Background: Despite much research effort, the major prognostic factor of oesophageal squamous cell cancer (ESCC) remains the pathological stage of the disease as defined by the TNM classification, whereas tumour grading is of limited value in this respect, mainly due to its low reproducibility. A better means for disease prognostication based on improved understanding of the pathogenetic mechanisms is urgently required.
Materials and methods: Among the cohort of 700 ESCC patients from the high-incidence area of China, previously subjected to extensive testing for Human papillomavirus (HPV) involvement by in situ hybridization (ISH) and PCR, a group of 273 patients was randomly selected for analysis of the primary tumour, adjacent mucosa and regional lymph nodes, by histopathology and quantitative image analysis. All these and the HPV data were subjected to extensive univariate and multivariate analysis to disclose independent predictors of progressive disease.
Results: For the analyses, the tumours were graded into two categories: well-moderately and poorly-differentiated. HPV DNA was detected in 116 (18.9%) of the carcinomas by ISH and in 15.2% by PCR. In univariate analysis, lymph node status (considered as the surrogate marker of progressive disease) was significantly (p < 0.01) predicted by the following nuclear parameters: nuclear area, G0/G1 ratio, HPV DNA status, integrated optical density (IOD), mean optical density (MOD) and S-Phase. In multivariate (stepwise backward LR) analysis, 6 variables remained as independent predictors of disease progression (at p < 0.05 level), the three most significant ones being nuclear perimeter, nuclear roundness and equivalent diameter (p < 0.01).
Conclusion: A series of quantitatively measured nuclear parameters seem to bear a close correlation with ESCC differentiation and progression in univariate analysis and some of these variables proved to be significant independent predictors of disease progression in multivariate modelling as well. These data clearly advocate the use of quantitative image analysis in searching for additional prognostic factors of ESCC.