Association of tumor size in pathological T4 colorectal cancer with desmoplastic reaction and prognosis

Abstract Background Tumor size in pathological T4 (pT4) colorectal cancer (CRC) is associated with oncological prognosis; however, its relation to epithelial‐mesenchymal transition (EMT)‐associated histology is unclear. We aimed to investigate the association of tumor size with oncological prognosis and EMT. Methods We performed a retrospective analysis of 95 patients with primary CRC who underwent radical surgery and were consecutively diagnosed with pT4. Results Both 3‐y disease‐free survival (DFS) and cancer‐specific survival (CSS) were significantly higher in patients with tumor size ≥50 mm than in those with tumor size <50 mm (P = .009 and P = .011, respectively). The independent factors identified in the multivariate analysis for DFS were pathological lymph node metastasis (hazard ratio [HR], 2.551; 95% confidence interval [CI], 1.031–6.315; P = .043), distant metastasis (HR, 2.511; 95% CI, 1.140–5.532; P = .022), tumor size (HR, 0.462; 95% CI, 0.234–0.913; P = .026), and adjuvant chemotherapy (HR, 0.357; 95% CI, 0.166–0.766; P = .008). The independent factors identified in multivariate analysis for CSS were tumor location (HR, 10.867; 95% CI, 2.539–45.518; P = .001) and tumor size (HR, 0.067; 95% CI, 0.014–0.321; P < .001). In pT4 CRC, smaller tumor size was associated with nonmature desmoplastic reaction and EMT‐related histology. Conclusions Tumor size ≥50 mm was associated with a better DFS and CSS than that of <50 mm, in patients with pT4 CRC. Smaller tumor size with advanced invasion likely reflects a more biologically aggressive phenotype in pT4 CRC.


| INTRODUC TI ON
In colorectal cancer (CRC), smaller tumor size has been reported to be associated with good survival and oncological prognosis. 1,2 Complete resection of smaller tumors could easily achieve remission compared to that of larger tumors, which have a higher risk of developing lymph node or distant metastasis. 3 However, several recent studies reported that tumor size was not associated with survival, 4,5 while others showed that a smaller size had poor survival. 6,7 Furthermore, smaller tumors did not show superior survival compared to medium-sized or larger tumors in patients with operable CRC. 5 In patients with stage IIC, smaller tumor size was also associated with poorer survival than larger tumor size. 7 Therefore, based on these recent findings, the association between tumor size and oncological prognosis is controversial; hence, no clear conclusions have been reached in CRC.
Cancer invasion and metastasis are affected by various stromal cells in the tumor microenvironment, as manifested by various pathological reactions. 8 Promotion of epithelial-mesenchymal transition (EMT), which allows epithelial cells to acquire the ability to invade and disseminate, can be morphologically identified on resected tissue specimens. In previous studies, EMT-associated histology was evaluated using pathological desmoplastic reaction (DR) and poorly differentiated clusters (PDCs), 9 and was found to be correlated with metastasis and worse oncological prognosis regardless of the CRC stage. 10 DR is morphologically categorized on the basis of keloidlike collagen and myxoid stroma, which are histological features that are closely related to the function of cancer-associated fibroblasts (CAFs) that play a main role in mediating the EMT program in the cancer microenvironment. 10,11 PDCs are defined as clusters of five or more tumor cells without gland formation and characterize the migratory phenotype of a tumor. 12 Therefore, DR and PDCs are thought to be morphological features of the tumor stroma that indicate EMT induction. 10,13 The pathological T4 (pT4) represents the most advanced tumor stage that is accompanied by serosa penetration (pT4a) and invasion of adjacent tissues/organs (pT4b), according to the TNM 8th edition. 14 In clinical experience, pT4 CRC tumors can be identified either as large and localized without metastases or as small but highly invasive. Such tumors may have oncologically different phenotypes in pT4 CRC. Tumor size in pT4 CRC is associated with oncological prognosis and may potentially be related to the morphological features. 15 However, no prior studies have investigated the relationship between tumor size and EMT-related histology evaluated by DR and PDCs. Therefore, this study aimed to evaluate the association of tumor size with oncological prognosis and EMT, evaluated based on DR and PDCs, in patients with pT4 CRC who underwent curative resection. We also aimed to determine the cutoff for tumor size based on survival, the EMT-associated histology assessed using DR and PDCs, and to analyze associations between tumor size and oncological prognosis.

| Patients and study design
We retrospectively analyzed all patients who underwent CRC surgery and were consecutively diagnosed with pT4 from June 2013 to March 2020, at the Gunma University Hospital in Japan. All cases with primary CRC were identified. Exclusion criteria were as follows: (a) patients who underwent preoperative treatments including chemotherapy and radiotherapy, and (b) patients who did not undergo radical surgery involving sites of distant metastasis.

| Evaluations of DR, PDCs, and EMT
An experienced pathologist, who was blinded to the patients' clinical history or outcomes, reviewed the primary tumors to evaluate pathological DR, PDCs, and EMT. Hematoxylin and eosin-stained glass slides of longitudinal sections of the deepest part of the tumor were microscopically scanned to evaluate DR and PDCs. Moreover, DR and PDCs were confirmed by evaluation by another pathologist who independently scored 50% of the cohort. The weighted κ values were good for grading DR and PDCs (κ = 0.8 and 0.7, respectively).
Each parameter was evaluated according to the criteria provided in previous reports. 10,12,16 DR was histologically classified using the following three categories: mature, intermediate, and immature DR. 16 The evaluation was based on the existence of keloid-like collagen and myxoid stroma, and the stroma was classified according to the most immature stromal area. Mature DR was diagnosed when fibrotic stroma was composed of fine, mature collagen fibers and did not contain keloid-like collagen or myxoid stroma. Intermediate DR was diagnosed when keloid-like collagen was present with mature stroma. Immature DR was diagnosed when the stroma with myxoid changes was present. PDCs were defined as clusters of five or more cancer cells infiltrating the stroma and lacking gland formation. 12 Tumors with ≤4, 5-9, or ≥10 clusters were classified as G1, G2, or G3, respectively. EMT-associated histology was based on DR and PDCs, and classified using three categories. 10

| Postoperative treatment and follow-up
We generally selected 5-fluorouracil-based chemotherapy for a total of 6 mo as postoperative adjuvant chemotherapy. At postoperative follow-up, blood tests, including measurement of tumor markers, were performed at 3 mo, and enhanced abdominal and chest computed tomography scan was conducted every 6 mo. Colonoscopy was also performed every 1-2 y. In cases of suspected recurrence, positron emission tomography was performed, and the postoperative recurrences were confirmed clinically, histologically, or by consecutive radiologic follow-up.

| Statistical analysis
Quantitative variables are expressed as median and range. Categorical

| Patient characteristics
During the study period, from June 2013 to March 2020, 140 patients underwent CRC surgery and were diagnosed with pT4. Of these, we excluded 15 patients who underwent preoperative treatments including chemotherapy and radiotherapy, and 30 who did not undergo radical surgery involving sites of distant metastasis.
Finally, this study included 95 patients who underwent radical surgery involving sites of distant metastasis and were consecutively diagnosed with pT4 for primary CRC.
In terms of tumor location, the cecum was the location in 10 cases

| Relationship between tumor size and postoperative recurrences and the cutoff value
The association between postoperative recurrences and tumor size, compared using the cumulative incidence rate, is shown in Figure 1.
The smaller the tumor was, the higher the rate of recurrence; the larger the tumor was, the lower the rate of recurrence (P = .012).
The ROC curve was used to determine the cutoff value for the tumor size that optimally predicted the development of postoperative recurrences. A cutoff value of 50 mm was selected, which maximized specificity and sensitivity (66.7% and 65.7%, respectively), for predicting postoperative recurrence based on tumor size ( Figure 2).
Tumor size <50 mm was detected in 36 cases (37.9%), and ≥50 mm was detected in 59 cases (62.1%). Comparison of clinicopathological characteristics between patients with tumors <50 and ≥50 mm is shown in Table 2. A significant difference between these groups was found in the pathological T stage, while no statistically significant difference was found for other factors. A proportion of pT4b was significantly higher in the group of patients with tumors ≥50 mm than in those with tumors <50 mm (P = .002).

| Prognosis according to tumor diameter
The overall 3y-DFS and 3y-CSS were 56.7 and 86.2%, respectively.

| Relation between EMT-related pathological factors and tumor size
The relation between EMT-related histology and tumor size is shown in Mature DR was significantly more abundant in patients with tumor size ≥50 mm than in those with tumor size <50 mm (P = .004).
Although the presence of PDCs did not show a significant association with tumor size (P = .128), EMT-related histology was significantly associated with tumor size (P = .033). In pT4 CRC, smaller tumor size was associated with nonmature stroma and EMT-related histology. factors secreted by cancer cells that support cancer invasion and metastasis; also, the heterogeneity of the CAFs may be manifested by the differential DR patterns. 13,18 Nonmature DR correlated with the morphology of CAFs that promotes invasion and metastasis of cancer cells. 19 Therefore, tumors with nonmature DR that gained an invasion ability at an early stage, characterized by smaller tumor size, may be at a high risk for metastasis and poor survival. In the Sakura trial, 17 DR was a poor oncological prognostic factor independent of the stage, and nonmature DR was the reason for recommending additional treatments, such as postoperative chemotherapy for stage II Nevertheless, our results suggest that in the case of large tumors, a better long-term survival may be expected than that with small tumors with R0 resection, even when resection of other organs is needed. Indeed, one of the main factors associated with long-term In addition to tumor size, pathological lymph node metastasis, distant metastasis, and adjuvant chemotherapy were associated with DFS, and tumor location was associated with CSS. The rectum has unique anatomic and physiologic features, which increase the risk of local spread and postoperative recurrence of rectal cancer compared to colon cancer. Therefore, as shown in our result, rectal cancer is reported to be associated with worse postoperative survival compared to colon cancer. 24 The presence of pathological lymph nodes metastasis is an important prognostic factor in CRC.

| Prognosis according to EMT-related pathological factors
Lymph node metastasis cause higher recurrence rates and shorter survival, making adjuvant chemotherapy important in improving the prognosis for these CRC patients with lymph node metastasis. 22,25 The goal of adjuvant chemotherapy is eradication of clinically occult micrometastases to increase the cure rate after a potentially curative resection for CRC patients. Remarkably, tumor size was an independent prognostic factor even including these previous prognostic factors; it is thought that the patients with smaller tumor sizes and a strong invasive tendency have very aggressive cancers.
This study had several limitations. First, it must be acknowledged that the study design was retrospective in nature and that it included a small sample size of CRC cases from a single institution. Because of the small sample size, no significant results were found to show that the EMT-histological findings were associated with an unfavorable prognostic factor in CRC patients, although a similar tendency was found in our study as in previous reports. 9,10 Second, although we compared the pathological results with tumor size, which was associated with postoperative oncological prognosis, we have not conducted molecular analyses in the present study. Hence, multicenter, large-scale prospective studies are needed to further confirm our results, and they should also tackle the molecular context of these tumors. Third, it might be biased to include patients with synchronous distant metastasis, even those achieved by curative resection.
However, we analyzed patients, excluding those with synchronous distant metastasis, and the results were similar to the present results. Despite these limitations, we believe our results provide clinically useful information. First, our study suggests that the small T4 tumors have acquired an early invasion ability and their oncological prognosis is poor; therefore, more aggressive treatments may be needed. Second, our study highlights that it is important to perform radical resection, because patients with large T4 tumors are likely to have long-term survival if R0 resection is achieved.
In conclusion, we found that tumor size significantly correlated with postoperative survival. Specifically, tumor size <50 mm significantly correlated with poor DFS and CSS. pT4 cases with tumor size <50 mm were associated with nonmature DR and also EMT-related histology. In these cases, patients may require multidisciplinary treatment before radical surgery to improve the postoperative oncological prognosis. Additionally, our results suggest that the timing of acquisition of metastatic and invasion ability may be associated with the oncological prognosis in CRC.

E TH I C S A PPROVA L
The study protocol was approved by the Institutional Review Board of Gunma University Hospital (approval no. HS2021-020). The requirement for informed consent was waived because the analysis was based on a retrospective record review.

D I SCLOS U R E
Funding: The authors have not received financial support from any funding sources for this study. Hiroshi Saeki https://orcid.org/0000-0001-7243-6960