Single cell mutational analysis of PIK3CA in circulating tumor cells and metastases in breast cancer reveals heterogeneity, discordance, and mutation persistence in cultured disseminated tumor cells from bone marrow

This study protocol was approved by Stanford’s Human Subjects Research and Institutional Review Board (Protocol 5630). Written informed consent was explained and signed by all participating patients prior to sample collection.

Single cell suspensions used for MagSweeper tumor cell isolation were prepared from primary and metastatic tissue from breast cancer patients. Tumor chunks were finely minced, gently pulled to release single cells or small cell clusters, filtered through a 70 micron mesh followed by centrifugation of the filtrate at 1900 g. The supernatant was discarded and the pellet was resuspended in 1x trypsin (Invitrogen/Life Technologies, Carlsbad, CA, USA) for 5–10 minutes. DMEM culture media with 10% FBS (Gibco/Life Technologies, Carlsbad, CA, USA) was added to stop the trypsin reaction. The target single tumor cells were labeled with EpCAM-conjugated microbeads and isolated by the MagSweeper as previously described [19, 21]. Individual tumor cells were aspirated under direct microscopic visualization (Axio Observer A1, Zeiss, Thornwood, NY, USA). Authenticated MCF7 and BT474 human breast cancer cell lines (ATCC, Manassas, VA, USA) were grown in DMEM and trypsinized to release single cells, which were then isolated by the MagSweeper and manually aspirated as single cells.

For immunostaining assays, EpCAM-captured cells were treated with DNase I Solution (StemCell Technologies, Vancouver, BC, Canada) to remove the DNA-linker on the magnetic microbeads, and placed on slides. Tumor cells were defined by immunostain assay [26‐28] as cells that stained positive for purified anti-cytokeratin (CK+) CAM 5.2 (BD Biosciences, San Jose, CA, USA) and DAPI nuclear stain (DAPI+) (VECTASHIELD Mounting Medium with DAPI, Vector Laboratories, Burlingame, CA, USA), and negative for CD45 (CD45-) (CD45 Ab-1/Bra55, NeoMarkers Lab Vision, Thermo Fisher Scientific, Kalamazoo, MI, USA). White blood cells (WBCs) from patients were collected using the MagSweeper and CD45 Dynabeads (Invitrogen, Carlsbad, CA, USA) and similarly immunostained.

Bone marrow aspirates obtained from the clinic were filtered through a 70 micron mesh to eliminate debris; the passed through solution was adjusted to 10 ml by adding DMEM. DTCs from the prepared bone marrow sample solution were isolated by the MagSweeper and hundreds of DTCs were directly cultured in DMEM with 10% FBS and 50 μg/ml of penicillin-streptomycin (Gibco/Life Technologies, Carlsbad, CA, USA). The cells were then identified by immunostaining assays. The DTCs were cultured for 21 days and single cells were assayed for mutation detection.

Blood sample collection and MagSweeper isolation of CTCs were isolated as previously described [19, 21]. White blood cells (n = 15, from the blood of Patient 12) were collected using anti-CD45 microbeads.

Single tumor cells from 17 breast cancer patients were lysed by incubation with a 1:10 dilution of proteinase K (Qiagen, Valencia, CA, USA) in individual PCR tubes for 20 min at 65°C in 10 μl of 1x GeneAmpPCR buffer II (Applied Biosystems/Life Technologies, Carlsbad, CA, USA). In one of the 17 patients, additional tumor cell clusters were tested from slides sectioned from formalin-fixed paraffin-embedded (FFPE) blocks of primary tumor and a lung metastasis with location confirmed by hematoxylin and eosin (H&E) stain. For tumor tissue fixed on slides, diluted proteinase K was placed on the targeted tissue section area of the slide. The solution was heated at 65°C for 20 min and collected for DNA mutation analysis. The DNA from single CTC, DTC, or tumor tissue cells was then pre-amplified for exons 9 and 20 of the PIK3CA gene, using Pfu Ultra DNA polymerase (Agilent Technologies, Santa Clara, CA, USA) with one set of primers for each exon (exon 9 forward primer: CTGTGAATCCAGAGGGGA, reverse primer: CAGAGAATCTCCATTTTAGCAC; exon 20 forward primer: GGAATGCCAGAACTACAATCTTTTG, reverse primer: CCTATGCAATCGGTCTTTGC). The reaction was amplified for 30 cycles at 94°C, 55°C, and 72°C for 30 sec per cycle for each temperature. The pre-amplified products were then diluted 1:10 in distilled water (DW) and 5 μl of the diluted products were used for a single 50 μl PCR using the same primers as above for exon 9, and an internal primer pair for exon 20 that decreased non-specific background amplification (forward primer: GTGGAATCCAGAGTGAGC, reverse primer: TTGCATACATTCGAAAGACC) [25]. PCR products were checked by 2% agarose gel against a GeneRuler 50 bp DNA Ladder (Frementas, Glen Burnie, MD, USA) and sequenced by BigDye Terminator v3.1 Cycle Sequencing Kits according to the manufacturer’s recommended protocol (Applied Biosystems/Life Technologies, Carlsbad, CA, USA). The sequenced results were analyzed with Sequencher 4.8 software (Gene Codes Corporation, Ann Arbor, MI, USA). DW was used as a negative template control. Single authenticated MCF7 and BT474 cells were used as amplification and sequencing controls: MCF7 cells contain a G1633A mutation in exon 9 and are wild type in exon 20; BT474 cells contain an exon 1 mutation but are wild type in exons 9 and 20. BT20 cells, which contain the A3140G mutation in exon 20, served as a positive control for sequencing of exon 20 [29]. Fifteen single WBCs isolated from blood samples also served as wild type control for sequencing.

From a group of 17 breast cancer patients, single tumor cells were collected using anti-EpCAM microbeads and the MagSweeper from 30 blood samples, a bone marrow biopsy, and six fresh tumor tissues (Table 1, and Table 2). CTCs from blood, DTCs from bone marrow, and tumor cells from fresh primary and metastatic tumors were defined by immunostain assay as cells that were CK+, CD45-, DAPI+. WBCs collected from the blood of one patient (Patient 12) using anti-CD45 microbeads for use as an additional wild type control for sequencing were confirmed by immunostaining to be CK negative/very weak, CD45 positive, and DAPI positive (Figure 1). Overall, 769 individual CTCs from blood, 75 DTCs from bone marrow, and 60 single TCs from fresh primary tumor tissue (chunk and core needle biopsy), lymph node metastasis, and a bone metastasis to the spine were collected. Of these, 185 CTCs, 24 DTCs, 33 single TCs from fresh primary and metastatic tissues (total 242 single cells) were collected for mutational analysis. Additional tumor cell clusters from two FFPE samples of primary tumor and a lung metastasis were also analyzed for PIK3CA mutations (Table 3).

The MagSweeper facilitates the isolation of viable target cells that can be grown in culture. Pooled bead-captured DTCs (Figure 2A) showed distinct morphological changes when grown over a 21-day cell culture period (Figure 2B and C). At the time of bone marrow biopsy, DTCs had round shapes when examined fresh and after capture by the MagSweeper (Figure 2A). However, onward from day 5, cell shapes started varying, with some changing to elongated or irregular shapes, although immunostaining assays at different stages demonstrated that all cultured cells remained CK+, CD45- and DAPI+, as expected of tumor cells (Figure 2B and C).

To first assess our mutation detection method, targeted DNA from single tumor cells or cells from cell lines was successfully pre-amplified by PCR and the expected bands were confirmed for PIK3CA exons 9 (216 bp) and 20 (269 bp) (Figure 3A). Using the pre-amplified PCR products as new DNA templates, exons 9 and 20 were separately amplified for sequencing, amplifying exon 9 using the original primers and exon 20 with internal primers. Single PCR bands for each exon verified the specificity of the amplification prior to sequencing (Figure 3B). A known heterozygous PIK3CA exon 9 mutation, G1633A (E545K), was identified in replicate samples of single MCF7 cells (positive control), single DTCs, and single CTCs, but was not detected in single WBCs (wild type/negative control) or single BT474 cells (which carry a PIK3CA mutation in exon 1, but no mutations in exons 9 or 20, another negative control) [30‐33] (Figure 3C). Note that A1634C (E545A) in the chromatogram in Figure 3B sometimes showed a pseudogene that may be co-amplified with these primers [34]. No exon 20 mutation was identified in negative control MCF7 and BT474 cells, any patient tumor cells, or captured WBCs among our samples, although sequencing of BT20 cells (positive control) detected the expected exon 20 mutation at A3140G (H1047R), confirming the accuracy of our sequencing method (Figure 3D).

PIK3CA mutation analysis was then performed on the 242 EpCAM-captured single tumor cells (185 CTCs, 24 DTCs, and 33 tumor cells). Three out of 17 (18%) patients showed the PIK3CA exon 9 G1633A mutation in tissue; no exon 20 mutations were detected (Table 1, Table 2 and Table 3). Of these three patients (Patient 12, Patient 14, and Patient 15), 33 individual tumor cells obtained from six fresh clinical samples were sequenced and three showed the exon 9 mutation: 8/8 (100%) single tumor cells from a spinal bone metastasis (Patient 12); 1/5 (20%) single tumor cells from a primary tumor core biopsy (Patient 14); and 3/5 (60%) single tumor cells from a metastatic lymph node (Patient 15). Not unexpectedly, and similar to this patient’s bone metastasis, all 24 (100%) sequenced single DTCs collected from a 3 ml aspirate of Patient 12’s tumor-replaced bone marrow contained the exon 9 mutation (the bone marrow aspirate had been done to determine if tumor overgrowth was causing her pancytopenia, and the DTCs detected were too numerous to count). Interestingly, the FFPE sample of Patient 12’s primary tumor was wild type, but a tissue section of her lung metastasis also showed the PIK3CA exon 9 mutation.

CTCs from Patient 12 were captured periodically over 14 months, during changing treatment of her metastatic breast cancer (Table 1, Table 2, Table 3 and Table 4, Figure 4). Remarkably, in this patient with mutant DTCs in her bone marrow and mutant tumor cells in her spinal bone metastasis and lung metastasis, CTCs in some blood draw samples were discordant and did not show the expected mutation. Notably, of ten blood samples collected from this patient over time, only nine blood samples contained capturable CTCs; of the 128 CTCs sequenced from these nine samples, mutant CTCs were detected in only two blood draws: at one and six weeks after bone marrow biopsy, with mutant cells comprising 50% (10/20) and 29% (2/7) of EpCAM-captured cells, respectively (Table 1, Table 2, Table 3 and Table 4, Figure 4). As expected, all 15 WBCs analyzed from this patient were wild type. Our single cell data thus indicate that not only may there be mutational heterogeneity within a sample, there may be genotypic discordance between CTCs, DTCs and metastases, or between CTC samples isolated at different time points.

In addition to mutational heterogeneity and discordance, this patient’s clinical pathology showed discordance between different sites when tested for the standard breast cancer biomarkers estrogen receptor (ER), progesterone receptor (PR), and HER2 growth factor receptor: primary tumor and lung were hormone receptor (ER and/or PR) positive, whereas bone marrow and bone were hormone receptor negative (Table 4). No tissue sample was HER2 positive.

Finally, when Patient 12’s DTCs were propagated in vitro, all cultured cells maintained the original PIK3CA exon 9 mutation when again tested on day 21, despite the observed morphological changes described above.