Author replyстатья из журнала
Аннотация: We are pleased to have the opportunity to reply to the comments by Mr. Promish regarding our article. The comments deal with the statistical method of probability based on Bayes' theorem. The Hazy Bayesian Inference (HBI) methodology of statistical analysis could clarify the ONM probability differences, as pointed out by Mr. Promish. Although we can understand Bayes' theorem, we regret that HBI would have been beyond our purposes for the study described in our article. Mr. Promish stated that "they evaluate nine variables for their potential as ONM predictors, selected five, and subjected those five to a multivariate analysis. However, they then used only a single criterion to segregate subjects into discrete 'ONM' and 'non-ONM' classes. In essence, having produced a range of 'scores' based on the multiplicity of possible variables, they then ignored the information contained in that range." We agree with Mr. Promish concerning the rule for multivariate analytic methods. But our main purpose was to evaluate the ranking and relevancy of significant clinical and microscopic factors predicting non-ONM and ONM populations for clinical use. Mr. Promish has presented the ONM probability for each of our cases calculated by the standard Bayesian rule. The values of P(ONM) are very interesting data for the cases we presented. For example, however, in one patient from the non-ONM population included in our retrospective study, we observed an endophytic type of growth, a Grade 4C mode of carcinoma invasion, a moderately differentiated type of cancer, slight lymphocytic infiltration, and the highest mitotic index. The subject's P(ONM) as calculated by Mr. Promish was 0.96. Why, then, has this patient not developed ONM? Contrary to that case, another patient in the ONM group had an exophytic type of growth, a Grade 2 mode of carcinoma invasion, a well-differentiated type of cancer, slight lymphocytic infiltration, and the highest mitotic index. His P(ONM) was 0.44. Why has he developed ONM? Although the ONM probability calculated by Dr. Promish ranged from 0.76 to 0.99 for patients with Grade 4C carcinoma invasion, as described in our article, 6 of 17 patients in our retrospective study who were determined to have Grade 4C carcinoma invasion were free from cancer metastases in the neck. Our current major concern is why the patients with Grade 4C carcinoma invasion have not developed metastatis in their lymph nodes. We thank Mr. Promish for kindly sharing critical comments. In addition to using our diagnostic method, in the future we would like to evaluate patients with clinically negative lymph node metastasis for relevance of ONM with reference to the Bayesian rule. Hiroyoshi Hiratsuka D.D.S. Ph.D.*, Kenji Nakamori D.D.S. Ph.D.*, Hajime Sunakawa D.D.S. Ph.D.*, Gen-iku Kohama D.D.S. Ph.D.
Год издания: 1998
Издательство: Wiley
Источник: Cancer
Открытый доступ: bronze
Том: 82
Выпуск: 9
Страницы: 1799–1799