The diversity of definitions for the rectosigmoid junction is becoming a major obstacle to the standardization of optimal treatment of rectal cancers. The aim of this study was to determine the average height of the sigmoid take-off and its association with individual factors.

Patients diagnosed with rectal and sigmoid colon cancer in our center from January 2010 to December 2018 were retrospectively enrolled in the cancer group. The results of 200 controls without colorectal disease were also reviewed (normal group). The distance of different landmarks and margins of cancer from the anal verge were retrieved from computed tomography (CT), magnetic resonance imaging (MRI), and endoscopy findings.

Method Patients All patients pathologically diagnosed with sigmoid or rectal cancer in Peking University Third Hospital from January 2010 to December 2018 were included in our study as the cancer group. Data regarding clinical characteristics and radiological results were retrospectively extracted from the department's database. Patients who were found on endoscopy to have cancer more than 20 cm from the anal verge were not included. Patients who received emergency surgery or palliative surgery, patients with a previous history of pelvic surgery, and patients without extractable radiological examinations were also excluded.

To evaluate the relationship between the height of the sigmoid take-off and all baseline covariates in normal patients, the results of 200 patients without any intestinal-related abnormalities who underwent pelvic MRI in our center from January 2019 to June 2019 were reviewed as a normal group.

Endpoints The sigmoid take-off is the radiological sign representing the transition from mesocolon to mesorectum. It was described as "the point that the surgeon felt there was mobility of the colon away from the vertebral column" . The point of the sigmoid take-off was determined by combining both the sagittal and axial planes in CT and MRI. The distance from the anal verge was measured in sagittal planes by curvilinear distances in accordance with the lumen of the rectum. The distances of the sacral promontory, third sacral segment, and superior and inferior margin of cancers from the anal verge were measured in both CT and MRI following the same standard. The distance of the anterior peritoneal reflection from the anal verge was also measured in MRI. CT and MRI scans were reviewed originally by a junior specialist, and the process of measurement was recorded in the form of photographs. All photographs were then reviewed simultaneously by a senior surgeon and a senior radiologist who were both on colorectal cancer with over fifteen years' experience and were blinded to clinical outcomes and endoscopy results. The data were accepted if both of the senior specialists agreed with the strategy of measurement. If either of the two senior specialists disagreed with the measurement of a certain patient, the strategy of the measurement would be adjusted by the junior specialist until a consensus was reached.

The heights of the superior and inferior margins of the cancers were compared with the heights of various anatomical landmarks in order to classify cancers into different groups according to different definitions of the rectum, and to determine how the distribution of cancers would vary using the different definitions. Both dual classifications (sigmoid colon cancer and rectal cancer) and triple classifications (sigmoid colon cancer, rectosigmoid cancer, and rectal cancer) were applied. In the dual classification, only the distance of the inferior margin of the cancer was compared with the height of certain landmarks. If the distance of the inferior margin was below the height of the landmark, the cancer would be classified as rectal cancer, otherwise, it would be classified as sigmoid colon cancer. In the triple classification, a cancer would be classified as rectal cancer if the height of its superior margin was below the height of the landmark. Rectosigmoid cancer represented cancers straddling the landmark. As in the dual classification, cancers, whose inferior margins were above the height of the landmark were defined as sigmoid colon cancers.

Statistical Analyses Demographic and clinicopathologic data were collected and analyzed in both the cancer group and the normal group. Mean values with standard deviation (SD) or median values with ranges were calculated as appropriate. The influence of factors sex and age on the distance of the sigmoid take-off in the normal group was evaluated by nonparametric statistics and Spearman's correlation analysis, respectively. The distances of the sigmoid take-off as measured in the MRIs of both groups were compared using propensity score matching method. Aside from age and sex, heights of sacral promontory, third sacral segment, and peritoneal reflection were included in constructing the propensity score. A 1:1 matching using both optimal and greedy algorithms without replacement was adopted with a caliper width of 0.1. Differences among results of CT, MRI, and endoscopy in the cancer group were tested using paired T-tests. Both univariate and multivariate linear regression analyses were carried out in both groups to explore the relationship between the height of the sigmoid take-off and individual factors including age, sex and body habitus. Selection of variables entering into the multivariable regression model was based on statistical significance (P< 0.100). All analyses were carried out with IBM SPSS Statistics version 22.0 for Windows. A two-tailed P value less than 0.05 was considered statistically significant.

Inclusion Criteria:

- All patients pathologically diagnosed with sigmoid or rectal cancer in Peking University Third Hospital from January 2010 to December 2018 were included in our study as the cancer group. Patients who were found on endoscopy to have cancer more than 20 cm from the anal verge were not included. Patients who received emergency surgery or palliative surgery, patients with a previous history of pelvic surgery, and patients without extractable radiological examinations were also excluded.

Exclusion Criteria:

• Patients who were found on endoscopy to have cancer more than 20 cm from the anal verge were not included.
• Patients who received emergency surgery or palliative surgery, patients with a previous history of pelvic surgery, and patients without extractable radiological examinations were also excluded.

Patients diagnosed with rectal and sigmoid colon cancer in our center from January 2010 to December 2018 were retrospectively enrolled in the cancer group. The results of 200 controls without colorectal disease were also reviewed (normal group).

We have uploaded our raw data to Mendeley Data.

In six month.

Mendeley Data.