New scoring system for the evaluation obstructive degrees based on computed tomography for obstructive colorectal cancer

Abstract

BACKGROUND

The degree of obstruction plays an important role in decision-making for obstructive colorectal cancer (OCRC). The existing assessment still relies on the colorectal obstruction scoring system (CROSS) which is based on a comprehensive analysis of patients’ complaints and eating conditions. The data collection relies on subjective descriptions and lacks objective parameters. Therefore, a scoring system for the evaluation of computed tomography-based obstructive degree (CTOD) is urgently required for OCRC.

AIM

To explore the relationship between CTOD and CROSS and to determine whether CTOD could affect the short-term and long-term prognosis.

METHODS

Of 173 patients were enrolled. CTOD was obtained using k-means, the ratio of proximal to distal obstruction, and the proportion of nonparenchymal areas at the site of obstruction. CTOD was integrated with the CROSS to analyze the effect of emergency intervention on complications. Short-term and long-term outcomes were compared between the groups.

RESULTS

CTOD severe obstruction (CTOD grade 3) was an independent risk factor [odds ratio (OR) = 3.390, 95% confidence interval (CI): 1.340-8.570, P = 0.010] via multivariate analysis of short-term outcomes, while CROSS grade was not. In the CTOD-CROSS grade system, for the non-severe obstructive (CTOD 1-2 to CROSS 1-4) group, the complication rate of emergency interventions was significantly higher than that of non-emergency interventions (71.4% vs 41.8%, P = 0.040). The postoperative pneumonia rate was higher in the emergency intervention group than in the non-severe obstructive group (35.7% vs 8.9%, P = 0.020). However, CTOD grade was not an independent risk factor of overall survival and progression-free survival.

CONCLUSION

CTOD was useful in preoperative decision-making to avoid unnecessary emergency interventions and complications.

Core Tip: This retrospective, single-center, observational study explores the relationship between computed tomography-based obstructive degree (CTOD) and the colorectal obstruction scoring system (CROSS) score and evaluates whether the CTOD could affect short-term and long-term prognoses. In this study of 173 patients, CTOD severe obstruction was an independent risk factor. Complications in the form of emergency interventions were more common in patients with CTOD-CROSS non-severe obstruction. The proposed CTOD may be useful in preoperative treatment decision-making and help prevent unnecessary emergency interventions and complications in certain patients.

INTRODUCTION

Colorectal cancer (CRC) is the most common cancer of the digestive system around the world and the fourth most common cause of cancer-related deaths worldwide[1,2]. Approximately 8%-29% of patients diagnosed with CRC present with acute colonic obstruction[3-6]. If obstructive CRC (OCRC) is not effectively decompressed, once intestinal perforation occurs, it can not only cause higher severe postoperative complications and death rates but also result in worse survival outcomes[7,8]. Patients with acute complete OCRC require emergent colectomy, with or without anastomosis, primary stoma formation, or colonic stent insertion. For patients with incomplete colorectal obstruction, the best treatment plan is formulated after preoperative bowel preparation, nutritional support, preoperative imaging, colonoscopy, and multidisciplinary consultation. Objective assessment of the degree of obstruction is particularly important in clinical decision-making.

The existing assessment still relies on the colorectal obstruction scoring system (CROSS) established by Japanese scholars to assess the degree of colon obstruction, which is based on a comprehensive analysis of patients’ complaints and eating conditions; however, data collection relies on subjective descriptions and lacks objective parameters[9,10]. There is no objective grading system for the degree of obstruction in patients with OCRC, which severely limits the objectivity of clinical decision making.

Patients with acute colorectal obstruction are unable to undergo colonoscopy and barium enema because of obvious complaints, poor general condition, and a lack of good bowel preparation. Even if completed, it can affect the results of colonoscopy and barium enema. However, there is the possibility of intestinal perforation. In the examination of OCRC, computed tomography (CT) can not only clearly display the air-fluid plane and the degree of dilatation of the intestinal canal, but also accurately reflect the changes occurring inside and outside the intestinal cavity, which helps clinicians judge the site, nature, and degree of obstruction and provides powerful evidence for diagnosis and treatment. According to the Consensus Conference of the 4^th Congress of the World Society of Emergency Surgery held in Campinas in May 2017, CT is the recommended imaging modality for OCRC[11].

Therefore, this study proposes to classify and grade the degree of OCRC using objective CT images to explore the relationship between CT-based obstructive degree (CTOD) and CROSS and to determine whether CTOD could affect the short-term and long-term prognosis. Integrating CTOD and CROSS to evaluate the application value of emergency interventions for OCRC.

MATERIALS AND METHODS

Study population

The clinicopathological data of 232 patients diagnosed with OCRC who underwent invasive intervention at the Department of Emergency Surgery, Fujian Medical University Union Hospital in China between January 2008 and February 2021 were retrospectively analyzed. Patients diagnosed with perforation before surgery or operatively (n = 7), those who did not undergo CT in our center (n = 29), those with incomplete CT images (n = 9), those discharged automatically without treatment (n = 3), those with malignancy of the terminal ileum (n = 1), or those who underwent stenting before admission (n = 1) were excluded. Ultimately, 173 patients were included in this study. The colonoscopy was performed as completely as possible. Staging was determined according to the 7^th edition of the Union for International Cancer Control tumor node metastasis classification system. The data used in this study were obtained after obtaining informed consent from the participants. This study was approved by the ethics committee of Fujian Medical University Union Hospital (approval No. 2022KY102). The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and its amendments.

The inclusion criteria were as follows: (1) Multidetector-row CT examination within 1 week before surgery; (2) Colorectal obstruction diagnosed through history collection, physical examination, and adjunctive examination; (3) Pathology confirmed the presence of cancer, including confirmed postoperatively; and (4) Age ≥ 18 years.

The exclusion criteria were as follows: (1) Preoperative CT imaging revealing perforation of the digestive tract; (2) Incomplete or poor quality; (3) Neoadjuvant chemoradiotherapy or chemotherapy; and (4) Pregnant or lactating women.

Classification and definition

The CROSS grade was as follows: Score 0, requiring a continuous decompressive procedure; Score 1, no oral intake; Score 2, liquid or enteral nutrients; Score 3, soft solids, low residue, and full diet with symptoms of stricture; and Score 4, soft solids, low residue, and full diet without symptoms of stricture.

In this study, we defined CROSS 0 as CROSS severe obstruction, CROSS 1-4 as non-severe obstruction. Emergency intervention was defined as an intervention performed within 24 hours of admission.

CT regions of interests annotated, reconstructions, and calculations

The regions of interests (ROIs) along the colorectal contour were manually annotated using ITK-SNAP software (version 3.6.0, open-source software; http://www.itksnap.org)[12] on CT scan images, including the site of obstruction and its proximal and distal bowel. The obstruction site was subdivided into parenchymal and nonparenchymal areas. Annotation was performed by two clinicians with extensive imaging experience, and the annotated images were reviewed by a senior clinician.

The ratio of proximal to distal obstruction (P/D ratio) was calculated as the ratio of the shortest diameter of the most extended proximal colon to the shortest diameter of the distal colon next to the obstruction site. Iterating through all slices of each case, the drawn obstruction region (label 1) was divided by the image threshold method into non-substantial parts (label 2), and the number of pixels of label 1 and label 2 was counted, which was recorded as number 1 and number 2, and number 2/number 1 was calculated, that is, the proportion of nonparenchymal areas at the site of obstruction (PNP). K-means clustering was performed on the P/D ratios and PNP, which are continuous variables. The final output was divided into three categories, defined as CTOD (Figure 1). The k-means method is performed by randomly selecting k objects. Each object initially represented the average or center of a cluster; for each remaining object, it was assigned to the nearest cluster based on its distance from the center of each cluster, and the average of each cluster was then recalculated. This process is repeated until the criterion function converges. This results in a higher degree of similarity within a cluster and a lower degree of similarity between clusters. In this experiment, k was set to 3.

Figure 1.

Workflow. Regions of interests (ROIs) segmentation: Using the ITK-SNAP software to annotate ROIs of the colon and rectum on computed tomography (CT) scan images, including the site of obstruction (red and light blue) and its proximal (green) and distal bowel segments (dark blue). The obstruction site was subdivided into parenchymal (red) and nonparenchymal areas (light blue). Processing: CT-based obstructive degree (CTOD) was obtained using k-means, the ratio of proximal to distal obstruction, and the proportion of nonparenchymal areas at the site of obstruction. Output: Compare the incidence of complications associated with CTOD. CTOD was integrated with the colorectal obstruction scoring system to analyze the effect of emergency intervention on complications. Long-term outcomes were compared between the groups. ROIs: Regions of interests; PNP: Proportion of nonparenchymal areas at the site of obstruction; P/D: Proximal to distal obstruction; CTOD: Computed tomography-based obstructive degree; CROSS: Colorectal obstruction scoring system.

In this study, we defined CTOD 3 as CTOD severe obstruction, CTOD 1-2 as non-severe obstruction. Integrating CTOD-CROSS, we categorize patients with both CTOD severe obstruction and CROSS severe obstruction into group A, those with CTOD severe obstruction and CROSS non-severe obstruction into group B, those with CTOD non-severe obstruction and CROSS severe obstruction into group C, and those with both CTOD non-severe obstruction and CROSS non-severe obstruction into group D.

Complication

The definition of each complication was based on previous reports[13-18]. Postoperative pneumonia was defined as newly developed infiltrate on chest radiography and positive bronchoalveolar lavage culture results. Complications were classified according to the modified version of the Clavien-Dindo classification system[19].

Follow-up

Postoperative follow-ups were performed every 3 months for 2 years and then every 6 months from years to 2-5. Most routine patient follow-up appointments included physical examinations, laboratory tests, chest radiography, abdominal ultrasonography, CT, and annual endoscopic examinations. Overall survival (OS) was calculated from the day of surgery or stenting until death or the final follow-up date, whichever occurred first. Progression-free survival (PFS) was calculated from the day of surgery or stenting until disease progression, tumor-related death, or the final follow-up date. The final follow-up was conducted in December 2021.

Statistical analysis

Statistical analyses were performed using statistical product and service solutions v.25.0 for Windows (statistical product and service solutions Inc., Chicago, IL, United States). All continuous variables are presented as the mean ± SD. χ² or Fisher’s exact test was used for categorical variables. Image threshold processing was performed using Python 3.7 software, and k-means clustering yielded the CTOD. All variables that yielded a P value ≤ 0.10 in univariate analysis and the factors considered important clinically important were included in the multivariate model. Subsequently, multivariate analysis was performed using logistic regression to determine the risk factors associated with postoperative complications. Cumulative survival rates were compared using the Kaplan-Meier method and the log-rank test. Independent variables that predicted survival were evaluated using the Cox proportional hazards model. Statistical significance was set at P value < 0.05.

RESULTS

The relation between CTOD grade and CROSS grade

The CTOD was divided into three categories according to the P/D ratios and PNP scores. There was a positive correlation between CTOD and CROSS grade (coefficient of contingency was 0.319, P = 0.003). With the increasing P/D ratio (CTOD1 vs CTOD2 vs CTOD3 = 2.857: 4.620: 7.120), the severity of OCRC increased (P < 0.001, showed in Supplementary Table 1). Similarly, a high CTOD score was associated with a low CROSS score.

The characteristics of CTOD grade

The overall complication rate in the CTOD severe group (CTOD 3) was significantly higher than that in the CTOD non-severe group (CTOD 1-2) (65.6% vs 44.0%, P = 0.027), with an increasing incidence of postoperative pneumonia (34.4% vs 14.9%, P = 0.010). Moreover, the total hospital stay (25.5 days vs 19 days, P = 0.004) and postoperative hospital stay (12.5 days vs 12 days, P = 0.021) in the severe CTOD group were significantly longer (Table 1).

Table 1.

Baseline characteristics of study patients stratified by computed tomography-based obstructive degree severity, n (%)

Characteristic	CTOD non-severe (CTOD 1-2)	CTOD severe (CTOD 3)	P value
PNP	0.0896	0.0689	0.405
P/D	4.080	7.120	< 0.001
Age (> 65 years)	71 (50.4)	15 (46.9)	0.722
Sex (female)	58 (41.1)	14 (43.8)	0.786
BMI (≥ 18.5 kg/m²)	108 (83.7)	23 (85.2)	0.850
ALB (≥ 35 g/L)	72 (51.1)	14 (43.8)	0.455
CEA (> 5ug/L)	71 (51.4)	18 (58.1)	0.505
Hb (< 9 g/L)	20 (14.2)	4 (12.5)	0.803
ASA 3-4	42 (29.8)	12 (37.5)	0.395
Surgical history	34 (24.1)	11 (34.3)	0.232
Comorbidities	79 (56.0)	18 (56.3)	0.982
SEMS	44 (31.2)	13 (40.6)	0.306
Left-sided	81 (57.4)	21 (65.6)	0.396
Surgical method			0.647
Stoma/SEMS	16 (11.4)	4 (12.5)
SBTS	33 (23.4)	10 (31.3)
Resection	92 (65.2)	18 (56.3)
Surgical time (minutes)	204.4	209.07	0.223
Bleeding (mL)	100	100	0.811
Total number of LN	19	19	0.675
Positive LN	1	1	0.184
Complication	62 (44.0)	21 (65.6)	0.027
SAE	20 (14.2)	5 (15.6)	0.834
Mortality within 30 days	1 (0.7)	0 (0.0)	1.000
Re-bowel obstruction	4 (2.8)	2 (6.3)	0.307
ICU stay	13 (9.2)	3 (9.4)	0.978
Pneumonia	21 (14.9)	11 (34.4)	0.010
Incision infection	21 (14.9)	3 (9.4)	0.415
pT 4 stage	51 (36.2)	11 (34.4)	0.981
pN 1-2 stage	81 (57.4)	18 (56.3)	0.981
pM 1 stage	42 (29.8)	5 (15.6)	0.104
Sepsis	5 (3.5)	1 (3.1)	1.000
Chemotherapy	74 (52.5)	14 (43.8)	0.502
Gas time (day)	3	3	0.869
Oral intake (day)	4	4	0.625
Total hospital stay (day)	19	25.5	0.004
Post-surgery stay (day)	12	12.5	0.021

CTOD: Computed tomography-based obstructive degree; PNP: Proportion of nonparenchymal areas at the site of obstruction; P/D: Proximal to distal obstruction; BMI: Body mass index; ALB: Albumin; CEA: Carcinoembryonic antigen; Hb: Hemoglobin; LN: Lymph node; ICU: Intensive care unit; ASA: American Society of Anesthesiologists; SAE: Serious adverse event; SEMS: Self-expandable metallic stents; SBTS: Stenting as bridge to surgery.

Univariate and multivariate analysis of CTOD on the complications

On univariate analysis, a more severe CTOD grade, older age, and stenting as bridge to surgery (SBTS) were significantly associated with complications (P < 0.10). Multivariate analysis showed that the CTOD severe group was an independent risk factor for complications [odds ratio (OR) = 3.390, 95% confidence interval (CI): 1.340-8.570, P = 0.010], whereas the CROSS grade was not (Table 2).

Table 2.

The univariate and multivariate logistic regression analysis of complication

	Univariate			Multivariate
	OR	95%CI	P value	OR	95%CI	P value
CTOD (3 vs 1-2)	2.433	1.091-5.423	0.030	3.39	1.340-8.573	0.010
ALB (≥ 35 vs < 35)	0.934	0.514-1.696	0.882
Age (≥ 65yrs vs < 65)	1.707	0.935-3.117	0.082	NA	NA	NA
BMI (≥ 18.5 vs < 18.5)	0.521	0.215-1.264	0.149	NA	NA	NA
Surgical method
Only stoma/SEMS	1	Ref.	0.219	NA	NA	NA
SBTS	2.683	0.868-8.295	0.086
Surgical resection	2.25	0.806-6.282	0.122

CTOD: Computed tomography-based obstructive degree; BMI: Body mass index; ALB: Albumin; SEMS: Self-expandable metallic stents; SBTS: Stenting as bridge to surgery; OR: Odds ratio; CI: Confidence interval; NA: Not appliable.

Stratification analysis based on the CTOD-CROSS grade

The integrated CTOD-CROSS grade showed strongly overlapping areas between these two classification systems (the ratio was 61.27%, Table 3). Comparing the complication rates of emergency intervention in groups A, B, C, and D, the results showed that the complication rate of emergency intervention in group D (CTOD-CROSS non-severe obstruction, i.e., CTOD1-2 to CROSS 1-4) was significantly higher than that of non-emergency intervention (71.4% vs 41.8%, P = 0.040) (Table 4). In group D, the incidence of pneumonia in emergency interventions was significantly higher than that in non-emergency interventions (35.7% vs 8.9%, P = 0.020). In group A (CTOD 3 to CROSS 0), there were seven cases (53.8%) of stenting as a bridge to surgery, one case (7.7%) of stoma alone, and five cases (38.5%) of resection. There were 4 cases (30.8%) of emergency interventions. In the SBTS group, the technical and clinical success rates of self-expandable metallic stents (SEMS) were 100% and 100%, respectively, with no stent-related complications, displacement, death, or intraoperative enterostomy.

Table 3.

The interaction between the two classifications of severity of obstruction

	CROSS non-severe (CROSS 1-4)	CROSS severe (CROSS 0)	Cumulation
CTOD non-severe (CTOD 1-2)	93 (group D)	48 (group C)	141
CTOD severe (CTOD 3)	19 (group B)	13 (group A)	32
Cumulation	112	61	173

CTOD: Computed tomography-based obstructive degree; CROSS: Colorectal obstruction scoring system.

Table 4.

Comparing the complication rates of emergency intervention in group A, B, C and D

Group	No-urgent intervention	Urgent intervention	P value
A	6/9 (66.7)	3/4 (75.0)	1.000
B	8/15 (53.3)	4/4 (100.0)	0.245
C	13/26 (50.0)	6/22 (27.3)	0.109
D	33/79 (41.8)	10/14 (71.4)	0.040

Survival outcomes

A total of 153 patients in the whole group were followed up, with a follow-up rate of 88.43% and a follow-up time of 1.5-143.2 months. In 153 patients undergoing resection, the results of Cox regression multivariate analysis showed that pT4 stage and pM1 stage were independent risk factors for OS and PFS, whereas CTOD was not an independent prognostic factor (Supplementary Table 2). The survival curve results showed no significant differences between SBTS and surgical resection in terms of OS and PFS (Figure 2).

Figure 2.

Overall survival and progression-free survival after surgery between the stenting as bridge to surgery and surgical resection groups. A: Overall survival; B: Progression-free survival. OS: Overall survival; PFS: Progression-free survival; SBTS: Stenting as bridge to surgery.

There were no statistically significant differences in the survival curves for different obstruction severities (CTOD 3 vs CTOD 1-2) (Supplementary Figure 1). Subgroup analysis showed no significant difference in OS and PFS between the emergency and non-emergency intervention groups in group D (Supplementary Figure 2).

DISCUSSION

Our center used the ITK-SNAP software to annotate the ROIs before surgery and then reconstructed the images in three dimensions to visualize the site and degree of obstruction of the OCRC. For further quantification, we combined the PNP and P/D ratio of CT images, two continuous variables, and performed k-means clustering to obtain the CTOD. The CROSS grade was based on whether the patients needed to undergo sustained stress reduction, whether they could eat, and how well they ate. It was a comprehensive indicator, but it was more subjective and dependent on the clinical experience judgment of the doctor in charge. However, the CTOD was graded by the numerical value measured by CT, which was objective, reproducible, and broadly applicable. Therefore, the results of this study explored the relationship between CTOD, which was obtained through objective measurement and calculation, and the long-term and short-term prognoses of OCRC, which were accurate and reliable. In multivariate analysis of complications, the commonly used CROSS grade was not an independent factor, but CTOD was an independent factor.

Whether the time to obstruction clearance is associated with a more favorable patient outcome remains controversial. de Roos et al[20] showed that surgical resection beyond 4 weeks after presentation seems to have better short-term and long-term outcomes for left-sided obstructive colon cancer (LSOCC). A Dutch multicenter randomized control trial comparing the clinical efficacy of stenting bridging surgery and emergency resection found no significant difference in 30-day or overall mortality[21]. There are literature reports exploring the relationship between the time interval and outcome of LSOCC after different bridging modalities to resection[22,23]. Ohki et al[10] compared the feasibility and safety of endoscopic placement of SEMS as a bridge to surgery in patients with OCRC at different levels of obstruction. They found high clinical (98.0% vs 98.4%) and technical (96.7% vs 97.8%) success rates in both groups (CROSS 0 vs CROSS 1 or 2). In this study, CT scans were completed within a week before intervention, not in real-time before the intervention. Indeed, proximal bowel distention may decrease after non-emergency intervention, so the initial CTOD cannot represent the current level of obstruction. On the other hand, for patients with non-severe obstruction found in this study, as indicated by CTOD-CROSS non-severe obstruction, non-emergency intervention can be initiated first to prepare for subsequent surgery. We found that the complication rate of emergency intervention was higher than that of non-emergency intervention in group D (CTOD-CROSS non-severe obstruction). Therefore, we could combine these two classification methods to screen out non-severe obstructive patients and subsequently sub-emergency patients. These times are used for nutritional support, correction of water-electrolyte imbalances, bowel preparation, intestinal tubing normalization, and more importantly, a multidisciplinary team consultation for preoperative staging and initiation of the initial neoadjuvant regimen if necessary. It should be noted that when bridging surgery is performed, the success rate of bridging, corresponding complications, and long-term oncological efficacy need to be carefully evaluated. Although there was no statistically significant difference in postoperative complication rates between emergency and non-emergency interventions in group A (CTOD-CROSS severe obstruction) in this study, this may be due to the low number of cases in this study. Therefore, this needs to be expanded further to address this problem. Patients with severe CTOD-CROSS obstruction often attract the attention of clinicians, with strong complaints, severe CT obstruction, and rapid disease progression. In a comprehensive assessment, regardless of the risk of infection or tumor spreading, more aggressive and early interventions are often taken to remove obstructions to minimize the risk of perforation than the serious consequences of perforation.

Many studies have shown that the prognosis for OCRC is poor[24,25]. Obstruction of stage II CRC is a high-risk factor and an indication for adjuvant chemotherapy according to the NCCN guidelines[26]. CRC complicated by obstructions of varying degrees has different requirements for the timing of intervention. Patients with acute severe obstruction face a dual challenge to identify a treatment regimen that will both relieve the obstruction and address the colorectal lesion, which is likely to be malignant. In these patients, emergency intervention to relieve the obstruction is often required. Emergency intervention measures include emergency surgical resection, with or without intestinal anastomosis, emergency ostomy, and placement of intestinal stents. Severe obstruction caused by CRC is associated with poorer survival rates[27,28]. Furthermore, in cases of acute severe obstruction, SBTS strategy can convert the treatment of OCRC from emergency surgery to elective surgery[29]. In contrast, non-severe intestinal obstruction often allows for more time for bowel preparation, nutritional support, multidisciplinary consultations, enabling the implementation of planned procedures. It is still necessary to closely observe the changes in the patient’s complaints and abdominal signs before planned surgery to avoid serious conditions such as undetected intestinal necrosis or even intestinal perforation. The results of this study showed that CTOD was an independent factor affecting short-term prognosis. Emergency intervention should be avoided for patients with non-severe obstruction. CTOD was not an independent prognostic factor for long-term survival. Moreover, in subgroup analysis, there was no difference in survival between emergency and non-emergency interventions among patients with non-severe obstruction. It may be possible to further explore the relationship between CTOD and long-term prognosis in studies with a longer timeline, larger sample size, and multicenter research.

The long-term oncological outcomes of OCRC have remained a hotspot for research in recent years. The results of this study showed that pT and pM stages were independent prognostic factors for OS and PFS. There were no significant differences between SBTS and surgical resection in terms of OS and PFS. Controversy remains regarding the effect of stenting on oncological outcomes. Several studies have reported possible adverse oncological outcomes of stent insertion from the perspective of pathophysiology[30-34]. Several studies comparing the oncological outcomes of stenting and emergency surgical resection have not found a significant difference in survival[35-37]. In addition, a propensity score matching study of the oncological outcomes of stenting bridging to surgery and emergency resection in obstructive left colon cancer did not differ in 3-year OS, DFS, or local recurrence rates[38]. Takahashi et al[38] found that circulating free DNA and circulating tumor DNA concentrations seven days after stent insertion were significantly higher than those in the transanal decompression tube group (992 ng/mL vs 308 ng/mL, P = 0.005; 83% vs 22%, P = 0.002), reflecting tumor damage caused by stent insertion. It may drive tumor cells to become more aggressive and metastatic and reduce the prognosis of such patients[30,33,38]. Maruthachalam et al[30] reported that eight of 20 patients with obstructive colon cancer had elevated CK20 mRNA levels after stent insertion. Sabbagh et al[33] showed that tumors undergoing surgical resection after stenting have worse histological features, such as tumor ulcers, ulceration around the tumor, nerve invasion, and lymph node violations, which may indicate worse oncological outcomes[31,32]. To address concerns about the impact of oncological outcomes of stenting, European Society of Gastrointestinal Endoscopy guidelines recommend stenting only in palliative settings or in cases of increased surgical risk (age ≥ 70 years or above American Society of Anesthesiologists III) as the preferred regimen and must be conducted in a specialized center[39]. In this study, no statistically significant difference was observed between SBTS and surgical resection in terms of OS and PFS.

However, several limitations of this study should be considered. First, because this study was single-center and retrospective in nature, the sample size was small. Therefore, a potential bias was inevitable. In the future, we can conduct multicenter, prospective studies to collect more data in a more controlled manner, thereby reducing the impact of retrospective bias. Building on this foundation, we can apply machine learning methods to train and validate the impact of CTOD on the short-term and long-term prognosis of OCRC, making the CTOD scoring more broadly applicable to different clinical populations. Second, the process of manually labeling ROIs for preoperative CT images is time consuming. The ROI annotation require some clinical and imaging experience, and there may be subjective and interobserver bias. However, in this study, ROI annotations were performed by two clinically experienced radiologists who were blinded to the patients’ clinical outcomes. Furthermore, the annotated images were reviewed by a senior clinician to minimize inter-observer bias as much as possible. With the advancement of imaging technology and artificial intelligence, in the future we can automate ROI annotations with advanced imaging technology or artificial intelligence-based tools, which will save time, effort and improve repeatability.

CONCLUSION

CTOD was an independent predictor of complications. Complications of emergency interventions were more common in patients with CTOD-CROSS non-severe obstruction. The proposed CTOD may be useful in preoperative treatment decision making to avoid unnecessary emergency interventions and complications in certain patients. The application of CTOD needs to be further confirmed by more multicenter prospective studies.

Data sharing statement

The datasets used in the current study are available from the corresponding author upon reasonable request.