After having observed that galectin-3 was always negative in our first series of benign cases, we then did not systematically analysed all the benign lesions for galectin-3 but focused on lesions which although unambiguously benign were highly cellular.
This prompts the debate of the cost-effectiveness of MDTs when used for routine benign cases.
For BTC versus benign cases, AUROC values for the seven discriminatory peaks were in the range 0.73 to 0.78 (Additional file 2: Table S3).
However, for late and benign cases CA125 performed better than E2F5.
For the histologically benign cases the mean age was 54.7 years ± 7.0 (minimum 47, maximum 72) and for the histologically malignant cases the mean age was 64.0 years ± 13.6 (minimum 33, maximum 86).
However, in terms of discrepant recall decisions between individuals and overall decisions, there were both over-reading (for the normal and benign cases) and under-reading errors (for the malignant cases) for many cases; although there was no statistical difference in the error rate between these three case categories.
For predicting BTC versus benign cases, the best model (((10*log878)*log1350*log5805^1.25)/(log2210^ 1.5 *log2554^(0.75)) > 5) gave a sensitivity, specificity, PPV and NPV of 79.5%, 83.9%, 86.1% and 76.5%, respectively, on the training set data, although could not be tested independently due to a lack of available samples from additional benign cases.
One possible explanation is that BMS patients belonging to that study had a slightly shorter disease duration and cognitive impairment was an exclusion criterion, whereas our patients were not cognitively assessed, thereby allowing for the inclusion of less "benign" cases.
This model however performed less well when tested for the discrimination of BTC and benign cases.
Ductal Carcinoma in Situ (DCIS); R1: reading 1; readingding 2. Suspicious microcalcifications are shown for 6 cases in Figure 1a c (benign cases) and Figure 2a c (malignant cases).
