br Results br This study included patients with
This study included 150 patients with tongue squamous cell carci-noma who were treated in Beijing Stomatological Hospital Aﬃliated with Capital Medical University from April 2015 to December 2017. The percentage of male patients was 53.33%. The average age was 58.01 ± 12.10 years. T1-stage disease was diagnosed in 28.67% of the patients, T2-stage disease was diagnosed in 47.33% of patients, and T3-stage disease was diagnosed in 24% of patients. Invasive growth was the most common tumor growth pattern, accounting for 62.67% of the tumors, followed by ulcerative type and exogenous type, accounting for 27.33% and 10% of the tumors, respectively. The mean DOI based on MRI in all patients was 11.75 ± 6.49 mm. The mean DOI of in-traoperative sections was 10.11 ± 5.90 mm, while the mean DOI based on analysis of pathological sections was 9.43 ± 5.57 mm (Table 1).
The overall diﬀerence in the DOI between MRI and pathological sections (DMP) was 2.32 ± 1.68 mm. The diﬀerence in the DOI be-tween intraoperative and pathological sections (DIP) was 0.68 ± 0.99 mm, and the diﬀerence in the DOI between MRI and in-traoperative sections (DMI) was 1.64 ± 1.32 mm (Table 2). The Bland-Altman scatter plot showed a relatively significant correlation within the 95% limits of agreement (LoA) among MRI and the analyses of intraoperative and postoperative pathological sections for the DOI at diﬀerent T stages (Fig. 3) and with diﬀerent tumor morphologies (Fig. 4).
To avoid uncertainty in the selection of the baseline and maximum depths of tumor invasion and to analyze the accuracy of MRI in as-sessing tumor margins, the DMP and DMI were compared with the corresponding two-dimensional analysis, and their consistency was verified. The results showed that Fluxametamide the mean two-dimensional tumor margin diﬀerence between MRI and intraoperative sections (TD-DMI) was 1.51 ± 0.90 mm, and the diﬀerence was 1.92 ± 1.41 mm be-tween MRI and pathological sections (TD-DMP). The Bland-Altman scatter plot showed a relatively significant correlation within the 95% LoA between the DOI diﬀerence (Fig. 5) and the two-dimensional analysis result (Fig. 6). The diﬀerence between the DMP and the two-dimensional analysis result was 0.41 ± 1.10 mm, and the diﬀerence between the DMI and the two-dimensional analysis result was 0.12 ± 0.79 mm (Table 3). M Multivariate linear regression was used to investigate the factors associated with the diﬀerences in DOI mea-surements and the diﬀerences in the two-dimensional analysis of tumor margins (Table 4). The results suggested prolactin the tumor growth pattern and T stage were significantly correlated with the upper diﬀerences. The diﬀerence increased with increasing T stage. For the DMP in T1 patients, the mean invasion depth was 1.48 ± 1.10 mm, while in T2 and T3 patients, the mean invasion depths were 2.08 ± 1.29 mm and 3.79 ± 2.00 mm, respectively. When the DMP was classified according to the type of tumor morphology, the results showed that the DOI of ulcerative masses was the largest at 3.72 ± 1.98 mm, followed by in-filtrating and exogenic tumors at 1.83 ± 1.23 mm and 1.53 ± 0.78 mm, respectively. We find the same results for the DMI and DIP, which are listed in Table 1. Based on the above statistical
Agreement on the infiltrating depth of a tongue tumor between MRI and in-traoperative and pathological sections.
Device pairings Mean diﬀerence ± SD 95% LoA
MRI: magnetic resonance imaging.
LoA: Limit of Agreement.
results, a multivariate linear regression analysis was performed to de-termine the exact value of the DMP when considering the infiltration depth and growth type, which had significant eﬀects. The following calculation was performed:
Based on the latest AJCC infiltration depth criteria of the eighth edition staging manual, when the pathological infiltration depths were 5 and 10 mm, the false-positive intervals of MRI infiltration depth were 1.314 to 3.035 mm and 1.779 to 3.5 mm, respectively, depending on the tumor morphology (Table 5).