The risk of colorectal cancer (CRC) is increased in patients with inflammatory bowel diseases (IBDs) compared with that in the general population [1-3], and dysplasia is considered as a precancerous lesion of CRC in long-standing ulcerative colitis (UC) or Crohn’s disease [4]. Traditionally, total proctocolectomy (TPC) has been regarded as a standard treatment for UC-associated dysplasia [4]. However, since early studies on the feasibility of polypectomy for dysplasia [5,6], subsequent studies have supported the therapeutic role of colonoscopic polypectomy for polypoid dysplasia in patients with colitis. The pooled incidence of CRC after endoscopic resection of polypoid dysplasia is 5.3 cases per 1,000 patient-years in patients with colitis [7], and the incidence of interval CRC was 2.5 cases per 1,000 patient-years in a recent large surveillance study on patients with colitis [8]. Therefore, recent guidelines suggest endoscopic resection as a key modality for the treatment of visible and endoscopically resectable dysplasia in these patients [9-11]. Endoscopic submucosal dissection (ESD) has been proposed as a resection technique for non-polypoid dysplasia [11,12], especially when conventional endoscopic mucosal resection (EMR) is not suitable for en bloc resection. However, only a small number of studies reported the feasibility of ESD for colitis-associated dysplasia to date. Herein, we review the feasibility and limitation of ESD for colitis-associated dysplasia.

The old terms to describe the gross appearance of colitis-associated dysplasia were notoriously confusing. The term “dysplasia-associated lesion or mass (DALM)” had been coined to describe endoscopically visible dysplasia showing heterogenous appearances, such as single polypoid mass, plaquelike lesions, and multiple polyps, in 1981 [13]. “Flat dysplasia” had been frequently used to describe invisible dysplasia detected by random biopsy in the past [14]. However, as more dysplasias have become visible and characterizable under endoscopy, the need for new and less-confusing classification systems has increased. The recent international consensus statement of the Surveillance for Colorectal Endoscopic Neoplasia Detection and Management in Inflammatory Bowel Disease Patients (SCENIC) suggested a new classification system for dysplasia identified on surveillance colonoscopy [10,11]. First, dysplasia should be divided into visible and invisible dysplasias. Therefore, flat dysplasia is no longer categorized as invisible dysplasia in the SCENIC classification. Second, the other imprecise terms, including DALM, adenoma-lesion or mass, and non-adenoma-like dysplasia, were abandoned. Instead, the Paris classification for sporadic colorectal neoplasia (CRN) was adopted to describe visible dysplasia. Third, two general descriptors for dysplasia, i.e., presence of ulcerations and distinctness of borders, should be considered. Presence of surface ulcerations and indistinct borders are currently suggested as features of endoscopically unresectable dysplasia [11]. Compared with previous descriptions for dysplasia, this new classification can be a more uniformed communication tool among endoscopists.

Because of the increased risk of lymph node metastasis, deep submucosal invasive CRC should be treated surgically instead of performing endoscopic resection if it can be predicted precisely [15]. However, the predictive factors of deep submucosal invasion are not well established in patients with long-standing colitis.

In sporadic CRN, type Vn Kudo pit pattern and invasive narrow band image (NBI) patterns, such as Sano type IIIB or Japan NBI expert team type 3, are closely associated with deep submucosal invasion [16-19]. According to a recent meta-analysis on deep submucosal invasion prediction in sporadic CRN, the pooled estimates of sensitivity were 77% for NBI and 81% for magnifying chromoendoscopy, while those of specificity were 98% for NBI and 95% for magnifying chromoendoscopy [20]. Neoplastic pit patterns are deemed to predict dysplasia or neoplasia less accurately in patients with IBDs than in those without. The diagnostic accuracy of neoplastic pit patterns in predicting sporadic CRN reached 98% in a study [21]. Neoplastic pit patterns predicted neoplasia, with sensitivity of 89% and specificity of 86%, in a meta-analysis that included studies based mostly on patients without colitis [22]. Conversely, neoplastic pit patterns could discriminate dysplasia from non-dysplasia, with a relatively low accuracy (62%–73%) [23-25], in patients with colitis, except in an early study that reported excellent diagnostic performance of neoplastic pit patterns (sensitivity of 94%, specificity of 93%, and accuracy of 92% for predicting dysplasia) [26]. Discrimination between carcinoma or high-grade dysplasia and low-grade dysplasia using NBI and magnifying chromoendoscopy is also challenging. A recent study reported sensitivity of 72.2%, specificity of 85.7%, and accuracy of 76% for differentiating carcinoma or high-grade dysplasia from low-grade dysplasia [27]. Given these data, invasive pit or NBI patterns may predict deep submucosal invasion less accurately in colitis-associated dysplasia/CRC than in sporadic CRN, although no comparative study is available yet.

Depression, presence of overlying ulceration, and failure to lift with attempted submucosal injection are suggested as gross features of submucosal invasion in dysplasia [11]. Of these findings, failure to lift with attempted submucosal injection may be often different from the positive non-lifting sign typically observed in deep submucosal invasive CRC in patients without colitis. A classic positive non-lifting sign is originally defined as non-lifting of the lesion with elevated surrounding mucosa after submucosal injection [28]. However, as the background of the lesion often has extensive scar changes, especially in patients with UC, not only the lesion itself, but also the surrounding area may not be lifted (diffuse non-lifting sign; Fig. 1). The diagnostic performance of these three gross features for predicting deep submucosal invasion is still unknown in patients with IBDs and thus needs to be investigated.

The techniques and equipment for ESD for sporadic CRNs have been described in many publications [29-31], and the basic steps of ESD for sporadic CRNs can also be applied for UC-associated dysplasia (Fig. 2). High-definition endoscope systems and a transparent cap attached to the distal end of the scope are necessary for ESD. Endoknives and electrosurgical units are also essential. Careful observation of the target lesion should be the first step of ESD. Dye chromoendoscopy and/or image-enhanced endoscopy is useful for delineating the margins and characterizing the lesion surfaces. Circumferential diathermic marks before submucosal injection are not generally required during ESD for sporadic CRN [29] because the margins of sporadic CRN are endoscopically distinguishable from the surrounding non-neoplastic mucosa. However, the surrounding mucosa of UC-associated dysplasia frequently shows chronic inflammatory or scar changes [32,33], which the operator may confuse with true dysplasia during a prolonged procedure. Therefore, we recommend making diathermic marks around the margins of dysplasia before injecting a submucosal solution, even if the borders are distinct (Fig. 3). A viscous solution, such as sodium hyaluronate solution, is preferred for submucosal injection, as it can provide a prolonged submucosal cushion [34]. Using an endoknife, a partial mucosal incision should be created at approximately 3–5 mm apart from the delineated margins. Circumferential mucosal incision created at once is not recommended for colorectal ESD because the submucosal solution will leak out easily. Trimming beneath the initial mucosal incision can expose the submucosal layer, and submucosal dissection should be followed. Once the submucosal layer is dissected enough to form a flap or a pocket, the operator can gently press and stretch the submucosal layer using the scope with a transparent cap. With this maneuver, the interface between the proper muscle and submucosal layers is visible via the transparent submucosal layer while submucosal dissection is performed.

The histological features of colitis-associated dysplasia, which are different from those of sporadic colorectal adenoma, can be summarized as follows: unclear delineation between neoplastic and non-neoplastic areas; variety in configuration, size, and diameter of the neoplastic glands; increased infiltration of mononuclear inflammatory cells at the laminar propria; mixtures of the dysplastic and non-dysplastic glands at the lesion surface; and neoplastic proliferation from the base of the gland to the surface [35,36]. These features themselves can confuse the pathologists with true dysplasia and reactive changes and are not actually universally observed in dysplasias. Consequently, the histological diagnosis of dysplasia occasionally shows discrepancies across pathologists, especially between unspecialized and specialized pathologists [37-39]. Current guidelines recommend that the diagnosis of dysplasia should be confirmed by a gastrointestinal (GI) specialist pathologist [9,10].

Similar to the studies on ESD for sporadic CRNs, the currently available studies on ESD for UC-associated dysplasia defined “complete (R0) resection” as en bloc resection with tumor-negative lateral and vertical margins at histology and “curative resection” as R0 resection with superficial (<1,000 mm from the muscularis mucosa) submucosal invasion, no lymphovascular involvement, and good cell differentiation [27,32,33,40]. Therefore, in addition to the grading for dysplasia, both completeness and curativeness should be assessed by a GI specialist pathologist who has expertise in interpreting dysplasia in patients with IBDs.

Conversely, whether biopsies should be obtained from the adjacent mucosa after removing dysplasias is not established yet. Recent studies mostly based on polypoid dysplasia showed that the yield of peri-lesional biopsies ranged from 0.3% to 5% [41-43], and one of them revealed that the diagnostic yield of peri-lesional biopsies from the mucosa not suspicious of dysplasia was 0% [43]. Therefore, peri-lesional biopsies after endoscopic resection of dysplasia may not be mandatory, especially if the peri-lesional mucosa appears normal under endoscopy. However, limited knowledge is currently known regarding the prevalence of invisible or endoscopically unsuspected dysplasia around non-polypoid dysplasia. Therefore, we recommend performing four-quadrant biopsies routinely from the margins of the postprocedural ulceration after ESD, until the yield of peri-lesional biopsies for non-polypoid dysplasias can be determined.

Although ESD has been suggested as a resection technique for non-polypoid dysplasia [11,12], its outcomes for non-polypoid dysplasia in patients with colitis were reported in several recent studies. The reported outcomes related to ESD for colitis-associated dysplasia or early CRC include the en bloc and R0 resection rates, curative resection rate, adverse events, local recurrence, and metachronous recurrence (Table 1).

Colectomy is indicated for the treatment of endoscopically unresectable dysplasias in patients with UC [9,11]. However, according to survey-based studies, a considerable proportion of patients with IBDs would refuse colectomy for the treatment of dysplasia [47-49], and the fear of permanent ostomy or complications was the main reason for refusal of surgical treatment [49].

TPC with ileal pouch-anal anastomosis (IPAA) has become a surgical procedure of choice for patients with UC, as the postoperative clinical and functional outcomes were stable during long-term follow-ups [50,51]. However, the mean daytime bowel frequency ranged from 5.7 at 1 year to 6.4 at 20 years [50]. Therefore, patients with UC in remission may not agree with the physician’s recommendation for surgery if they have dysplasia, not cancer. In a Korean multicenter retrospective study [52], 415 patients with UC underwent colon surgery, and most of them (92.2%) underwent TPC. Medical intractability of UC (n=270, 65.1%) was the most common indication of surgery, followed by dysplasia or malignancy (n=52, 12.5%). The overall postoperative complication rate was 34.7%. Ileus (n=21), bleeding (n=16), and anastomotic leakage (n=15) were common early postoperative complications. Pouchitis (n=48) was the most common late complication. In another largescale (n=3,707) study including 2,953 patients with UC and 150 patients with Crohn’s disease, the early and late complication rates in TPC with IPAA were 33.5% and 29.1%, respectively [53]. The authors also reported a perioperative mortality rate of 0.1%.

Therefore, the decision to undergo colectomy for dysplasia should be highly individualized, and the patients’ clinical status, life expectancy, and preferences must be considered [54]. The final decision should be balanced between these factors and the risk of CRC.

The procedure-related outcome measures of ESD for colitis-associated dysplasias, such as the en bloc and R0 resection rates and incidence of complications, are comparable with those of ESD for sporadic CRNs. However, such data are based on small studies reported by groups with a high level of expertise [27,32,33,40]. Moreover, the local and metachronous recurrence rates are relatively high, and no long-term outcome data are available yet. Nonetheless, given the functional outcome and risk of complications related to colectomy, ESD can be considered feasible for dysplasia that cannot be removed using the conventional EMR techniques.

The most appropriate conditions for attempting ESD for colitis-associated dysplasia can be suggested as follows (Table 3). First, the peri-lesional mucosa should be in remission endoscopically. Second, the patients’ colitis itself should be medically tractable. Third, any of the endoscopic findings indicating possible invasive cancer should be absent, although the diagnostic performance of invasive pit or vascular patterns, surface ulcerations, and the non-lifting sign has not been determined yet. Fourth, an expert endoscopist highly skilled in both colorectal ESD and surveillance colonoscopy for patients with colitis should assess the lesion and perform ESD. The endoscopists in the currently available studies already had expertise in colorectal ESD before performing ESD for colitis-associated dysplasia [27,32,33,40]. However, the qualification needed for performing ESD of colitis-associated dysplasia among endoscopists is not currently available. A recent Korean study reported that the first ESD for colitis-associated dysplasia had been attempted after approximately 200 cases of ESD experiences [33]. Fifth, a GI pathologist specialized in interpreting the histology of dysplasia should be available. Finally, both an IBD physician and an IBD surgeon should be available to provide proper management after curative or non-curative ESDs, respectively. Therefore, we strongly suggest that ESD for colitis-associated dysplasia be performed by an ESD expert endoscopist in a center with a well-organized IBD team.

ESD has been suggested as one of the therapeutic options for endoscopically resectable dysplasia in patients with colitis. Recent studies performed by expert centers support the feasibility of ESD for the treatment of colitis-associated dysplasia in terms of the en bloc and R0 resection rates and the risk of procedure-related complications. The relatively high incidence of colectomy-related complications justifies selecting ESD as a primary treatment for endoscopically resectable dysplasia that cannot be effectively treated using the conventional EMR technique. However, the local and metachronous recurrence rates after ESD for colitis-associated dysplasia are relatively high, and long-term therapeutic outcome data are still lacking. In addition, the lesion and patient characteristics suitable for performing ESD are not well established in colitis-associated dysplasia. Therefore, the therapeutic decision for colitis-associated dysplasia, which potentially needs ESD, should be made after individualized and multidisciplinary approach in close collaboration with a multidisciplinary IBD team. The lesion characteristics, patients’ general medical conditions and preferences, and status of colitis must be taken into consideration. After curative ESD, meticulous colonoscopic surveillance is mandatory to monitor for local and metachronous recurrences of dysplasia. Further studies should be conducted to establish the long-term therapeutic outcomes of ESD, such as the incidence of post-ESD CRC and the risk of colectomy after ESD.