A random sample of SCC tumors from our TMAs were selected to calculate tumor staging using our linear regression model when considering only E-cadherin score.

The tables show the score for phospho-Rb S249, p39, E-cadherin and the patient´s tumor stage (first fourth columns, respectively). The fifth column shows the stage predicted by our linear regression analysis the when the three scores are combined. The Error (sixth) columns in all five tables show the difference between the vendor score and our linear regression analysis score, a value of 0 indicating identical scores, while a number other than 0 means that the two scores were different by a magnitude equal to the number indicated in the column. Finally, all three scores combined accurately predicted the tumor stage in 5 out of 7 samples. A linear regression model having only linear terms can result in positive or negative predictions if no other restrictions are present. In contrast, the three scores combined were able to accurately predict tumor stage in 5 out of 7 samples, and in the two samples in which the prediction was not perfectly accurate, they overestimated tumor stage by only a magnitude of 1. This suggests that the combination of phospho-Rb S249, p39, and E-cadherin scores may have strong predictive power for assessing stage.