of the limited impact by nCLE on patient management, a recent study by Napoleon et al. focused on the diagnosis of serous cystadenoma, a benign pancreatic cyst, rather than on the diagnosis of malignant or potentially malignant pancreatic cysts. Thus, it is still to be elucidated whether nCLE can change the current diagnostic algorithm of PCN.