We selected the top 20 ranked amplicons in the FFPE sample for this analysis. In 19 of these 20 amplicons the overlap was >0.9 with the same ADM2-defined interval in the sorted fresh frozen sample. These intervals included a series of focal amplicons on chromosomes 2 and 9 that highlighted known and putative oncogenes (Figure 3). One striking example was a highly focal amplicon that selleck chemicals Cisplatin targeted a single gene, BCL11A, and was detected in both matching samples. Figure 2 Whole genome comparison of aCGH results with matching sorted FFPE and FF samples. Figure 3 Gene-specific comparison of aCGH results with matching sorted FFPE and FF samples. We then assessed the global utility of our FFPE assays with different tissues including TNBCs, bladder carcinoma, glioblastoma, and small cell carcinoma of the ovary (SCCO) (Figures S2, S3, S4, S5, S6) and verified selected aberrations by FISH (Figure S7).

These samples were obtained from multiple tumor banks and contained variable amounts of debris and non-tumor cells. We used single parameter DNA content assays to detect and sort the diploid, aneuploid, and 4N cell populations present in each sample. In each case we were able to discriminate homozygous and partial deletions, and map breakpoints and amplicon boundaries to the single gene level in the sorted samples regardless of tumor cell content. These include potentially clinically relevant aberrations such as focal amplicons of EGFR, USP25, and CCND1, and homozygous deletions in PARD3, CDKN2A, and PTEN. These latter aberrations included single exon deletions.

One striking exception was SCCO a rare tumor that presents in very young women and girls [28]. The SCCO genomes did not contain any focal amplicons or homozygous deletions. However the resolution of our assays with FFPE samples allowed us to map a 1p36.22 breakpoint created by a single copy loss to the CASZ1 locus, a zinc finger gene implicated in neuroblastoma [29] (Figure S6). Sorted FFPE Input for NGS Current NGS protocols typically require larger amounts of genomic DNA template as input. Furthermore widely used methods preselect samples with high (e.g. >70%) tumor content and are dependent on genomic DNA templates of highly uniform quality as inputs for library construction [8], [30]. The small fragment sizes of DNAs typically isolated from routine FFPE samples are not suited for linear amplification with highly processive enzymes such as phi29.

Therefore we investigated the use of single primer isothermal amplification Anacetrapib (SPIA) (NuGEN Ovation) to generate templates from sorted FFPE samples that are suitable for aCGH and NGS. To rigorously test this method we compared aCGH data from matching FF, non-amplified FFPE, and SPIA FFPE samples. We collected aliquots of 10,000, 25,000 and 50,000 nuclei during sorts of individual pancreas FFPE samples. Each sorted aliquot was extracted, amplified, labeled, and then hybridized to 400 k CGH arrays.