Introduction

A new wave of change in the field of colorectal neoplasms and colonoscopy is here, leading to the work in colonoscopic resection to begin anew. Nonpolypoid colorectal neoplasms (NP-CRN) are now widely recognized in Western countries (Fig. 36.1).¹ This finding requires numerous paradigm shifts in our clinical, educational, and research activities. The detection and diagnosis of NP-CRN necessitates the search for subtle nonprotruding findings during colonoscopy.² Treatment of NP-CRN requires mucosal resection technique, which is more complex than polypectomy. Because nonpolypoid neoplasms can be found throughout the gastrointestinal (GI) tract, all endoscopists must become familiar with endoscopic mucosal resection (EMR) techniques, and future generations must understand its theoretical basis, obtain the dexterity to perform it, and perfect it. Knowledge about the biology and outcomes of the management of nonpolypoid GI neoplasms is to be acquired. Within the same period of time, new technology—high definition, high magnification, and image-enhanced endoscopy³—that is useful for the detection, diagnosis, and treatment of NP-CRN has become available and is increasingly employed. The technique of submucosal dissection, which has become a standard resection technique for early gastric cancer in Japan, is also being applied for the resection of certain types of colorectal neoplasms to achieve R0 resection—curative en bloc resection with all margins being free of neoplasms (Fig. 36.2)⁴—without surgery.

Fig. 36.1 A, Nonpolypoid colorectal neoplasm (NP-CRN) (slightly elevated) in the sigmoid colon. B and C, The border and surface pattern of the lesion were examined with image-enhanced endoscopy techniques using narrow band imaging (B) and diluted indigo carmine (C). The lesion was resected using mucosal resection technique; pathology showed adenoma.

Fig. 36.2 General classification of methods of resection of colorectal lesions.

Advances in the ability to detect, diagnose, and treat all types of colorectal neoplasms are important. These advances provide hope that the benefit of colonoscopy screening to prevent colorectal cancer, one of the most common causes of cancer death worldwide, can be extended further. In the United States, it was estimated that there would be approximately 146,970 new cases of colorectal cancer in 2009 and 49,920 deaths caused by the disease. The lifetime risk of developing colorectal cancer in the U.S. population is about 6%, with 90% of cases occurring in individuals older than 50 years. The U.S. incidence of colorectal cancer is slightly higher in men than in women, but because women live longer than men, the total number of cases is higher in women. Colorectal cancer incidence and mortality also vary by race and ethnicity, with the highest rate occurring in African Americans; an intermediate rate occurring in whites and Asian/Pacific Islanders; and the lowest rates occurring in American Indians, Alaska Natives, and Hispanics.⁵ Most colorectal cancer deaths are believed to be preventable with screening colonoscopy and polypectomy.⁶ The paradigm that a significant majority of colorectal cancers are thought to arise through the adenoma-to-carcinoma sequence⁷ has been expanded with mismatch repair, serrated, and hybrid pathways.⁸

Neoplasms can be both polypoid and nonpolypoid, and NP-CRN has a higher risk to contain high-grade dysplasia or submucosal invasion at the time of colonoscopy compared with polypoid neoplasm.^1,9,10 It has been estimated that 25% to 40% of adults older than 50 years in the United States have at least one adenoma and that a small fraction of these adenomas progresses to cancer. Because it is impossible to predict which adenoma will become malignant, physicians attempt to remove all adenomas during colonoscopy. The National Polyp Study, which showed that removal of adenomas during screening colonoscopy can decrease the subsequent development of colorectal cancer by 90% compared with historical controls, provided a level of support to this current standard of practice.⁶ It is our dream to be able to prevent the development of advanced cancer safely and efficaciously in every patient we treat.

Endoscopically, adenomas and early colorectal cancer can be classified as polypoid (protruded) and nonpolypoid (superficial) types (Fig. 36.3).¹¹ Colonoscopic polypectomy can be used to remove polypoid lesions. Colonoscopic mucosal resection, using the inject and cut technique, is a safe and efficacious technique used to remove nonpolypoid or sessile lesions. Colonoscopic submucosal dissection technique, which is used to remove diseased mucosa by dissecting through the middle to deeper layers of the submucosa, can refine our ability to remove lesions that are difficult, if not impossible, to remove using mucosal resection technique. In this chapter, we provide an in-depth survey of colonoscopic resection of polypoid and nonpolypoid colorectal lesions.

Fig. 36.3 Macroscopic classification of early neoplastic lesions of the colon and rectum. This classification provides a more precise schematic description of early neoplastic lesions. In addition to the commonly described pedunculated, semipedunculated, and sessile lesions, the classification provides the appropriate descriptors for flat and depressed lesions. The classification is particularly useful for the endoscopist in deciding treatment strategy of early colorectal carcinomas because the risk of submucosal invasion of these lesions corresponds to the endoscopic appearance and size. Image-enhanced endoscopy is useful to study the surface details and border of the nonprotruding type.

Differential Diagnosis

Colonic lesions are classified as either epithelial or nonepithelial. Epithelial lesions include neoplastic adenomas (serrated, tubular, tubulovillous, villous), carcinomas, and nonneoplastic polyps (hyperplastic, juvenile, hamartoma, inflammatory). Occasionally, numerous lesions are encountered during colonoscopy of patients who have polyposis syndromes, including familial adenomatous polyposis (adenomas), Peutz-Jeghers syndrome (hamartomas), juvenile polyposis (juvenile polyps), or Cowden’s syndrome (hamartomas). In addition, patients with hereditary nonpolyposis colorectal cancer may harbor multiple advanced neoplastic lesions. Nonepithelial colonic lesions typically arise in the submucosal, muscularis propria, or serosal layers of the colonic wall and include lipomas, leiomyomas, carcinoids, lymphomas, and metastatic tumors. Indentations of the colonic wall by adjacent organs or endometrial implants on the serosa can also have the appearance of subepithelial lesions.

Careful endoscopic observation of the surface features of the lesion can often allow differentiation of epithelial from nonepithelial origin because nonepithelial lesions are usually covered by normal mucosa. With current videocolonoscopes, and especially with the addition of image enhancement and magnification endoscopy,¹² it is increasingly possible to distinguish reliably among hyperplastic polyps, adenomatous polyps, and superficial early adenocarcinomas. Colonoscopic resection may be viewed as a diagnostic procedure. Removal of the entire lesion, when possible, provides the most rigorous evidence that a malignancy was not missed, as it might be because of sampling error with standard biopsies. These procedures provide the definitive treatment when the lesions are removed completely. Subepithelial lesions sometimes can be removed safely when they are located above the muscularis propria, as evidenced by their endoscopic appearance, response to submucosal saline injection, and, if needed, endoscopic ultrasound (EUS). Generally, biopsy should be performed if possible in lesions that are not amendable to endoscopic resection to ascertain their histology, and the lesions should be endoscopically marked for surgical planning with submucosal tattoo and radiopaque clips.

Clinical Features and Pathology

The macroscopic classification of adenomas and early colorectal neoplasms is crucial in the discussion of diagnosis and treatment of early colorectal cancer.^11,13 The classification by the Japanese Society for Cancer of the Colon and Rectum can provide a common descriptor of adenomas and early colorectal cancer (mucosal or submucosal cancer, regardless of lymph node status). The Paris classification has been promoted for worldwide use¹⁴—although a derivative from the Japanese categorization, its slight alterations in morphologic definitions make it difficult to infer directly the knowledge that has been meticulously collected by our Japanese colleagues. The application of a standard classification of colorectal lesions is the first step in stratifying which lesions are more likely to contain advanced pathology.

Based on their endoscopic appearance, we classify adenomas and early colorectal cancers as polypoid and nonpolypoid. The polypoid type consists of pedunculated, semipedunculated, and sessile polyps. The nonpolypoid type consists of superficially elevated, flat, and depressed lesions. Excavated superficial colorectal neoplasms are rarely observed. Superficially elevated nonpolypoid lesions are differentiated from sessile polypoid lesions both endoscopically (the height of the lesion is less than half the diameter) and histologically (the thickness of the lesion is less than twice that of the adjacent normal mucosa).¹⁵ The term flat is often used to describe superficially elevated lesions. In the colon, however, flat generally connotes that the surface is flat, rather than that the lesion is at the same level as the surrounding mucosa (in the colon and rectum, in contrast to in the esophagus, early neoplastic lesions are rarely at the same level as the surrounding mucosa.)

In the Japanese literature and increasingly becoming common worldwide, flat lesions (IIa) larger than 10 mm are also called laterally spreading tumors (LST), although the term refers more to the growth pattern rather than the endoscopic appearance. In the United States, these large flat lesions are often called carpet lesions. LST with nodular or coarsely granular surfaces are called LST-granular. Other LST are called LST-nongranular. This distinction is important because LST-nongranular are more likely to contain invasive cancer, which is more difficult to assess endoscopically. LST-nongranular are also more difficult to resect and may require en bloc resection to ensure cure. In the Paris classification, semipedunculated lesions are lumped into the pedunculated category, and flat lesions are defined as lesions with height less than a closed biopsy forceps. Depressed nonpolypoid lesions, although rare, accounted for almost one-third of the invasive early cancers that were resected endoscopically by Kudo and colleagues,¹⁶ who observed more than 14,000 colorectal lesions. Confirmatory reports have been published.^12,17 Epidemiology studies of flat and depressed lesions in Western countries have also shown that depressed lesions have a high likelihood of containing invasive cancer (Table 36.1).¹⁸ More than 40% of small (6 to 10 mm) depressed lesions contain submucosal invasive cancer; virtually all large (>2 cm) depressed lesions have submucosal invasion (Table 36.2).¹⁶ In comparison, submucosal invasive cancer is rare in flat lesions smaller than 10 mm. The risk increases to about 30% in LST larger than 2 cm. Protruding (polypoid) lesions have the lowest rate—slightly greater than 2%—of submucosally invasive cancer.¹⁶

Indications and Contraindications

Instruments

Electrocautery

High-frequency electrical current is employed to facilitate cutting and to coagulate vessels at the resection margin. Colonoscopic resection is typically performed with a monopolar snare. The metallic conducting snare serves as the active electrode, and the circuit is completed via a conducting grounding pad that is affixed to the patient’s skin. In the case of polypectomy, when the snare grasps the polyp, electrical current is applied as the snare is closed to transect the stalk. Electrical current traveling through tissue heats it. The amount of heat transferred to each point in the tissue (per unit time) is given by the product of the square of the current density and the resistance. The current density is the amount of current passing through a unit area. Although the same total current passes through the stalk of the polyp and the grounding pad, the current density is much higher at the stalk because the cross-sectional area is smaller. As a result, the stalk is cauterized while the bowel wall and the rest of the patient’s body generally are left untouched (Fig. 36.4).³⁸

Fig. 36.4 The concept of electrocautery is important to master. In this case, application of understanding of the concept of current density—the amount of current per unit area—is shown. A, Current density is highest at the plane with the smallest area, in this case, cross-sectional area that has been strangulated by the loop (arrow). The ideal way to snare a pedunculated polyp that has been looped is to tighten the snare as much as possible to make the snared plane smaller than the plane that has been looped. B, Care is given to snare with adequate distance from the deployed loop. C, In this case, the snare was not closed tighter than the loop, and cautery also occurred at looped area. This case also provides a reason to deploy the loop slightly away from the wall of the colon to prevent accidental coagulation necrosis.

High-frequency electrical current greater than 300 kHz (300,000 cycles/sec) is used because lower frequencies can stimulate muscles, nerves, and the heart. The effect of the current on the tissue depends on the temperature achieved in the tissue, which depends on the shape of the electrode, the duration and waveform of the current, and the voltage. At temperatures between 50° C and 70° C, irreversible cell damage occurs. Between 70° C and 100° C, the tissue is coagulated; collagen is converted to glucose, and the glucose causes the coagulated tissue to become sticky. At temperatures greater than 100° C, the tissue is desiccated; intracellular and extracellular water is vaporized, and the tissue dries out, shrinks, and becomes sticky. During polypectomy, these effects are visualized as a shrinking and whitening of polyp stalks. If low peak voltage (<200 V) is used, the tissue is devitalized, coagulated, and desiccated. As the tissue is dehydrated, the resistance increases until current can no longer flow and no further heat ensues. In this mode, there is little or no cutting; in some clinical situations, the snare can become entrapped in the desiccated tissue.

For efficient electrosurgical cutting, temperatures greater than 500° C are required. At these high temperatures, intracellular and extracellular water is vaporized rapidly, and the cellular architecture is disrupted by steam pressure. Electrosurgery units generate vaporization by producing electrical arcs that jump between the active electrode (snare) and the tissue. The electrical arcs focus the current on a small area on the tissue and result in a locally very high current density and temperature. At least 200 V are required to generate these electrical arcs; if the tissue is partially desiccated, the voltage required is substantially higher. Electrosurgical cutting is ideally performed without contact between the electrode and the tissue, leaving room for proper arcing to form; cutting occurs without mechanical pressure as tissue is vaporized.

During colonoscopy, the snare is typically in contact with the tissue. This contact can lead to lower resistance and high current, causing an inappropriate decrease in voltage that prevents electrical arcs from forming. The result is an unintended coagulation effect without cutting. Newer cautery units, such as the ERBE (Marietta, GA), automatically supervise the initial phase of cutting and provide the necessary current and voltage to ensure proper arcing. During electrosurgical cutting, there is also a coagulation effect, which depends on factors such as the electrode thickness and voltage. Thin electrodes (e.g., a needle-knife), low voltages, and rapid cutting result in little coagulation. Thicker electrodes (e.g., a thick snare loop) and higher voltages result in more intense coagulation. Uncontrolled rapid cutting with minimal coagulation can occur if mechanical pressure (e.g., by closing the snare rapidly) is applied; this phenomenon is also familiar to endoscopists from endoscopic retrograde cholangiopancreatography sphincterotomy, where it is called a “zipper” cut.

Most electrocautery units allow the endoscopist to choose between various cautery modes that are designed to produce predominantly cutting, predominantly coagulation, or a blend of both. However, factors such as the power level, tissue resistance, and speed of snare closure can alter the effect of the cautery. The “cutting” mode delivers a continuous sine-wave voltage pattern. When sufficient voltage is delivered to create electrical arcing, electrosurgical cutting ensues. “Coagulation” and “blended” modes actually deliver a higher voltage, but the delivery is intermittent: short pauses separating brief bursts of delivery. The duty cycle (the total proportion of time that current is delivered) is typically between 5% and 50% in coagulation modes and between 50% and 80% in blended modes. The higher voltages allow deeper coagulation during cutting, particularly in “coagulation” mode. An additional effect, fulguration (spray coagulation), can occur with very high voltages and low duty cycles when a series of random electrical arcs carbonize the tissue.

The use of submucosal injection during mucosal resection promotes a better distribution of current because the current can fan out from the resection site to the wide saline cushion. This fan out reduces thermal damage to the larger submucosal vessel and part of the colon wall immediately beneath the lesion.³⁹

Techniques

Adequate bowel preparation is important; split dosing of bowel preparation is very useful to clean the bowel adequately for the detection and diagnosis of NP-CRN and resection of all types of appropriate lesions. Techniques to prevent looping of the colonoscope during insertion are essential.⁵⁰ Familiarity with the patient, staff, equipment, and accessories is required. Various techniques are available to perform sophisticated colonoscopic polypectomy, mucosal resection, and submucosal dissection. These techniques, designed to increase the safety of resection, have allowed resections of lesions that in the past would have been accomplished through surgery. Interpretation of the pathology is vital, and preparation of the pathologic specimen requires the participation of the endoscopists at the completion of the procedure.

Estimation of Malignant Potential

Colonoscopic resection should be performed in patients whose lesions are most likely to benefit. Equally important, resection should be performed only when the endoscopist has the knowledge and expertise and supporting staff to complete the procedure. The assessment to determine the most likely pathologic findings and the depth of invasion is important in planning for colonoscopic resection. Lesions assessed to be noninvasive are most likely to benefit from endoscopic treatment. Lesions with minimal or moderate risk for submucosal invasion can be treated with endoscopy, provided that the endoscopist believes that the lesion can be safely removed in its entirety and that the potential benefits of endoscopic treatment outweigh the risks. Patients whose lesions are strongly suggestive of invasion should be referred to surgery after a confirmatory biopsy because endoscopic resection would expose them to unnecessary risks. Colonoscopic resection of neoplasms with massive submucosal invasive cancer is generally difficult; has a high risk of bleeding, perforation, recurrence, and metastasis; and should be avoided. Sometimes it is appropriate, after assessment of the lesion, to reschedule the patient for a dedicated resection procedure. Rescheduling allows appropriate discussion of the risks and benefits with the patient and ensures availability of the necessary equipment, endoscopy time, and personnel for the procedure.

Colonoscopy, in some cases with image enhancement, is generally adequate to assess cancer depth (Fig. 36.5). High magnification colonoscopy can be beneficial but generally is not available in Western countries. Colonoscopy with ultrasound is generally unnecessary because it does not provide a clear advantage in assessing depth of invasion over imaging with image enhancement. The assessment is based on several characteristics.

Fig. 36.5 Indigo carmine spray can be very useful for detecting and classifying flat and depressed lesions. A, Spraying a dilute solution of indigo carmine improves visualization of the grooves on the mucosal surface of the colon (innominate grooves). B, These grooves are also observed in hyperplastic lesions, as in this 15-mm lesion. C, Innominate grooves are not observed in neoplastic lesions. Indigo carmine also improves visualization of the border of this 5-mm depressed adenoma.

(From Palo Alto.)

Appearance

The classification of a lesion helps in providing a general stratification of the risks for the lesion to contain high-grade dysplasia or submucosal invasive carcinoma. NP-CRN has up to 10 times higher risk of containing in situ or submucosal invasive carcinoma than a polypoid lesion of similar size. The larger the size, the more likely is submucosal invasion. The general appearance of the lesion can provide further clues regarding the likelihood for invasive cancer (Figs. 36.6 and 36.7). Firm consistency, adherence, ulceration, and friability are findings suggestive of invasion. In addition, the appearance of expansion of normal tissue immediately surrounding the lesion may indicate the presence of cancer creeping into the surrounding submucosa. Converging folds (two or more) toward the lesion can also predict submucosal invasion. Saitoh and colleagues²³ reported the diagnostic operating characteristics of endoscopic findings of depressed colorectal lesions. They observed that patients with one or more findings of expansion appearance, deep depression surface, irregularity of depression surface, or converging folds toward the lesions are more likely to have deep submucosal invasion; the presence of one or more of these findings allowed endoscopists to determine deep submucosal invasion with a 91% accuracy. These investigators used indigo carmine to improve visualization of lesions, as described subsequently.

Fig. 36.6 Pseudodepression versus true depression. A, This adenomatous lesion has small depressions where indigo carmine solution pools. Small depressions without sharp borders, often referred to as pseudodepressions, are typically found in noncancerous adenomas. B and C, Extensive depressions with sharp borders are worrisome findings because they are often observed in cancerous lesions such as these two early adenocarcinomas.

(A and C, From Palo Alto; B, from Tokyo.)

Fig. 36.7 Careful observation of the shape of a lesion is crucial to develop an appropriate treatment strategy. A, This sessile lesion has a full or expansive appearance and converging folds, suggesting the presence of invasive carcinoma in the deeper layers of the submucosa. B, Converging folds toward the lesion and central depression are present. The presence of converging folds is specific for invasion deep into the submucosa or beyond. Both of these lesions were treated surgically.

(From Tokyo.)

Image-Enhanced Endoscopy

Image-enhanced endoscopy is an integral part of colonoscopic resection of lesions.³ It is part of standard practice in many progressive endoscopy units. Two types of image enhancement are available: equipment-based and dye-based. Equipment-based enhancement includes optical (Narrow Band Imaging; Olympus) or software (Fujinon Intelligent Chromo Endoscopy; Fujinon, and Tone Enhancement; Pentax) methods. Dye-based enhancement includes diluted indigo carmine and crystal violet; dyes are not used in most Western countries (Fig. 36.8). The equipment-based method is part of newer colonoscope systems, and dye-based image enhancement can be readily made available in any unit at low cost. Although equipment-based enhancement is sufficient in most cases, it cannot completely replace dye-based enhancement using diluted indigo carmine. The assessment of depth requires pooling of indigo carmine, and complex surface pattern analysis may require confirmatory dye spray.

Fig. 36.8 Close-up observation of the surface of lesions using magnification endoscopy with crystal violet chromoscopy can improve endoscopic assessment. A, A 7-mm superficial elevated lesion with depression. B, Crystal violet has been applied. C, On close examination using magnification endoscopy (100×), the pit pattern was consistent with a deeply submucosal invasive pattern. The patient underwent surgical resection where a deep submucosal invasive adenocarcinoma metastatic to local lymph nodes (2/7) was removed.

(From Tokyo.)

Equipment-based and dye-based image enhancements are complementary to each other. First, image enhancement is useful to confirm the presence of a lesion. Second, it may be required to visualize and to define the border of NP-CRN. Third, image enhancement may be needed to visualize details of the surface pattern of the lesion because neoplasms cause abnormal growth patterns of glands, pits, and microvessels. Equipment-based enhancement allows endoscopists to visualize the microvessels that encircle the glands. Indigo carmine, by pooling in the pits, permits endoscopists to visualize the glands, which surround the pits. Image enhancement provides the opportunity to obtain endoscopic diagnosis: hyperplastic, adenoma or serrated adenoma, slightly invasive carcinoma, and advanced carcinoma. Indigo carmine chromoscopy is simple to use. A few milliliters of diluted solution of indigo carmine (0.1% to 0.4%) can be gently sprayed directly from a syringe through the accessory channel onto the tissue surrounding the lesion; direct spraying may cause minor bleeding, which can obscure visualization of the pit pattern. Indigo carmine solution is not absorbed but rather pools in mucosal crevices and depressions.

Magnification Endoscopy

The ability to select lesions that warrant endoscopic removal is attractive (see Fig. 36.5). Hyperplastic lesions are left alone, adenomas and superficial early carcinomas are resected endoscopically, and invasive cancers are resected surgically. Closer examination of the surface mucosa using magnification endoscopy (100×) allows visualization of the pit pattern, which can provide insights as to the pathology of the lesion. Pit patterns may reflect the tangential structure of the glands of the lesion.⁵¹ As the structural organization of the glands becomes disordered or even absent, as it does in invasive carcinoma, the lesion, as seen magnified from its surface, may have a disorderly pattern. Adenomas have sulci patterns. Hyperplastic lesions have a specific, orderly, enlarged round, oval, or stellar pit pattern.⁵² The original classification of the pit pattern may be too complex to apply in routine clinical practice. A simpler classification that groups the patterns as nonneoplastic, noninvasive, and invasive has been reported but is yet to be used widely.

Nonlifting Sign

Observation of the lesion during and after submucosal saline injection during mucosal resection is a simple but important method to assess the potential for deeply invasive carcinoma (Fig. 36.9).^53–55 Lesions may not lift because of desmoplastic reaction, invasion from the lesion itself, or submucosal fibrosis from prior multiple biopsy, cautery, India ink injection, or ulceration. Several studies have reported the diagnostic operating characteristics of the nonlifting sign. The positive predictive value of the nonlifting sign is approximately 80%.⁵⁶ There is a correlate to the nonlifting sign when submucosal injection is not used; it is typically very difficult to capture in the snare deeply invasive lesions. Difficulties encountered during attempted snare resection should alert the endoscopist to the possibility of deep invasion.

Fig. 36.9 A, Several previous attempts at endoscopic resection of this sessile rectal lesion were unsuccessful. The lesion has a full appearance that suggests invasive cancer, but prior pathology revealed only villous adenoma. B, Nonlifting sign. Injection of saline into the submucosa beneath this lesion does not result in lifting of the lesion; instead, a backflow of injected saline is seen jetting toward the endoscope. Analysis of the surgically resected specimen showed adenocarcinoma invading down to the muscularis propria.

(From Palo Alto.)

Endoscopic Ultrasound

Generally, endoscopic ultrasound (EUS) is not used in differentiating nonpolypoid mucosal lesions from submucosally invasive ones.^57–59 The information needed to decide whether or not to perform mucosal resection can usually be collected by observation during conventional colonoscopy, indigo carmine chromoscopy (with standard magnification), and tests for the nonlifting sign. EUS has been used to locate the blood vessels of large sessile or LST lesions. In this study, the information collected did not change the risks of postpolypectomy bleeding significantly.⁶⁰

Techniques for Resection

Polypectomy

Diminutive (Baby) Polyps

Diminutive polyps can be removed with various techniques, including single or repeated use of cold biopsy, hot biopsy (Fig. 36.10), cold snare,⁶¹ hot snare (Fig. 36.11), or fulguration.⁶² The optimal technique for complete eradication of all polyp tissue is unknown; the effect of routine removal during colonoscopy of diminutive (≤5 mm) polyps on colon cancer mortality is also unknown. However, documentation of adenomas is important for stratification of need for follow-up colonoscopy, which is based on assessment of risk of developing colorectal cancer.

Fig. 36.10 Application of hot biopsy forceps to remove and cauterize a 5-mm adenoma. A, The lesion was captured using the forceps, and coagulation current was applied while the mucosa was tented up to form a pseudopedicle. Tenting causes the cautery effect to be concentrated at the pseudopedicle. B, The base of the lesion, which was previously the pseudopedicle, has been coagulated.

(From Tokyo.)

Fig. 36.11 Standard hot snare polypectomy of a thin-stalked pedunculated polyp. A, Snare loop has been positioned in the middle of the stalk of the polyp. B, As electrocautery was applied, the polyp was slightly tented, and care was taken to avoid having the head of the polyp touch the opposite wall. Touching the opposite wall could reduce the coagulation at the lower end of the stalk and might cause a contralateral burn. C, A stalk remnant with a well-coagulated surface is visible. Leaving some stalk remnant allows application of an endoloop for prevention of bleeding. Alternatively, had bleeding occurred immediately after snaring, the bleeding stalk remnant could have been cauterized further by grasping it with the snare.

(From Palo Alto.)

Pedunculated, Semipedunculated, and Sessile Lesions

Pedunculated and semipedunculated lesions may be resected by snare-loop polypectomy at the middle or upper stalk. Sessile lesions can be resected by a similar technique at the base. Large polyps (>2 cm) or polyps with a thick stalk carry a higher risk of immediate or delayed bleeding. Prophylactic treatment to strangulate the blood vessels before resection prevents immediate or delayed bleeding (Fig. 36.12).^{46,47,63–66}

Fig. 36.12 Use of the endoloop to prevent postpolypectomy bleeding. The endoloop is used like a snare except it can be detached after its deployment at the base of the polyp. A, The endoloop has been applied at the base of a large pedunculated lesion. The electrocautery snare has been placed above the loop with sufficient room to prevent the endoloop from slipping off after transection. B, Resection site immediately after resection. A small blood vessel is visible. There was no bleeding after resection. The diagnosis was lipoma.

(From Palo Alto.)

One prophylactic method is application of endoloops, which are detachable loops that are applied to the base of the polyp stalk to strangulate the vessels supplying the polyp. Iishi and colleagues⁶³ reported the use of endoloops in 47 patients and compared the rate of bleeding with 42 patients who did not have endoloops placed before polypectomy of pedunculated polyps with heads larger than 1 cm. No immediate or delayed bleeding was observed among patients who had endoloops; five (12%) patients in the control group had bleeding (one immediate, four delayed). Di Giorgio and colleagues⁶⁷ reported similar results (0% vs. 12% in patients who had and did not have endoloops placed) in a randomized trial with more patients. Another prospective, randomized study of large pedunculated colon polyps showed significantly lower and less severe delayed bleeding in the patients who underwent prophylactic detachable snare placement at the stalk base followed by conventional polypectomy and clip application in the residual stalk compared with the patients who underwent epinephrine injection alone followed by conventional polypectomy.⁶⁶ However, placement of the endoloop in pedunculated polyps that have short stalks can be difficult (Fig. 36.13). Massive bleeding can occur if the endoloop slips right after snare polypectomy.^48,68 In addition, the endoloop is made of nylon, which can be too floppy for ensnaring the polyp.

Fig. 36.13 Application of the endoloop can be difficult, especially with large pedunculated polyps in a narrow and tortuous sigmoid colon. The double-channel technique, also called lift and ligate, can be useful in these cases. A, The endoloop has been placed over the large polyp head, and a large grasper has been passed through the second accessory channel. B, The grasper was used to pull the polyp toward the endoscope and facilitate deployment of the endoloop. The polyp appeared dusky shortly after endoloop application, indicating that there was sufficient tension to cause ischemia of the polyp (not shown). Snare polypectomy was subsequently performed. The snare was placed well above the endoloop to prevent slippage of the loop.

(From Palo Alto.)

Because massive colonic bleeding can easily cause immediate loss of visualization, other prophylactic techniques may be more appropriate (Figs. 36.14 and 36.15).⁶⁹ Seitz and colleagues⁶⁵ used diluted epinephrine to inject to the base of the stalk before polypectomy. Because the effect of epinephrine is transient, the site was then clipped. Other authors have reported safe use of single or multiple deployment of endoclips before polypectomy (Fig. 36.16).^63,70 The clips did not conduct current, perhaps because care was taken to avoid contact between the snare and the clips. En bloc resection is a key component to precise pathologic staging in cases of polyps containing invasive cancer. A case series of giant pedunculated polyps described a greater than 80% volume reduction after 4 to 8 mL of 1 : 10,000 epinephrine injections into both the polyp head and the stalk. The dramatic volume reduction decreased the need for piecemeal resection.⁷¹ In addition to reducing bleeding, the prophylactic use of the endoloop or endoclip at the base of large pedunculated polyps to facilitate en bloc rather than piecemeal resection has been described. The loop and let go technique may be useful for the treatment of large pedunculated lesions, especially colonic lipomas (Fig. 36.17).⁷²

Fig. 36.14 An alternative to endoloops for preventing postpolypectomy bleeding is the endoclip. A, Single endoclip has been applied at the lower part of a long polyp stalk. B, The polyp appeared dusky after the endoclip placement, indicating effective strangulation of the stalk. The snare was subsequently positioned above the clip, and electrocautery was applied. It is important to avoid touching the clip with the snare during electrocautery application.

(From Palo Alto.)

Fig. 36.15 Multiple endoclips can be used to prevent postpolypectomy bleeding from a thick-stalked polyp. A, Two endoclips were placed on opposite sides of the base of the stalk, causing the polyp to become dusky. B, A third endoclip was placed on the stalk remnant after transection to prevent bleeding. A sizable blood vessel was visible but was not bleeding. A fourth clip was subsequently placed on the vessel (not shown).

(From Palo Alto.)

Fig. 36.16 Application of endoloops or endoclips may be difficult when the stalk is short. Diluted epinephrine can also be used to prevent immediate bleeding after polypectomy. A, Diluted epinephrine (1 : 10,000) was injected into the base of the stalk before resection. An endoclip had been previously applied to an area believed to contain a blood vessel. B, The effect of epinephrine was appreciated with the pale-appearing mucosa seen surrounding the resected site. Multiple endoclips were applied in a zipperlike sequence to clamp any possible large vessels.

(From Palo Alto.)

Fig. 36.17 Endoscopic polypectomy of colonic lipoma—well known to be benign—is known to carry a high risk of perforation. The ligate and let go technique is safe and efficacious to treat colonic lipoma.⁷² A, Typical appearance of a colonic lipoma. Originally, the lesion was sessile (not shown), but by repositioning the patient to make it hanging, the lesion became pedunculated because its weight had pulled it from its point of attachment. B, Ligation of the point of attachment using an endoloop caused the lesion to become ischemic. C, Biopsy on biopsy showed the naked fat, confirming the diagnosis. There was no residual on follow-up colonoscopy a few months later.

Mucosal Resection

Various resection techniques have been described.^27,37 The standard inject and cut technique, also known as saline-assisted polypectomy, is most common. The inject, lift, and cut technique is also popular in Japan. The simple suction technique has been used in many patients, some of whom had exceptionally large sessile or flat lesions. The mucosal resection technique using a ligation device is particularly useful for resection of submucosal lesions in the rectum.

Inject and Cut Technique

The inject and cut technique requires submucosal injection to lift the diseased mucosa. The key aspects of submucosal injection are to inject a sufficient amount and to recognize the presence of the on-lifting sign. The ideal solution, which would form a substantial bulge and would not dissipate quickly, has not been defined. Physicians in the United States routinely use saline (Figs. 36.18 and 36.21); physicians in Japan use Griseol, a mixture of saline and glycerol with a small amount of indigo carmine. Indigo carmine aids in the assessment of depth during and after resection. The remaining submucosa is blue-green in color, and deeper resections yield visualization of muscularis propria, fat, or other organs. Standard 25-gauge sclerotherapy needles are used. Tumor seeding has been reported in only one patient.⁷³ In our experience, performance of safe and effective mucosal resection demands that all necessary equipment and a trained assistant be present so that resection can be performed immediately after injection. We use a stiff standard snare. Whenever possible, we perform en bloc resections; if we must remove lesions piecemeal, we attempt to do so during a single session. We are able to resect most colorectal lesions using the inject and cut technique.

Fig. 36.18 The inject and cut endoscopic mucosal resection (EMR) technique. A, The lesion is carefully examined to assess for the potential of malignancy and depth of invasion. B, Indigo carmine was used to define the border of the lesion and to gain insight from the surface appearance. The lesion has a depressed surface, suggesting a high likelihood of malignancy. C, Saline solution (approximately 5 mL) was injected into the submucosa, lifting the lesion. D, A stiff snare was used to capture the entire lesion with a small rim of surrounding normal mucosa. E, The resection site was carefully examined by spraying additional diluted indigo carmine solution to examine for evidence of residual component. F, The lesion was carefully removed, without breaking it apart, using a Roth net. Maintaining the lesion as a single specimen is important to allow the pathologist to determine the resection margin if cancer is found. After the case had been completed, the specimen was pinned at its periphery onto a piece of wood with thin needles (not shown). EMR specimens are sectioned at 2-mm intervals to provide accurate staging.

(From Palo Alto.)

Fig. 36.19 The fulcrum technique was used in this case to ensure that the snare was placed in a plane parallel to the fold. Saline had previously been injected to lift the lesion. A, The tip of the snare was impacted slightly to the right of the lesion. The snare was slowly opened as the colonoscope was gently turned to the left. Close coordination between the endoscopist and the assistant is required to perform this maneuver. B, The snare was pushed toward the lesion, and air was suctioned slightly to draw the lesion into the snare. The snare was then slowly closed. Before cautery application, the snare was moved back and forth while examining the effect on the wall to ensure that muscularis propria was not entrapped in the snare.

(From Palo Alto.)

Fig. 36.20 The maneuvers to prevent perforation can be exemplified in these series of pictures, although the risk here was not perforation but bleeding from transection of underlying hemorrhoids. A, A 2-cm flat lesion with its distal margin at the dentate line. A large amount of saline had been injected to lift the lesion. B, The snare captured the lesion. C, The lesion was tented high into the lumen while the snare was loosened slightly. This maneuver was performed to release potentially entrapped structures deeper than the submucosa. D, Two separate resections were performed without bleeding. A small amount of residual tissue was found between the resection sites and was eliminated using the argon plasma coagulator (at 60 W). Histology showed a villous adenoma. On follow-up 1 and 2 years later, there was no evidence of recurrence (not shown).

(From Palo Alto.)

Fig. 36.21 Piecemeal resection of a large sessile lesion. A biopsy had been performed to document that the lesion was a large villous adenoma, as was suspected from its endoscopic appearance. A, The lesion was located adjacent to the ileocecal valve. B, A large amount (40 to 50 mL) of saline mixed with a few drops of 0.2% indigo carmine was injected into the submucosa to lift the lesion. Efforts were made to inject so that the lesion does not tilt away from the colonoscope. C, The first piece was resected using a stiff snare at the edge of the lesion. By resecting the lateral margin first, the endoscopist developed an understanding of the depth of the lesion and the location of the submucosal plane. D, A small piece of the lesion had been resected. The visible bluish layer was the submucosa. E, Subsequent resections were performed without proceeding deeper than the submucosa. The remaining large piece of the lesion was snared. Note the position of the snare: parallel with the bowel wall. F, Appearance after piecemeal resection, which was completed approximately 40 minutes later. Argon plasma was applied to ensure complete eradication of the villous adenoma. G and H, The area of prior resection on follow-up 3 months later. There was no residual adenoma.

Simple Suction Technique

The simple suction technique, developed by Soehendra and colleagues,^37,74,75 uses a special stiff 0.4-mm monofilament snare. The construction of this snare allows consistent placement of the snare parallel to the bowel wall and, with slight pressure, capture of the diseased mucosa. Piecemeal resections are performed without submucosal injections.

Submucosal Resection with Ligation

Submucosal resection with ligation⁷⁶ can be particularly useful for resection of submucosal lesions, such as carcinoid tumors in the rectum (Fig. 36.22).^28,77 After the endoscope is fitted with the ligation device, the target area is ligated, with or without prior deep submucosal injection. Standard polypectomy is performed below the rubber band. Small submucosal lesions may require prior markings at their periphery, achieved by brief bursts of cautery using the tip of a snare, because such lesions may be difficult to find after the ligation device has been fitted to the endoscope.

Fig. 36.22 Endoscopic submucosal resection of a small carcinoid tumor in the rectum using a band ligation device. A, Lesion. B, After deep submucosal injection of saline and placement of a band using the ligation device, a snare is seen transecting the lesion as cautery is applied. The lesion was contained within the resected specimen with a clear margin (not shown). By cutting immediately below the band, a deeper resection can be performed.

(From Tokyo.)

Other Techniques

Other techniques that have been described but are not widely used include mucosal resection with cap^78–80; inject, lift, and cut technique (double-channel EMR technique); and use of short endoclips to position the snare (Fig. 36.23).⁸¹ Although popular and efficacious for resection of superficial early cancer in the esophagus and stomach, EMR with cap or inject, lift, and cut can be risky to use in the colon.⁷⁸ The thin muscularis propria of the colon can easily be suctioned into the cap, potentially leading to perforation.

Fig. 36.23 India ink injection is used to mark an ulcerated depressed lesion (inset). India ink solution was injected submucosally at three separate points. The depressed lesion was later shown to be an adenocarcinoma invading deep into the submucosa. Locations of the tattoos (distal, proximal, same level of the lesion) must be documented precisely, especially in cases of flat and depressed lesions. These lesions are often not palpable by the surgeon. Sometimes an endoclip is used in addition to India ink to mark the site on radiographs taken immediately after colonoscopy; this can assist in surgical planning, particularly when laparoscopic resection is contemplated in cases in the left colon. In these cases, the clip, as seen in the abdominal x-ray, can provide the precise location of the lesion.

(From Palo Alto.)

Submucosal Dissection

Endoscopic submucosal dissection (ESD) is the latest technique available for the resection of large nonpolypoid GI neoplasms.⁸² The technique was initially developed for the resection of early gastric cancer that meets specified criteria in which the risks of lymph node metastasis approaches nil. However, in the colon and rectum, most neoplasms are benign and can be resected with minimal risks of recurrence. The indications for colorectal ESD are relatively uncommon (see Box 36.1).⁴ Colorectal ESD is generally performed using various endoscopic knives, including the flex knife, bipolar knife, hook knife, and flush knife. The submucosal injectants used have also varied: saline with and without diluted epinephrine, glycerol, and hyaluronate. After generous submucosal injection, the procedure is begun with a circumferential cut to isolate the lesion with 3 or 4 mm of surrounding normal mucosa. After further submucosal injection, the submucosa under the lesion is dissected until the submucosa is divided throughout. The lesion is removed en bloc. Numerous techniques have been described. By slow dissection of the submucosa under direct visualization, lesions that otherwise would be unable to be captured by a snare or lesions with submucosa fibrosis can also be removed.

The results of colorectal ESD performed by expert endoscopists are very encouraging. Saito and colleagues⁸³ reported the results of 200 colorectal ESDs measuring 38 mm on average, which took a median time of 90 minutes to resect. These investigators were able to obtain an en bloc resection rate of 84% with curative resection of 83%. They had a 5% perforation rate; except in one case (0.5%), the perforations were able to be treated by endoscopic clipping. Intraprocedure bleeding occurred in most patients and was treated endoscopically. Delayed bleeding occurred within the first 5 days after the procedure. Other authors have reported similar data showing high rates of en bloc resection and perforation, which can typically be treated with clipping, bowel rest, and antibiotics.⁸⁴

The advantages of colorectal ESD include the ability to obtain a complete and accurate histologic assessment because the tissue is submitted with normal lateral margin and in one piece. In addition, the risk of recurrence in resection of benign lesions with clear margins is nil, and in lesions with minimal submucosal invasive adenocarcinoma, risk of recurrence can be estimated and stratified. Certain cases of well-differentiated adenocarcinoma that are slightly invasive into the submucosa without lymphatic or vascular invasion have minimal risks of invasion. Patients who have smaller risks of metastasis than surgery may be advised to undergo follow-up only. In Japan, ESD for the colon and rectum is indicated as long as the depth of invasion is less than 1000 µm from the muscularis mucosa.⁸⁵

At the present time, the disadvantages of colorectal ESD are many, and they have become the barriers to its application in the Western countries. Colorectal ESD has a high risk of perforation (on average 5% to 6%, but up to 10%, in expert hands), although most perforations can be treated using endoscopic clipping, antibiotics, and bowel rest. Colorectal ESD requires additional training (25 to 40 gastric ESDs) and specialized devices and takes prolonged procedure time. In addition, patients are admitted to the hospital for approximately 5 days for supportive care and to ensure that complications occur in a monitored setting. Although the current technique and technology of colorectal ESD may pose some limitations, efforts are under way to simplify this modality. These efforts are aimed at eliminating the need for submucosal dissection, creating a modified technique of colorectal ESD that is less difficult and time-consuming.

ESD with snaring, described by Toyonaga and coworkers,⁸⁶ uses standard snaring technique after circumferential incision and trimming. The same authors have also described the results of EMR with a small incision. In this technique, after submucosal injection, a small mucosal incision was made using the tip of the snare. Snaring was then performed by pushing the snare lightly to the incision. These maneuvers, in effect, allowed the tip of the snare to be fixed and the opened snare to capture the surrounding normal mucosa of the lesion. At the present time, ESD with snaring and EMR with small incision techniques have been reported to lead to shorter procedure times, although data on bleeding and complication rate are still limited.

Management and Surveillance after Polypectomy

Limited data are available to guide recommendations for surveillance after resection of adenomatous polyps.⁹¹ The current joint guideline from the American Cancer Society, the U.S. Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology published in 2008 is helpful in directing the interval for patients who have adenomatous lesions.⁹² Patients who have one or two small (<1 cm) tubular adenomas should have follow-up colonoscopy at 5 to 10 years. Patients who have advanced (villous histology or high-grade dysplasia) or multiple (three or more) adenomas should have follow-up at 3 years. Patients with numerous adenomas, large sessile adenoma, malignant adenoma, or an incomplete colonoscopy should have follow-up examinations at short intervals. Numerous studies have shown that large sessile, flat, and depressed adenomas can be removed endoscopically with relatively good success.^{16,65,75,93–95} These studies were conducted by expert endoscopists in referral centers, however.

After piecemeal resection, a repeat colonoscopy is typically performed in 3 to 6 months to assess for local recurrence. The postmucosectomy scar site should be examined carefully; image-enhanced endoscopy techniques may be useful to show the presence of the innominate grooves across the scar and normal pit or microvessel patterns. A resected specimen that was noted to contain adenoma at the index examination may contain advanced histology at a follow-up examination.^65,96 Khashab and colleagues⁹⁷ reported a high predictive value for long-term eradication in cases where the postmucosectomy scar site showed both normal macroscopic and microscopic (biopsy) findings. In cases with residual neoplasia, appropriate therapy with biopsy or repeat EMR is prudent, and another surveillance colonoscopy should be performed at 6 months. Subsequent examinations should be performed every 3 to 6 months until long-term eradication is confirmed, and then the patient should resume surveillance at the recommended guideline intervals. The rationale for such an intensive follow-up schedule is the relatively high rate of local recurrence described in the literature, particularly after piecemeal polypectomy.

A series of approximately 300 polyps larger than 3 cm that were resected endoscopically showed a recurrence rate of 17%, with most recurrent lesions successfully treated endoscopically.⁶⁵ In our efficacy study of standardized inject and cut EMR technique on 125 nonpolypoid lesions (117 flat and 8 depressed), we identified a 10% rate of local recurrence at the prior EMR site at the first surveillance colonoscopy, with ultimate eradication following one or two additional colonoscopies. The recurrence is typically small. No lesion required surgery. Over a 4.5-year follow-up period, no patient developed or died of advanced colorectal cancer or distant metastasis.² The management of patients who have polypoid lesions containing invasive carcinoma is not straightforward. The risk of metastases of T1 lesions is approximately 10% to 15%.^98,99 Because most patients do not develop metastases, the decision of whether to perform surgery is complex. Immediate surgery performed shortly after initial local resection of T1 lesions confers a disease-free 5-year survival rate significantly higher than that of patients who had surgery only after local recurrence or lymph node metastasis from rectal cancer.¹⁰⁰

Given the pathology findings, the endoscopist must assess whether the risk for lymph node metastasis is lower than that of partial colectomy. Various stratification methods have been reported; most distinguish patients at high risk and low risk to develop recurrence or metastasis. Published data often are collected from a few patients; even though they indicate that the absolute fraction of patients who developed metastases was low, great care must be exercised because the upper limit of the confidence interval of this fraction is often higher than the risk of surgery. Many studies grouped different types of lesions. The study by Haggitt and colleagues¹⁰¹ that reported the level of invasion as the major prognostic factor is often cited to direct the management of polypoid lesions with submucosal invasion (Fig. 36.24). In this retrospective study, the authors concluded a low risk of metastasis or local recurrence when the level is less than 4. Generalizing their conclusions to endoscopic practice at large should be cautioned, however. The study involved a small number of patients—some had endoscopic treatment and others had surgery. The study lumped all patients who had sessile lesions that contained submucosal invasion into one group (level 4). Finally, Haggitt and colleagues¹⁰¹ did not report the outer limit of the 95% confidence interval of their data (Table 36.3). When analyzed, patients with level 0 to 3 of invasion had up to 2.74% risk of developing metastasis, whereas patients with level 4 had up to 91% risk of favorable outcomes.

Fig. 36.24 Haggitt and colleagues¹⁰¹ stratified the level of cancer submucosal invasion by the following criteria: level 0, carcinoma in situ (i.e., no extension below the muscularis mucosa); level 1, carcinoma invading through the muscularis mucosa but limited to the head of the polyp (i.e., above the junction between the adenoma and its stalk); level 2, carcinoma invading the level of the neck (i.e., the junction between adenoma and its stalk); level 3, carcinoma invading any part of the stalk; and level 4, carcinoma invading into the submucosa of the bowel wall below the stalk. In malignant sessile lesions, invasive carcinoma is considered as level 4.

Table 36.3 Statistical Limitations of the Haggitt* Study

CI, confidence interval.

* See reference 101.

Because the risk of morbidity from surgery is less than 1% in patients with minimal comorbidity, the data by Haggitt and colleagues¹⁰¹ do not fully support that these patients should be uniformly advised to have watchful waiting. Similarly, a patient who has high risk for surgery with more than 10% risk of mortality may have a higher risk of dying from surgery than from developing metastasis (see Table 36.3). Detailed pathologic studies from Japan have found that the absolute depth of invasion should be considered because there is a gradual increase in the risk of lymph node metastasis with the depth of submucosal invasion.^16,102 Kitajima and colleagues⁸⁵ pooled the data from six institutions in Japan and classified the invasion according to Fig. 36.25. For pedunculated polyp, submucosal invasion was measured from an imaginary line drawn from the level 2 line of Haggitt’s study to the deepest portion of the submucosa invasion. For NP-CRN, the muscularis mucosa was used as baseline, and the depth of invasion was the vertical distance from this line to deepest portion of invasion. When the muscularis mucosa could not be identified, the top of the lesion was used as baseline. For pedunculated lesions, they found that 0 of 53 patients (95% confidence interval 0% to 7%) with invasion limited to the head had lymph node metastasis. For the nonpedunculated lesions, 0 of 123 patients (95% confidence interval 0% to 3%) with depth of invasion less than 1000 µm had lymph node metastasis at the time of surgery. Similar to the data by Haggitt and colleagues,¹⁰¹ this study was limited by the small sample size and length of follow-up.

Fig. 36.25 Stratification of submucosal invasion of early colorectal cancer according to the General Rules for Clinical and Pathological Studies on Cancer of the Colon, Rectum and Anus. A, For nonpolypoid and sessile lesions, the vertical depth of the carcinoma is measured from the muscularis mucosa to the deepest point of invasion. B, If the muscularis mucosa has been destroyed by the carcinoma, the measurement is made from the surface of the lesion. Endoscopic therapy without surgery is accepted when the depth of the well-differentiated carcinoma without lymphatic or vascular involvement is less than 1000 µm and the lateral and vertical margins are negative. C, For pedunculated and semipedunculated lesions, the imaginary line is drawn according to level 2 invasion by Haggitt et al¹⁰¹: The lesion is considered to be head invasion only when the carcinoma did not infiltrate below the line. Endoscopic therapy is accepted without surgery in head invasion of well-differentiated carcinoma without lymphatic and vascular involvement.

Kikuchi and colleagues¹⁰² stratified the risk of metastasis with the depth of carcinoma invasion from the muscularis mucosa and followed the patients for at least 5 years. Sm1 is slight mucosal invasion from the muscularis mucosa to a depth of 200 to 300 µm, sm2 is intermediate invasion, and sm3 is invasion near the inner surface of the muscularis propria. The investigators found that 64 patients with sm1 invasion (75% of whom had either semipedunculated or sessile lesions) had no evidence of metastasis during follow-up of at least 5 years. Other authors have developed stratification systems that combine depth and histology.^98,103 Favorable histology includes well-differentiated and moderately differentiated adenocarcinoma with carcinoma cells at least 2 mm from a clearly visualized margin. Unfavorable histology includes poorly differentiated adenocarcinoma, mucinous adenocarcinoma, and signet-ring cell carcinoma or one with adenocarcinoma cells within 2 mm from a clearly visualized margin. If the margins could not be assessed, the lesion was classified as unfavorable histology. No patients with favorable histology treated with endoscopic polypectomy had an adverse outcome. However, this study also involved only a few patients.

Additional factors have been suggested to be important in the patient risk stratification for local recurrence or metastasis. Tanaka and colleagues¹⁰⁴ reported that the absence of lymphatic or vascular involvement is highly predictive of a successful outcome. Masaki and Muto¹⁰⁵ showed that unfavorable histology (presence or absence of small nests of cancer cells with poorly differentiated or mucinous histology) at the invasive margin is predictive of adverse outcome. The management of small depressed lesions depends on the histology, lymphovascular invasion, and depth of invasion. Lesions larger than 1 cm often contain invasive cancer; patients who have such lesions are often referred directly to surgery.¹⁶

At the present time, the stratification of patients with lesions containing submucosal invasive cancer is not resolved. Meticulous examination of the specimen by an experienced pathologist in close communication with the endoscopist helps to optimize the decision whether to recommend surgery. A patient with submucosal invasion must be well informed of the risks and benefits of endoscopic therapy and surgery, even in cases where the risk of metastasis is small.

Complications

Complications of polypectomy and mucosal resection include bleeding, transmural burn, and perforation. Familiarity with the endoscopic findings, symptoms, and signs of complications and treatment of complications is a prerequisite for performance of colonoscopic polypectomy and mucosal resection.

Postpolypectomy Bleeding

Postpolypectomy bleeding can occur during or after the procedure. The reported incidence varies according to the definition of bleeding, and the size and type of lesions resected. Significant bleeding was observed in 0.4% of hot biopsy specimens of diminutive polyps in one series. Other authors reported no complications in a series of more than 900 hot biopsy specimens of diminutive polyps. The overall risk is approximately 1% to 2% for snare polypectomy. Rosen and coworkers¹⁰⁶ reported a 0.4% risk of bleeding requiring hospital admission in a retrospective study involving 4721 patients who had polypectomies. Nivatvongs¹⁰⁷ reported 10 episodes of bleeding requiring blood transfusions among 1172 patients. Soehendra and coworkers⁷⁵ reported a 24% risk of bleeding in a series of 176 large (>3 cm) polypectomies. Most of these hemorrhages occurred during the procedure, and all were successfully treated by endoscopic methods.

A variety of techniques are useful, including application of the endoclip or endoloop, use of APC, injection of diluted epinephrine, cauterization using monopolar or bipolar instruments, and repeat application of the snare or hot forceps biopsy to grasp the remnant stalk of pedunculated polyp. In cases in which endoscopic treatment fails, selective angiogram with the application of absorbable gelatin sponge (Gelfoam) emboli or surgery is employed. Our preferred technique for minor bleeding is APC; for significant bleeding, we use the endoclip or endoloop with or without prior injection of diluted epinephrine (Fig. 36.26). Parra-Blanco and colleagues¹⁰⁸ reported a case series showing the efficacy of endoclips to treat postpolypectomy bleeding. In cases with delayed bleeding, we typically would purge the bowel using 4 to 6 L of polyethylene glycol solution within 3 hours, immediately followed by a colonoscopy,¹⁰⁹ although Rex and colleagues¹¹⁰ have reported successful colonoscopic treatment of delayed postpolypectomy bleeding without prior bowel purge.

Fig. 36.26 Principles of endoscopic clipping in prevention and treatment of postpolypectomy bleeding. A, Spurting arterial bleeding that occurred a few days after snare polypectomy. B, The endoclip has been well positioned to capture the bleeding vessel. To obtain the best control and leverage, the tip of the endoclip applicator was positioned very close to the tip of the colonoscope. Application of the endoclip will be more secure when normal underlying tissue is also captured during endoclip deployment. In this case, the endoclip was positioned slightly distal to the bleeding vessel to allow for capturing tissue underneath the vessel. C, To ensure capturing the vessel and its underlying tissue, the clip was pushed down, the lumen was suctioned, and the clip was slowly closed. D, The endoclip has been placed securely. Sequential application of multiple endoclips may be required in postpolypectomy bleeding from a thick stalk (not shown).

(From Palo Alto.)

Future Trends

Advances in the technology and technique of colonoscopy have allowed us to manage increasingly complex colorectal lesions via endoscopy. Learning and using these new technologies and techniques may be demanding, but they enable us to perform endoscopic resections of lesions (Fig. 36.27) that previously would have required major abdominal surgery. The more sophisticated approaches can also improve recognition of the nonpolypoid lesions that were previously underappreciated in the Western countries and can assist in their treatment. Further developments will increase the potential of colonoscopic polypectomy, mucosal resection, and submucosal dissection. Extensive and long-term databases, using a common terminology for endoscopy and pathology,^115,116 are needed. Simplifications in the technology and techniques of the resections would allow them to be used more widely and ultimately benefit more patients.

Fig. 36.27 En bloc resection of a large rectal localized lymphoma using the insulated-tip (IT) knife. A, The lesion seen after indigo carmine spray. B, IT knife is shown. C, Submucosal injection with diluted epinephrine and indigo carmine has been performed, and a circumferential incision has been made surrounding the lesion. D, After resection, the lesion is oriented, stretched, and pinned on a piece of Styrofoam by the endoscopist. The lesion was subsequently immersed in formalin. The pathologist sections the specimen at 2-mm intervals and notes their positions relative to each other. A precise determination of lateral and vertical depth of invasion can be reported.

(From Tokyo.)