Tuesday, November 30, 2010

Tumor for Tumor

A new paper appeared in Clinical Cancer Research on fighting tumor with Tumor derived vaccines from dendrite cells. Dendritic cells are critical to the human body's immune system, helping identify targets, or antigens, and then stimulating the immune system to react against those antigens. The new research grew dendritic cells from a sample of a patient's blood, mixed them with proteins from the patient's tumor, and then injected the mixture into the patient as a vaccine. The vaccine then stimulated an anti-tumor response from T-cells, a kind of white blood cell that protects the body from disease.
Purpose: To determine whether an autologous dendritic cell (DC) vaccine could induce antitumor immune responses in patients after resection of colorectal cancer metastases and whether these responses could be enhanced by activating DCs with CD40L.
Experimental Design: Twenty-six patients who had undergone resection of colorectal metastases were treated with intranodal injections of an autologous tumor lysate– and control protein [keyhole limpet hemocyanin (KLH)]–pulsed DC vaccine. Patients were randomized to receive DCs that had been either activated or not activated with CD40L. All patients were followed for a minimum of 5.5 years.
Results: Immunization induced an autologous tumor-specific T-cell proliferative or IFNγ enzyme-linked immunospot response in 15 of 24 assessable patients (63%) and a tumor-specific DTH response in 61%. Patients with evidence of a vaccine-induced, tumor-specific T-cell proliferative or IFNγ response 1 week after vaccination had a markedly better recurrence-free survival (RFS) at 5 years (63% versus 18%, P = 0.037) than nonresponders. In contrast, no association was observed between induction of KLH-specific immune responses and RFS. CD40L maturation induced CD86 and CD83 expression on DCs but had no effect on immune responses or RFS.
Conclusion: Adjuvant treatment of patients after resection of colorectal metastases with an autologous tumor lysate–pulsed, DC vaccine–induced, tumor-specific immune responses in a high proportion of patients. There was an association between induction of tumor-specific immune responses and RFS. Activation of this DC vaccine with CD40L did not lead to increased immune responses. Clin Cancer Res; 16(22); 5548–56. ©2010 AACR.
Citation: Richard J. Barth, Jr., Dawn A. Fisher, Paul K. Wallace, Jacqueline Y. Channon, Randolph J. Noelle, Jiang Gui, and Marc S. Ernstoff, 'A Randomized Trial of Ex vivo CD40L Activation of a Dendritic Cell Vaccine in Colorectal Cancer Patients: Tumor-Specific Immune Responses Are Associated with Improved Survival', Clin Cancer Res November 15, 2010 doi:10.1158/1078-0432.CCR-10-2138

Thursday, August 5, 2010

Somatic Mutations in Four Human Cancers

[Source : From mass genomics]
In a letter to Nature this week, a group from Genentech presents an elegant analysis of 2,576 somatic mutations across 441 tumors comprised of breast, lung, ovarian, and prostate cancer types and subtypes. Using something called “mismatch repair detection” (MRD) technology, the authors surveyed 1,507 candidate genes spanning some 4 megabases of sequence, largely comprised of known cancer genes and “druggable” genes. MRD apparently uses E. coli to isolate amplicons that contain mutations relative to a reference sequence, which are then assessed for variations by a resequencing tiling array. Matched normal samples were also screened (in pools of five) to eliminate germline events.

I admit to knowing little about MRD or its capabilities, but I’m very familiar with the validation platform (Sequenom), which has proven its value in the HapMap, Cancer Genome Atlas (TCGA), and 1,000 Genomes projects.
Significantly Mutated Genes

Any doubts I had concerning a study from the private sector were quickly swept away, not just by the quality of the journal, but by the analysis that the authors presented. Simply put, I was enchanted. Figure 1, for example, illustrates the significantly mutated gene (SMG) analysis with a grid of eight bubble plots, one per cancer subtype. Significant genes are notable not just by their position on the Y-axis (Mutation q-score), but the size of the bubble, which corresponds to the number of mutations.

The set of SMGs varied across type and subtype, but some patterns immediately jump out. PIK3CA and TP53 were the most significant across three breast cancer subtypes. TP53, in fact, was significant across all eight subtypes, most strikingly in lung and ovarian cancer. KRAS stood out in pancreatic cancer and lung adenocarcinoma, but not squamous lung carcinoma.

On average across all tumors studied here, the authors found 1.8 protein-altering mutations per megabase, with the highest rates seen in lung adenocarcinomas (3.5/Mb) and squamous carcinomas (3.9/Mb). The lowest mutation rate (0.33/Mb) was in prostate tumors, 75% of which harbored the TMPRSS2-ERG gene fusion. These patterns are consistent with Figure 1, where prostate shows a sparse handful of significant genes, while lung cancers have large and diverse sets of them.
Integrated Copy Number and Mutation Analysis

Next, the authors integrated their mutations with Agilent 244K array CGH copy number data to identify genes that were significantly altered, either by mutation, copy number, or both. In Figure 2a, the authors plotted significantly altered genes by their copy number gain or loss, which nicely separated oncogenes and tumor-suppressor genes. The integrated analysis identified 35 additional cancer genes including STK11, EPHB1, and notably GNAS (the G-protein alpha subunit). GNAS proved an important finding, as it was mutated and amplified across several human cancers.
Pathway-based and Recurrency Analyses

The integrated dataset identified two pathways - RTK signaling and RAS/MAPK as the most significantly altered across all tumor types. Furthermore, when the authors compared their dataset with the COSMIC database and the findings of recent cancer sequencing studies, they pinpointed novel recurrent mutations in several genes including HER2, NOTCH4, and PIK3R1.

The authors conclude that their study “represents a substantial expansion of the knowledge base of cancer somatic mutations,” and I tend to agree. They not only generated a rich dataset, but also analyzed and presented it in comprehensive fashion. Furthermore, they (perhaps unsurprisingly) identify numerous cancer genes that are druggable targets, thereby translating these findings into actionable information.


Kan Z, Jaiswal BS, Stinson J, Janakiraman V, Bhatt D, Stern HM, Yue P, Haverty PM, Bourgon R, Zheng J, Moorhead M, Chaudhuri S, Tomsho LP, Peters BA, Pujara K, Cordes S, Davis DP, Carlton VE, Yuan W, Li L, Wang W, Eigenbrot C, Kaminker JS, Eberhard DA, Waring P, Schuster SC, Modrusan Z, Zhang Z, Stokoe D, de Sauvage FJ, Faham M, & Seshagiri S (2010). Diverse somatic mutation patterns and pathway alterations in human cancers. Nature PMID: 20668451

Friday, April 16, 2010

Foundation Medicine: a new genomic approach to cancer diagnosis

[Reposted from Times Online]


On June 25, 2000, Bill Clinton welcomed Francis Collins and Craig Venter to the White House to announce the completion of the first draft of the human genome. Ten years on, scientists have taken many important steps towards understanding what this sequence means for human health, yet most of the medical benefits still lie in the future. Except, that is, in one important disease. Cancer.

Cancer is at root a disease of the genes, caused when mutations trigger uncontrolled cell division. Understanding the genome is thus a highly valuable weapon against tumours, as a knowledge of the molecular defects that drive them can allow doctors to attack them with targeted therapies. In the past decade, several of these smart drugs have hit the market, often transforming the prognosis for patients.

Imatinib (Glivec or Gleevec), for instance, has turned chronic myeloid leukaemia into a controllable disease. Patients with colon cancer can benefit from cetuximab (Erbitux) only if their tumours do not carry a mutation in a gene called KRAS. Trastuzumab (Herceptin) is often highly effective against breast cancers with a HER2 mutation, while gefitinib (Iressa) and erlotinib (Tarceva) work against lung tumours with an EGFR mutation. And most recently,an experimental drug called PLX4032 has had hugely promising results against melanomas with a particular mutation of the BRAF gene, offering hope for treatment of a cancer with a terrible prognosis.

The move towards such targeted therapies is certain to be a dominant theme of oncology over the next decade. But if they are truly to transform patient care, they need to be accompanied by simple diagnostics -- the tests that identify the genetic subtype of a patient's tumour, so that doctors can choose the appropriate therapy. As Sir John Bell has pointed out, access to such tests is currently patchy, particularly on the NHS.

That is where an exciting new company may soon come in. Foundation Medicine, based in Boston, has today announced it has raised $25 million in funding to develop a one-stop-shop service for genomic diagnosis of cancer. It is going to be a while before it has a product to offer doctors -- its executives are currently engaged in a "listening tour" while they develop their services. But its idea is one that has great potential to help patients to benefit from advances in genomic science.

Foundation is advised by several luminaries in this field, including Eric Lander, director of the Broad Institute, a pioneer of the Human Genome Project, and an adviser to President Obama. Other consultants include Levi Garraway, Matthew Meyerson and Todd Golub, of the Dana Farber Cancer Institute and Harvard Medical School.

Its goal is to broaden significantly the scope of genomic cancer diagnosis, so that samples of each patient's tumours can be analysed not just for one or two genes that might be salient to their treatment, as can happen at present, but for dozens.

A few leading-edge cancer centres, such as Dana Farber and Massachusetts General Hospital (which I visited recently), are starting to do this sort of testing already. Foundation, though, will offer an outsourced service that is suitable for any hospital, to widen access to such diagnosis to patients being treated at ordinary centres that lack the facilities or expertise to do such tests themselves.

Alexis Borisy, Foundation's chief executive, described the idea to me as follows:

"The vision that motivates us at Foundation is that we believe more and more cancer patients are going to benefit from an understanding of the core molecular aberrations that are driving their specific tumours. The technology has moved to a point where it's feasible to decipher those in an individual, patient-specific manner, and to do that in day-to-day oncology practice.

"Our aim is to make that technology available to oncologists, to have an off-the-shelf service for oncologists that they can have complete confidence in."

Dr Meyerson told me:

"Through Foundation Medicine, we’re going to make genomic testing available to all cancer patients. It will, if you like, democratise access to this sort of thing. That is our goal. My hope is we will be doing comprehensive testing on every patient’s cancer."

The notion of testing tumours for multiple mutations is important because of the pace at which scientists are unravelling cancer genomics. It is already emerging that some mutations that are common in one cancer type, such as BRAF in melanoma, also occur at lower frequencies in others -- in this case, in lung cancer. That means that a BRAF inhibitor, though designed for melanoma patients, might also work against lung cancers with the same mutation.

The problem is that while testing lung tumours for EGFR mutations alone makes sense, as that gene is commonly defective in that cancer type, it is not cost-effective to perform the BRAF test in isolation to catch perhaps 1 per cent of patients who might carry it. A reasonably-priced "multiplex text" that covers dozens of genes, however, could be applied to every tumour to provide useful diagnostic clues.

As Borisy put it:

"Some mutations might have been validated in one type of tumour, but could also be relevant and actionable in other settings. In the next couple of years, literally tens of thousands of cancer genomes are going to be analysed. We're going to discover actionable genes that are present in maybe 1 per cent, 0.5 per cent of tumours. At this low frequency, we won't be able to find these by testing one aberration at a time."

Many more rare mutations of this sort are going to emerge as projects like the International Cancer Genome Consortium examine the genomic architecture of cancer over the coming years. Standardized tests that look at dozens, hundreds even, of mutations are going to be necessary if we're to make full clinical use of these insights -- and services like Foundation's are likely to make those accessible.
There's another issue here: gene patenting. As I wrote when a US court last month struck down Myriad Genetics' patent on the breast cancer risk genes BRCA1 and BRCA2, broad patent rights over genes have the power to severely restrict multiplex testing of the type Foundation intends to offer. It just isn't practical to pay royalties to dozens of different patent holders over dozens of different mutations for which you might test.

Posted by Mark Henderson on April 15, 2010 in Cancer , Genetics |