Table 1 SARIFA/TAF overlap: pathologic review
Category | Overlap: SARIFA/TAF (in %) | Reasons for discrepancy | |
|---|---|---|---|
1) Wulczyn et al.19 | Supplement 1: closest to centroids | 25/25 (100%) | 1. intervening stroma or inflammation between tumor cells and adipocytes 2. mucinous tumors where no vital tumor cells were directly adjacent to adipocytes but only mucin 3. inflammatory infiltrate or peripheral ganglia in adipose tissue that mimic cancer cells (this may be also associated with poor prognosis, at least shown for inflamed adipose tissue43) Also it must be kept in mind that TAF are random tiles and do not necessarily depict the invasion front; however, the tumor-fat interface is often the invasion front. |
Supplement 1: randomly sampled TAF patches | 16/25 (64%) | ||
Supplement 2: practice patches | 13/25 (53%) | ||
Supplement 3: assessment patches | 63/100 (63%) | ||
2) L’Imperio et al.26 | Fig. 1: Examples of TAF | 3/3 (100%) | |
Supplementary online content | 6/7 (86%) | ||
3) Krogue et al.27 | Fig. 2: Temporal validation | 0/5 (0%) | |
Fig. 2: External validation 1a | 3/5 (60%) | ||
Supplementary Fig. S2 | 8/25 (32%) | ||
Note: In 3), a similar feature to TAF was identified and described as “predominantly adipose and inflammatory cells with occasional tumor cells.” Accordingly, the overlap between SARIFA and this feature was assessed even though this feature did not necessarily match the exact TAF classification as in 1). Overall, our findings indicate a relevant histological overlap between SARIFA and TAF—while there are also some subtle differences. | |||
