Biomedical Engineering Reference
In-Depth Information
only partially degraded genomic DNA can be prepared. While it is possible
to perform matrix-CGH from such archived collections, the success of these
studies is highly dependent on the quality of the fixative (buffered formalin
is required) and the duration of the fixation. As this information is often not
available, the success rate of studies on archived material can be quite limited.
For labelling of genomic DNA probes, a number of different protocols are
used, e.g. incorporation of dye-labelled nucleotides by nick-translation, uni-
versal PCR or random primer extension. Apart from labelling of very small
DNA amounts, e.g. from microdissected tumor samples with universal PCR
reactions, random labelling currently is the most widely used protocol for
matrix-CGH. In comparison to expression profiling by DNA microarrays,
matrix-CGH has a much higher demand with respect to sensitivity and sig-
nal linearity. This is due to the necessity to measure subtle ratio differences
for the detection of monoallelic gains or losses of
<
0.5 (0.5 for a deletion
and 1.5 for a trisomy compared to 1.0 for a balanced state). Therefore tiny
variations in signal intensities have to be detected with high accuracy. Ad-
ditionally, in contrast to constitutional diseases, ratio differences in primary
tumor specimens are diminished according to the sample's content of 'con-
taminating' stromal tissue (fibrocytes, leukocytes, vessel endothelial cells) or
adjacent non-malignant cells exhibiting a normal diploid karyotype. For reli-
able genomic profiling, the proportion of cells to be analyzed, e.g. the content
of tumor cells, should be at least 50%.
Fig. 12.2.
Matrix-CGH profile of the tumor cell line HL60. BAC and PAC clones
are arranged in chromosomal order along the X-axis. Linear normalized ratios of
HL60 and male-control DNA fluorescence signals are shown on the Y-axis. All the
genomic imbalances characteristic of HL60 are detected
