Macroscopic characterization

Macroscopic characterization of a lesion provides information about its endoscopic appearance, its malignant potential, and the skill level required for its resection. There are 2 macroscopic types: (1) type 0, the superficial lesions; and (2) types 1 to 5, the advanced cancers ( Box 2 ). Type 0 includes the colorectal polyps, herein called lesions because the term polyp connotes abnormal growth of tissue projecting from mucous membrane. Type 0 can be classified into protruding or polypoid, and nonprotruding or nonpolypoid. Polypoid lesions are further subtyped into pedunculated (0–Ip) and sessile (0–Is). Nonpolypoid lesions can be subtyped into the superficial elevated (0–IIa), flat (0–IIb), and depressed (0–IIc) types. In the colon and rectum, the superficial elevated types are commonly classified together with flat lesions because of their shapes being tabletop flat, and because the true colorectal flat (0–IIb) lesions are rare. Flat colorectal neoplasms equal to or larger than 10 mm are called lateral spreading tumors (LSTs). LSTs are differentiated into granular LST (LST-G) and nongranular LST (LST-NG).

Macroscopic characterization provides superior information about the risk of early cancer compared with size of the lesion. NP-CRNs have higher risk of early cancer than the polypoid-CRNs, irrespective of size. The observation that the larger the polypoid neoplasms the higher the risk of deeper submucosal invasion does not apply for NP-CRNs. In NP-CRNs, the difference in the size of lesions with superficial and deep submucosal invasion is not pronounced. LST-Gs with even-sized nodules have a low risk (<2%) of submucosal invasion, irrespective of size, whereas LST-Gs with mixed-sized nodules have a higher risk of submucosal invasion (7.1% for lesions <20 mm and 38% for lesions >30 mm), with the point of invasion usually located under the largest nodule. The LST-NGs, particularly those that have a thinner center (LST-NG with pseudodepression) also have a high risk of submucosal invasion: 12.5% for lesions less than 20 mm to 83.3% for lesions more than 30 mm. In such lesions the points of invasion are typically multifocal. Although depressed lesions are uncommon (1%–6% of NP-CRNs), their risks of submucosal invasion are the highest: the overall risks are reported to be 27% to 35.9% compared with 0.7% to 2.4% in the flat lesions.

In addition to the macroscopic classification, there are several other important clues to submucosal invasion in NP-CRN, including fold convergence, expansion appearance, and ulceration ( Box 3 ).

Microscopic characterization (optical diagnosis)

Optical diagnosis is a real-time, in vivo microscopic characterization of a lesion at the tissue level during endoscopy that is based on objective criteria and is reproducible. Objectivity based on standardized and reproducible criteria is important to bring the validity of the optical diagnosis to the same level as pathologic diagnosis. Both electronic and dye-based image-enhanced endoscopy techniques are conceptually similar to the application of the various stains in pathology and can delineate the neoplastic borders. They can enhance optical diagnosis and provide information on the depth of invasion.

In practice, we routinely use image-enhanced endoscopy to make an optical diagnosis. Following a detailed high-definition white light inspection of the lesion, we perform image-enhanced endoscopy using narrow band imaging (NBI) to systematically characterize the lesion color and mucosal surface and microvascular patterns. We apply the Narrow Band Imaging International Colorectal Endoscopic (NICE) classification to make an optical diagnosis (type 1, hyperplastic; type 2, adenomatous or superficial invasive carcinoma [<1000 μm]; or type 3, deep submucosal invasive cancer) ( Fig. 1 ). Lesions that are a good match with the criteria are considered to have high confidence. We spray diluted indigo carmine (0.2%) to confirm the diagnosis of a depressed lesion, which is characterized by dye pooling into the depressed area. The indigo carmine solution can also help delineate the borders of the lesion before and during resection. We do not routinely apply the Kudo classification, although it can be useful to confirm the optical diagnosis. We use the clinical classification: hyperplastic, adenoma, and deeply invasive, as proposed by Matsuda and colleagues.

NICE classification.

Optical diagnosis is vital to the management of complex colorectal lesions. All of the lesions shown in this article were optically diagnosed before resection. Deep submucosal invasive lesions are generally not resected by endoscopy because their resections are associated with increased risks of complications, and they have a high risk of metastasis.

The nonlifting sign

The nonlifting sign, which is assessed during the submucosal injection, provides further opportunity to assess for the possibility of submucosal invasion. The nonlifting sign has shown a predictive positive value of 80%, negative predictive value of 96%, and an overall accuracy of 94.8%. It can be falsely positive because of submucosal fibrosis (secondary to prior biopsies, incomplete resection, or cauterization), desmoplastic reaction under the lesion, prior India ink tattooing, or presence of ulceration.

Personnel

Personnel who are proficient in the use of the range of equipment and the intricacies of the technical procedure details, and who remain calm to manage complications are critical assets in performing advanced resections.

Equipment

We use high-definition adult colonoscopes with a water jet channel. We often use the therapeutic upper endoscopes when the lesion is located in the left colon. Pediatric colonoscopes featuring a smaller diameter may be useful to retroflex in the right colon to manage lesions extending over folds. Distal attachments can be helpful to improve visualization behind folds and stabilize the endoscope tip. We use CO ₂ insufflation because it has been shown to improve patient comfort during and after the procedure, and it decreases the risk of tension pneumoperitoneum if a perforation were to occur. The accessories that we routinely use are shown in Box 5 .

Choice of sedation

The ideal sedation is conscious (moderate sedation). Deep sedation with propofol may result in labored breathing, which makes the colon move significantly and interferes with the EMR. Furthermore, patient repositioning, which is often required in order to optimize the procedure, may be difficult in obese patients.

Injection

Injection techniques

We use a dynamic submucosal injection technique ( Fig. 2 ) that is designed to create a generous bulge under the lesion. In this technique, after puncturing the mucosa and inserting the needle into the submucosa, a small amount of saline is injected to confirm insertion into the submucosal layer, followed by rapid large-volume injection. Unlike the variceal injection technique, in which the needle is kept stationary during the injection (static injection technique), dynamic submucosal injection involves moving the needle within the submucosa by slowly deflecting the tip of the endoscope or gently pulling the catheter back to create the desired mound of submucosal fluid cushion with injection. At the same time, air is suctioned out of the lumen. In essence, the submucosal bulge is sculpted.

The dynamic injection technique. The dotted arrow shows the planned path of the needle that was to be followed during injection ( A ). Diluted indigo carmine solution in saline was injected rapidly while lifting the tip of the endoscope and suctioning the lumen ( B ). Further injection ( C ). After 8 to 10 mL of injection, there was a large submucosal cushion ( D ). A stiff 10-mm snare was used to capture the lesion and surrounding normal tissue ( E ). After resection ( F ).

It is important to appreciate the difference between the dynamic and static injection techniques. In the static technique, after inserting the needle into the submucosa, the needle is kept stationary until an adequate amount of fluid is slowly infused. In this technique the lumen is maintained insufflated to allow visualization of the lesion during injection. A common disappointment after the static injection technique has been inadequate thickness of the submucosal cushion, which is unsatisfactory for successful resection, because the injected saline spreads thinly throughout the area rather than collecting as a mound under the lesion.

The inject-and-cut EMR

The inject-and-cut EMR is a simple technique that is widely used for removal of large flat or sessile lesions. Lesions larger than 20 mm cannot be removed in a single piece (en bloc resection), and in such cases piecemeal resection is required.

We use a stiff snare for EMR after dynamic submucosal injection. If an LST-G mixed lesion with a large nodule is encountered, it is important to resect the large nodule in a single piece and submit it in a separate bottle of formalin after fixing it because of the likelihood of advanced features or invasive carcinoma underneath such a lesion. Our preference is to resect LST-NGs en bloc if they are less than 20 mm.

After capturing the lesion with a snare, the lumen is insufflated with air to stretch the wall, and the snare is lifted up while the snare is slightly loosened to release any entrapped muscularis propria. The lesion is then strangled and transected using blended current: 35 W in the left and 40 W in the right side of the colon. In large lesions requiring piecemeal resection, successive resections should be performed using the free resection margin from prior resection as an anchor point. Additional submucosal injections can be performed through the already-visible submucosa to create a cushion if necessary for safe resection. It is important to keep the snare parallel to the wall during piecemeal resection to avoid entrapping muscle when a snare is deployed at an angle.

Prevention of recurrence using argon plasma coagulation

A major drawback of piecemeal resection is the risk of leaving residual tissue at the end of resection, which can ensure recurrence in 100% of cases. We resect all visible neoplastic tissue. Application of argon plasma coagulation (APC) to the borders and bridges of the resected area may burn microscopic residual tissue to reduce the risk of recurrence.

EMR-C

EMR-C with a prepositioned snare in the colon has been regarded as unsafe because of the theoretic risk of entrapment of thin colon wall during suction of the lesion into the cap resulting in perforation, although this technique is widely used in the resection of lesions of the rectum where the wall is thick.

Underwater EMR

Binmoeller and colleagues observed that the muscularis propria of the colon and rectum that are filled with water stay circular during endoscopic ultrasonography. Thus, he proposed to use water immersion to eliminate the need for submucosal injection during mucosal resection. To perform this technique once the lesion is identified, the air is suctioned and 500 mL to 1 L of water are infused until complete filling of the lumen is achieved. Once underwater, the margins of the lesion are marked using APC. The resection is performed using a 15-mm duckbill snare with blended current. To maximize tissue capture a torque-and-crimp technique is used. APC can be delivered underwater as well. Note that the lesions are resected in piecemeal fashion.