Abstract
Background
Several targeted therapies have been approved for treatment of solid tumors. Identification of gene mutations that indicate response to these therapies is rapidly progressing. A 34-gene next-generation sequencing (NGS) panel, developed and validated by us, was evaluated to detect additional mutations in community-based cancer specimens initially sent to our reference laboratory for routine molecular testing.
Methods
Consecutive de-identified clinical specimens (n = 121) from melanoma cases (n = 31), lung cancer cases (n = 27), colorectal cancer cases (n = 33), and breast cancer cases (n = 30) were profiled by NGS, and the results were compared with routine molecular testing.
Results
Upon initial mutation testing, 20 % (24/121) were positive. NGS detected ≥1 additional mutation not identified by routine testing in 74 % of specimens (90/121). Of the specimens with additional mutations, 16 harbored mutations in National Comprehensive Cancer Network guideline genes. These various additional mutations were in gene regions not routinely covered, in genes not routinely tested, and/or present at low allele frequencies. Moreover, NGS yielded no false negatives. Overall, NGS detected mutations in 59 % of the genes (20/34) included in the panel, 75 % of which (15/20) were detected in multiple tumor types. Mutations in TP53 were found in 51 % of tumors tested (62/121). Mutations in at least one other (non-TP53) gene present in the panel were detected in 64 % of cases (77/121).
Conclusion
This assay provides improved breadth and sensitivity for profiling clinically relevant genes in these prevalent solid tumor types.
Similar content being viewed by others
This study demonstrates that the 34-gene next-generation sequencing mutation profiling test for solid tumors can detect additional mutations not identified by routine (single-gene or mutation) molecular testing. |
The breadth of additional mutations identified by the 34-gene panel suggests a need to assess the response characteristics of tumors harboring various mutations in parallel or alternative signaling pathways, or co-occurring mutations; and the potential benefits of broader mutation profiling. |
1 Introduction
Cancer has become an increasing focus for drug development in recent years, with the number of new cancer drugs in development tripling from 2001 to 2010 [1]. The often modest survival benefit and high degree of adverse effects from standard chemotherapy and radiation treatment have led researchers to focus new drug development on treatments that target specific signaling molecules or whole regulatory pathways. The efficacy of many targeted drugs correlates with molecular biomarkers, which are now commonly used to help select patients for treatment [1–3]. The list of biomarkers shown to predict the therapeutic response of solid tumors has steadily increased over the years. Such markers include somatic mutations in drug targets and components of receptor tyrosine kinase (RTK) and phosphatidylinositol 3-kinase (PI3K) signaling pathways (e.g., EGFR, KRAS, BRAF, ERBB2, KIT) [4–9]. Traditional testing for mutations is based on treatment selection targets in individual genes or mutational hotspots related to the specific tumor and drug [10–15].
Several factors argue for more comprehensive mutational profiling of tumors. For example, the mutational heterogeneity of solid tumors has broadened the scope of cell signaling pathways targeted by new therapeutics [16–18]. Identifying pathway alterations could steer the clinician toward or away from drugs that target the affected pathways. Furthermore, approved biomarkers and drug treatments for one tumor type may have potential applications in other tumor types. One example of successfully applying a single targeted treatment across multiple tumor types and biomarker indications is the use of the tyrosine kinase inhibitor (TKI) imatinib (imatinib mesylate). Imatinib was originally developed to target c-abl in chronic myeloid leukemias (CMLs) harboring the Philadelphia chromosome (BCR/ABL1), but its indications have since been expanded to include the treatment of certain gastrointestinal stromal tumors (GISTs) harboring KIT or PDGFRA mutations, as well as advanced or metastatic melanomas harboring KIT mutations [19, 20]. Other approved targeted treatments may have clinical utility in additional tumor types. Vemurafenib and dabrafenib, for example, are approved for treatment of V600 mutation-positive melanoma but have been recently recommended as options for BRAF V600 mutation-positive non-small cell lung cancer (NSCLC) [21, 22]. The rapid pace of drug development and overlap in affected regulatory pathways among cancers suggest a potential benefit of prospective profiling of tumors for targets that might not yet be clinically actionable for a given tumor type.
Another important utility for mutation profiling is to identify mutations in signaling proteins “downstream” from therapeutic targets, which can cause acquired drug resistance during targeted treatment [23–27]. In addition, treatment-naïve patients frequently have concurrent mutations in multiple genes that may have implications for prognosis or treatment decisions [28–34].
Given the overlap in genes found to be mutated in multiple solid tumor types, we aimed to develop a “universal” solid tumor mutation profiling assay. A set of 34 cancer-associated genes was carefully selected on the basis of NCCN guideline recommendations [10–15], potential for informing therapy selection in solid tumor patients, and prevalence of somatic mutations in solid tumors. Twenty-two (65 %) of the 34 genes have mutations with known or potential clinical significance in at least two solid tumors [35–42]. Moreover, according to data from The Cancer Genome Atlas (TCGA; http://www.cBioPortal.org, accessed in October 2013), 26 of the 34 genes are mutated in at least 1 % of melanoma, lung cancer, and colorectal cancer (CRC) cases [43, 44]. To demonstrate the applicability of this sequencing panel in clinical samples, we used it to retrospectively profile a set of formalin-fixed, paraffin-embedded (FFPE) tumor specimens (n = 121) initially submitted to our clinical laboratory for routine molecular testing. The 34-gene next-generation sequencing (NGS) test results were compared with the initial laboratory test results to ascertain whether broader mutational profiling could provide additional information relevant to therapy selection.
2 Methods
2.1 Patients and Specimens
This study comprised de-identified FFPE samples submitted to the Quest Diagnostics Nichols Institute (San Juan Capistrano, CA, USA) for tumor marker analysis from October 2013 through April 2014. The study was exempt from Institutional Review Board oversight, as determined by the Western Institutional Review Board. The diagnosis of each cancer was confirmed by pathology review. A total of 133 FFPE samples were initially collected, 12 of which were excluded because of insufficient DNA (<10 ng). The remaining 121 samples that were analyzed included CRC (n = 33), lung cancer (n = 27), melanoma (n = 31), and breast cancer (n = 30).
2.2 DNA Extraction
Hematoxylin and eosin–stained slides of each sample were reviewed by a pathologist to identify tumor-rich areas and estimate the tumor fraction. After manual macrodissection, total DNA was extracted from one to five 10 µm unstained sections, depending on the tumor area, using a DNA extraction kit (Roche Molecular Diagnostics, Indianapolis, IN, USA). DNA was quantified with a Qubit DNA HS assay kit (Life Technologies, Carlsbad, CA, USA).
2.3 NGS Analysis
2.3.1 Ion Torrent PGM Library Preparation
A polymerase chain reaction (PCR) amplicon library was generated from each extracted DNA sample. Targeted regions within the 34 genes (see Table S1 in the Electronic Supplementary Material) were chosen on the basis of known hotspot regions, regions with reported clinically relevant mutations, regions with mutations that may be considered for clinical trial enrollment, and/or whole genes where significant hotspot regions are not well identified. These regions were amplified using 231 primer pairs in two primer pools. To accommodate the short fragment sizing of FFPE DNA, the amplicons were designed to be 126–183 bp.
Briefly, PCR was performed in 2 × 10 µL reactions, each containing 10–40 ng of genomic DNA; forward and reverse primers (primer concentrations optimized for balanced amplification); 440 µM (each) dATP, dCTP, dGTP, and dTTP; 5 mM MgCl2; 57 mM KCL; and 0.6 units of Gold Polymerase (Celera, Alameda, CA, USA). The PCR amplification was carried out under the conventional conditions. Two PCR products from each sample were pooled. Sequencing adaptors with short stretches of index sequences (Ion Xpress™ Barcode Adapters 1-16 Kit, Life Technologies) that enabled sample multiplexing were ligated to the amplicons by using the PrepX PGM DNA library kit on the Apollo 324 system (Wafergen Biosystems, Fremont, CA, USA). The amplicon library was then nick-translated using Platinum PCR SuperMix High Fidelity under the following conditions: one cycle at 72 °C for 20 min and one cycle at 95 °C for 5 min. The library was quantified using the Qubit DNA HS assay kit.
2.3.2 Sequencing Template Preparation
Pooled libraries were created by diluting four patient samples with distinct barcoding in a final library concentration of 10 pmol/L. Each library pool contained a positive control DNA sample harboring eight variants with known frequencies (Horizon Diagnostics, Waterbeach, Cambridge, UK). Each library pool was subjected to emulsion PCR (E-PCR) using an Ion OneTouch 2 template kit on an Ion OneTouch 2 system (Life Technologies), following the manufacturer’s protocol. The Qubit Ion Sphere quality control kit (Life Technologies) was used to estimate the percentage of the Ion Sphere particles (ISPs) with amplified template DNA. Enrichment of ISPs was achieved using the Ion OneTouch kit on the IT OneTouch ES (Life Technologies), following the manufacturer’s protocol.
2.3.3 Sequencing
Enriched ISPs were subjected to sequencing on an Ion 318 Chip using the Ion PGM sequencing kit (Life Technologies) per the manufacturer’s instructions.
2.3.4 Data Analysis
Sequence reads from the Ion Torrent PGM were aligned with Ion Torrent Suite software version 3.4 (Life Technologies). Variants were called using the Ion Torrent variant caller. Metrics derived from the BAM files included region of interest (ROI) coverages (reads with ≥Q20 average Q-score). Samples with ≥95 % amplicons having ≥300 reads were considered passed. Valid variant calls required position coverage ≥300 Q20 reads, variant frequency ≥4 %, and variant coverage in both read directions. Proximity to homopolymer regions and read strand bias were also considered. These metrics were integrated into a custom software application, CLS Mutation Review, which also directly accesses the Integrated Genome Viewer (IGV) for focused visualization of aligned reads. Manual review of all variants was performed in this CLS Mutation Review application. Population variants and positions with known technical issues (e.g., single-base insertions or deletions (INDELs) at homopolymer regions and positions with common low-frequency artifacts) identified during assay validation studies were tagged within the application to aid the review.
2.4 Initial Molecular Laboratory Tests
2.4.1 Sanger Sequencing
Sanger sequencing assays were performed by standard methods. The specific genes and exons that were tested depended on the cancer type and included EGFR, KRAS, HRAS, NRAS, BRAF, PIK3CA, TP53, and KIT.
2.4.2 Immunohistochemistry
Estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) immunohistochemistry (IHC) testing was performed using standard assays. The stained slides were reviewed and classified as ER and PR positive when ≥1 % of tumor cells showed positive nuclear staining. For the HER2 IHC assay, slides were classified as negative, equivocal, or positive on the basis of the College of American Pathologists (CAP) guidelines.
2.4.3 Fluorescence In Situ Hybridization
Commercial kits were used to assess HER2 gene amplification (Vysis PathVysion™ HER-2 DNA Probe Kit; Abbott Laboratories, Des Plaines, IL, USA), ALK translocation (Vysis LSI ALK Break Apart Rearrangement Probe Kit; Abbott Laboratories), and ROS1 translocation (Cytocell ROS1 Breakapart kit; CytoCell, Compiegne, France).
3 Results
3.1 Patient Specimen Characteristics
Patient characteristics are shown in Table 1. Melanoma samples were more often from men, while lung cancer specimens were more commonly from women. The median patient ages for each cancer type ranged from 57 to 68 years. The most common method for collecting melanoma, lung cancer, and CRC specimens was tissue resection, while most breast cancer tissues came from biopsies. Stage IV cancer patients accounted for about a third each of the melanoma and CRC samples. More than half (57 %) of breast cancer specimens were stage I. The most frequent tumor grade designations were poorly differentiated in lung cancer (37 %) and breast cancer (40 %), and moderately differentiated in CRC (39 %); no grading information was available for most (84 %) of the melanoma specimens.
3.2 Characteristics of the 34-Gene NGS Mutation Assay
3.2.1 Reproducibility
The inter-assay precision of the sequencing performance was assessed by testing three clinical FFPE specimens containing known variants (six single-nucleotide variants [SNVs], one insertion [INS] and two deletions [DEL]) with mutation frequencies between 4.6 and 62.3 % over three different runs. For intra-assay precision, three FFPE specimens made from cell lines harboring known variants (two SNVs and one DEL) with mutation frequencies between 4.8 and 43 % were assayed within a single run. All expected low-frequency variants were detected in inter- and intra-assay replicates. The standard deviations [SDs] of the variant frequencies from the inter- and intra-assays for all variants that were tested ranged from 0.11 to 4.5 % and from 0.2 to 1.1 %, respectively, indicating that the 34-gene NGS mutation test is highly reproducible (data not shown).
3.2.2 Analytical Sensitivity
To assess the assay’s analytical sensitivity, three clinical FFPE specimens with known mutation frequencies (14–78 %) were mixed with DNA from FFPE tissue that did not contain mutations in the regions of interest. The clinical FFPE specimens were added at percentages ranging from 2.5 to 78.0 %. Variants in nine target regions were detected, seven of which (two INDELs and five single-nucleotide polymorphisms [SNPs]) were detected in the mixed sample at or near 5 % (5–7 %). Variants were not detected when the expected frequency was ≤2.5 %. Although INDELs were detectable at or near 5 %, the reported frequencies of some deletions tended to be lower than expected (2.5–5 %). Therefore, the analytical sensitivity of this assay was defined as 5 % for SNPs and 10 % for INDELs (data not shown).
3.3 Results of the 34-Gene NGS Mutation Assay
3.3.1 Amplicon Coverage
All 121 specimens were sequenced at an average of 845K total AQ20 reads, with an average of 3658 reads per amplicon. On average, 98.6 % (range 95–100 %; SD 2.63) of amplicons met the minimum per-amplicon criterion of ≥300 reads, indicating adequate amplicon coverage for all specimens.
3.3.2 Frequency and Breadth of Mutations Detected by NGS
Before reporting of variants, all non-coding or synonymous variants were filtered out of the dataset. We also conservatively excluded any potential population polymorphisms (i.e., those with population allele frequencies ≥0.1 % from the dataset [1000 Genomes]). After exclusion of these population polymorphisms, 83 % of the specimens (100/121) had at least one mutation detected by the 34-gene assay, including 23/31 melanoma specimens (74 %), 31/33 CRC specimens (94 %), 21/27 lung cancer specimens (78 %), and 25/30 breast cancer specimens (83 %). Melanoma, lung cancer, and breast cancer specimens most commonly had a single mutated gene (Fig. 1). CRC specimens, in contrast, most often harbored mutations in two or more genes. Specimens rarely harbored more than three mutated genes, regardless of tumor type. Overall, 20 (59 %) of the 34 genes in the NGS panel had mutations detected in at least one sample. Of those, BRAF, PIK3CA, PIK3R1, PTEN, and TP53 were mutated in all four tumor types; seven were mutated in at least three tumor types; and 15 were mutated in at least two tumor types (Fig. 2). The only genes mutated in just one tumor type were RET and Notch1 (CRC), STK11 (lung cancer), and ERBB2 and ERBB4 (breast cancer).
The mutations detected by the 34-gene NGS assay are depicted by tumor type and sex in Fig. 3a–e and Table S2 in the Electronic Supplementary Material. A total of 172 non-population polymorphism variants were detected in the study. Fourteen genes were mutated in tumor specimens from both male and female patients. Among genes with at least five mutation instances, DDR2 mutations were detected only in females (7 % of overall female specimens; 13 % of female CRC; 18 % of female lung cancer) and PIK3CA mutations were detected in more male than female CRC specimens (39 % male versus 7 % female). Previously reported confirmed somatic variants accounted for 81 % (139/172) and 72 % (100/139) of the somatic variants were known hotspot variants (Table 2). A total of 7.6 % of these variants (13/172) had no previously reported somatic variant reported at the position and had no predicted functional effect. The remaining 8.7 % of the variants (15/172) that were detected had similar position variants reported as confirmed somatic and/or had predicted functional effects on the protein level (Table 2, Tables S3–S6).
In melanoma samples, BRAF mutations were the most common (39 %; nine hotspot and one with predicted functional relevance); these included single-nucleotide substitutions in eight specimens (seven at V600E, GTG>GAG; one at K483Q) and dinucleotide substitutions in five (V600E, GTG>GAA [n = 4]; V600K, GTG>AAG [n = 1]). In addition, 19 % of the melanoma specimens (6/31) had single-nucleotide NRAS substitutions (hotspot variants; Q61K [n = 2], Q61R [n = 3], and G12R [n = 1]), one of which also had two co-occurring BRAF mutations. Additional genes with mutations found in melanoma are listed in Table S3 and included PTEN (13 %; three inactivating, one variant of unknown significance [VUS]), MAP2K1 (10 %; one functionally relevant, two VUS), PIK3CA (6 %; all predicted functionally relevant), and single instances (3 %) of IDH1 (hotspot variant), MET (predicted functional relevance), EGFR (hotspot position but unreported amino acid change), FBXW7 (previously reported somatic with functional impact), PIK3R1 (VUS), and SMO (VUS). TP53 mutations were found in 7/31 melanoma samples (23 %) (Fig. 3b; Table S3).
In CRC samples, KRAS was the most frequently mutated gene (30 %; nine hotspot and one with predicted functional relevance). All KRAS mutations in this group were single-nucleotide substitutions (G12V [n = 2], G12D [n = 2], G12C [n = 1], G13D [n = 3], A146T [n = 1], and L19F [n = 1]). In addition, eight of the 33 samples (24 %) harbored single-nucleotide hotspot PIK3CA substitutions (one each of G106R, R38H, E453K, H1044R, N1044K, E545K, Q546H, and C420R). BRAF hotspot mutations were detected in three of the CRC samples (9 %; V600E [n = 2] and G466V [n = 1]). Additional mutations were found in DDR2 (2/33; no hotspots exist for this gene), IDH1 (2/33; one hotspot, one early termination), NOTCH1 (2/33; one previously reported somatic, one VUS), PTCH1 (two hotspot), NRAS (one hotspot), PIK3R1 (one previously reported somatic), RET (one previously reported somatic), and PTEN (two previously reported somatic). TP53 was also mutated in 19 (58 %) CRC samples. A comprehensive list of genes and variants detected in CRC are provided in Fig. 3c and Table S4 in the Electronic Supplementary Material.
Of the 27 lung cancer specimens, 10 (37 %) harbored single-nucleotide hotspot KRAS substitutions: G12 V (n = 4), G12C (n = 1), G13C (n = 1), G12A (n = 1), G12S (n = 1), Q22K (n = 1), and A146T (n = 1). In addition, three (11 %) harbored EGFR mutations (two hotspot E746_A750del and one previously reported somatic R680Q) and three (11 %) contained DDR2 mutations (R31L, K392N, and H92Y). Two lung cancer samples (7 %) had BRAF mutations (one previously reported somatic G464V and one variant with predicted functional effect, E501V). Mutations co-occurring with KRAS mutations were found in EGFR (one previously reported somatic), DDR2 (one), BRAF (one previously reported somatic), PIK3CA (one hotspot), PIK3R1 (one VUS), PTCH1 (one previously reported somatic), and TP53 (two hotspot and three previously reported somatic) (Fig. 3d). A previously reported somatic PTEN mutation was found in one specimen, co-occurring with a DDR2 mutation, and one specimen harbored a likely benign STK11 mutation with unknown significance. TP53 was mutated in 44 % (4 hotspot, 7 previously reported somatic, and 1 frameshift variant not previously reported) of the lung cancer specimens. A comprehensive list of genes and variants detected in lung cancer are provided in Fig. 3d and Table S5 in the Electronic Supplementary Material.
Of the 30 breast cancer specimens, 12 (37 %) harbored PIK3CA hotspot mutations: H1047R (n = 8), H1043L (n = 1), M1043V (n = 1), E452K (n = 2), and E452Q (n = 1); one of these samples had two co-occurring PIK3CA hotspot mutations (E542Q and H1047R). Two specimens (7 %) harbored ERBB2 hotspot mutations, one of which had two co-occurring ERBB2 hotspot mutations (L755S and S310Y); both specimens also harbored co-occurring PIK3CA hotspot mutations (M1043V and E452K). Additional genes mutated in breast cancer specimens included 7 % with PTCH1 (one hotspot and one VUS) and single instances (3 %) of PTEN (previously reported somatic), MAP2K1 (previously reported somatic in position, but different missense variant), PIK3R1 (VUS), ERBB4 (VUS), and SMO (VUS). TP53 was also mutated in 14 breast cancer specimens (47 %; four hotspot, eight previously reported somatic, one in position with previously reported somatic, and one VUS). A comprehensive list of genes and variants detected in breast cancer are provided in Fig. 3e and Table S6 in the Electronic Supplementary Material.
3.3.3 Comparison of Variants Detected in Genes Assayed in Initial Laboratory Tests Versus the 34-Gene NGS Mutation Test
The results of the initially ordered routine tests (Table 3) were compared with those of the 34-gene NGS assay in the lung cancer, CRC, and melanoma samples (91 cases in total). Breast cancer specimens were not included in this analysis, because no sequencing tests are routinely offered for breast cancer management. Of the 91 cases, 24 had positive results with the initial routine tests. The NGS assay detected mutations in 33 of the 91 cases (36 %) in the same genes, nine more than the routine tests. Six of these nine were attributed to expanded content within these genes. No mutations detected by the routine tests were missed by NGS.
Additional variants detected within this set of genes were found in all 3 tumor types (Table 4 and Table S7 in the Electronic Supplementary Material). Of three melanoma samples that were negative by the initial Cobas® 4800 BRAF V600 mutation test but positive by NGS (Table 4), two were BRAF V600E dinucleotide mutations (GTG>GAA), the same region covered by initial testing. One was in an expanded BRAF region in NGS, a BRAF K483Q mutation (a hotspot position, but only K483E is a previously reported confirmed somatic variant). In CRC specimens, three additional variants were found, all due to expanded regions of KRAS and PIK3CA not covered in the initial testing. Two had non-codon 12/13/61 KRAS mutations (A146T hotspot and L19F previously reported somatic), and one had an exon 7 PIK3CA mutation (C420R hotspot variant). In lung cancer specimens, one had the hotspot exon 19 deletion, E746_A750del, but was detected at 6 % frequency, too low to detect by initial testing that utilizes Sanger sequencing. Additional variants were found in three other lung cancer specimens because of expanded regions of EGFR and KRAS not being covered in initial testing. One had an EGFR exon 17 variant (previously reported somatic R680Q), and one had a non-codon 12/13/61 KRAS (A146T hotspot).
3.3.4 NGS Detection of Mutations in Genes Not Tested in Initial Laboratory Assays
NGS detection of mutations in genes not detected by the initial laboratory assays, including genes or gene regions not covered by the initial test, are described. These data on the sample and gene level are shown in Fig. 3. Gene- and variant-level information is shown in Table S7 in the Electronic Supplementary Material, and detailed per-sample data can be found in Tables S3–S6.
3.3.4.1 Melanoma
Melanoma samples were all submitted for BRAF and/or KIT mutation testing. The NGS assay detected non-BRAF/KIT mutations in 13/20 initially negative samples (65 %) (Fig. 3b). Five of 10 specimens (50 %) testing positive for BRAF mutations on the initial test also harbored additional co-occurring mutations in PTEN, MET, PIK3R1, EGFR, and/or TP53.
3.3.4.2 Colorectal Cancer
All 33 CRC specimens in the study were submitted for KRAS mutation testing, and 24 % (8/33) were also submitted for PIK3CA, BRAF, and/or NRAS mutation testing. The NGS assay detected mutations in 22/24 specimens (92 %) that were negative on the initial laboratory test (Fig. 3c). Of eight specimens that were positive for KRAS in the initial routine tests, NGS testing revealed that five had additional co-occurring in PIK3CA, DDR2, PTCH1, and/or TP53. One BRAF-positive specimen was identified by initial testing and had no additional co-occurring mutations detected by NGS.
3.3.4.3 Lung Cancer
NGS mutation profiling detected at least one mutation in 75 % of the specimens (15/20) with negative initial test results (EGFR, KRAS, and/or ALK rearrangement testing) (Fig. 3d). Mutations detected in genes beyond KRAS or EGFR in the initially negative specimens included BRAF, NRAS, PIK3CA, DDR2, MET, PTEN, and TP53. Furthermore, 5/7 specimens (71 %) that were initially positive for EGFR mutation, KRAS mutation, or ALK rearrangement also harbored at least one co-occurring mutation (EGFR R680Q, BRAF G464V, KRAS Q22K, and/or PTCH1 or TP53) detected by NGS profiling.
3.3.4.4 Breast Cancer
We examined NGS mutation profiling results in breast cancer specimens submitted for HER2, ER, and PR testing. Twenty-eight of the 30 breast cancer specimens were tested for HER2 status by IHC and/or fluorescence in situ hybridization (FISH). Of the 22 HER2-negative specimens, 10 (45 %) harbored PIK3CA hotspot mutations and two (9 %) had ERBB2 hotspot mutations (Fig. 3e). Two of the above mentioned HER2 negative specimens had co-occurring PIK3CA and ERBB2 mutations, one of which had dual ERBB2 hotspot mutations. Another of these PIK3CA-positive specimens had a co-occurring BRAF mutation not previously reported but with predicted functional effect significance. Additional mutations detected in HER2-negative specimens included single instances of co-occurring BRAF hotspot and MAP2K1 mutations, a previously reported PTEN somatic, and a PTCH1 hotspot. The remaining mutations that were identified were variants of unknown significance. TP53 mutations were found in 47 % of the breast cancer specimens (14/30), both HER2 negative and positive. Only one (17 %) of the six HER2-positive specimens had an additional mutation detected, which was a PIK3CA hotspot.
4 Discussion
The intent of this study was to demonstrate the improved diagnostic yield achieved by utilizing an NGS mutation profiling assay with higher sensitivity and expanded content beyond routine tests that were actually ordered by treating physicians in community cancer specimens. The results demonstrated that this approach provided additional information relevant to treatment selection. This additional information was elucidated by sequencing genes or gene regions not typically tested in a given tumor type; identifying co-occurring gene mutations made possible by parallel gene sequencing; and the increased sensitivity of NGS technology over standard Sanger sequencing. Moreover, every sequencing-based result from routine testing was also detected by NGS, meaning that no mutations were “missed” by this technology.
Molecular tumor profiling is becoming important in the management of advanced cancer [45]. NGS is widely used in cancer research settings and is also an attractive technology in the clinical laboratory because it can detect multiple variants in different targets in a single assay [46, 47]. However, the most common clinical specimen type, FFPE tissue, has limitations due to degradation of DNA, often limited tumor tissue, and at times a low fraction of tumor cells in the sample. Our aim was to develop a highly sensitive comprehensive mutation profiling assay to simultaneously detect a broad range of gene mutations that are currently, or are likely to become, predictors of therapeutic response in the solid tumors most commonly evaluated in the clinical laboratory. The resulting 34-gene mutation profiling assay encompasses all genes recommended for the various solid tumors [10–15], as well as genes undergoing clinical trials and/or demonstrating clinical or preclinical evidence supporting an association with clinical response to targeted therapies. This test was validated according to Clinical Laboratory Improvement Amendments (CLIA) regulations and demonstrated assay sensitivity of 5 % and a minimum DNA input of 5 ng. Concordance studies yielded 100 % accuracy. In addition, specificity was determined by assaying 20 normal FFPE tissue specimens that yielded no reportable variants detectable above 4 % allele frequency (data not shown).
In this study, we conservatively removed all variants with minor allele frequencies (MAF) ≥0.1 % of the general population (1000 Genomes database), thereby excluding population polymorphisms from this dataset. Of 121 clinical specimens, 59 % (20) of the 34 genes included in the assay harbored mutations in at least 1 tumor type. Furthermore, most (15/20; 75 %) of these affected genes had mutations in at least two tumor types. This finding helps support the use of a single mutation profiling assay for multiple tumor types.
The NGS assay painted a clearer picture for 15 cases by identifying 16 mutations in NCCN guideline genes (three BRAF in melanoma; two KRAS, two BRAF, and one NRAS in CRC; and two EGFR and six KRAS in lung cancer) that were not detected by the routine clinical tests that the samples were submitted for. This was due to either the specimen not being submitted for testing of that analyte or the limitations of routine testing. One exon 19 deletion in EGFR was detected at 6 % allele frequency by NGS but was missed by routine laboratory testing because of lesser sensitivity. Two melanoma specimens were positive for the dual nucleotide BRAF V600E variant but were missed by the Cobas® 4800 BRAF V600 mutation test, a phenomenon that has been previously reported [48]. Those remaining were due to the specimen not being initially tested in that gene or the gene region not being covered by routine testing. Overall, NCCN guideline–supported treatment response indications were revealed in six of these cases upon NGS profiling that were not found upon initial testing. Three cases had mutations that indicate sensitivity to BRAF inhibitors (two melanoma) or EGFR TKIs (one lung), and three cases had mutations that indicate reduced sensitivity to anti-EGFR therapies (three CRC). NCCN guideline interpretation would not have been altered for the other nine cases.
Also of note was the frequency of co-occurring mutations detected by NGS. While most specimens had only one of the panel genes mutated, at least 27–55 % of each tumor type had ≥2 panel genes mutated and 7–15 % of each tumor type had ≥3 panel genes mutated in a single tumor specimen. Many of these co-occurring mutations existed in specimens with NCCN guideline gene mutations. One ALK rearrangement–positive lung cancer specimen also contained a codon 22 KRAS mutation that has a predicted functional impact on the protein level (cBioPortal.org and 1000 Genomes annotation). Although the clinical significance of codon 22 mutations is not well understood, it is notable that activating KRAS mutations have been associated with lack of response to crizotinib in ALK-positive patients [49, 50]. One HER2-positive breast cancer specimen also harbored a PIK3CA hotspot mutation. HER2-positive tumors with PIK3CA mutations have been associated with lack of response to conventional anti-HER2 therapies (e.g., trastuzumab and lapatinib) [51], while trastuzumab emtansine (Kadcyla, Genentech) may be efficacious [52]. However, clinical studies with conflicting results substantiate the need for further studies before treatment-related biomarker status can be defined [57]. Although numerous co-occurring mutations were observed, the effects on treatment response have yet to be conclusively determined for most.
Many specimens that had tested negative upon initial laboratory testing harbored mutations in other genes or expanded gene regions. Of the 172 variants detected throughout the study, 33 (19 %) were NCCN guideline-recommended biomarkers for the tumor type they were detected in (12 melanoma, 11 CRC, and 10 lung cancer). Beyond the NCCN guideline-recommended biomarkers, 60 variants (35 %) met inclusion criteria for treatment-related phase 2 or 3 clinical trials (https://clinicaltrials.gov, accessed from March 2015 through January 2016). These included mutations in DDR2 that were notably found only in tumors from female patients (three lung cancer and two CRC). Of the remaining 88 variants (51 % of the total variants) that did not have known NCCN guideline or clinical trial implications, 76 had a predicted functional impact on the protein level. Functional impact was determined by 1000 Genomes annotation (pathogenic or likely pathogenic clinical significance, deleterious SIFT score, possibly or probably damaging PolyPhen score); cBioPortal.org annotation (Mutation Assessor prediction of medium or high); or inactivation by frameshift or early termination (Tables S3–S6 in the Electronic Supplementary Material).
Notably, not all NCCN guideline–recommended biomarker genes for a particular tumor type were ordered for each specimen submitted, and in fact some of these biomarkers were elucidated by broader NGS profiling. This of course does not rule out that ancillary testing may have been ordered through an alternative clinical laboratory testing source. Furthermore, most variants with clinical trial eligibility but no NCCN guideline recommendations were not detected by routine clinical laboratory testing. Therefore, the broader profiling capabilities offered by this NGS mutation profiling test have the potential to offer additional treatment possibilities for a significant number of cancer patients.
The limitations of this study include the small number of specimens per cancer type; lack of comparison with germline testing as a means to confirm the somatic status of the variants that were detected; lack of clinical information such as treatment regimens and responses; lack of data on the clinical impact of some of the mutations detected by NGS; and lack of knowledge on ancillary tests not performed through Quest Diagnostics. Future studies with larger sample sizes and follow-up treatment data will provide better estimates of how often this panel will identify informative mutations in specific cancer types.
5 Conclusion
This study was unique in that it represents a broad comparison in multiple tumor types with the specific mutation assays that were ordered by the treating physicians in the community. It was not simply a method comparison of NGS profiling on a technical level, but a comparison of information provided by genes and test types selected by the treating physicians for each submitted specimen. This comparison demonstrated that (1) there is community variation in the set of genes ordered for molecular testing, even within the same tumor types; (2) adding newer NCCN guideline biomarkers altered the predicted treatment response interpretation for some specimens that were submitted on the basis of older NCCN guideline recommendations (e.g., KRAS mutation only versus KRAS, BRAF, and NRAS mutation testing to predict anti-EGFR monotherapy response in CRC); (3) molecular testing with improved sensitivity and fewer technical limitations improved detection of NCCN guideline biomarkers critical for treatment response prediction in real community specimens (e.g., NGS versus EGFR Sanger sequencing for lung cancer and BRAF Cobas® for melanoma); and (4) mutations associated with treatment-related clinical trial studies were detected in at least 33 % of the specimens (40/121) in this study.
Furthermore, nearly half of all samples that were tested harbored mutations in genes that were not originally tested for, and a significant proportion of specimens also harbored co-occurring mutations. Our findings support the need for further studies into (1) the benefits of broader mutation profiling; (2) response characteristics of tumors harboring various mutations in parallel or alternative signaling pathways, or co-occurring mutations; and (3) a better characterized demonstration of the clinical benefit of narrow versus broad mutation profiling in all cancer types.
References
Arrowsmith J. A decade of change. Nat Rev Drug Discov. 2012;11(1):17–8.
Bailey AM, Mao Y, Zeng J, Holla V, Johnson A, Brusco L, et al. Implementation of biomarker-driven cancer therapy: existing tools and remaining. Gaps Discov Med. 2014;17(92):101–14.
Collins I, Workman P. New approaches to molecular cancer therapeutics. Nature Chem Biol. 2006;2(12):689–700.
Paez JG, Jänne PA, Lee JC, Tracy S, Greulich H, Gabriel S, et al. EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy. Science. 2004;304(5676):1497–500.
Lynch TJ, Bell DW, Sordella R, Gurubhagavatula S, Okimoto RA, Brannigan BW, et al. Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small cell lung cancer to gefitinib. N Engl J Med. 2004;350(21):2129–39.
Chapman PB, Hauschild A, Robert C, Haanen JB, Ascierto P, Larkin J, et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med. 2011;364(26):2507–16.
Rosell R, Carcereny E, Gervais R, Vergnenegre A, Massuti B, Felip E, et al. Erlotinib versus standard chemotherapy as first-line treatment for European patients with advanced EGFR mutation-positive non-small-cell lung cancer (EURTAC): a multicentre, open-label, randomised phase 3 trial. Lancet Oncol. 2012;13(3):239–46.
Mendelsohn J, Baselga J. The EGF receptor family as targets for cancer therapy. Oncogene. 2000;19(56):6550–65.
Carvajal RD, Antonescu CR, Wolchok JD, Chapman PB, Roman RA, Teitcher J, et al. KIT as a therapeutic target in metastatic melanoma. JAMA. 2011;305(22):2327–34.
National Comprehensive Cancer Network. NCCN breast cancer guidelines (version 2.2015). http://www.nccn.org/professionals/physician_gls/pdf/breast.pdf. Accessed 7 April 2015.
National Comprehensive Cancer Network. NCCN colon cancer guidelines (version 2.2015). http://www.nccn.org/professionals/physician_gls/pdf/colon.pdf. Accessed 7 April 2015.
National Comprehensive Cancer Network. NCCN melanoma guidelines (version 3.2015). http://www.nccn.org/professionals/physician_gls/pdf/melanoma.pdf. Accessed 7 April 2015.
National Comprehensive Cancer Network. NCCN non-small cell lung cancer guidelines (version 5.2015). http://www.nccn.org/professionals/physician_gls/pdf/nscl.pdf. Accessed 7 April 2015.
National Comprehensive Cancer Network. NCCN thyroid carcinoma guidelines (version 2.2014). http://www.nccn.org/professionals/physician_gls/pdf/thyroid.pdf. Accessed 7 April 2015.
National Comprehensive Cancer Network. NCCN soft tissue sarcoma guidelines (version 1.2015) http://www.nccn.org/professionals/physician_gls/pdf/sarcoma.pdf. Accessed 7 April 2015.
Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144(5):646–74.
Fisher R, Pusztai L, Swanton C. Cancer heterogeneity: implications for targeted therapeutics. Br J Cancer. 2013;108(3):479–85.
Burrella RA, Swantona C. Tumor heterogeneity and the evolution of polyclonal drug resistance. Mol Oncol. 2014;8(6):1095–1111.
Peng B, Lloyd P, Schran H. Clinical pharmacokinetics of imatinib. Clin Pharmacokinet. 2005;44(9):879–94.
Guo J, Si L, Kong Y, Flaherty KT, Xu X, Zhu Y, et al. Phase II, open-label, single-arm trial of imatinib mesylate in patients with metastatic melanoma harboring c-kit mutation or amplification. J Clin Oncol. 2011;29(21):2904–9.
Venook AP, Arcila ME, Benson AB III, Berry DA, Camidge DR, Carlson RW, et al. NCCN Working Group report: designing clinical trials in the era of multiple biomarkers and targeted therapies. J Natl Compr Canc Net. 2014;12(11):1629–49.
Sánchez-Torres JM, Viteri S, Molina MA, Rosell R. BRAF mutant non-small cell lung cancer and treatment with BRAF inhibitors. Transl Lung Cancer Res. 2013;2(3):244–50.
Kelloff GJ, Sigman CC. Cancer biomarkers: selecting the right drug for the right patient. Nat Rev Drug Discov. 2012;11:201–14.
Yu HA, Arcila ME, Rekhtman N, Sima CS, Zakowski MF, Pao W, et al. Analysis of tumor specimens at the time of acquired resistance to EGFR-TKI therapy in 155 patients with EGFR-mutant lung cancers. Clin Cancer Res. 2013;19(8):2240–7.
Wagle N, Van Allen EM, Treacy DJ, Frederick DT, Cooper ZA, Taylor-Weiner A, et al. MAP kinase pathway alterations in BRAF-mutant melanoma patients with acquired resistance to combined RAF/MEK inhibition. Cancer Discov. 2014;4(1):61–8.
Narita Y, Okamoto K, Kawada MI, Takase K, Minoshima Y, Kodama K, et al. Novel ATP-competitive MEK inhibitor E6201 is effective against vemurafenib-resistant melanoma harboring the MEK1–C121S mutation in a preclinical model. Mol Cancer Ther. 2014;13(4):823–32.
Wagle N, Emery C, Berger MF, Davis MJ, Sawyer A, Pochanard P, et al. Dissecting therapeutic resistance to RAF inhibition in melanoma by tumor genomic profiling. J Clin Oncol. 2011;29(22):3085–96.
Roock WD, Vriendt VD, Normanno N, Ciardiello F, Tejpar S. KRAS, BRAF, PIK3CA, and PTEN mutations: implications for targeted therapies in metastatic colorectal cancer. Lancet Oncol. 2011;12(6):594–603.
Chaft JE, Arcila ME, Paik PK, Lau C, Riely GJ, Pietanza C, et al. Coexistence of PIK3CA and other oncogene mutations in lung adenocarcinoma—rationale for comprehensive mutation profiling. Mol Cancer Ther. 2012;11(2):485–91.
Janku F, Wheler JJ, Naing A, Falchook GS, Hong DS, Stepanek VM, et al. PIK3CA mutation H1047R is associated with response to PI3K/AKT/mTOR signaling pathway inhibitors in early-phase clinical trials. Cancer Res. 2013;73(1):276–84.
Wang L, Hu H, Pan Y. PIK3CA mutations frequently coexist with EGFR/KRAS mutations in non-small cell lung cancer and suggest poor prognosis in EGFR/KRAS wild type subgroup. PLoS One. 2014;9(2):e88291.
Jänne PA, Shaw AT, Pereira JR, Jeannin G, Vansteenkiste J, Barrios C, et al. Selumetinib plus docetaxel for KRAS-mutant advanced non-small-cell lung cancer: a randomised, multicentre, placebo-controlled, phase 2 study. Lancet Oncol. 2013;14(1):38–47.
Ludovini V, Bianconi F, Pistola L, Chiari R, Minotti V, Colella R, et al. Phosphoinositide-3-kinase catalytic alpha and KRAS mutations are important predictors of resistance to therapy with epidermal growth factor receptor tyrosine kinase inhibitors in patients with advanced non-small cell lung cancer. J Thoracic Oncol. 2011;6(4):707–15.
Boland JM, Jang J, Li S, Lee J, Wampfler AM, Erickson-Johnson JA, et al. MET and EGFR mutations identified in ALK-rearranged pulmonary adenocarcinoma: molecular analysis of 25 ALK-positive cases. J Thoracic Oncol. 2013;8(5):574–81.
Forbes SA, Beare D, Gunasekeran P, Leung K, Bindal N, Boutselakis H, et al. COSMIC: exploring the world’s knowledge of somatic mutations in human cancer. Nucleic Acids Res. 2015;43(Database Issue):D805–11.
Lovly C, L Horn, W Pao. Molecular profiling of lung cancer. My Cancer Genome. 2015. http://www.mycancergenome.org/content/disease/lung-cancer. Accessed 7 April 2015.
Balko JM, Mayer I, Levy M, Arteaga C. Molecular profiling of breast cancer. My Cancer Genome. 2015. http://www.mycancergenome.org/content/disease/breast-cancer. Accessed 7 April 2015.
Chan E. Molecular profiling of colorectal cancer. My Cancer Genome. 2015. http://www.mycancergenome.org/content/disease/colorectal-cancer. Accessed 7 April 2015.
Maki RG, Keedy V. Molecular profiling of gastrointestinal stromal tumor (GIST). My Cancer Genome. 2014. http://www.mycancergenome.org/content/disease/gist. Accessed 7 April 2015.
Lovly C, Pao W, Sosman J. Molecular profiling of melanoma. My Cancer Genome. 2014. http://www.mycancergenome.org/content/disease/melanoma. Accessed 7 April 2015.
Espinosa AV, Gilbert J. Molecular profiling of thyroid cancer. My Cancer Genome. 2015. http://www.mycancergenome.org/content/disease/thyroid-cancer. Accessed 7 April 2015.
US National Institutes of Health. https://ClinicalTrials.gov. Accessed 7 April 2015 through 21 December 2015.
Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio Cancer Genomics Portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2012;2(5):401–4.
Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signa. 2013;6(269):pl1.
Pleasance ED, Cheetham RK, Stephens PJ, McBride DJ, Humphray SJ, Greenman CD, et al. A comprehensive catalogue of somatic mutations from a human cancer genome. Nature. 2010;463(7278):191–6.
Singh RR, Patel KP, Routbort MJ, Reddy NG, Barkoh BA, Handal B, et al. Clinical validation of a next-generation sequencing screen for mutational hotspots in 46 cancer-related genes. J Mol Diagn. 2013;15(5):607–22.
Kanagal-Shamanna R, Portier BP, Singh RR, Routbort MJ, Aldape KD, Handal BA, et al. Next-generation sequencing-based multi-gene mutation profiling of solid tumors using fine needle aspiration samples: promises and challenges for routine clinical diagnostics. Mod Pathol. 2014;27(2):314–27.
Qu K, Pan Q, Zhang X, Rodriguez L, Zhang K, Li HR, et al. Detection of BRAF V600 mutations in metastatic melanoma comparison of the Cobas 4800 and Sanger sequencing assays. J Mol Diagn. 2013;15(6):790–5.
Doebele RC, Pilling AB, Aisner DL, Kutateladze TG, Le AT, Weickhardt AJ, et al. Resistance to crizotinib in patients with ALK gene rearranged non-small cell lung cancer. Clin Cancer Res. 2012;18(5):1472–82.
Anaia S, Suzukia K, Ijichia K, Haradaa T, Toyokawac G, Setoc T, et al. A case of crizotinib-resistant lung adenocarcinoma harboring a KRAS mutation and an EML4–ALK fusion gene. J Med Cases. 2014;5(12):631–3.
Berns K, Horlings HM, Hennessy BT, Madiredjo M, Hijmans EM, Beelen K, et al. A functional genetic approach identifies the PI3K pathway as a major determinant of trastuzumab resistance in breast cancer. Cancer Cell. 2007;12(4):395–402.
Baselga J, Verma S, Ro J, Huober J, Guardino E, Fang L, et al. Relationship between tumor biomarkers (BM) and efficacy in EMILIA, a phase III study of trastuzumab emtansine (T-DM1) in HER2-positive metastatic breast cancer (MBC). J Cancer Res. 2013;73:Abstract LB-63.
Duzkale H, Shen J, McLaughlin H, Alfares A, Kelly MA, Pugh TJ, et al. A systematic approach to assessing the clinical significance of genetic variants. Clin Genet. 2013;84(5):453–63.
Bai Y, Kim JY, Watters JM, Fang B, Kinose F, Song L, et al. Adaptive responses to dasatinib-treated lung squamous cell cancer cells harboring DDR2 mutations. Cancer Res. 2014;74(24):7217–28.
Chan, E. KRAS c.436G>A (A146T) mutation in colorectal cancer. My Cancer Genome. 2014. http://www.mycancergenome.org/content/disease/colorectal-cancer/kras/28. Accessed 15 Dec 2015.
Akagi K, Uchibori R, Yamaguchi K, et al. Characterization of a novel oncogenic K-ras mutation in colon cancer. Biochem Biophys Res Commun. 2007;352:728–32.
Cescon DW, Bedard PL. PIK3CA genotype and treatment decisions in human epidermal growth factor receptor 2–positive breast cancer. J Clin Oncol. 2015;33(12):1318–21.
Kancha RK, von Bubnoff N, Bartosch N, Peschel C, Engh RA, Duyster J. Differential sensitivity of ERBB2 kinase domain mutations towards lapatinib. PLoS One. 2011;6(10):e26760.
Saito T, Mitomi H, Imamhasan A, Hayashi T, Kurisaki-Arakawa A, Mitani K, Takahashi M, Kajiyama Y, Yao T. PTCH1 mutation is a frequent event in oesophageal basaloid squamous cell carcinoma. Mutagenesis. 2015;30(2):297–301.
Acknowledgments
The authors thank Jeff Radcliff (Quest Diagnostics) for critical review and suggestions on the manuscript, Melissa Smith (MedIncite) for assistance with manuscript preparation, Anne Lin (Quest Diagnostics) for specimen management, Vicki Hartman and Susan-Maria Miranda for coordination of clinical specimens, and Diedre Nguyen for acquisition of clinical result data.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
Heather Sanders, Kevin Qu, Hairong Li, Lin Ma, Cindy Barlan, Xi Zhang, James Prentice, David Wolfson, Beryl Crossley, Anthony Sferruzza, Feras Hantash, and Frederic Waldman are employees of the Department of Hematology and Oncology, Quest Diagnostics Nichols Institute. John Sninsky, David Ross, Andrew Grupe, and Joseph Catanese are employees of Quest Diagnostics.
Funding
All studies were funded by the Quest Diagnostics Nichols Institute.
Ethical approval and informed consent
The study protocol was reviewed by the Western Institutional Review Board and was determined to be research that is exempt from Institutional Review Board oversight, according to the criteria set forth in 45 CFR 46, “Protection of Human Subjects.” The NGS studies were performed on remnants of specimens collected for routine clinical care that would otherwise have been discarded. Clinical information used in the analyses was recorded in such a manner that subjects could not be identified by investigators, directly or through identifiers linked to the subjects.
Additional information
H. Sanders and K. Qu contributed equally to this work and share first authorship.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
About this article
Cite this article
Sanders, H., Qu, K., Li, H. et al. Mutation Yield of a 34-Gene Solid Tumor Panel in Community-Based Tumor Samples. Mol Diagn Ther 20, 241–253 (2016). https://doi.org/10.1007/s40291-016-0197-0
Published:
Issue Date:
DOI: https://doi.org/10.1007/s40291-016-0197-0