This article is mentioned in:

Introduction

Lung and colon cancers are among the main causes of death in developed countries. The life expectancy of patients is very limited, especially in metastatic disease. Nevertheless, in recent years, the development of targeted therapies (Tyrosine Kinase inhibitors or inhibitors of receptors which hyperactivate survival pathways) has shown great therapeutic promise. For example, patients with a mutated EGFR gene and wild-type KRAS gene lung tumor are eligible for gefitinib therapy (1). In colon cancer, patients with wild-type KRAS and BRAF could be treated with panitumumab or cetuximab (2). In skin cancer, vemurafenib (3) or imatinib (4) can be used to treat mutated BRAF (V600E) or c-KIT melanoma, respectively. As the efficacy of these targeted therapies depends on specific genetic abnormalities, molecular diagnosis has become essential for the treatment of cancers. Since 2008, molecular biology platforms have screened for genetic alterations in EGFR (exons 18–21), KRAS (codons 12 and 13) and BRAF (codon 600). At the beginning, the gold standard was Sanger sequencing, but the technique has low sensitivity and is expensive. With the increasing number of samples and gene alterations to screen for, alternon-amplified techniques were developed. For example, allelic discrimination (5) was developed to screen for a specific mutation quickly at a relatively low price. In parallel, screening technologies, such as High Resolution Melting (6) or fragment analysis (for indel alterations), were developed to use with Sanger sequencing, but only in the presence of potential mutations in the region of interest. Over the years, the number of genes with genetic alterations that could be targeted by therapies has increased rapidly. This medical progress has led to the need for more and more molecular diagnoses. This need has now been met by the recent development of Next Generation Sequencing (NGS), which has revolutionized molecular diagnosis. Indeed, the sequencing capacity allows the analysis of dozens of genes on multiplexed samples. In this paper, we describe the results we obtained with the Truseq Cancer Panel. In addition to the routinely detected mutations, NGS analysis, thanks to its high sensitivity, revealed new mutations in routinely analyzed genes.

Materials and methods

Patients and DNA samples

Eighteen tissue samples with >400 ng gDNA (Table I) from patients treated at the Centre Georges-François Leclerc between 2009 and 2013 were randomly chosen. Genomic DNA was extracted from FFPE tissues with either the QIAamp DNA mini kit (Qiagen, Heidelberg, Germany) or the Maxwell 16 FFPE Plus LEV DNA purification kit (Promega, Madison, USA). The samples had already been genotyped by allelic discrimination, fragment analysis and Sanger sequencing. Written consent was provided by all patients, and the researchers obtained authorization from the diagnostic centers to use the tumor samples.

Table I Clinical details of studied patients.

Table I

Clinical details of studied patients.

Patients	Organ of origin	Histology	Age (years)
L1	Lung	Keratinizing poorly differentiated squamous carcinoma	61
L2	Lung	Moderately differentiated adenocarcinoma	67
L3	Lung	Poorly differentiated adenocarcinoma	62
L4	Lung	Adenocarcinoma	50
L5	Lung	Adenocarcinoma	69
L6	Lung	Adenocarcinoma	55
L7	Lung	Mucus-secreting adenocarcinoma	68
L8	Lung	Acinar differentiated mucus-secreting adenocarcinoma	86
C1	Colon	Moderately differentiated adenocarcinoma	55
C2	Colon	Adenocarcinoma	64
C3	Colon	Moderately differentiated adenocarcinoma	59
C4	Colon	Adenocarcinoma	72
C5	Colon	Adenosquamous adenocarcinoma	70
C6	Colon	Poorly differentiated adenocarcinoma	55
C7	Colon	Adenocarcinoma	80
C8	Colon	Poorly differentiated adenocarcinoma	72
C9	Colon	Well differentiated infiltrating lieberkunien adenocarcinoma	58
C10	Colon	Colloidal adenocarcinoma	67

Whole genome amplification

The Repli-g FFPE kit (Qiagen) was used to amplify 300 ng of gDNA from patients L7 and L8: 10 μl of gDNA solution were mixed with 8 μl of FFPE buffer, 1 μl of ligation enzyme and 1 μl of FFPE enzyme. The solution was then incubated at 24°C for 30 min, at 95°C for 5 min and then kept at 4°C. The Repli-g master mix was prepared by mixing, per sample, 29 μl of Repli-g Midi reaction buffer and 1 μl of Repli-g Midi DNA Polymerase. This second mixture (30 μl) was added to the gDNA solution. This solution was then incubated at 30°C for 2 h, 95°C for 10 min and kept at 4°C. Amplified DNA was stored at −20°C. Thanks to this protocol, we obtained 6900 and 6400 ng of amplified gDNA.

Preparation of libraries

Libraries were prepared with the Truseq Cancer Panel (Illumina, San Diego, USA) by following the manufacturer protocol. Briefly, 400–1250 ng of gDNA in 5 μl water was hybridized with an oligo pool. Then, unbound oligos were removed, and extension-ligation of bound oligos was followed by PCR amplification. PCR products were cleaned and checked for quality using Tapestation analysis (Agilent). The PCR product size had to be around 350 bp. Before sequencing, the libraries were normalized by the normalization process of the Truseq Cancer Panel.

Sequencing with MiSeq device

As each library possessed a specific primer index combination (i5 and i7), the libraries were pooled for 2 sequencing runs (pool no. 1, 10 libraries; pool no. 2, 9 libraries). For the MiSeq sample sheet, each sample was identified by its specific index combination. Libraries were paired-end sequenced with 2×151 bp cycles.

Analysis of obtained sequences

At the end of the run, sequences were aligned to the human genome reference hg19. Generated BAM files were analyzed with the Genome Golden Helix software (Golden Helix, Bozeman, USA). A genetic variation was defined by a Q-score above 30 (except for indel alteration).

Results

All mutations detected with standard methods were detected with NGS

In the routine diagnosis of lung or colon carcinomas, mutations in KRAS (exon 2), EGFR (exon 18–21), BRAF (exon 15) and HER2 (exon 20) genes are analyzed using three different methods: allelic discrimination for targeted mutations, fragment analysis for the screening of indel variations and Sanger sequencing for non-targeted mutations or characterization of indel abnormalities detected by fragment analysis. All mutation hotspots are analyzed one by one. Over the years, more and more genes and mutation hotspots will need to be explored. For example, exons 3 and 4 of KRAS, and exons 2–4 of NRAS and HRAS need to be analyzed before anti-EGFR antibody can be prescribed for colon cancer (7,8). In the first step of our study, we used the Truseq Cancer Panel kit to sequence samples that had already been analyzed in routine diagnosis. We then compared the results obtained by NGS with the results of the routine diagnosis at the same mutational hotspots. All the mutations detected in the routine diagnosis were also detected by NGS (Fig. 1 and Table II). Moreover, mutations not found in routine diagnosis were detected by NGS. These included a 15-nt deletion (c.2235_2249delGGAATTAAGAGAAGC) in two lung carcinomas classified as wild-type using routine methods (patients L1 and L6). In the L1 sample, another mutation in the KRAS gene (G12D) was also identified. In patient L6, this 15-nt deletion in EGFR was concomitant with a V600E BRAF mutation.

Figure 1 All the mutations detected during routine diagnosis were also detected by NGS. (A) G12D KRAS mutation detected by allelic discrimination (upper panel) and by NGS (lower panel) for patient C4. Orientation of the KRAS gene needs to be completed to obtain the coding sequence. (B) A 24-nt deletion in exon 19 of EGFR obtained with Sanger sequencing (upper panel) for patient L3. Unlike the Sanger method, NGS analysis (lower panel), made it easy to identify the exact deletion (23 nt + 1 nt). The double blue arrow shows the start of the deletion. (C) V600E BRAF mutation detected by allelic discrimination (upper panel) and by NGS (lower panel) for patient L6. Orientation of the BRAF gene needs to be completed to obtain the coding sequence. The arrows indicate the orientation of each gene.

Table II Comparison of results obtained routinely and with NGS.

Table II

Comparison of results obtained routinely and with NGS.

	KRAS		BRAF		EGFR		HER2
Patients	Routinea	NGS	Routineb	NGS	Routinec	NGS	Routined	NGS
L1	WT	G12D	WT	WT	WT	15-nt E19	WT	WT
L2	WT	WT	WT	WT	15-nt E19	15-nt E19	WT	WT
L3	WT	WT	WT	WT	24-nt E19	24-nt E19	WT	WT
L4	WT	WT	WT	WT	L858R	L858R	WT	WT
L5	WT	WT	WT	WT	15-nt E19 T790M	15-nt E19 T790M	WT	WT
L6	WT	WT	V600E	V600E	WT	15-nt E19	WT	WT
L7	G12C	G12C	WT	WT	WT	WT	WT	WT
L8	WT	WT	WT	WT	WT	WT	WT	WT
C1	WT	WT	WT	V600R	ND		ND
C2	WT	WT	WT	WT	ND		ND
C3	G13D	G13D	WT	WT	ND		ND
C4	G12D	G12D	WT	WT	ND		ND
C5	WT	WT	WT	WT	ND		ND
C6	WT	WT	WT	WT	ND		ND
C7	G13D	G13D	WT	WT	ND		ND
C8	WT	WT	WT	WT	ND		ND
C9	WT	WT	WT	WT	ND		ND
C10	WT	WT	WT	WT	ND		ND

a Allelic discrimination (codons 12 and 13),

b Allelic discrimination (codon 600),

c Allelic discrimination (codons 719, 790, 858 and 861), fragment analysis and Sanger sequencing (exons 19, 20 and 21)

d fragment analysis and Sanger sequencing (exon 20).

{ label (or @symbol) needed for fn[@id='tfn5-ijo-45-03-1167'] } ND, not determined.

{ label (or @symbol) needed for fn[@id='tfn6-ijo-45-03-1167'] } Characters in bold show the discordance between routine and NGS analyses.

In colon cancer, a ‘common’ 15-nt deletion in the EGFR gene was detected only with NGS

Up to now, rare mutations of the KRAS gene have not been routinely analyzed in lung carcinomas. In our small population, a Q61H mutation in the KRAS gene was found in the sample L8. This mutation was localized in exon 3, which is not routinely analyzed in lung cancer. No other alteration was found in the routinely analyzed genes (HRAS and NRAS).

In colon cancer, only the genes KRAS, BRAF and very recently NRAS and HRAS are studied. Concerning rare mutations of KRAS, NRAS and HRAS, we detected a Q61K mutation in the NRAS gene in patient C5. As numerous genes were sequenced by the Cancer Panel kit, we analyzed the results obtained for the PIK3CA, HER2 and EGFR genes, which are routinely analyzed in lung carcinomas. No mutations were detected in exon 20 of HER2, or in exons 18, 20 or 21 of EGFR. In exon 20 of PIK3CA, an H1047L mutation was detected in patient C9. Concerning exon 19 of EGFR, a 15-nt deletion (the same as that observed in lung carcinomas) was detected in three patients (C4, C6 and C7). As this region was not routinely analyzed for colon cancer, we decided to perform both fragment analysis and Sanger sequencing. Neither fragment analysis, nor Sanger sequencing was able to detect the 15-nt deletion in exon 19 of EGFR in colon cancer (Fig. 2A). In contrast, NGS sequencing detected the deletion in >8% of sequenced fragments (Fig. 2B) for one patient. The two other patients harbored the mutation in approximately 4% of read sequences. Among these three patients, only one did not present a concomitant KRAS mutation.

Figure 2 A 15-nt deletion detected in exon 19 of EGFR from colon carcinomas. (A) Fragment analysis and Sanger sequencing of EGFR exon 19 in a colon carcinoma. No alteration was observed in the region. (B) In the same sample analyzed with NGS, a 15-nt deletion (c.2235_2249delGGAATTAAGAGAAGC) was observed in about 8% of the read sequences. The arrow indicates the orientation of the gene.

WGA does not alter the NGS sequencing results

An important limitation in routine diagnosis is the quantity of gDNA extracted from FFPE samples and another paraffin block cannot be obtained in most cases. To counteract this limitation, we tested the impact of Whole Genome Amplification (WGA) on two samples of gDNA obtained from FFPE tissues. We then performed allelic discrimination on non-amplified gDNA and amplified gDNA (Fig. 3A). A KRAS G12C mutation was detected in both the amplified and non-amplified sample from patient L7. For patient L8, no KRAS G12C mutation was observed in either sample. In NGS analysis, the KRAS G12C mutation was also observed in patient L7 (non-amplified and WGA) but not in patient L8 (Fig. 3B). We also analyzed other routinely studied genes to compare the sequences before and after WGA. Whatever the gene analyzed, no point mutation was induced by the WGA (e.g., with the V600E BRAF and L858R EGFR hotpoint mutations in Fig. 3C). Even the rare mutation Q61H of KRAS was detected in both non-amplified and WGA gDNA from patient L8 (Fig. 3D). Moreover, the variant allele fraction was not modified after amplification.

Figure 3 Whole Genome Amplification (WGA) of gDNA from FFPE did not alter the mutation profile of patients. (A) Routine G12C KRAS allelic discrimination of gDNA from patients L7 and L8 before and after WGA. (B) Analysis, by NGS, of KRAS exon 2 for the same patients. The profiles obtained with NGS technology are the same as with routine allelic discrimination: L7 WGA signal was lower than the non-amplified L7 signal. Nevertheless, this minor difference would not have affected the diagnosis. (C) WGA did not create genetic alterations, for example with the V600E BRAF and L858R EGFR hotpoint mutations. (D) The rare Q61K KRAS mutation observed in patient L8 was conserved by WGA. The arrows indicate the orientation of each gene.

NGS analysis revealed cancer susceptibility SNP and genetic alterations in some genes

The Illumina Cancer Panel kit studies exons with mutation hotpoints of 48 genes. We therefore analyzed all covered sequences for the 8 lung carcinomas and 10 colon carcinomas. Twenty-eight genetic alterations and three SNP related to cancer susceptibility or different protein activities were found (Table III). The most frequently altered gene was TP53 with nine alterations detected in nine patients. Double mutations in the colon cancer susceptibility genes APC and SMAD4 were detected in two patients (C3 and C8, respectively), suggesting a familial risk of colon cancer in these patients. For the patient with the APC mutation, we detected a concomitant c-MET activating mutation E168D. Concerning patient C8, we found a large number of alterations in different genes (c-KIT, c-MET, FBXW7, FGFR3, FLT3, IDH1, KRAS, RB1, SMAD4 and TP53), suggesting high genetic instability in this SMAD4 mutated tumor. Moreover, thanks to the non-targeted analysis, we detected two BRAF exon 15 mutations, N581S and V600R, which induce intermediate and strong activation of the protein, respectively. Furthermore, two mutations with unknown impact were detected in PIK3CA and PTEN. Concerning SNP, two patients (L7 and C4) harbored the breast cancer susceptibility ATM F858L SNP, and one patient (C10) had the rare c-KIT M541L SNP, which may influence the response to imatinib. Finally, 10 patients (8 heterozygotes and 2 homozygotes H472H) harbored the KDR Q472H polymorphism, which has been reported to increase tumor microvasculature.

Table III Exonic SNP and genetic alterations in other analyzed genes.

Table III

Exonic SNP and genetic alterations in other analyzed genes.

Genes	Nucleotide variation	Protein sequence variation	Patients	Impact
APC	c.2626C→T	R876X	C3	Loss of function (familial mutation)
	c.3944C→T	S1315X	C3	Loss of function (somatic mutation) (26)
ATM	c.2572T→C	F858L	L7, C4	Breast cancer susceptibility SNP (23)
BRAF	c.1742A→G	N581S	C2	Intermediate activated (27)
	c.1798_1799GT→AG	V600R	C1	Strongly activated (28)
c-KIT	c.1621A→C	M541L	C10	SNP with a potential effect on imatinib response (29)
	c.2146G→A	D716N	C8	Possible resistance to imatinib (24)
c-MET	c.504G→T	E168D	C3	Activated (19)
	c.1156C→A	L386I	C8	Unknown (never observed)
FBXW7	c.832C→T	R278X	C8	Uncertain significance (30)
FGFR3	c.1196_1197GC→AG	R399H	C8	Unknown (31)
FLT3	c.2039C→T	A680V	C8	Activated (32)
IDH1	c.290G→A	G97V	C8	Loss of wild-type function (33)
KDR	c.1416A→T	Q472H	H: L1-L3, L5, L7-L8, C5-C6 O: C7, C9	Increased activity SNP (25)
KRAS	c.408T→A	S136K	C8	Unknown
PIK3CA	c.2176G→A	E726K	C2	Unknown (34)
PTEN	c.563A→T	D187V	L1	Unknown
RB1	c.2074_2075insATGA	Y692FsX2	L2, L5	Loss of function
	c.2119T→C	S707P	C8	Unknown
SMAD4	c.1009G→A	E337K	C8	Unknown
	c.1082G→A	R361H	C8	Loss of function (35)
TP53	c.310C→T	Q104X	L1	Unknowna
	c.523C→T	R175V	C8	Unknown
	c.527G→T	C176F	L5	Partially functional/deleteriousa
	c.536A→G	H179R	C3	Non-functional/deleteriousa
	IVS5+2T→G	G187Fs	C1	Unknown
	c.709A→C	M237L	C10	Partially functional/deleteriousa
	c.743G→A	R248Q	C6	Non-functional/deleteriousa
	c.742C→T	R248W	C5	Non-functional/deleteriousa
	c.830G→T	C277F	C7	Non-functional/deleteriousa

a From IARC database (36).

Discussion

Molecular diagnosis is the current challenge in cancer management. Indeed, with the increased number of targeted therapies and resistance mechanisms developed by cancer cells, the molecular analysis of tumors is a very important task to achieve optimal cancer therapy. Sanger sequencing, even when accompanied by alternon-amplified technologies, such as allelic discrimination or high resolution melting technology, has shown its limits. Today, next-generation sequencing is providing exciting new perspectives. In this study, we tested Truseq Amplicon technology for the analysis of mutation hotspots of 48 genes in gDNA extracted from FFPE samples. All of the mutations detected by routine Sanger sequencing, allelic discrimination or fragment analysis (in KRAS, BRAF, EGFR genes) were also identified with NGS analysis. Moreover, other alterations at the mutation hotspots of the routinely analyzed genes were also detected. These additional alterations included a G12D KRAS, a V600R BRAF and a 15-nt deletion in exon 19 of EGFR. The additional deletion in the EGFR gene was concomitant with the G12D KRAS mutation in one patient, and with a V600E BRAF mutation in another patient. KRAS, BRAF and EGFR mutations are normally exclusive (9) but concomitant KRAS and EGFR mutations have already been described (10,11). The identification of concomitant mutations should increase with the higher sensitivity of NGS technologies. Nevertheless, as it may be impossible to confirm these ‘new’ mutations using routine techniques, their clinical relevance and even their existence may be debatable (12). In the same way, some mutations are detected by NGS in very few of the read sequences. For example, the 15-nt deletion found in three colon carcinomas was detected in less than 8% of the read sequences, and among these three colon carcinomas, one had no KRAS/BRAF mutation, one had a concomitant G12D KRAS mutation, and one also had a G13D KRAS mutation. This observation is quite disturbing, and raises two questions: was the 15-nt deletion true, and if so, was this alteration clinically relevant given the small number detected. Today, the only way to have an answer would be to treat these patients with EGFR tyrosine kinase inhibitors or to observe anti-EGFR antibody resistance in these patients. To date, only patients with KRAS (13), HRAS or NRAS mutations (7,8) can be diagnosed as immediately resistant. Concerning our three colon carcinomas, two may benefit from treatment with EGFR TKI as the G13D KRAS mutation does not seem to interfere with the inhibition of the EGFR pathway (14).

With the increase in the number of genes to be analyzed for molecular diagnosis, the quantity of gDNA obtained from FFPE tissues will rapidly become a major problem, especially for lung carcinomas. In this work, we tested Whole Genome Amplification in two lung carcinomas and analyzed the resulting samples by NGS. The genetic profile obtained before and after WGA was qualitatively the same and quantitatively close. Indeed, only the intensity of the G12C KRAS mutation in patient L7 was slightly lower in the amplified sample. The strong similarity between amplified and non-amplified samples is in accordance in very recent studies, which showed that WGA can be safely used for diagnosis (15,16). Moreover, through this experiment, we showed that Truseq Amplicon technology is compatible with samples treated by WGA.

Among the genes or codons studied in the panel but not analyzed in routine molecular diagnosis, we detected 27 different alterations in 14 genes. Of these, 9 were detected in the TP53 gene, which is the most frequently altered gene in cancer (17). Eight mutations were in the DNA binding domain of the protein, indicating that these mutations are deleterious. One mutation occurred in a splice site, inducing a frameshift that may not be deleterious (18). We detected 2 genes with a double mutation, APC and SMAD4. The presence of two mutations in these two colon cancer predisposition genes indicated that these patients could have been members of families with a high risk of colon cancer. Both patients harbored mutations in the c-MET genes. The APC mutated patient had the activating mutation E168D (19), making him/her eligible for crizotinib therapy, which is generally used in lung cancer (20). Concerning the SMAD4 mutated tumor, we detected eight other altered genes, suggesting high genetic instability in this tumor type (21,22). The impact of most of these alterations is unknown. Nevertheless, the activating A680V mutation of FLT3 may be targeted by anti-FLT3 therapies currently in clinical development for the treatment of leukemia.

Three SNP modifying protein sequences were found. ATM F858L polymorphism, detected in two patients, is associated with an increased risk of breast cancer (23), but the small number of patients in our study does not allow us to draw any conclusions with regard to the predisposition for colon and lung cancer. Then, c-KIT M541L polymorphism was found in only one patient. The impact of this polymorphism in not known, but it has been suggested that it may affect the response to imatinib (24). Finally, KDR Q472H polymorphism was the most interesting alteration. Indeed, tumors with histidine show higher vascularization than do tumors with glutamine (25). In the Caucasian population, the frequency of each genotype is 58, 36 and 6% for Q472Q, Q472H and H472H, respectively. In our small population of tumors (n=18), we found enrichment of the histidine allele in 55.5% of our tumors (44.5% Q/Q, 44.5% Q/H and 11% H/H). In lung carcinomas we observed an enrichment of heterozygous tumors (75%), whereas in colon, the enrichment concerned homozygous H/H tumors. Moreover, during the analysis, we found that the presence of each allele was not 50/50 in heterozygous tumors, but varied from 9 to 96% of the read sequences. This raises the question of true polymorphism or a selection of tumor cells with a high angiogenic capacity. To answer this question, it would be necessary to analyze the constitutive DNA of each patient. According to the study by Glubb et al, patients with heterozygous or homozygous histidine tumors could be more sensitive to inhibiting treatments of VEGFR2 or to bevacizumab.

In conclusion, the analysis by NGS of FFPE lung and colon carcinomas identified the alterations highlighted by routine molecular diagnosis techniques. Thanks to its higher sensitivity, NGS analysis revealed new mutations that were not detected routinely. The impossibility to confirm the presence of these mutations by another technology is problematic, and the only way to answer this question is by conducting clinical trials that compare treatments of patients diagnosed by routine techniques or by NGS. Finally, the use of NGS in routine practice could revolutionize the management of cancer patients. Indeed, simultaneous analysis of numerous genes could identify drug-sensitive alterations generally observed in other cancer types (for example a c-MET alteration in a colon carcinoma that would be treated with crizotinib in lung cancer). Nevertheless, high throughput studies that combine NGS analysis and clinical trials need to be performed before NGS analysis can be generalized in routine molecular diagnosis.

Acknowledgements

We thank Philip Bastable for editing the manuscript.

References