Cancer is caused by altered regulation and function of multiple cellular pathways. Clonal accumulation of alterations in these different pathways leads to unregulated cell proliferation and cancer. The molecular alterations relevant to each patient’s cancer are unique and it is difficult to identify these changes at a functional (protein) level.
Scientists at Foundation Medicine, Inc. are leading a transformation in cancer care. They are innovating and commercializing products that help clinicians to select appropriate treatment options for each patient, informed by a thorough understanding of the molecular changes specific to their disease.
Cancer is a disease of the genome
Therapy targeting the alterations that drive an individual patient’s cancer can yield dramatic results (Figure 1). The approach taken at Foundation Medicine is to perform comprehensive genomic profiling of clinical samples to identify all classes of genomic alterations in hundreds of genes relevant to cancer. This is the minimum required to inform selection of targeted therapies.
Figure 1. Dramatic results observed with targeted therapy. A V600E mutation was detected in BRAF for a patient with advanced stage melanoma, whose CT scan is shown (A) before and (B) after targeted treatment with Vemurafenib [1].
Delivering comprehensive genomic profiling to the clinic
Traditional approach to cancer diagnosis
Traditionally, molecular characterization of a patient’s cancer has been a low priority test performed in a disease-specific manner after pathologic review of a tumor specimen. For example, for advanced non–small cell lung carcinoma (NSCLC), the 2011 National Comprehensive Cancer Network® (NCCN®) guidelines for molecular profiling recommended multiple assays for genomic alterations in only 3 genes: KRAS (base substitution in codon 12), EGFR (in-frame indels, base substitutions, and copy number alterations), and EML4-ELK (rearrangements). This required multiple traditional assays to detect all of the relevant alterations—Sanger sequencing, PCR, and mass spectrometry to identify base substitutions; cytogenetics, fluorescence in situ hybridization (FISH), and immunohistochemistry (IHC) to detect rearrangements; gel sizing and PCR to detect indels; FISH and IHC to detect copy number alterations.
Limitations of traditional diagnostics methods
Low Sensitivity: These traditional methods typically demonstrate low sensitivity, limiting detection of the genomic alterations being investigated. The utility of Sanger sequencing for characterizing routine samples is limited (Figure 2A), because many base substitutions and indels occur at a minor allele frequency of
5–20% due to lowpurity of tumor tissue, which is often ≤40% in routine clinical samples.
Poor Accuracy: The ability of traditional assays to accurately detect genomic alterations in tumor samples is limited by their technology.
- Assays that detect single alterations have a low diagnostic yield. The need for serial testing leads to inefficient use of precious tissue and time. These types of assays also result in incomplete molecular profiling of many samples and each new diagnostic target requires development and validation of new assays.
- Many assays have limited sensitivity to the relevant alterations. Sanger sequencing cannot reliably detect short variants below 20% minor allele frequency (MAF), while FISH does not reliably detect complex rearrangements.
- Reproducibility of FISH and IHC between testing sites can be affected by inter-operator and reagent (e.g., antibody) variability. The nature of these assays makes it difficult to include quantitative, process-matched controls that provide minimum performance specifications.
Requirements for the ideal clinical cancer diagnostics assay
Clinically relevant—compatible with formalin-fixed, paraffin-embedded (FFPE) tissues, core needle biopsies, and fine needle aspirates
Comprehensive—a single test must be capable of detecting all classes of genomic alterations across all cancer-relevant genes in all samples
Accurate—highly sensitive and specific, delivering minimal false negative results in low purity samples and no false positives
High quality—delivers highly reproducible results from limited DNA (~50 ng) isolated from low (~20%) purity FFPE tumor specimens in ≤14 days from sample receipt
Validated—must be comprehensively validated
Next generation sequencing for molecular diagnostics
When correctly implemented, next generation sequencing (NGS) methods provide millions of individual observations that are representative of the genetic makeup of a sample. Application of NGS to clinical samples for reliable patient diagnosis requires meaningful performance specifications and extensive optimization of sample preparation and data analysis workflows.
Figure 2. Quantitative variant detection to 5% MAF with next generation sequencing. Tumor samples are typically only 20–40% pure because of admixture with normal tissue, resulting in a low frequency of the potential alterations being interrogated. (A) Sanger sequencing cannot reliably detect base substitution or indels below a minor allele frequency (MAF) of 20%. (Sanger trace adapted from Tsiatis et al, 2010, [2].) (B) Using NGS, variants at 5% MAF and below can be detected quantitatively. (Figure courtesy of Foundation Medicine, Inc.)
Hybridization capture: Hybridization capture of target genomic regions allows NGS to be used for comprehensive profiling of clinical oncology specimens. Hybrid capture can provide deep, uniform sequencing coverage representative of the sample input. It allows detection of all classes of genomic alterations across hundreds of genes and provides the performance metrics required for ensuring assay specificity and sensitivity.
Table 1. Hybridization capture facilitates comprehensive genomic profiling of clinical cancer specimens.
|
|
Whole genome
sequencing
|
Hybridization
capture
|
PCR
|
|
Preserves input complexity
|
Yes |
Yes |
No
|
|
Detects substitutions and indels
|
Yes |
Yes |
Yes |
|
Detects CNAs
|
Yes |
Yes |
No
|
|
Detects genomic rearrangements
|
Yes |
Yes |
No
|
Cost effective
(Turnaround time)
|
No
(>4 weeks)
|
Yes
(~2 weeks) |
Yes
(~2 weeks) |
Foundation Medicine’s approach to molecular diagnostics
Use of hybrid capture for targeted sequencing: Research scientists at Foundation Medicine use hybridization capture to focus sequencing efforts on genomic regions of interest. They have performed substantial development and optimization of hybrid capture to ensure uniform and reproducible genomic profiling of hundreds of cancer genes. A major change the scientists implemented, to obtain more uniform coverage, was to switch from using RNA probes to individually synthesized DNA probes (xGenLockdown Probes from IDT). This change also provided deeper coverage of the targeted sequences. Figure 3 shows the improvements achieved.
Figure 3. High uniformity and deep coverage achieved with xGen Lockdown Probes. Use of individually synthesized xGen Lockdown Probes, coupled with optimization of the hybridization reaction resulted in (A) more uniform and (B) deeper coverage than observed with mass-synthesized RNA probes. Coverage was highly uniform from 20% to 80% GC. (C) A plot of coverage location and depth for anaplastic lymphoma kinase (ALK) intron 19 shows the improvement that xGen Lockdown Probes provide over the previously used RNA probe protocol.
Reproducibility and dynamic range: Any reliable diagnostic method needs to be highly reproducible. Due to the varying purity of malignant cells in tumor tissue samples that can dilute the signals of focal amplifications and homozygous deletions, effective molecular diagnostics tests require reproducible coverage over a broad dynamic range. Foundation Medicine scientists found that xGen Lockdown Probes offer both (Figure 4A).
Figure 4. Reproducibility and broad dynamic range provide confidence. (A) Targeted RNA-seq data shows reproducible per exon coverage of technical replicates of Universal Human Reference RNA (Agilent). (B) Targeted DNA-seq data of per target coverage plotted against genomic position of a 100% pure breast cancer cell-line and a mixture of 20% breast cancer cell-line with 80% of a matched normal cell-line demonstrates the how broad dynamic range can enable detection of copy number alterations in a sample of 20% tumor purity.
FoundationOne for solid tumors: Using xGen Lockdown Probes, Foundation Medicine developed FoundationOne, a comprehensive genomic profile for all solid tumors, to simultaneously detect all clinically relevant classes of genomic alterations in a single assay. FoundationOne interrogates the entire coding sequence of 315 cancer-related genes plus introns from 28 genes commonly rearranged in solid tumors. The assay, which has a 14-day median turnaround time from sample receipt, requires only small amounts of tissue from samples, such as routine FFPE samples, needle biopsies, and fine needle aspirates (≥50 ng of DNA).
This validated assay delivers high accuracy that is achieved by the high, uniform coverage of xGen Lockdown Probes—>99% of exons covered >100X in routine clinical samples. Computational biology algorithms used to develop the assay were validated for high accuracy in clinical samples with extensive stromal contamination [3].
FoundationOne provides more comprehensive data than traditional assays
Identification of new markers: Molecular profiling of an NSCLC tumor specimen by traditional assays (standard of care testing) was negative for all genes covered by the 2011 NCCN guidelines for advanced NSCLC. No genomic alterations were found for KRAS, EGFR, or EML4-ALK. However, subsequent testing with FoundationOne revealed a novel KIF5B-RET fusion in the same specimen (Figure 5). At the time of testing, RET was only known to be recurrently rearranged in thyroid cancer. RET rearrangement profiling has since been added to the NCCN guidelines.
Figure 5. Novel KIF5B-RET fusion detected in non–small cell lung cancer.
Detection of clinically relevant mutations: Traditional, disease-specific testing for HER2/ERBB2 across 27 tumor types using IHC/FISH identified amplifications as the only alteration and only in gastroesophageal, breast, and esophageal cancers (Figure 6A). When the same samples were tested with FoundationOne, clinically relevant alterations—rearrangements, indels, base substitutions, and amplifications—were detected in HER2/ERBB2 in all 27 tumor types (Figure 6B).
Figure 6. More clinically relevant alterations detected with comprehensive molecular testing. (A) Twenty-seven tumor types were tested for alterations in HER2/ERBB2 using immunohistochemistry and fluorescence in situ hybridization (FISH). Only 3 of the tumor types (gastroesophageal junction, breast, and esophagus) showed amplification in HER2/ERBB2. (B) The same 27 tumor types were tested using FoundationOne. The comprehensive test detected various categories of clinically relevant alterations for all of the tumor types.
Comprehensive genomic profiling critical to tumor assessment
The usefulness of traditional, single-alteration molecular tests is limited by their inability to deliver the most informative results. NGS can enable comprehensive genomic profiling; however, clinical application requires significant investment in infrastructure in addition to substantial effort to optimize protocols and bioinformatics methods. Using FoundationOne, Foundation Medicine is able to identify clinically meaningful alterations in the genomes of up to 85% of patients with cancer. The scientists believe that comprehensive genomic profiling must be completely integrated into the pathologic assessment of tumor specimens.