The Future of Genomics and Proteomics in DIPG
The sequencing of the human genome has resulted in a revolution in biology and medicine that offers enormous possibilities leading to an improved understanding, diagnosis, and treatment of human diseases. These advances have largely been achieved in two major domains. First, the development of new technologies that can rapidly and cheaply analyze DNA, RNA, and proteins has led to an explosion in our understanding of the building blocks of the cells, and how they interact with each other through the process of growth and development. The second major advance has occurred in the area of bioinformatics. Developments in computational sciences have permitted the storage and analysis of the billions of fragments of tumor data that result from these technologic advances, and provide the opportunity to place them in the context that better approximates complex biologic systems.
Thus, we now have the opportunity to examine the genome of cancer as well as begin to understand it. Here we will be to review the technical advances, specifically genomics and proteomics, and place these in the context of future therapeutic developments for diffuse intrinsic pontine gliomas (DIPG). These important advances are just now being applied to DIPG and like most other advances, will take some time before their impact is felt in the clinical treatment of DIPG.
In its simplest form, genomics refers to the reading of the genetic code of cells. DNA is the genetic material that acts as the blueprint for making new cells as well as all of the information needed to maintain the current ones. Tumors often alter their DNA (the blueprint), and divide into multiple "daughter" cells which then inherit the same altered DNA, leading to the propagation of the cancer. When a cell divides, the two new cells formed are called daughter cells and when they divide, four daughter cells will be created etc. Alterations in the DNA occur via changes in the genetic code which is made up of four nitrogen bases identified by the letters: A (adenine), G (guanine), T (thymine), and C (cytosine). These 4 letters spell out all of the proteins that need to be made (called the coding regions) as well as the intervening sequences of DNA that contain the control regions (called the non-coding regions) that determine when proteins are made. When an error occurs in the code of a cell, not only does it have the potential to affect that cell, but that error is also transmitted to every new daughter cell. Thus, errors can accumulate, increasing the malignant phenotype of the tumor, as well as its resistance to therapy.
How do Mutations in the DNA Sequence Cause Cancer?
There are generally two types of genes that can sustain mutations leading to cancer: tumor suppressor genes and oncogenes. In normal cells, tumor suppressors make proteins that keep cells "in check" (e.g. suppress tumors). However, if errors in the DNA of tumor suppressor gene(s) are acquired, and these errors destroy the function of the tumor suppressor, then tumors are no longer suppressed, and the cell is not kept "in check" any longer. While tumor suppressors cause cancer by their absence, oncogenes cause cancer by their presence. Mutations occur to oncogenes that impart new abilities of the protein to cause cancer. The alteration in the DNA of oncogenes results in a protein that instructs cells to continuously divide, with uncontrolled proliferation.
The cell that possesses these types of mutations determines the kind of cancer observed. When these mutations happen in blood cells for example, leukemia results. If they happen in a brain cell of the pons, the patient is diagnosed with a diffuse pontine glioma. Thus, we consider cancer (tumor) a result of the accumulation of abnormalities in the DNA of cells. The major types of alterations of the DNA that can occur in pediatric brain stem gliomas (and all other cancers) are described below.
Mutation: Genes contain the sequence code necessary to make proteins, and the proteins make up all cells and tissues. Mutations in the DNA occur such that the sequence (blueprint) for a protein has an error in it resulting in defective functioning. In other words, mutations in the genetic code result in no protein being produced, or a defective protein being made. Mutations may involve alternations in only one letter of the code (point mutation), many letters (nucleotides), or the entire gene (deletions, amplifications, see below). If the altered protein has a critical role in cell proliferation, then the first step toward development of a tumor has occurred.
Translocation: Translocations of the DNA occur when certain parts of the DNA coding for a protein, break into two and rejoin at a site belonging to a different protein. This can sometimes result in a new molecule that tells the cell to do the wrong thing. For example, if the protein responsible for rapid proliferation of immune cells during infection accidently gets linked to the gene that makes astrocytes (the cells of the brain that hold everything in place), you may create a new "molecule" that accidentally tells astrocytes to rapidly divide.
Deletions and amplifications: Deleting or amplifying regions of the DNA is a common way in which errors can be introduced into the genetic material of the cell. If a piece of DNA coding for the molecules that stop cellular proliferation is lost (deleted), then uncontrolled cell division can result. In a similar way, if a segment of DNA that codes for the protein that makes cells start to divide is amplified, then the cell is stimulated to proliferate and tumor growth ensues.
Truncation (partial deletion of a gene): Many genes are organized such that one part possesses the functional part of the molecule (stop or start cell division) while the other end possesses the control sequences. Thus, knocking out bits of a gene, rather than the whole thing, can sometimes have a significant impact on the function of that molecule. For example, if the control region of a molecule is lost, it may continue to activate the cell even at times when it should be in the "off" position.
Epigenetic control of methylation and acetylation: Some cells are out of control not as a result of some assault on the integrity of the DNA but because a perfectly normal protein is expressed in the wrong place or at the wrong time. For example, during early embryogenesis, the fertilized egg must rapidly divide billions of times as the fetus grows. This is an example of normal proliferation. If an astrocyte that has finished dividing accidently turns on the embryonic signal for a cell that is supposed to be dividing, then it will begin to divide even in the absence of any mutations, translocations, deletions, or amplifications. The process of abnormal expression or timing of a normal gene may be of particular importance for pediatric cancers where the proliferation of many cell types associated with growth and development are still active.
The field of genomics is typically divided into two major components: structural and functional. Advancements in the evaluation of the structural organization of the DNA required technology that could rapidly, inexpensively and reproducibly analyze the genetic code of both normal and abnormal tissue. This would then allow for the identification of the mutations, translocations, amplifications, or deletions of the DNA discussed above. Improvements in this technology are happening rapidly, and genomic analysis can now be done at both large and small academic institutions around the world. Many of these techniques are performed on small inert platforms (often called "chips") where millions of reactions can be performed simultaneously. The complexity of each chip determines how much information can be gathered from the sample being tested. Chips are available for the study of DNA (the genetic material), RNA (the intermediate code derived from the DNA) and proteins (the translated end product of the RNA code into amino acids). Simply finding an alteration in the DNA, RNA, or protein however does not prove that it is responsible for disease. Thus, clinicians and scientists must map the genetic abnormalities identified, onto the tumors ability to divide, infiltrate, and escape treatment. For example, a mutation in a protein not expressed in the tumor is unlikely to be responsible for the tumor. Thus, each abnormality in the DNA, RNA, and protein must be assessed for its active role in the tumor. Once identified, the relevant abnormalities can then be considered for targeting with drugs or other therapies.
Changes in DNA, RNA, or epigenetic events may be responsible for the abnormal function associated with many tumors, but the analysis of the proteins themselves is the most direct method to assess for critical changes in a cell’s function. Recall that the purpose of DNA is to provide the code (via RNA) for all proteins. The field of proteomics uses a variety of techniques that allow for the separation of the thousands of proteins in a cell, as well as the structure and function of many of these molecules. Proteomics is usually divided into two critical phases. First is the separation of the different proteins in a sample—typically achieved by their size and overall charge (basic or acidic); and second is the identification of the different proteins—usually achieved through a technique called mass spectroscopy. As with genomic analysis, tissue is necessary for proteomics; however because many tumors shed their proteins in the blood or cerebral spinal fluid (CSF), these samples can sometimes act as surrogates to tumor tissue. Our increasing ability to identify the location, quantity, and activity of different proteins in a tumor sample has provided the pharmaceutical industry with the targets on which tumor specific inhibitors can be developed. In fact, a number of these drugs are already being used for a variety of different adult cancers and have started testing in pediatric patients as well.
Genomics of DIPG
A primary requirement for genomic analysis of cancer, is actual tumor material. While the biopsy of pontine gliomas was frequently performed in the 1970s (before any of the current genomic techniques were available), a change in national policy occurred in the 1980s for sound scientific and clinical reasons. The diagnosis of DIPG was becoming easier to define radiographically through CT scans in the 1970s and in particular with MRI scans in the 1980s. Furthermore, with the very poor prognosis of these tumors with or without biopsy, the lack of justification for a biopsy, in which patients could experience significant neurosurgical damage, resulted in a moratorium on this procedure. Over the intervening 30 years, innumerable clinical trials of radiation alone or in combination with chemotherapy, biologic therapy, anti-angiogenic (anti blood vessel) therapy, gene therapy, immunotherapy, etc. have been performed. None of these approaches have significantly altered the outcome of this disease when compared to treatment with radiation therapy alone. Because none of these children had a biopsy, the reason these combination therapies failed remains unknown. When biopsy was performed, this was generally done because the tumor was atypical and histologic assessment was needed. The majority of these atypical lesions were discovered not to be DIPG. These studies were important as they demonstrated the relative safety of biopsy in this region.
Adult DIPG and animal models of pontine gliomas (not all of them are diffuse and intrinsic) are helping guide our understanding of the important genomic changes that help maintain a tumor’s growth and resistance to therapy. Unlike most diseases, adults rarely get DIPG and in the few reports of this disease in these patients, the clinical course appears different than in children—a finding that suggests that DIPG prefer the environment of the pediatric pons. While it is quite easy to start tumor growth in animals for lung, breast, prostate, or colon cancer, mice do not develop DIPG spontaneously. Fortunately, a number of groups have been working on the development of animal models and the first reports of possible contenders are now available. While these models are likely to be useful in extending our understanding of this disease, they are unlikely to provide all of the answers. As we discovered many years ago, we have cured just about every tumor type in mice many times over and yet those same results have often not been realized in humans. They may, however, become good animal models for “proof of principle discoveries” once pediatric DIPGs are analyzed for their genomic changes.
To overcome the lack of fresh tumor material derived from the time of diagnosis of patients with DIPG, many centers have initiated genomic analysis of autopsy material. While these studies will provide some important insight into the biology of these tumors, they all suffer from the fact that the molecular analyses performed post-mortem are altered by the initial treatment of the tumor with the variety of procedures mentioned above (radiation, chemotherapy etc.), and by the relatively limited sample size of these studies. For example, in a study of 11 cases (9 autopsy and two newly diagnosed DIPG), abnormal expression of PDGFRa in 4 cases and PARP1 in 3 cases were identified. Equally important in these studies was the identification that the genetic abnormalities in DIPG were different from malignant gliomas in other parts of the brain. Thus, molecular profiles of supratentorial malignant gliomas cannot always be used to identify appropriate pathways for the treatment of DIPG, which perhaps helps to explain the three decades of failed clinical trials. These studies also support a commonly held belief in the field that DIPGs are a heterogeneous population of tumors, and that one treatment is not likely to be useful in all cases.
Another approach to the molecular classification of DIPG has used imaging. Both MRI/MRS and PET/SPECT can identify individual markers in a tumor without the need for biopsy. For example, an 11year old female with a large pontine tumor demonstrated strong uptake of In-111-pentreotide, which identifies the presence of somatostatin receptors in the tumor. Similarly, PET imaging allows for the detection of glucose metabolism in tumors and can provide some important metabolic information in DIPG. As new pathways in DIPG tumors are identified, the ability to follow tumor growth or response with these types of imaging markers will likely make these modalities of greater importance in the near future.
With significant advances in neurosurgical technique and previously performed biopsies (in selected cases), the ability to safely biopsy brainstem tumors is becoming better recognized. In a landmark study demonstrating the safety of biopsy of DIPG, 24 consecutive children successfully underwent this procedure in Paris. Not only did the patients not suffer long-term consequences of the biopsy, in two patients, a diagnosis other than a DIPG was identified. The rapidity of improved neurosurgical techniques is now opening the door to direct administration of therapy into the brainstem, not just biopsy. As our improved molecular understanding of these tumors continues, the ability to administer drugs directly into the pons will likely play a greater role in treatment. A major regulator of cellular proliferation known as p53, has been extensively evaluated in both newly diagnosed and autopsy cases of DIPG, as well as other brainstem tumors. This critical regulatory gene was abnormal in over half of the cases in two different studies. Unfortunately, there are currently no drugs targeting p53. While only limited information on the molecular phenotype of pontine gliomas is currently available, the opportunity to change this is rapidly approaching.
We now find ourselves at an important crossroads to the molecular classification of DIPGs. For the last 30 years, it has been felt that the diagnosis of DIPG is easily made by imaging and clinical evaluation. The risks of biopsy within the pons were felt not to justify routine biopsy, and the moratorium on biopsy was considered appropriate. Today, with improved neurosurgical techniques and the availability of sophisticated genomic technologies that can derive extensive data from very small biopsy samples, the tide has turned. DIPG is not a single disease caused by a single mutation. Rather, there are a large number of abnormal pathways that likely account for these tumors and only by identifying them can we expect to develop the kinds of interventions that will be successful.