Management of T1 Colorectal Cancer

Colorectal cancer (CRC) is a common carcinoma, and its early detection and diagnosis are becoming possible with the widespread use of colonoscopy. Surgical intestinal resection (SR) with lymph node dissection is the standard treatment for CRC invading the muscularis propria (T2) or more. Intramucosal CRC, which reportedly has no lymph node metastasis (LNM), can be treated using endoscopic resection (ER) alone. In contrast, approximately 10% of patients with submucosal invasive (T1) CRC have LNM [1‒4], indicating that SR is necessary for approximately 10% of patients with T1 CRC, while local excision, such as ER, is sufficient for the other 90%.

Advances in endoscopic equipment and techniques have made complete ER for T1 CRC possible [5‒16]. Consequently, narrowing down the list of T1 CRCs with a low risk of LNM that can be treated using ER alone and reducing the number of unnecessary surgeries that are physically burdensome is necessary, considering the aging population worldwide and the trend toward minimally invasive treatment. This review discusses the management of T1 CRC by highlighting the latest findings related to LNM prediction, treatment methods, and prognosis.

Although LNM is observed in approximately 10% of patients with T1 CRCs as a whole [1‒4], its risk can be stratified using various histopathological factors. Many studies have reported that lymphovascular invasion and histological tumor grade are associated with LNM [4]. Recently, the association with tumor budding was also recognized [17], and the International Tumor Budding Consensus Conference has recommended that tumor budding should be included in guidelines and pathological reports, particularly in stage II and pT1 CRCs, since it is an independent predictor of LNM [18]. A recent meta-analysis including 16 observational studies with 10,181 patients, among whom 1,307 had LNMs, reviewed the histopathological risk factors for LNM [19]. Lymphovascular invasion (odds ratio [OR]: 7.42, p < 0.001), tumor budding (OR: 4.00, p < 0.001), submucosal invasion depth ≥1,000 μm (OR: 3.53, p < 0.001), high tumor grade (OR: 3.75, p < 0.001), polypoid growth pattern (OR: 1.59, p = 0.040), and rectal location of tumor (OR: 1.36, p = 0.003) were associated with LNMs. Although the risk of LNM and indications for surgery were determined based on these factors according to the current guidelines [1, 20, 21], their specificity remains unsatisfactory.

Recently, a nomogram for predicting LNM in T1 CRC was developed based on real-world data, including the risk factors described in the Japanese Society for Cancer of the Colon and Rectum (JSCCR) guidelines [22]. Artificial intelligence (AI) systems for predicting LNM in T1 CRC have also been developed, including a machine-learning artificial neural network model using data on histopathological findings, characteristics of patients, and tumors [23, 24]. These new tools showed a diagnostic performance that exceeded the current guidelines under the same conditions; therefore, they are expected to improve specificity in the diagnosis of LNM in T1 CRC.

Although conventional guideline-based diagnosis and these novel AI systems employ histopathological findings diagnosed by human pathologists, interobserver disagreement among pathologists exists in the diagnosis of each finding, which is problematic in terms of consistency and accuracy. Additional immunohistochemical staining, such as D2-40 for lymphatic invasion and Victoria blue or Elastica Van Gieson staining for vascular invasion, may improve the diagnosis of lymphovascular invasion [25]. A systematic review showed that the diagnostic OR of additional immunohistochemical staining for LNM was higher than that of hematoxylin and eosin staining alone in multivariate analysis (5.95 vs. 1.89, p = 0.01) [26]. Pooled κ values of lymphovascular invasion using hematoxylin and eosin and additional immunohistochemical staining were 0.37 and 0.62, respectively. Furthermore, a “pathologist-independent” AI system, which is a diagnostic system constructed by performing deep learning on scanned whole-slide images without any conventional histological risk assessment, has also been developed [27‒29].

Regarding tumor location, rectal T1 cancer is more malignant than colonic T1 cancer. The recurrence rate after SR with lymph node dissection for T1 cancer was reported to be 4.2–4.5% in the rectum, which is higher than 1.5–1.9% in the colon [2, 3]. Although no significant difference was found in the local recurrence rate (0.6 vs. 0.3%, p = 0.52), the frequency of distant recurrence was higher in the rectum than in the colon (4.5 vs. 1.6%, p < 0.05) [3].

Regarding the macroscopic tumor type, the frequencies of LNM and recurrence of pedunculated type T1 CRC are lower than those of non-pedunculated type T1 CRC [30‒32]. Malignancy differs between head and stalk invasion among pedunculated type T1 CRCs. LNM incidence was significantly lower in head invasion cases than in stalk invasion cases (0.0 vs. 6.2%, p = 0.02) [32]. However, LNM or recurrence can occur in patients with head invasion of pedunculated type T1 CRC if any risk factors of LNM, such as lymphovascular invasion, are positive, as well as stalk invasion of pedunculated type T1 CRC [4, 31, 32].

Endoscopic findings under white-light endoscopy or chromoendoscopy without magnification suggestive of deep T1 CRC have been reported. Regarding the polypoid type, the characteristics of deep T1 CRC include an expansion appearance (protrusion and overextension of the tumor and/or surrounding normal mucosa, resembling a submucosal tumor), tumor stiffness or unevenness in the comprehensive view, coarse surface findings (surface roughness) in the surface property, converging folds toward the tumor (two or more mucosal folds converging toward the tumor), poor extension of the surrounding colonic wall, and stiffness or deformity of the colonic lumen among properties of the tumor surroundings. Regarding the flat and depressed types, an expansion appearance, tumor stiffness or unevenness in the comprehensive view, protrusion on the depression surface, uneven depression surface, strong redness in the surface property, converging folds toward the tumor, colonic wall deformity, stiffness of the colonic lumen, table-like protrusion in the tumor surroundings, and negative air deformity (tumor deformity not observed as air decreasing by the aggregated submucosal cancer mass) are characteristics of deep T1 CRC. However, the accuracy of deep T1 CRC diagnosis based on these findings was 74.7%, even when diagnosed by Japanese experts [33]. Magnifying chromoendoscopy and narrow-band imaging yielded higher sensitivity for the diagnosis of T1 and deep T1 CRCs than white-light endoscopy or chromoendoscopy based on macroscopic morphological features [34]. The type V_N pit pattern and Japan Narrow-band imaging Expert Team classification type 3 are good predictive indicators of deep T1 CRC [35]. Although these endoscopic observation techniques cannot assess the condition of the deepest part of the tumor invading the submucosa, endoscopic ultrasonography enables objective visualization of tumor invasion based on the histological structure of the colorectal wall and is useful for predicting vertical margins in ER and diagnosing invasion depth [36].

Although the treatment principle for T1 CRC is SR with lymph node dissection, ER is indicated when the risk of LNM is low, and en bloc resection is possible considering the tumor size and location [1], specifically for intramucosal and slightly T1 cancer, regardless of the size and macroscopic type. However, accurate preoperative diagnosis of the invasion depth is currently challenging. Therefore, ER is sometimes performed with the intention of achieving total excisional biopsy, in which curability and the necessity of additional SR with lymph node dissection are assessed based on histopathological examination of the resected specimens. The establishment of ER as a safe, minimally invasive, and effective treatment option and the advent of endoscopic full-thickness resection (EFTR) have made this strategy acceptable and highly recommended for highly selected T1 CRC, particularly for rectal lesions [9]. Many studies have reported that preceding ER with complete resection of T1 CRC did not affect the prognosis of patients who underwent additional SR [12‒16]. The latest large multicenter study from Japan showed that performing ER before SR for T1 CRC did not adversely affect long-term prognosis, including 5-year overall survival rate [16].

Complete resection is the most important procedure when ER is performed as a total excisional biopsy. However, incomplete resection, such as piecemeal resection of the cancerous component, sometimes causes tumor implantation into the intestinal wall and local recurrence, making it difficult to perform accurate histopathological evaluation and provide appropriate additional treatment. Additionally, performing excessive ER for T1 CRC with a high risk of LNM renders the procedure meaningless and increases patient burden and healthcare costs. Therefore, case selection is important, and ER should be performed in cases that are expected to have a low risk of LNM and can be resected en bloc. Among the reported histopathological risk factors for LNM [19], only the invasion depth can be diagnosed preoperatively, and achieving complete resectability using ER is affected by the invasion depth. Therefore, preoperative evaluation of the invasion depth is important in addition to the diagnosis of T1 CRC. Invasion depth and the distance between the tumor invasive front to the muscle layer (tumor-free distance) can be measured using endoscopic ultrasonography [36]. Tumor-free distance ≥1 mm is reportedly a significant predictor of vertical margin ≥500 μm after endoscopic submucosal dissection (ESD), which may be useful for preoperative assessment to achieve R0 resection.

Cold forceps polypectomy and cold snare polypectomy, which do not involve electrocautery and are expected to pose a low risk of postprocedural bleeding, have been widely adopted. Although cold polypectomy is an easy and time-efficient technique, resected specimen contains very little submucosal tissue. Therefore, the technique should not be indicated for lesions where cancer is suspected but only for lesions smaller than 10 mm with a preoperative diagnosis of adenoma [37].

Endoscopic mucosal resection (EMR) is a common resection technique that uses a snare after a submucosal injection. Recently, underwater EMR (UEMR), in which snaring is performed after the lumen is filled with water without submucosal injection, has become popular [38]. A key attractive aspect of UEMR is that it does not involve the submucosal injection process, which is the most technically demanding process in conventional EMR and can affect performance. A systematic review comparing the results of conventional EMR and UEMR showed that the en bloc and R0 resection rates were significantly higher with UEMR than with conventional EMR (70.2 vs. 58.1%, risk ratio: 1.21, and 58.1 vs. 44.6%, risk ratio: 1.25, respectively) [39]. However, only a few reports are available on the curability of UEMR for T1 CRC. Fukuda et al. retrospectively compared the results of UEMR and conventional EMR for T1 CRC [40] and found that the en bloc and R0 resection rates were 77% and 55%, respectively. Moreover, further studies are required to adopt UEMR for T1 CRC.

EMR can achieve complete resection for lesions measuring <20 mm but is difficult for those >20 mm in size [5, 41]. ESD enables en bloc resection regardless of the tumor size [5]. Although ESD has been prevented from being disseminated because of its higher risk of complications and longer procedure time, it has been standardized through the development of strategies and devices. Recently, a large Japanese multicenter prospective study reported that the en bloc resection rate of ESD for colorectal lesions measuring ≥20 mm was 97.0% with a mean procedure time of 80.6 min [6]. Intraprocedural and postprocedural perforations occurred in 2.6% and 0.6% of cases, respectively. This study also showed a favorable long-term prognosis [7]. Although ESD has proven to be a useful resection technique, the quality of ESD is important for T1 CRC treatment since specimens containing insufficient submucosa or those with strong thermal denaturation pose a risk of positive vertical margin or inaccurate histopathological evaluation [8].

Per anal endoscopic myectomy [10] or endoscopic intermuscular dissection [11], which is a technique for dissecting between the inner circular and outer longitudinal muscle layers in the lower rectum, was devised as an application of ESD. This technique is a more reliable VM0 resection method than ESD for lower rectal T1 cancer (shown in Fig. 1).

The development of reliable endoscopic closure techniques and devices has led to the emergence of EFTR, in which the entire intestinal wall can be resected endoscopically. EFTR can be a promising complete resection method for cancers with invasion into deep submucosa or muscularis propria and those with severe fibrosis. FTRD (Ovesco Endoscopy AG, Tübingen, Germany) is available in some countries and regions for resecting lesions <20 mm in diameter, and its effectiveness has been reported in the Dutch colorectal eFTR registry, with an R0 resection rate of 88.2% [42].

Although piecemeal resection and positive horizontal margins are associated with recurrence, they can be cured using additional ER [43]. However, a positive vertical margin is also estimated to be associated with a high risk of recurrence, which has a significant impact on poor prognosis; therefore, an additional SR is strongly recommended [1, 13, 17, 20, 21].

The frequency of recurrence after ER alone for low-risk T1 CRC was low and not significantly different from that in SR cases [12, 30, 44]. In contrast, high-risk T1 CRCs treated with ER alone without additional SR have a high frequency of recurrence [3, 13, 30, 44]. In the JSCCR guideline, additional SR with lymph node dissection is recommended after en bloc ER with negative vertical margin if any of the following findings is observed during histopathological examination: submucosal invasion depth of ≥1,000 μm; lymphovascular invasion positive; poorly differentiated adenocarcinoma, signet-ring cell carcinoma, or mucinous carcinoma; and budding grade of BD2/3 at the deepest invasion site (shown in Fig. 2) [1]. However, the incidence of LNM has been reported to be 1.2–1.9% in cases with submucosal invasion depth of ≥1,000 μm without other risk factors (lymphovascular invasion, unfavorable tumor grade, and high-grade tumor budding) [1, 30, 45]. Therefore, there may be room for the relaxation of the “submucosal invasion depth of ≥1,000 μm” criterion. Nevertheless, the validity of the current JSCCR guidelines has been demonstrated in a large-scale multicenter study where the disease-free survival rate after local resection alone for T1 CRC without risk factors was 99.2% [46].

The European Society of Gastrointestinal Endoscopy recommendation is similar to the JSCCR guidelines, where an en bloc R0 resection of a colorectal lesion with superficial submucosal invasion, which is well to moderately differentiated and shows no lymphovascular invasion or grade 2 or 3 budding, should be considered a low-risk (curative) resection, and no further treatment is generally recommended [20]. In the National Comprehensive Cancer Network guideline, unfavorable histological findings are defined as poor tumor grade and lymphovascular invasion, in addition to resection margin involvement [21].

Although reducing the number of unnecessary surgeries for T1 CRC is an important issue, considering the burden on patients, T1 CRCs have a good prognosis if surgically resected; therefore, the necessary surgical treatment should not be delayed. Furthermore, the necessity of surgery should be based on the risk of LNM and the patient’s life expectancy, physical condition, social characteristics, and wishes.

For reference, the postoperative mortality rate of CRC was reported to be 1.4%, 1.8%, and 0.3% for right hemicolectomy, left hemicolectomy, and low anterior resection (LAR), respectively, according to the National Clinical Database [47]. However, postoperative decline in the quality of life should also be considered, particularly in rectal cancer cases. LAR syndrome has been reported to occur in approximately 80% of patients after LAR. The most frequently reported outcomes were incontinence (97%), stool frequency (80%), urgency (67%), evacuatory dysfunction (47%), gas-stool discrimination (34%), and a measure of quality of life (80%) [48].

The frequency of recurrence after SR with lymph node dissection for T1 CRC is reportedly 1.9–5.6% [2, 3, 12‒14]. SR and additional SR after ER can control local recurrence rather than distant metastatic recurrence [3, 13]. More than 80% and 90% of the recurrences were detected within 3 and 5 years after SR, respectively [1, 12, 17, 30, 44]. LNM (p = 0.0008) and histological tumor grade (p = 0.041) are independent risk factors for recurrence after the curative resection of T1 CRC, and the time to recurrence is more likely to be shorter in patients with LNM than in those without LNM [2].

Surveillance is aimed to improve the prognosis of patients through early detection and treatment of recurrences. Although no clear consensus has been reached regarding the frequency and modalities, patient burden, test accuracy, optimal intervals, and cost-effectiveness should be considered when designing surveillance strategies. Furthermore, intraluminal, extraluminal, and distant organ evaluations using colonoscopy, computed tomography, and tumor markers should also be conducted for a minimum of 5 years [1].

For the diagnosis of LNM in T1 CRC, the AI system reportedly has a diagnostic performance that exceeds the current guidelines [23, 24, 27‒29]; however, these data were recorded retrospectively. The system needs to be prospectively validated in real-world clinical practice, in which recurrence rates, long-term prognosis, and the diagnosis of LNM should be assessed. Recently, a novel method for predicting LNM in T1 CRC using genomic diagnostics has been developed. Ozawa et al. developed a microRNA signature by combining the expression status of five markers (MIR32, MIR181b-1, MIR193b, MIR195, and MIR411) [49]. This approach enabled the prediction of LNM using biopsy samples from untreated patients with T1 CRC.

Regarding treatment, ESD is undoubtedly a useful ER method for colorectal neoplasia [6, 7]; however, challenges remain in its widespread practice with high quality. Therefore, a simpler and more reliable ER method for en bloc resection is expected to be established, and EFTR may be possible.

The treatment of the lower rectum requires special consideration. Although rectal surgery is sometimes hesitant due to concerns about postoperative complications and colostomies, the recurrence rate is high if surgery is not performed. Chemoradiotherapy after local excision for high-risk rectal T1 cancer has been reported to reduce recurrence as well as total mesorectal excision, with preservation of anal function and without severe complications [50, 51]. Therefore, this strategy is acceptable when complete ER is performed for T1 rectal cancer, followed by adjuvant chemoradiotherapy for patients with high-risk LNM based on histopathology.

Figure 3 shows the management of T1 CRC based on current guidelines [1]. The establishment of more reliable treatment methods and accurate diagnosis of LNM in the near future is expected to provide less invasive treatment for patients with low-risk T1 CRC.

Performing complete resection of T1 CRC and assessing the risk of LNM based on histopathological examination is currently possible with the development of conventional ER methods and the advent of new ER techniques. Although prospective validation is required, advanced technologies using AI and genomics can enable achieving a high yield of LNM diagnoses, thereby providing a less invasive treatment for patients with T1 CRC.

The authors have no conflicts of interest to declare.

This study was not supported by any sponsor or funder.

Hidenori Tanaka wrote the initial draft. Ken Yamashita, Yuji Urabe, Toshio Kuwai, and Shiro Oka assisted in revising the manuscript and commented on the previous versions of the manuscript. Shiro Oka critically reviewed and approved the final version of the manuscript. All authors have approved the manuscript and agreed to its submission for publication.