Abstract: Objective:To investigate the accuracy of highresolution magnetic resonance imaging (MRI) in diagnosis of the lymph nodes metastases (LNMs) and stage of rectal cancer (RC).
Methods:The retrospective crosssectional study was conducted. The clinicopathological data of 65 patients with RC who were admitted to the Peking University People′s Hospital between April 2014 and April 2016 were collected. The results of postoperative pathological examination: of 65 patients with RC, 24 had positive LNMs and 41 had negative LNMs; N0, N1 and N2 were respectively detected in 41, 14 and 10 patients. Imaging data were captured using 3.0Tesla MRI and body phasedarray coil. Diagnostic criteria for LNMs of RC: criterion 1: irregular lymph node boundaries and signal characteristics were not considered; criterion 2: heterogeneous signal of lymph nodes and boundary characteristics were not considered; criterion 3: irregular lymph node boundaries and / or heterogeneous signal of lymph nodes. The American Joint Committee on Cancer (AJCC) cancer staging manual (7th edition) have established the N staging scheme for RC. Lymph nodes distribution according to the rectal lymphatic drainage: D₁ was located in fascia of the mesorectum; D_1a above the level of tumor; D_1b at the level of tumor; D 1c under the level of tumor; D₂ around the superior rectal artery and root of inferior mesenteric artery; D₃ on the side of pelvic cavity. Observation indicators: (1) Efficiency and comparison of LNMs of RC diagnosed by highresolution MRI according to the 3 criteria (postoperative pathological result as a gold standard). (2) Efficiency and comparison of N stage of RC diagnosed by highresolution MRI according to the 3 criteria (postoperative pathological result as a gold standard). (3) Efficiency of LNMs of RC diagnosed by highresolution MRI according to the maximum short diameter of lymph nodes: ① maximum short diameter distribution of positive and negative LNMs of RC; ② sensitivity, specificity, accuracy and consistency of LNMs diagnosed using different maximum short diameter of lymph nodes as a threshold (postoperative pathological result as a gold standard); ③ comparison of accuracy of LNMs diagnosed using maximum short diameter of lymph nodes with highest diagnostic accuracy as a threshold and using the 3 criteria; ④ sensitivity, specificity, accuracy and consistency (postoperative pathological result as a gold standard) of LNMs diagnosed using maximum short diameter of lymph nodes with highest diagnostic accuracy as a threshold combined with the highest efficiency in the (1), and its comparison in accuracy of LNMs with highest diagnostic accuracy as a threshold. (4) Distribution of LNMs of RC diagnosed by highresolution MRI according to the 3 criteria. (5) Followup situations. Followup using outpatient examination and telephone interview was performed to detect patients′ postoperative survival up to October 2016. The sensitivity, specificity, accuracy and comparison of ratio were respectively done by the chisquare test. Kappa test was used for consistency, κ≤0.40 was used as low consistency, 0.40<κ≤0.60 as moderate consistency, 0.60<κ≤0.80 as higher consistency, and κ>0.80 very high consistency.
Results:(1) Efficiency and comparison of LNMs of RC diagnosed by highresolution MRI according to the 3 criteria: accuracies of LNMs of RC diagnosed by highresolution MRI according to the criterion 1, 2 and 3 were respectively 93.8%, 87.7% and 90.8%, showing very high, higher and very high consistencies compared with postoperative pathological results (κ=0.87, 0.74, 0.81, P<0.05), and with no statistically significant difference in diagnostic accuracy among them (χ²=1.495, P>0.05). (2) Efficiency and comparison of N stage of RC diagnosed by highresolution MRI according to the 3 criteria: accuracies of N stage of RC diagnosed by highresolution MRI according to the criterion 1, 2 and 3 were respectively 87.7%, 83.1% and 84.6%, showing the same higher consistencies compared with postoperative pathological result (κ=0.77, 0.68, 0.72, P<0.05), and with no statistically significant difference in N stage among them (χ²=0.567, P>0.05). (3) Efficiency of LNMs of RC diagnosed by highresolution MRI according to the maximum short diameter of lymph nodes: ① maximum short diameter distribution of positive and negative LNMs of RC: maximum short diameter ranges were respectively 3-18 mm in positive LNMs and 1-9 mm in negative LNMs, and maximum short diameter <3 mm and ≥10 mm were respectively negative and positive LNMs. ② Efficiency of LNMs of RC diagnosed using different maximum short diameter of lymph nodes as a threshold: diagnostic accuracy of 70.8% was the highest when maximum short diameter >7 mm was used as a standard of positive LNMs, showing a low consistency compared with postoperative pathological result (κ=0.29, P<0.05). ③ Comparison of accuracy of LNMs diagnosed using maximum short diameter of lymph nodes >7 mm as a threshold and using the 3 criteria: there was a statistically significant difference among them (χ²=15.637, P<0.05); accuracies of LNMs of RC diagnosed by highresolution MRI according to the criterion 1, 2 and 3 were higher than that diagnosed using maximum short diameter of lymph nodes >7 mm as a threshold (χ²=10.354, 5.656, 6.923, P<0.05). ④ Comparison of accuracy of LNMs diagnosed using maximum short diameter of lymph nodes >7 mm combined with the criterion 3 as a threshold and using maximum short diameter >7 mm as a threshold: the criterion 3 was used as a threshold because there was no statistically significant difference in diagnostic accuracy among the 3 criteria (P>0.05). Diagnostic accuracy was 78.5% when maximum short diameter >7 mm combined with the criterion 3 as a threshold, showing a low consistency compared with postoperative pathological result (κ=0.36, P<0.05), with no statistically significant difference in diagnostic accuracy compared with maximum short diameter >7 mm as a threshold (χ²=0.154, P>0.05). (4) Distribution of LNMs of RC diagnosed by highresolution MRI according to the 3 criteria: positive LNMs of RC diagnosed by highresolution MRI located mostly in D₁ (76.1%-83.1%)and D_1b (77.8%-81.4%). (5) Followup situations: of 65 patients, 54 were followed up for 6-25 months, with a median time of 14 months. During the followup, 7 patients had distant metastases and 47 had tumorfree survival.
Conclusions:There are higher accuracies of LNMs and N stage of RC diagnosed using preoperative highresolution MRI. Diagnostic accuracy of LNMs of RC cannot be improved when characteristics of lymph node morphology and size are used as a diagnostic standard. The positive LNMs of RC locate mostly in D₁ and D_1b.