[Prognostic significance and comparison of methods for calculating tumor budding in gastric cancer].

Objective: To evaluate the degree of tumor budding by three methods in gastric cancer and to study its relationship with clinical and morphological characteristics and its impact on overall survival. Choose the best way to determine the tumor budding in gastric cancer.

Material And Methods: In the surgical material from 172 patients with verified gastric cancer of the tubular histological subtype, tumor budding was identified and counted using three methods: H. Ueno, L. Wang, E. Karamitopolou. Histological samples stained with H&E, as well as using antibodies to the immunohistochemical marker PanCK, were studied. The calculation was carried out in the zone of maximum budding "hot spot" with an assessment of 5-10 visual fields and a choice of the average value. The results of calculating the tumor budding were compared with the main clinical and morphological characteristics of gastric cancer using statistical analysis. Overall survival data were obtained for 144 patients.

Results: The degree of tumor budding determined by any of the methods is significantly associated with the macroscopic form according to the classification of R. Bormann (=0.023/p=0.047/=0.039), morphological type according to the classification of P. Lauren (=0.001/=0.000/=0.001), the presence/absence of signet-ring cells (=0.008/=0.008/=0.017), with the differentiation level (=0.000/=0.000/=0.001), with the presence/absence of tumor emboli in the vascular lumen (=0.020/=0.022/=0.019), TNM classification parameters and the clinical stage of the oncological process (=0.000/=0.000/=0.000). With an increase of tumor budding, the risk of death increases in 1.77/2.12/1.81 the time (=0.03/=0.01/=0.02) set by any of the methods.

Conclusion: When using any of the methods for assessing the degree of tumor budding, it has been shown that it is a strong negative prognostic factor in stomach cancer. Although the L. Wang method has the greatest negative predictive value (RR=2.12), we believe that the H. Ueno method should be used, since it is the least labor-intensive.