Abstract P1-04-04: Towards methodology of detecting PIK3CA mutations on single cell level by massive parallel sequencing

Background: The oncogene PIK3CA encodes the p110α catalytic subunit of phosphatidylinositol 3-kinase. Gain-of-function mutations occur in 20 – 40% of the diagnosed breast cancer cases and may predict response to PI3K/AKT/mTOR inhibitors. We have developed a methodology to determine the PIK3CA exon 9 and 20 mutational status on single cell level, allowing analysis of low-input patient material such as circulating tumor cells (CTC). Methods: Single and groups of cultured breast cancer cells (BT20, MDA-MB-361 and MCF7) were isolated and recovered from spiked blood samples using the CellSearch and DEPArray platform respectively, and subjected to whole-genome amplification. Additionally, Formalin-fixed, paraffin-embedded (FFPE) tissue sections from 3 breast cancer patients were collected, of which two were mutant (MT) and one carried the wild-type (WT) genotype, as previously determined by external laboratories. Targeted PIK3CA PCR for exon 9 and 20 with unique multiplex identifiers (MIDs) allowed mixing of 32 samples and differentiation in the sequence data. Massive parallel sequencing was carried out using Titanium Amplicon chemistry (Lib-A) on a 454 GS Junior and data were analyzed using the amplicon variant analyzer (AVA) software. Results: A throughput of 120.464 high-quality reads with a mean coverage depth of 987 reads (range 288 – 5979) allowed the reliable assessment of variants, which were concordant with NCBI SNP/COSMIC databases and external laboratories. In BT20, the variants and frequencies were the missense P539R (48,83%) and H1047R (45,36%) mutations. Missense E545K mutation was detected in MDA-MB-361 (33,33%) and MCF7 (70,83%). Above mentioned mutations were detected in 73,3% and 100% of analyzed single and groups of tumor cells, respectively. Inability to determine the mutational status in some single cells was due to loss of amplicon during purification. In the MT FFPE tissue sections, two variants were detected and were concordant with previously obtained results. The variants and frequencies detected were E542K (14,12%) and E545K (21,38%), and correlated well with the tumor cell content of the actual sample (40% and 50%, respectively). Additionally, variant analysis showed 2 different SNPs in exon 20, which are classified as the synonymous mutations T1025T and A995A. In the WT sample no PIK3CA mutations were detected, suggesting a 100% specificity on the obtained 454 data. Conclusions: Our study delivers a semiautomated workflow using next-generation sequencing for the assessment of PIK3CA gain-of-function mutations on FFPE tissue sections and on single DEPArray purified cells from CellSearch pre-enriched whole blood samples. We are now implementing this workflow in the analysis of circulating tumor cells from patients with metastatic breast cancer. At the time of submitting, a cohort of 40 patients with luminal breast cancer was selected to assess the PIK3CA mutational status in the primary tumor, CTC and (if present) metastatic compartment. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P1-04-04.