ABSTRACT Background

Although anti-EGFR therapy has established efficacy in metastatic colorectal cancer, only 10-20% of unselected patients respond. This is partly due to KRAS and BRAF mutations, which are currently assessed in the primary tumor. To improve patient selection, assessing mutation status in circulating tumor cells (CTCs), which possibly better represent metastases than the primary tumor, could be advantageous. We investigated the feasibility of KRAS and BRAF mutation detection in colorectal CTCs by comparing three sensitive methods and compared mutation status in matching primary tumor, liver metastasis and CTCs.

Methods

CTCs were isolated from blood drawn from 49 patients before liver resection using CellSearch™.

DNA and RNA was isolated from primary tumors, metastases and CTCs. Mutations were assessed by co-amplification at lower denaturation temperature (COLD)-PCR (Transgenomic™), real-time PCR (EntroGen™), and nested Allele-Specific Blocker (ASB-)PCR and confirmed by Sanger sequencing.

Results

In 43 of the 49 patients, tissue RNA and DNA was of sufficient quantity and quality. In these 43 patients, discordance between primary and metastatic tumor was 23% for KRAS and 7% for BRAF mutations. RNA and DNA from CTCs was available from 42 of the 43 patients, in which ASB-PCR was able to detect the most mutations. Inconclusive results in patients with low CTC counts limited the interpretation of discrepancies between tissue and CTCs.

Conclusion

Determination of KRAS and BRAF mutations in CTCs is challenging but feasible. Of the tested methods, nested ASB-PCR, enabling detection of KRAS and BRAF mutations in patients with as little as two CTCs, seems to be superior.

INTRODUCTION

The introduction of new drugs such as monoclonal antibodies directed against EGFR has improved the life expectancy of colorectal cancer patients³⁵⁸. Unfortunately, only 10 - 20%

of unselected metastatic colorectal cancer patients respond^358-360, which is partly due to activating mutations in genes downstream of the EGF-receptor, such as KRAS and BRAF.

KRAS mutations are present in 30 - 40% of colorectal cancer patients³⁶¹. Extensive descriptions of the inactivity of anti-EGFR therapy in KRAS-mutated patients4,334,362-365

show that these agents generate a response in 17 - 40% of patients with KRAS wild-type (wt) tumors versus 0%

of patients with KRAS mutated (mt) tumors⁴. Based on these results, the European Medicines Agency has approved the use of monoclonal antibodies against EGFR solely for patients with KRASwt tumors³⁶⁶.

Another potentially predictive mutation for response to EGFR-inhibiting therapy is BRAF, which is present in ~10% of colorectal cancer patients³⁶¹. Evidence for the predictive value of BRAF mutations is not as abundant as for KRAS^367-368, and a mutated BRAF is not yet an exclusion criterion for this therapy. However, if the same mechanism applies, which is very likely given the similar important role of BRAF and KRAS in the EGFR pathway, BRAF mutations could be as an important predictive factor as KRAS.

Although KRAS and BRAF mutation status are currently determined in the primary tumor, primary tumor tissue is not always available, is of insufficient quality, or has been obtained years before the diagnosis of metastatic disease. Importantly, mutation status of the primary as well the metastatic lesions can change over time and during the course of therapy^369-370. In this regard, ideally, mutation status of the patient’s actual metastatic tumor load would be assessed right before treatment is started. However, metastatic tissue is often hard to obtain, and usually only through invasive and painful procedures. These drawbacks could potentially be overcome by assessing mutation status of circulating tumor cells (CTCs), which can be present in the peripheral blood of metastatic colorectal cancer patients^371-372.

Here we describe the use of three techniques to assess KRAS and BRAF mutation status in enriched CTC fractions, consisting of CTCs and >100-fold excess DNA from leukocytes⁹. A fast nested Co-amplification at Lower Denaturation temperature (COLD-)PCR combined with Surveyor®/WAVE® denaturing High-Performance Liquid Chromatography (HPLC) followed by sequencing (Transgenomic®, Omaha, NE), a commercially available real-time PCR kit (EntroGen™, Tarzana, CA), and a nested Allele-Specific PCR with a Blocking reagent (ASB-PCR)³⁷³ were tested and compared for their ability to detect mutations in CTCs of colorectal cancer patients with liver metastases undergoing partial liver resection. In addition, the mutation status in matched primary and metastatic tumor tissue was determined and correlated to each other and to CTC mutation status.

METHODS

Patients and ethics statement

From 63 patients with metastatic colorectal cancer, 2 x 30 mL blood samples were taken for CTC enumeration and characterization by way of venipuncture before liver metastasis resection and prior to tumor manipulation. This study was approved by Leiden University Medical Center and Erasmus University Medical Center Institutional Review Boards (METC P05.182), and all patients were enrolled in Erasmus MC, Rotterdam, Netherlands after written informed consent was obtained.

Cancer cell lines and synthetic DNA

Colorectal cancer cell lines HCT116 and breast cancer cell line SK-BR-3 were obtained from ATCC (Manassas, VA) and cultured under recommended conditions. HCT116 was previously established to harbor a heterozygous G13D, G>A KRAS mutation³⁷⁴, and SK-BR-3 is BRAF and KRAS wild-type²²⁶. For cell line experiments, cells were harvested at log phase and counted by Improved Neubauer Hemacytometer (Hausser Scientific, Horsham, PA). A range of synthetic DNA concentrations as supplied in the EntroGen kit (see next) were tested in the EntroGen kit and in the ASB-PCR to assess assay detection limits.

Enumeration and isolation of CTCs

Two samples of 30 mL blood from 49 metastatic colorectal cancer patients about to undergo liver metastasis resection were drawn in CellSave™ tubes (Veridex LLC, Raritan, NJ) for CTC enumeration or EDTA tubes for CTC isolation. Prior to enumeration and isolation, a density gradient-based enrichment step was applied as described before^337-338. For CTC isolation, samples were then processed on the CellTracks™ AutoPrep System (Veridex LLC) using the CellSearch™ Profile kit (Veridex LLC). For CTC enumeration, samples were processed using the CellSearch™ Epithelial Cell Kit (Veridex LLC) and CTC counts were determined on the CellTracks™ Analyzer (Veridex LLC) according to the manufacturer’s instructions.

mRNA and DNA isolation from CellSearch-enriched CTC fractions and tissue

After removal of the supernatant using a MagCellect Magnet (R&D Systems, Minneapolis, USA), the cells in the enriched CTC fractions were lysed by adding 250 μL of Qiagen AllPrep DNA/RNA Micro Kit Lysis Buffer (RLT+ lysis buffer) (Qiagen, Valencia, CA) and stored immediately at -80°C until DNA and RNA isolation was performed with the AllPrep DNA/RNA Micro Kit (Qiagen) according to the manufacturer’s instructions and as described before³⁸. For analysis of primary and metastatic tumors, total RNA and DNA was isolated from fresh frozen tissue with RNA-Bee (AMSBIO, Abingdon, UK) as described before²⁹⁰ and from FFPE tissue with the

column-based High Pure RNA Paraffin Kit (Roche Applied Science, Penzberg, Germany) according to the manufacturer’s instructions. After DNA and RNA extraction from CellSearch-enriched CTCs and tissues, quality and quantity checks were performed by evaluating the levels of a set of reference genes by real time PCR as described before^38,375. For tissue RNA, this was preceded by measurements with a Nanodrop Spectrophotometer and agarose gel electrophoresis, and for genomic DNA (gDNA) by measurements with the Quant-iT PicoGreen dsDNA reagent (Life technologies, Carlsbad, CA).

Fast COLD-PCR & Surveyor/WAVE technology

Fast COLD-PCR (Transgenomic, Omaha, NE) exploits the observation that a single-nucleotide mismatch along a double-stranded DNA sequence results in a change of the melting temperature (T_m) for that sequence, so that when PCR denaturation temperature is set to a temperature slightly lower than T_m, DNA amplicons differing by a single nucleotide are selectively denatured and amplified, enriching the sample for mutated alleles³²⁹. Assay sensitivity was established by the manufacturer to be 0.1% (http://www.transgenomic.com/lib/ps/602136-00.pdf).

For COLD-PCR, 10ng gDNA from CellSearch-enriched CTC fractions, consisting of CTCs and

>100-fold excess DNA from leukocytes, was selectively pre-amplified with KRAS and BRAF primers by two rounds of COLD-PCR as described before³⁷⁶. Amplified exons were analyzed by Surveyor digestion using the Transgenomic SURVEYOR kit, in which the surveyor nuclease digests mismatch-containing DNA. After purification of digested DNA using the QIAquick PCR purification kit (Qiagen), samples were analyzed by WAVE technology according to the manufacturer’s instructions³⁷⁶. For sequencing, amplified exons were fractionated by denaturating HLPC, and fractions of interest were again amplified by PCR. PCR products were purified as above and sequenced using the Applied Biosystems PRISM® 3730 sequencer.

EntroGen KRAS/BRAF kit

EntroGen provides a commercially available real-time PCR kit for KRAS and BRAF mutation detection, which uses allele-specific primers without employing a pre-amplification step.

For the EntroGen assay, 50 ng gDNA from tissues was amplified with gene specific primers and the KRAS codons 12, 13 and 61 and BRAF V600E mutations were detected using allele-specific probes (EntroGen), as described previously³⁷⁷. For CTC samples, a nested PCR was first performed on 10ng gDNA extracted from CellSearch-enriched CTCs, followed by a second PCR with the allele-specific probes as described before³⁷⁷.

ASB-PCR assay

ASB-PCR is an assay developed to suppress the amplification of primer:template mismatches.

Two key features of the assay are a mutant-specific primer that is shortened at its 5’-end to reduce the T

m to approximately 10°C below the annealing temperature of the assay, and the use of a blocking oligonucleotide, which has a sequence complementary to the wild-type sequence but is phosphorylated at the 3’-end to prevent its extension. The combination of these two modifications results in suppression of the amplification of wild-type allele, and the assay has been reported to be capable of detecting mutant alleles with DNA inputs between 2 and 250 pg among a thousand-fold excess of wild-type DNA³⁷³.

ASB-PCR was performed essentially as described before³⁷³. Briefly, cDNA was synthesized using 5 µL CTC RNA with Invitrogen’s cDNA Reverse Transcriptase kit (Life Technologies, Carlsbad, CA). All 10 µL of the resulting cDNA was then used in a nested PCR using 2x Expression Master Mix (TaqMan Gene Expression MasterMix kit, Life Technologies) and 2 µL 0.5 µM Primer Mix in a 20 µL reaction volume. For CTC samples, 2 µL PCR product from the first amplification (above) was used for the second round of amplification; while with DNA isolated from tissues, 5 10 ng gDNA was used, both in a 20 µL assay with 2 x TaqMan Gene Expression Mastermix and 10x Primer/Probe/Blocker mix (sequences and cycling conditions in Supplementary Table 1).

TOPO TA cloning and Sanger sequencing

PCR products from samples that were estimated to have sufficient mutation were cloned with the TOPO® TA Cloning® Kits for Sequencing (Life Technologies) according to the manufacturer’s instructions. Positive colonies were amplified individually under the same PCR conditions as the first round of amplification of the nested-PCR. Amplified products were purified using QIAquick PCR Purification kit (Qiagen) and sent to GENEWIZ (South Plainfield, NJ) for Sanger dideoxy terminator sequencing³⁷⁸.

RESULTS