22
IASLC ATLAS OF ALK TESTING IN LUNG CANCER
with the Vysis LSI ALK Break Apart FISH Probe Kit, and the criteria should be validated when a different analytic reagent or biologic specimen is used. The probe size may differ among probe designs, and a larger probe size results in both a larger signal size and a shorter distance required for the definition of a split.
Scoring
A minimum of 50 tumor cells is needed when there is one scorer and a minimum of 100 tumor cells is needed when there are two scorers (see more information in the “Cutoff Value” section). Specimens with fewer assessable cells are not suitable for FISH analysis (Camidge 2010). The signal pattern for each tumor cell should be recorded on a scoring worksheet (Figure 8). Scoring is likely to be more accurate when it is done while viewing the tissue under a microscope with single (red and green) and dual interference filters. Image-based scoring has limitations and the image must represent all section depths in order to avoid false interpretation of isolated signals (Figure 9). Copy number gain of native ALK is common in NSCLC (Figure 10) and there is no indication that it is associated with protein overexpression. At this time, we do not recommend that ALK copy number be routinely included as part of scoring.
Cell
Cell 1 Cell 2 Cell 3 Cell 4 Cell 5 Cell 6 Cell 50
Fused signal
2 2 2 1 0 1 2
5' signal
0 1 0 1 1 0 0
3’ signal
0 1 1 0 2 0 0
Pattern
Fused Split Isolated 3' Isolated 5' Split Fused Fused
Summary of Scoring: Total # of cells scored: 50 Total # of cells with fused pattern: 19 Total # of cells with split pattern: 22 Total # of cells with isolated 3' pattern: 5 Total # of cells with isolated 5' pattern: 4 Total # of cells with rearrangement-positive patterns: 22+5=27 Rearrangement-positive cell rate: 27÷50 x100=54%
Figure 8. An example of a worksheet for scoring cells.
Note: The signal size in captured images tends to be 16 planes at 0.4 µm= slightly larger than that seen A No z-stacking 6.4 µm z-stacking B on actual examination under fluorescent microscopy. When Figure 9. Image-based analysis is difficult if z-stacking (consolidation of multiple focus levels into one plane) does not represent the whole depth of the section. scoring is done on captured images, the distance between signals may be underestimated, which may compromise the results of analysis. Another pitfall of using images for scoring is that image capturing often consolidates multiple focus levels into one plane, and as a result, a vertically split signal along the z-axis of the tissue plane may be indistinguishable from a fused signal.