Endoscopic approach to large non-pedunculated colorectal polyps

Abstract

Large non-pedunculated colorectal polyps ≥20 mm (LNPCPs) constitute approximately 1% of all colorectal polyps and present a spectrum of risks, including overt and covert submucosal invasive cancer (T1 colorectal cancer (CRC)). Importantly, a curative resection may be achieved for LNPCPs with superficial T1 CRC (T1a or T1b <1000 µm into submucosa), if an enbloc R0 excision (clear margins) with favourable histology is achieved (ie, absence of high-grade tumour budding, lympho-vascular invasion, and poor differentiation). Thus, while consensus recommendations advocate for endoscopic resection as the primary treatment option for LNPCPs, thorough optical assessment is imperative for selecting the most suitable ER strategy. In this review, we highlight the critical components of optical evaluation that assist in predicting the risk of T1 CRC, including morphology (Paris and LST classifications), surface pit/vascular pattern (JNET and Kudo classifications), and lesion location. Different resection modalities, including endoscopic submucosal dissection and endoscopic mucosal resection are discussed, along with important considerations that may influence the resection strategy of choice, such as access to the LNPCP and submucosal fibrosis.

Introduction

Endoscopic resection (ER) stands as the cornerstone in the management of benign colorectal polyps, directly contributing to a reduction in the incidence of colorectal cancer (CRC).¹ The vast majority of these polyps (>90%) lack advanced pathology, are smaller than 10 mm, and can be effectively removed through cold-snare polypectomy (CSP).² However, large non-pedunculated colorectal polyps ≥20 mm (LNPCPs) constitute approximately 1% of all colorectal polyps and present a spectrum of risks, including overt and covert submucosal invasive cancer (T1 CRC). Although consensus recommendations advocate for endoscopic resection (ER) as the primary treatment option for LNPCPs, supported by robust evidence highlighting its superior safety and cost-effectiveness compared to surgery,^3–5 thorough optical assessment is imperative for selecting the most suitable ER technique for these lesions.

This is especially crucial as the scope of ER indications has expanded from benign polyps to those harbouring T1 CRC, which may potentially be curable through endoscopic means with careful en-bloc resection, and if horizontal and vertical margins are clear (R0 excision), and the histology is favourable. Favourable histology (low-risk T1 CRC) includes the absence of high-grade tumour budding, lympho-vascular invasion, and poor differentiation. For patients with T1a and T1b CRC plus low-risk histological features, the lymph node and metastases risk (LNM) is 0-0.8% and 2.9%–5.9%, respectively, with a collective recurrence risk of 1.2%.⁶ Conversely, in patients with T1a or T1b CRC with unfavourable histology, the LNM risk increases to 3.7%–23.9% with a recurrence risk of 9.5% (without salvage surgery).⁶ Therefore, R0 resection of concerning lesions is paramount to provide accurate histological staging, whilst also offering a chance of curative resection for low-risk T1 CRC lesions. To achieve an en-bloc R0 excision, endoscopic submucosal dissection (ESD) often supersedes the conventional endoscopic mucosal resection (EMR) technique. Furthermore, in instances where the risk of lymph node metastasis (LNM) is deemed significant, ESD does not preclude adjuvant surgery with lymph node resection, and thus can also be considered a staging procedure.

Thus, meticulous endoscopic evaluation and characterization of LNPCPs become indispensable in ensuring the selection of the most appropriate ER technique for each lesion. While distinguishing between benign colorectal polyps and advanced colorectal cancers infiltrating the muscularis propria is generally straightforward, identifying those with T1 CRC poses a greater challenge due to the heterogeneity in size and morphology of LNPCPs. In this review, we aim to elucidate the decision-making process behind piecemeal and en-bloc ER strategies, delineate various ER techniques, outline important adjunctive therapies, and discuss post-resection surveillance protocols.

Piecemeal versus En-bloc resection

Optical evaluation of the colorectal lesion

To guide any ER approach, accurate assessment and diagnosis of the colorectal lesion are of primary importance. An endoscopists optical evaluation should include the lesion’s size, location, morphology, granularity, microvascular, and surface pit patterns. These assessments can be achieved using high-definition white light examination (HD-WLE), narrow band imaging (NBI), or blue-light imaging (BLI) combined with image magnification.

Accurate optical evaluation can help the endoscopist predict histology and identify overt (presence of optical features of T1b CRC or deeper) or covert (absence of overt optical features of CRC, but other high-risk features may be present) cancer. If high-risk features of T1 CRC are detected, en-bloc resection should be performed (features are discussed below). If the risk of T1 CRC is deemed low, piecemeal endoscopic mucosal resection (pEMR) is suitable.

First, lesion gross morphologic assessment is simple and incredibly informative. There are 2 primary endoscopic morphologic classification systems: (A) Paris classification⁷ and (B) laterally spreading tumour (LST) classification⁸:

• (A) Paris classification describes 3 major polyp morphologies:

  • (1) Polypoid: 0-Ip (pedunculated) or 0-Is (sessile)

  • (2) Non-polypoid: 0-IIa (minor elevation) or 0-IIb (flat) or 0-IIc (minor depression)

  • (3) Excavated: 0-III (ulcerated).

• (B) LST (defined as: colonic polyps that are >10 mm with either a 0-IIa or 0-Is morphology that extend laterally along the colonic wall). Two distinct LST phenotypes with 4 subtypes are described:

  • (1) LST-granular (LST-G; nodular surface appearance):

  • (1) Homogeneous LST-G.

  • (2) Mixed LST-G (LST-MG).

  • (2) LST-non-granular (LST-NG; smooth surface appearance):

  • (3) LST-NG pseudo-depressed (PD).

  • (4) LST-NG flat elevated (FE).

There is now robust data that describes the risk of submucosal invasion based on lesion size and morphology. Homogeneous LST-G colonic lesions are considered to have the lowest risk of T1 CRC (<2%, regardless of size), followed by LST-NG (FE) lesions (10.4% in lesions ≥2 cm).^9,10 In contrast, LST-MG polyps have a significantly higher risk of T1 CRC, especially if the polyp size is ≥2 cm and the lesion is in the rectosigmoid location (22.3%–38%).^10,11 The highest-risk lesions are LST-NG (PD) polyps and en-bloc resection should always be considered in this lesion phenotype (41.4% in lesions ≥2 cm, irrespective of location).¹¹

Similarly, using the Paris classification, colonic polyps with depressed (0-IIc) morphology are strongly associated with T1 CRC (61%).^12–15 In 2017, Burgess et al. assessed the morphology of 2277 polyps that were ≥2 cm and found that 0-Is non-granular and 0-IIa+Is non-granular lesions had a substantially higher risk of covert T1 CRC (OR 22.5, 95% CI 7.07–71.6 and OR 14.4, 95% CI 4.53–45.5, respectively).¹⁶ More recently, O’Sullivan et al. assessed a similar cohort of Australian patients (n = 2451 LNPCPs) and reaffirmed that depressed (0-IIc) and nodular (0-IIs or IIa-Is) lesions had a far higher risk of T1 CRC when compared with flat lesions (OR 35.7, 95% CI 22.6–56.5 vs. 3.5, 95% CI 2.6–4.9).¹¹ Similar to the data described by D’Amico et al., O’Sullivan et al. also reported that the rectosigmoid location (vs. proximal location) and polyp size ≥4 cm were both associated with an increased risk of T1 CRC (OR 3.19, 95% CI 2.47–4.13 and OR 1.75, 95% CI 1.35–2.26, respectively).^11,17

Colonic lesion size, location and/or morphology alone do not have enough discriminant value to reliably predict risk of T1 CRC. Endoscopic surface pattern assessment is of equal importance and can increase the accuracy of colorectal lesion diagnosis. There are 3 primary endoscopic surface pattern classifications: (A) NBI International Colorectal Endoscopic (NICE) classification,^18,19 (B) Japanese NBI expert team (JNET)²⁰ and (C) Kudo pit pattern classification²¹:

• (A) NICE classification (see Table 1; does not require magnification endoscopy or dye):

  • (1) Type 1 (serrated class: either hyperplastic or sessile serrated polyp).

  • (2) Type 2 (adenoma).

  • (3) Type 3 (deep T1 CRC and overt T1 CRC).

• (B) JNET classification (see Table 2; requires magnification endoscopy):

  • (1) JNET 1 (same as NICE Type 1: either hyperplastic or sessile serrated polyp).

  • (2) JNET 2A (regular vessel and surface pattern: indicative of low-grade dysplasia).

  • (3) JNET 2B (variable vessel calibre, irregular vessel and surface pattern: indicative of high-grade dysplasia or superficial T1 CRC).

  • (4) JNET 3 (same as NICE Type 3: deep T1 CRC and overt T1 CRC)

• (C) Kudo pit pattern classification (see Table 3; requires dye and magnification endoscopy) describes 6 polyp pit patterns:

  • (1) Type 1: normal colonic mucosa.

  • (2) Type 2: asteroid appearance (= hyperplastic polyp).

  • (3) Type IIIS: tubular or round pit smaller than normal pit (= tubular adenoma).

  • (4) Type IIIL: tubular or round pit larger than normal pit (= tubular adenoma).

  • (5) Type IV: gyrus/dendritic (= tubulovillous or villous adenoma).

  • (6) Type VI: irregular arrangement (= neoplastic or invasive).

  • (7) Type Vn: loss or decrease of pits with amorphous structure (= neoplastic or invasive).

Table 1.

NBI International Colorectal Endoscopic (NICE) classification^11,12.

Variable	Type 1	Type 2	Type 3
Colour	Same or lighter than the background	Brown relative to background	Brown or black relative to background
Vessels	None or isolated lacy vessels	Brown vessels surrounding white structures	Had areas of disrupted or missing vessels
Surface pattern	Dark or white spots of uniform size	Oval, tubular, or branched white structures	Amorphous or absence of pattern
Most likely histology	Hyperplastic or serrated polyps (sessile serrated polyp)	Adenoma to superficial T1 CRC	Deep T1 CRC

Table 2.

Japanese Narrow Band Imaging Expert Team Classification¹³.

Characteristics	Colours	JNET 2A	JNET 2B	JNET 3
Vessel pattern	Invisible	Regular calibre Regular distribution	Variable calibre, irregular distribution	Loose vessel areas, interruption of thick vessels
Surface pattern	Regular dark or white spots similar to surrounding mucosa	Regular tubular or branched or papillary	Irregular or obscure	Amorphous areas
Most likely histology	Hyperplastic polyp or sessile serrated polyp	Low-grade dysplasia	High-grade dysplasia or superficial T1 CRC	Deep T1 CRC

Table 3.

Kudo Pit Pattern Classification¹⁴.

Types	Features	Interpretation
I	Round, normal	Normal
II	Asteroid	Hyperplastic
IIIS	Tubular or round pit smaller than normal pit	Tubular adenoma
IIIL	Tubular or round pit larger than normal pit	Tubular adenoma
IV	Gyrus/dendritic	Tubulovillous or villous adenoma
VI	Irregular arrangement	Neoplastic
VN	Loss or decrease of pits with amorphous structure	Neoplastic

Colonic lesions with a JNET 2B surface pattern or Kudo pit pattern type VI carry a higher risk of superficial T1 CRC (features of covert T1 CRC). As such, these lesions should be considered for en-bloc resection irrespective of lesion location, size, or morphology.²⁰ If the colonic polyp is assessed to be NICE Type 3 or have a JNET 3 surface pattern or Kudo pit pattern type Vn pattern (features of overt T1 CRC), a multidisciplinary team review should be undertaken to discuss the benefit of endoscopic versus surgical resection.

Traditionally, the accuracy of optical diagnosis for T1 CRC was thought to be suboptimal. Recently, endoscopic surface pattern assessment of flat (0-IIa) LNPCPs was deemed to be highly accurate.²² Vosko et al., in a cohort of 1583 patients, noted that in Paris 0-IIa LNPCPs, the sensitivity and specificity for predicting cancer was 91% and 96%, respectively. Conversely, surface pattern assessment for T1 CRC in nodular lesions was found to be less accurate (sensitivity 53%, specificity 94%, missed T1 CRC 6%).²² Therefore, in nodular lesions, overall morphology and location are increasingly important. As mentioned above, nodular lesions in the rectosigmoid location carry a higher risk of covert T1 CRC and as such, Paris 0-Is or LST-MG lesions in the left colon should be considered for en-bloc resection irrespective of surface pattern.

Given the complex interaction of LNPCP characteristics when predicting T1 CRC risk, algorithms have been designed with the hope of improving endoscopist identification of covert and overt T1 CRC. A promising algorithm is the Colorectal Neoplasia Endoscopic Classification to Choose the Treatment (CONECCT) classification.²³ The CONECCT classification combines covert (macroscopic features) and overt signs of carcinoma (irregular pit or vascular patterns). Both the sensitivity and negative predictive values for T1 CRC were impressively 100%. Therefore, the use of this classification may decrease unnecessary surgery and non-curative piecemeal endoscopic mucosal resection (pEMR). However, the specificity and positive predictive value were both low (26.2% and 11.6%, respectively), and therefore equally this may lead to many ESDs being performed for benign lesions. Importantly, the CONECCT classification does not include location of lesion, which remains an important consideration.

Recently, O’Sullivan et al. also proposed a simplified T1 CRC decision-making algorithm.¹¹ In this Australian cohort, there was a high prevalence of T1 CRC in depressed LNPCPs (62%). It was proposed that in the absence of surface features of deep T1 CRC (and irrespective of location), en-bloc resection should always be performed for this lesion phenotype. For flat LNPCPs, the prevalence of T1 CRC was unsurprisingly low (4.3% granular, 1.8% non-granular) and the algorithm suggests that pEMR is safe and suitable for these lesions irrespective of granularity or location. For nodular lesions, location is proposed as the primary discriminatory variable, as those in the rectosigmoid location carry a high risk of covert T1 CRC and therefore should always be considered for en-bloc resection. Comparatively, for those in the proximal colon, nodular non-granular LNPCPs carry a high risk of T1 CRC (20%), while nodular granular LNPCPs carry a low risk (5%). Therefore, nodular granular LNPCPs are appropriate for pEMR whilst nodular non-granular LNPCPs should undergo en-bloc resection.

In summary, endoscopic features associated with an increased risk of superficial T1 CRC (covert T1 CRC) in the absence of features concerning for deep T1 CRC (overt T1 CRC) should be resected en-bloc. However, for en-bloc resection of colorectal lesions >20 mm, ESD is required. Ultimately, endoscopists should be aware of the endoscopic features associated with superficial T1 CRC and the rationale for en-bloc resection. When identified in the community or in a non-expert centre, these patients should be referred to a dedicated centre with appropriate endoscopic and colorectal surgical expertise.

Standard techniques for endoscopic resection

Endoscopic mucosal resection (EMR)

Endoscopic mucosal resection (EMR) is a highly efficient and refined procedure, one that has been honed over the past decade to become the most embraced primary modality for LNPCP resection, with a striking 2.48% of all colonoscopies in the United States now incorporating EMR.²⁴ Noteworthy advancements, including the adept recognition and management of deep mural injury (DMI),^25,26 alongside innovations such as mechanical closure of post-EMR defects in the proximal colon to mitigate the risk of clinically significant post-EMR bleeding,²⁷ have solidified EMR’s position as a safer and more economical option compared to ESD and surgical interventions.^5,28–31

The technique entails several key steps: (1) a submucosal injection of a saline/colloid solution and blue dye, often with diluted epinephrine (referred to as chromo-injectate), (2) systematic snare-based tissue resection, including a 2- to 3-mm margin of normal tissue, (3) transection of the captured tissue in 1–3 pulses of fractionated current, (4) irrigation of the defect post-resection for better identification of residual polypoid tissue and DMI, (5) subsequent resections following the submucosal plane, aligning the snare with the advancing mucosal defect edge and including a 2- to 3-mm margin of submucosal tissue, and (6) repeated submucosal injections every 2–4 resections until complete removal of polypoid tissue is achieved (Figure 1).^32–37

Figure 1.

Endoscopic mucosal resection (EMR). (A) Granular, Paris IIa, JNET IIa LNPCP in the tranverse colon. (B) No concerning areas noted on NBI. (C) Submucosal injection performed to lift the lesion. (D) Post-EMR defect. (E) Post-EMR defect following STSC to the margin. (F) Endoscopic clip closure.

Due to its similarities with cold-snare polypectomy (CSP), EMR is easily learned and enables rapid piecemeal resection of extensive LNPCPs, making it particularly suited for adenomatous lesions devoid of overt T1 CRC or with a low risk of covert T1 CRC. Additionally, for LNPCPs measuring up to 25 mm, EMR may facilitate en-bloc R0 excision, serving as an alternative to ESD for such lesions. However, in these cases, ensuring a clear vertical margin typically necessitates a large submucosal injection, employment of a 20 mm or 25 mm snare, and carries a heightened risk of significant DMI (Type III-IV).^25,26

Current guidelines also advocate for thermal ablation of the resection margin to reduce the risk of recurrence to 2%–5%.³⁸ This may be achieved with snare-tip soft coagulation (STSC) to create a 3–5 mm rim of ablated tissue (Erbe VIO Soft Coag: 80W, Effect 4; Erbe, Tubingen, Germany). Moreover, for non-lifting or fibrotic areas, adjunctive techniques such as cold avulsion with adjuvant snare-tip soft coagulation (CAST) or hot avulsion can be employed.³⁹ Collectively, these techniques have significantly enhanced the success of EMR for complex lesions, including those previously attempted,⁴⁰ as well as those situated at challenging locations like the anorectal junction⁴¹ or ileocecal valve.⁴²

Endoscopic submucosal dissection (ESD)

Endoscopic Submucosal Dissection (ESD) represents a significant stride in endoscopic innovation, originally conceived as a curative approach for early gastric cancer, yet readily adapted to the colorectum. Theoretically, the size of an LNPCP does not preclude the possibility of achieving an en-bloc R0 excision through ESD. This technique is ideal for lesions exhibiting overt features of T1 CRC or those harboring a heightened risk of covert T1 CRC, offering the potential for organ-sparing, curative resection without undue over-treatment of benign lesions.

While ESD encompasses various iterations, fundamental principles consist of (1) utilization of a distal cap attachment to facilitate access to the submucosal plane, (2) a generous submucosal injection of chromo-injectate to elevate the lesion and expand the submucosal layer, (3) creation of a mucosal incision, and (4) meticulous dissection beneath the lesion within the submucosal plane using a slender (0.3-mm diameter) electrosurgical knife until en-bloc removal is achieved. In certain scenarios, dissection may be performed in a retroflexed position, leveraging enhanced stability and optimizing endoscope alignment parallel to the cutting plane, particularly in rectal procedures (Figure 2).

Figure 2.

Endoscopic submucosal dissection (ESD). (A) Granular Paris IIa+Is, JNET IIa with focal IIb LNPCP in the rectum. (B) Disruption of pit and vascular pattern on NBI. (C) Lesion marked. (D) Submucosal dissection. (E) Final post-ESD defect. (F) Specimen ex-vivo.

Several technical refinements have emerged to bolster safety and efficiency in ESD. The adoption of a variety of traction methods, such as rubber bands,⁴³ multi-loop devices,⁴⁴ adaptive traction devices,⁴⁵ and clips with snares or thread, improve the speed of submucosal dissection and reduce the risk of muscle injury or perforation. Innovative approaches to ESD, such as tunnelling or pocket creation are also effective in select lesions.^46,47 Paramount to the procedure is the meticulous management of bleeding, especially concerning large submucosal vessels, notably prevalent in the rectum. Vigilant control of bleeding is imperative, as it can hinder visualization, exacerbating the risk of muscle or further vessel injury.

EMR versus ESD

The main consideration between EMR and ESD, comes down to the need for an en-bloc R0 excision, with the objective being complete removal and precise pathological evaluation.⁴⁸ One of the key benefits of en-bloc resection is that it reduces the risk of incomplete removal of the lesion or incomplete retrieval of all fragments for analysis.⁴⁸ This is of particular importance for lesions with suspected T1 CRC.

ESD, compared with EMR, has a higher risk of complications including perforation and bleeding. It is also more costly related to more frequent post-procedural hospitalizations. To optimize patient care and resection outcomes, ESD should be performed in a tertiary centre with local expertise and a collaborative colorectal surgical department. It is well documented that the risk of perforation is associated with endoscopist experience (1.4-4.2% vs. 20.4%) and failure of en-bloc resection is also seen in less experienced endoscopists (OR 3.0, 95% CI 1.7–1.8).^49,50 The opportunity to obtain expert ESD training in North America is however lacking. There are currently no standardized guidelines for colorectal ESD training nor any recognized credentialing programs. For safe and effective colorectal ESD to be performed, improved access to training is integral in North America.

The utility of an en-bloc resection for benign lesions is less clear. A recent randomized controlled trial (RCT) conducted by a seasoned team in France undertook a comparative analysis of EMR and ESD for LNPCPs measuring 25 mm or larger, with the exclusion of rectal lesions.⁵¹ While the recurrence rates were significantly lower in the ESD cohort (0.6%) compared to EMR (5.1%), the absolute rates in both groups were small. Furthermore, recurrences typically manifest as diminutive, low-grade lesions that are readily manageable endoscopically, often with adjunctive techniques such as CAST.⁵² While there was no difference in the rates of surgical intervention, ESD typically took at least three times longer than EMR, and carried a higher risk of adverse events, such as perforation, delayed bleeding, and post-polypectomy syndrome (35.6% vs. 24.5%).⁵¹ Furthermore, in contexts where endoscopy waitlists are extensive, the substantial investment of resources and time in ESD may be deemed inefficient when EMR outcomes remain equivalent.

The opportunity cost associated with protracted ESD procedures, often requiring hospital admission, may inadvertently impede endoscopy access for others, an essential consideration in a global milieu urging optimal resource utilization. Given the predominance of benign lesions in the colon outside the rectum, EMR should persist as the primary choice. It offers a safe, efficient, cost-effective, outpatient-centred approach, effectively mitigating opportunity costs. Nevertheless, for select cases, ESD is an invaluable asset essential for optimizing patient outcomes. Lesions exhibiting endoscopic evidence of superficial T1 CRC or those harboring a high risk for covert T1 CRC, particularly bulky rectal lesions, carry a substantial pretest probability of potentially curable cancer. Thus, ensuring their en-bloc R0 resection through techniques like ESD becomes imperative.⁵³

Cold-snare polypectomy (CSP)

Cold-snare polypectomy (CSP) represents an approach to tissue transection devoid of electrocautery. Employing specialized stiff thin-wire snares (with a wire diameter of ≤0.3 mm), CSP harnesses shearing mechanical forces during snare closure to achieve precise excision. It can be safely executed with or without submucosal injection, although injectate administration can aid in delineating lesion margins. The incorporation of dilute epinephrine serves to mitigate intraprocedural bleeding, facilitating the differentiation between residual polypoid tissue and the expanding resection defect. Echoing the principles of EMR, CSP entails the careful positioning of the snare over the target area, encompassing a healthy margin of surrounding tissue. Upon closure of the snare handle to the point of resistance, the target tissue is captured, followed by complete transection upon confirming appropriate tissue capture. In cases where tissue transection proves challenging, the endoscopist may apply traction against the endoscope by gently pulling on the snare catheter or opt for snare placement revision.

Post-transection, irrigation of the resection defect aids in submucosal expansion and facilitates thorough defect inspection. Oozing from the resection site is anticipated and typically requires no intervention. Cold-snare protrusions, comprising compressed submucosa and muscularis mucosae, are benign entities devoid of adverse outcomes or recurrence implications. Given its shallower depth of resection, CSP finds optimal utility in managing sessile serrated lesions (SSLs), with safe treatment attainable through this approach.⁵⁴ A recent meta-analysis found that CSP for SSLs carried a technical success rate of 100%, adverse event rate of 2%, and residual or recurrence rate of 3.1%.⁵⁵

When comparing CSP to EMR for LNPCPs, a recent multicentre RCT from Germany found lower rates of adverse events in the CSP group (1% vs 7.9%, P = .001). However, there was a greater risk of recurrence (23.7% vs 13.8%, P = .020). In multivariable analyses, a lesion diameter of >4cm was an independent predictor for residual adenoma (OR 2.47).⁵⁶ This is consistent with modelling data, which shows that a 40 mm polyp has a risk of incomplete resection between 40.74% and 60.60%.⁵⁷ While the results of this RCT are promising, generalizability is limited due to only 30.5% of EMR patients undergoing thermal ablation of the margin, and 36.8% of lesions being SSLs. More recently, an RCT from Australia which enrolled adenomas and where STSC was performed to the EMR-defect margin, found significantly lower rates of recurrence with EMR than CSP (1.1% vs 18.4%, P < .001). As expected, rates of post-EMR bleeding were higher with EMR (7.8% vs 1.1%, P = .034).⁵⁸ While these RCTs have underscored the superior safety profile of CSP in comparison to EMR, they have also brought to light a significant drawback: a markedly higher risk of recurrence, rendering its application for adenomatous lesions unjustifiable.

Modified techniques for endoscopic resection

Hybrid EMR/ESD

Hybrid EMR/ESD procedures blend components of EMR and ESD components to either achieve an en-bloc resection or aid in piecemeal resection amidst significant submucosal fibrosis. In the case of LNPCPs sized between 25 and 30 mm, hybrid EMR may facilitate quick en-bloc R0 excision as compared to ESD. This involves performing a circumferential mucosal incision around the lesion following chromo-injectate submucosal injection. Once the incision is completed, the lesion often contracts towards the centre. This facilitates direct snare placement within the submucosal plane surrounding the lesion and enables tissue capture beneath the lesion. Piecemeal hybrid EMR may also be beneficial for larger lesions with fibrosis, such as those in a background of ulcerative colitis. In a similar fashion, a circumferential mucosal incision exposes the submucosal plane around the entire LNPCP, facilitating snare placement and resection, albeit in this case in a piecemeal fashion (Figure 3). Hybrid ESD primarily refers to salvage completion EMR in the event that ESD is unable to be completed, often due to severe fibrosis or poor access.⁵⁹ In a recent meta-analysis, Hybrid ESD was still able to achieve an en-bloc resection rate of 81.6%. Complications, recurrences, and need for surgery occurred in 7.7 %, 4.5%, and 3.6% respectively.

Figure 3.

Hybrid and underwater EMR of a dysplastic lesion in a patient with ulcerative colitis. (A) Non-granular, Paris IIb, JNET IIa descending colon LNPCP. (B) Same lesion visualized with NBI. (C) Lesion marked with diathermic markings then submucosal injection performed. (D) Circumferential incision using the tip of the snare. (E) Resection with underwater EMR. (F) Final defect following hybrid/underwater EMR.

Underwater EMR

This technique involves the instillation of saline to cover the lesion and/or associated colonic segment. Due to the ensuing buoyancy, LNPCPs float away from the muscularis propria. By doing so, this approach enables the en-bloc R0 excision of LNPCPs up to 30 mm in size, without elevating the risk of DMI. In a recent meta-analysis of 7 RCTs with 1458 patients (Underwater-EMR: 739, Conventional-EMR: 719), the pooled rate of en-bloc resection was significantly higher with U-EMR vs C-EMR (70.2% vs 58.1%). R0 resection rates were also higher with U-EMR (58.1% vs 44.6%). While there was no difference in perforation or bleeding, recurrence at surveillance colonoscopy was significantly lower with U-EMR than with C-EMR (RR 0.62).⁶⁰ However, these studies did not utilize thermal ablation of the margin as part of C-EMR, and therefore were not comparing U-EMR to the accepted standard of care (EMR with thermal ablation of the margin).

Endoscopic intermuscular dissection (EID)

Endoscopic intermuscular dissection (EID) entails dissection between the circular and longitudinal muscularis propria layers, primarily performed in the rectum below the peritoneal reflection. It can be utilized in patients with extensive fibrosis or with suspected deep T1 CRC who are poor surgical candidates.⁶¹ In a cohort study of 67 patients, the rates of overall technical success, R0 resection, and curative resection were 96%, 81%, and 45%, respectively. Only minor adverse events occurred in 8 patients (12 %).⁶² Although the rate of curative resection was high (45%), their definition included all T1 lesions, not just those <1000 μm in depth into the submucosa. This is a controversial area, with Japanese guidelines reporting a depth of submucosal invasion ≥1000 μm as a risk factor prompting surgery.⁶³ Furthermore, ESGE considers sm²/sm³ tumours (≥1000 μm) as high-risk lesions. However, they suggest that following an en-bloc R0 resection of a rectal lesion meeting the single high-risk criterion of submucosal invasion deeper than 1000 μm, options other than surgery could be discussed.⁶⁴ While further studies are needed to demonstrate its utility in this setting, EID may emerge as a definitive procedure for frail and elderly T1 CRC patients who are non-operative candidates.

Endoscopic full-thickness resection (EFTR)

Endoscopic full-thickness resection (EFTR) methods encompass various techniques, including intentional full-thickness EMR, ESD extending into full-thickness defect, or use of a full-thickness resection device (FTRD). These are typically utilized in the rectum, below the peritoneal reflection, where closure is typically unnecessary. In other (intraperitoneal) colonic locations, closure with clips or endoscopic suturing devices is necessary. FTRD is often limited to lesions close to 15 mm in size. It involves suctioning or pulling the lesion into a cap, including the muscularis propria, followed by clip closure and snare resection over the clip. Similar to EID, it is reserved for cases with suspected deeper T1 CRC or severe fibrosis. Overall clinical success has been shown to be up to 91%, technical success 83%, and R0 excision 81%.⁶⁵

Important considerations when planning and performing endoscopic resection

The decision between piecemeal and en-bloc resection strategies may also be intricately influenced by various lesion-related characteristics, including challenging access to the LNPCP. This may be due to endoscope-related issues (eg, looping or angulation) or certain anatomical sites (eg, anorectal junction, ileocecal valve, appendiceal orifice, or peri-anastomosis). Although these factors may render ESD unsafe or infeasible, they also confer a higher risk of technical failure, adverse events, and adenoma recurrence with EMR, as highlighted by the SMSA score, which considers lesion access and site among other factors.⁶⁶

Access

Poor access often complicates attempts at endoscopic resection. Techniques aimed at enhancing access include patient repositioning (left lateral, supine, right lateral, prone), and abdominal pressure application. Changing to a gastroscope may improve flexibility and passage through tight angulations, facilitate retroflexion, and enables devices to exit the channel at a different orientation than a colonoscope. Other methods to improve access include instillation of saline to elevate a lesion, utilization of traction (see above), or use of commercially available overtube devices such as Lumendi⁶⁷ or Pathfinder⁶⁸ to prevent looping and provide scope stability.

Complex anatomical locations

While EMR has demonstrated efficacy at anatomically challenging locations such as the anorectal junction,⁴¹ ileocecal valve,⁴² appendiceal orifice,⁶⁹ and anastomoses,⁷⁰ data on ESD is still emerging. Regardless, special considerations are warranted when approaching LNPCPs at these sites.

Anorectal junction (ARJ)

The ARJ is located within 2 cm of the pectineal line.⁷¹ Endoscopic resection at the ARJ is technically challenging due to the presence of distinctive anatomic and physiologic characteristics,⁷² including unique innervation and lymphovascular supply secondary to a rectal venous plexus. Thus, specific measures are required to limit pain and bleeding from resecting over the dentate line and haemorrhoidal vessels. There is also a theoretical risk of systemic bacteraemia because of direct drainage into the systemic circulation.⁷³ Surgery is ideally avoided in this area due to the risk of impairing anal function, which can significantly influence the patient’s quality of life.⁷⁴ The chromo-injectate in this location should include a local anaesthetic to mitigate peri and post-procedural anorectal pain. Another consideration is the risk of stricture formation due to the small luminal diameter in this area, reaching up to 74.2% in cases with an ER defect ≥90% of circumference.⁷⁵ In these cases, topical nitrates should be considered to reduce anal spasms, steroid to reduce the risk of fibrosis, and facilitation of early endoscopic dilation is key. Despite the challenges, the ARJ location should not preclude either EMR or ESD, especially considering the feasibility of retroflexed views that offer endoscope stability.

Ileocecal valve (ICV)

Lesions involving the ICV may extend into the terminal ileum, necessitating meticulous assessment due to difficulties in distinguishing adenomatous tissue from normal villiform ileal mucosa. Employing a distal cap attachment and initiating resection within the terminal ileum are favourable approaches. In cases of ileal infiltration, resection should begin within the ileum, followed by the anterior and posterior angles of the ICV lips. A limited submucosal injection is recommended in the distal terminal ileum and lips of the ICV because overexpansion of the submucosa may obstruct the lumen and cause the snare to “bounce” off the surface of the valve. A small (10- or 15-mm diameter), stiff, thin-wire snare is preferred at this location because of the relatively small luminal diameter. At the end of the resection, thermal ablation to mitigate the risk of recurrence should be considered, however, the ileal margin avoided due to the risk of stricture formation. When performed systematically in this manner, the technical success of EMR in this region can be upwards of 93.9%, and recurrence as low as (4.6%).⁴² Although these are not Western studies, a Japanese group has reported their outcomes for ESD performed for LNPCPs involving the ICV.^76–78 They demonstrated an en-bloc resection rate of 95% and R0 excision rate of 89%.⁷⁸ In cases of terminal ileal involvement, a hybrid approach can also be utilized, as a mucosal incision performed with a snare-tip or ESD knife within the terminal ileum, can aid in margin orientation and facilitate polypoid tissue protrusion out into the caecum.

Appendiceal orifice (AO)

Adenomatous lesions of the appendiceal orifice may be effectively managed with EMR. An underwater approach is often preferable to facilitate AO opening and easing EMR execution.⁷⁹ In one study assessing outcomes of EMR at the AO, complete clearance of visible adenoma was achieved in 92.6% of cases. Reassuringly, there were no cases of post-EMR appendicitis. Of these patients, 91% avoided surgery to longest follow-up. Prior appendectomy status assessment is crucial for AO lesions. In cases without a prior appendectomy, where the distal margin LNPCP cannot be cannot be visualized or if more than 50% of the circumference of the orifice is involved, surgery (appendectomy) should be considered.⁶⁹ In cases with a prior appendectomy, ESD may be an alternative, as it facilitates a deeper resection, with studies from Japan demonstrating curative resection rates of 70.4%.⁸⁰

Anastomoses

Anastomotic lesions pose challenges due to associated scarring and fibrosis. Adjunctive techniques are frequently required for these lesions (see below). In a small case series from Australia, peri anastomotic LNPCPs were removed successfully by EMR, but with the majority (90%) requiring CAST. While DMI type II was frequent (40%), there were no cases of DMI III-V, nor any series adverse events. Reassuringly, at 6-month surveillance, there was no recurrence.⁷⁰

Addressing submucosal fibrosis

Submucosal fibrosis significantly impacts resection outcomes across all ER modalities. Prior resection attempts or the presence of tattoo should raise suspicion of fibrosis,⁸¹ necessitating adjunctive techniques for complete resection.

Hot avulsion and CAST

Hot avulsion utilizes biopsy forceps with electrocautery to remove residual adenomatous tissue that is not amenable to snare capture whilst providing thermal ablation at the time of avulsion. In 2014, Veerappan et al reported on the safety and efficacy of hot avulsion in adjunct to snare polypectomy for non-lifting colonic polyps in 20 patients.⁸² In this cohort, the rate of recurrence was reported to be 15%, however, recurrences were all successfully treated with repeat hot avulsion. In a more recent study of hot avulsion, the rates of recurrence (17.52%) were similar to a cohort of patients that did not require hot avulsion post-EMR (16.02%; P = 0.76).⁸³ While rates of recurrence appear high, this was performed in an era where thermal ablation to the post-EMR defect was not routinely performed. More recently, CAST became an alternative to hot avulsion and involves cold-avulsion of residual adenomatous tissue with a biopsy forceps, followed by snare-tip soft coagulation of the avulsed area. Although it may be laborious and best reserved for small areas of residual adenoma, it has been reported to be highly effective. For instance, in a recent study of previously attempted lesions, CAST was employed in 46.2% of cases, with overall EMR technical success reaching 95.6% and recurrence rate of 7.8%.⁴⁰ It is important to note that the process of CAST renders it difficult to assess the underlying muscle for injury (DMI type II) and thus any such areas should be managed with targeted clip closure.⁸⁴ Further studies comparing CAST and hot avulsion are required to determine the most optimal technique for the management of small areas of residual adenoma.

Submucosal release

In scenarios featuring submucosal fibrosis with substantial residual polypoid tissue, CAST, and hot avulsion may be cumbersome. Hence, a submucosal release can be performed to facilitate snare capture. The method involves using a cutting current with the snare-tip to dissect through fibrotic bridges. This works by providing access to the submucosal plane adjacent to the residual tissue, and ultimately facilitates snare-capture. It is important to ensure that an additional submucosal lift with chromo-injectate is performed to avoid inadvertent DMI. While this technique is increasingly being utilized, there is no literature highlighting its effectiveness.

Surveillance and follow-up

After EMR and ESD, surveillance typically entails serial assessments at 6 months (SC1), 18 months (SC2), and 36 months (SC3) post-resection. ESD exhibits near-zero recurrence rates, whereas the recurrence rates following EMR with thermal margin ablation are around 5%. The authors of a recent EMR versus ESD RCT suggested to initiate surveillance colonoscopy at 36 months (SC3) post-ESD, given the minimal recurrence risk.⁵¹ Similarly, a recent study from Australia shows that it may be possible to skip to the 36-month surveillance (SC3) post-EMR if SC1 reveals no recurrence, as a clear scar at SC1 is durable at SC2.⁸⁵ We suggest that SC1 remains vital post-ESD, as up to 16% of patients will have synchronous disease, which is often overlooked or not attended due during the index EMR or ESD.⁸⁶ This underscores the importance of a thorough assessment via the initial 6-month surveillance colonoscopy (SC1).

Recurrence management, typically modest upon SC1 assessment, warrants similar endoscopic resection techniques as delineated in this review. Modalities typically utilized include EMR with adjunct techniques such as CAST, hot avulsion, and submucosal release. CSP, due to its superficial depth of resection, is often inadequate, while ESD is frequently unnecessary. Recent studies have demonstrated that recurrence can be managed in a single session, with subsequent recurrence rates remaining low.⁵²

In instances of non-curative resections, patients typically require surgical intervention and/or adjuvant chemo-radiotherapy. Comprehensive assessment through imaging modalities such as a CT CAP and MRI pelvis become imperative to evaluate for loco-regional and distant disease extent. Comparatively, in non-operative candidates, a conservative endoscopic surveillance approach may be utilized. In a recent meta-analysis of 2961 T1 CRC patients, disease-specific survival was similar at 5 years.⁸⁷ Thus, in comorbid non-operative candidates, a non-curative resection may not influence survival.

Supplementary material

Supplementary material is available at Journal of the Canadian Association of Gastroenterology online.

Author contributions

Study concept and design: Jeffrey D Mosko. Drafting of the manuscript: Sunil Gupta, Tony He. Critical revision of the manuscript for important intellectual content: Sunil Gupta, Tony He, Jeffrey D. Mosko.

Supplement sponsorship

This article appears as part of the supplement “27th Anniversary Key Topics in Gastroenterology in 2024.” The symposium and this supplement were funded by grants from the following sponsors:

• Platinum: Abbvie Canada, Janssen Inc, Pfizer Canada, Takeda Canada, Fresenius-Kabi

• Silver: Boston Scientific, Ferring Pharmaceuticals, Organon, Eli Lilly and Company

Conflicts of Interest

S.G. and T.E.: No conflicts of interest. J.D.M.: Speaker Honouraria/Consulting Fees from Boston Scientific, Pendopharm, Medtronic, Fuji.

In addition to this COI statement, ICMJE disclosure forms have been collected for all co-authors and can be accessed as supplementary material.

Data availability

There are no data associated with this review.