By Turna Ray
Canadian researchers investigating the use of massively parallel sequencing to guide treatment for tumors in the tongue have concluded that the technology may become a viable tool in the personalization of treatments for rare cancers.
"We think that the concept of determining a complete genomic analysis to aid in clinical decision making will become routine in a few years' time," Steven Jones, lead author of the study and head of bioinformatics at Canada's Michael Smith Genome Sciences Center, told Pharmacogenomics Reporter. "In the meantime, this [study] has the biggest potential in understanding the molecular underpinnings for rarer cancers where established protocols often don't exist and the lack of patient numbers prevents any systematic identification of therapeutics through phase type trials."
The study, published in Genome Biology earlier this month, combined analysis of a patient's tumor genome and transcriptome sequences to guide treatment of tongue adenocarcinoma with Pfizer's Sutent (sunitinib) and Bayer's Nexavar (sorafenib). Before guiding treatment based on sequencing data in the trial, the study subject was given Tarceva (erlotinib) based on tumor pathology, but he failed to respond to this treatment.
Specifically, the researchers analyzed the copy number variation, gene expression, and protein-coding mutations in the tumor to determine that known targets of sunitinib and sorafenib, including the RET oncogene, were up-regulated, which indicated that the patient would respond to this course of treatment. Ultimately, the analysis generated "clinically useful information and provided the rationale for a therapeutic regime that, whilst not curative, did establish stable disease for several months," the authors wrote.
"This approach is of particular relevance for rarer tumor types, where the scarcity of patients, their geographic distribution, and the diversity of patient presentation mean that the ability to accrue sufficient patient numbers for statistically powered clinical trials is unlikely," they said. "The ability to comprehensively genetically characterize rare tumor types at an individual patient level therefore represents a logical route for informed clinical decision making and increased understanding of these diseases."
The PGx-guided approach stabilized the disease for a few months, after which the patient's cancer gained resistance to treatment through new pathways. Jones said that based on these findings, a near-term goal for cancer researchers would be to start characterizing the different mechanisms for therapeutic resistance, particularly since disease metastases can occur through different genomic pathways requiring new treatment strategies.
In the single-patient study, researchers identified in the subject's pretreatment tumor more than 7,600 genes with copy number gain; more than 1,000 genes that showed increased expression relative to genes expressed in the blood and in unrelated tumors; and four genes that contained somatic protein-coding mutations. In particular, genes with protein products that are targets of the RET inhibitors Sutent and Nexavar were highly overexpressed or amplified, so the researchers first treated the patient with Sutent, which resulted in stable disease for four months.
When the lesions began to grow again, the researchers treated the patient with a combination of Nexavar and sulindac, which stabilized the cancer for another three months. But after that period, the cancer progressed and new lesions appeared.
In the recurring metastasis, the researchers found 7,288 genes within copy number amplicons, 385 genes exhibiting increased expression linked to other tumors, and nine new somatic protein-coding mutations. According to the paper, these amplifications and mutations rose through activated MAPK and AKT pathways, which are known to confer resistance to the treatments given to the patient.
While the researchers acknowledged in the paper that the observed tumor changes in this one patient may be due to "many other changes" and that their understanding of cancer biology "is far from complete," these early-stage findings may have potential for future commercial application. Jones said his research team has discussed the work with Pfizer, maker of Sutent, "and our current goal is to derive Sutent resistance in xenograft models followed by genome sequencing to examine how recurrent and similar the genetic changes in response to therapy are."
[ pagebreak ]
Pfizer is already working with Genomic Health to individualize Sutent treatment in Stage I-III, localized clear cell-type renal carcinoma (PGx Reporter 01/09/08) The drug is currently indicated as a treatment for advanced kidney cancer, and for treating gastrointestinal stromal tumors when Gleevec cannot be taken or is not working. The vascular endothelial growth factor/angiogenesis inhibitor is in clinical trials for the treatment of gastric, liver, prostate, breast, lung, and colorectal cancer.
"I am not sure if this [strategy] would be something companies would develop drugs around initially," Jones said. "But fundamentally we are deriving important information around oncogenic mechanisms and responses to therapy that will allow the existing drugs to be used more efficiently and effectively, and identify the sub-types of cancer where new therapeutics are actually needed."
Finally, based on this clinical work, the study authors recommended that cancer researchers build repositories that not only contain genomic and transcriptomic information of various cancers, but also catalog which markers are linked to which oncologics.
Tumors in the tongue represent between 20 percent and 25 percent of salivary gland tumors. Cancers originating in the tongue are usually detected in advanced stages, particularly if tumors form in the base of the tongue, and in many cases patients experience disease metastasis to the neck. A combination of radiotherapy and surgery is the standard treatment for this type of cancer.
In the study published in Genome Biology, Jones and colleagues conducted genomic and trancriptomic analysis of a 78-year-old Caucasian man who in 2007 had a mass on the base of his tongue, which doctors identified through biopsy as a papillary adenocarcinoma. This tumor subsequently spread to his neck and lungs. After a fresh tissue needle biopsy of a lesion on the lung, researchers focused their analysis on genetic changes impacting cellular function, such as copy number variation and sequences of protein products.
"Our initial analysis of sequence alignments identified 84 DNA putative sequence changes corresponding to non-synonymous changes in protein-coding regions present only within the tumor, of which four were subsequently validated to be somatic tumor mutations by Sanger sequencing," the researchers write in the paper, which also discusses observed somatic mutations in two tumor suppressor genes,TP53 (D259Y), and a truncating mutation in RB1 (L234*).
In order to profile the expression of tumor transcripts, the researchers conducted whole-transcriptome shotgun sequencing. Since there wasn't any normal tissue available for comparison, the researchers compared gene expression changes in the patient's leukocytes with a compendium of 50 tumor-derived WTSS datasets. "This compendium approach allowed us to identify a specific and unique molecular transcript signature for this tumor, as compared to unrelated tumors, enriched in cancer-causing events specific to the patient’s tumor and therefore should represent relevant drug targets for therapeutic intervention," the researchers explained.
The comparison revealed 3,064 genes that were differentially expressed in the patient's lung tumor versus genes expressed in the blood or in the compendium datasets. "This analysis provided insight into those genes whose expression rate was likely to be a driving factor specific to this tumor and not identifying genes that correlate simply with proliferation and cell division," researchers wrote. "It is conceivable that such an approach, coupled with a greater understanding from multiple tumor datasets, could be replaced by the absolute quantitation of oncogene expression as a means to determine clinical relevance."
After the genomic and transcriptomic analysis of the cancer tumor, researchers correlated the mutated, amplified or differentially expressed genes with known cancer pathways from the Kyoto Encyclopedia of Genes and Genomes database and to drug targets in the DrugBank database. This search yielded 15 amplified, overexpressed, or mutated genes in cancer pathways that appeared to be targetable by approved drugs, including Sutent, Nexavar, motesanib, and sulindac.
Sutent and Nexavar are both approved for renal cancer, while Amgen's motesanib is in clinical trials for thyroid cancer, GIST, and NSCLC. These drugs interact with similar pathways, including RET, VEGFRs, and KIT. Sulindac, marketed by Merck as Clinoril, is a nonsteroidal anti-inflammatory drug that inhibits the MAPK3 (ERK1) pathway.
[ pagebreak ]
As the researchers compared databases of genes to drugs, the "RET proto-oncogene emerged as a gene of particular interest to us, as it was present in a region of genomic amplification and was abundantly expressed," researchers wrote in the paper.
The study subject, before being treated with Sutent and Nexavar based on sequencing data, was given the EGFR inhibitor Tarceva, after tumor pathology revealed EFGR overexpression. However, after six weeks of treatment the pulmonary nodules continued to grow. The subsequent genomic and transcriptomic analysis offers a window into why the patient failed to respond to Tarceva.
"The high expression of RET (which like EGFR, activates the RAS/ERK pathway) provides a plausible explanation of the failure of erlotinib to control proliferation of this tumor. PTEN loss has also been implicated in resistance to the EGFR inhibitors [Iressa (gefitinib)] and [Tarceva] to which the tumor was determined to be insensitive," the researchers wrote in the paper. "Lastly, the mutated RB1 may also play a role in the observed Tarceva insensitivity, as the loss of both RB1 and PTEN as seen in this tumor has previously been implicated in gefitinib resistance."
Once the researchers integrated copy number, gene expression, and mutational data, they were able to build a hypothesis regarding the mechanisms driving the patient's tumor, and identify drugs that might target the cancer. Before administering Sutent, the researchers used FISH and immunohistochemical analysis to confirm PTEN and RET status in tumor cells.
After 28 days of treatment with Sutent, the patient's lung metastases had shrunk by 22 percent and there were no new lesions. In contrast, while on Tarceva, the patient's metastases had grown by 16 percent. When the cancer began showing signs of growth after four months of Sutent treatment, the patient was switched to Nexavar and sulindac — treatments that were also consistent with his genomic profile. On this regime the patient remained stable for three months, after which new lesions appeared on the tongue, skin, neck, and lungs.
Ultimately, the published study offers insight into how to use genomic strategies to manage therapeutic resistance incurred by cancer patients. "In some ways the resistant forms can represent a 'new disease type' with possibly many treatment options becoming available in these new forms that weren't present in the pre-treatment tumors," Jones said.
In the study, transcriptome and genomic sequencing of a skin nodule revealed nine new non-synonymous protein-coding changes that were not present in the pre-treatment tumor or in the patient's DNA.
"In the tumor recurrence, 0.13 percent of the genome displayed high levels of amplification, compared to 0.05 percent in the initial tumor sample," the researchers reported in the paper. "Also, 24.8 percent of the initial tumor showed a copy number loss whereas 28.8 percent of the tumor recurrence showed such a loss." Additionally, the researchers identified eight regions where the copy number status shifted from a loss to a gain in the tumor recurrence and twelve regions where the copy number changed from a gain to a loss.
The BCGSC researchers identified 459 differentially expressed genes in the metastatic skin nodule versus the blood/compendium, and as many as 209 genes overlapped with the differentially expressed genes in the lung tumor versus the blood/compendium set. Between the skin and the lung metastasis there were 6,440 differentially expressed genes. Finally, the researchers identified various drugs and combination treatments that target 23 amplified, overexpressed, or mutated genes in cancer pathways.
In particular, "pathway analysis … shows IL8 signaling to be significant in the [Sutent]-resistant skin tumor compared to the lung tumor," the researchers reported. "Though the mechanism of resistance is still unclear, IL8 has been observed to transactivate EGFR and downstream ERK, stimulating cell proliferation in cancer cells."
Based on these findings, while the researchers found Nexavar and Sutent to have some efficacy targeting tumor pathways and stabilizing diseases for a period, they concluded that a cocktail of drugs would be necessary to impact the complex interplay of genomic pathways and control advancement and spread of the tumor.
"We propose that complete genetic characterization in this manner represents a tractable methodology for the study of rare cancer types and can aid in the determination of relevant therapeutic approaches in the absence of established interventions," the researchers concluded in the paper. "We envisage that as sequencing costs continue to decline that whole-genome characterization will become a routine part of cancer pathology."
According to Jones, in the early days of next-generation sequencing, this experiment would have cost upwards $200,000. "But nowadays I am assuming it would be in the realm of a few tens of thousands" of dollars," he said, "which is not a huge amount in the context that cancer drugs themselves may be costing over $5,000 per month."