Q&A: Louis Staudt on Molecular Classification of Cancer

Louis Staudt says his goal is to “continue to define [molecular] cancer subtypes because they are not going away, they are a part of nature”Cancers are inherently diverse, capable of arising from cells at various stages of development and driven by numerous molecular mechanisms. Louis M. Staudt, M.D., Ph.D., has pioneered molecular classification of cancers, specifically lymphomas. His discoveries in defining cancer subtypes have been instrumental in the progress of diagnostics and precision medicine, specifically his characterization of molecular subtypes within diffuse large B-cell lymphomas. For these pivotal discoveries, Dr. Staudt has been recognized by this year’s AACR Award for Outstanding Achievement in Blood Cancer Research. As the chief of the NCI Lymphoid Malignancies Branch, Dr. Staudt’s current work explores new ways to define and target lymphoid malignancies, including through epigenetic and immunologic mechanisms. Speaking with Blood Cancer Discovery’s Abigail Cleveland, Dr. Staudt shares his career experience on the path to scientific discovery, perspective on molecular classification of cancers and precision medicine, and thoughts on the future of biology-informed chemotherapy-free therapies.Following your training at the University of Pennsylvania, you moved to the Nobel laureate David Baltimore’s lab at the Whitehead Institute. What was it like to be part of that scientific community?That was the intellectual high point of my life. It was at the time when the lab discovered NF-κB and the role of RAG proteins in VDJ recombination. Our lab sat on the floor with Robert Weinberg’s lab [who first identified the human oncogene RAS], where I learned about cancer biology. We also had a shared group meeting with Phillip Sharp [who received the Nobel Prize for discovering “split genes”], where we learned about basic mechanisms of transcription and splicing. I was even at a lab meeting when they disclosed that the mechanism by which splicing occurs is through a lariat structure of RNA. It was an exciting combination of new technologies and the biological questions the leaders of these labs wanted to investigate.I was interested in studying transcription factors in B cells. At that time, gel shift had just been developed to visualize—in gel electrophoresis—the binding of a transcription factor to its DNA motif. By doing this, I was able to observe a lymphoid-restricted transcription factor that bound to the octamer DNA motif, which we called Oct-2. Oct-2 ended up having an important role in germinal center B cells, which is what I currently study.However, the most important lesson I learned from David Baltimore was a requirement to think creatively and in ways that would amuse your fellow biologists. That aspiration to be creative was his main gift to me, and I aim to do that in my own lab.“The use of new tools is powerful when you have an inherent idea about what it is you are looking for.”Was there a specific driving question that led you to do cancer research?I was fortunate to get a job at the NCI. There, I worked on projects that developed from my postdoc work. For example, I cloned a new transcription factor, Oct-3, now called Oct-4, which is a central transcription factor for embryonic stem cells [and now defined as one of the Yamanaka factors; ref. 1].Working at one of the largest cancer centers in the world, the NCI, I considered how I could apply what I knew about B cells to the mission of that institute. Therefore, I simply asked, “What are the cancers of B cells?” And those are the lymphomas. I went to the library and read the current textbook chapters on lymphomas, and there did not appear to be a molecule in sight, rather the focus was on descriptive pathology and histology. So, the subject seemed ripe for the picking. I just had to identify relevant molecules.Why study such a genetically complex cancer?I could have worked on any type of lymphoma, but the most common type is diffuse large B-cell lymphoma (DLBCL). It was clear from everything I was reading that DLBCL was considered a wastebasket diagnosis, based mainly on the presence of sheets of large cells. Some patients responded to chemotherapy and could be cured by it, but others succumbed to the disease. It was a clearly heterogeneous disease, and I had a hunch that it was a genetically complex cancer. I was motivated to figure out how that complexity would relate to treatment response, which brought me to gene expression profiling of cancer.Did the heterogeneity and complexity of DLBCL make it more difficult to cross-pollinate ideas between subtypes—or between blood cancers and solid tumors?Lymphocytes are fundamentally different than hepatocytes, brain cells, or epithelial cancers. Therefore, the pathways that are critical in other cell types, such as MAP kinase, do not play a vital role in lymphomas. Some important pathways are exclusive to lymphomas because they use mechanisms that are unique to B cells responding to antigens in their environment.Specificity was right up my alley. I could explore those lymphoma-driving pathways and then use various strategies to test whether those pathways were essential in normal B cells. I viewed the fact that there was not that much cross-pollination as giving me a larger open field to run in.“Cancer is complicated. You have to embrace its complexity to find the right regimen of targeted agents that will leave the cancer with no way out.”Tell us about your experience collaborating with Ash Alizadeh and Patrick Brown to develop the lymphocyte-specific microarray, Lymphochip?A colleague at the NIH happened to visit Patrick Brown [who invented the microarray] at Stanford University and said, “You will not believe what Pat can do. He can look at the activity of thousands of genes at the same time!” That resonated with what I imagined might cause some tumors to respond to chemotherapy and others not, namely the activity of particular genes and the transcription factors that contol them. A light bulb went off, and the very next day I called Pat and told him about my idea to make a microarray enriched for genes we knew were expressed in lymphocytes, which he enthusiastically endorsed. I proposed the idea to five students who came through my lab, but the only one to embrace the project was Ash Alizadeh [who is now a medical oncologist and geneticist at Stanford University].The Lymphochip was a microarray enriched for genes expressed in lymphoid cells. Back in the day, you could deplete RNA of genes expressed in all cells by subtractive hybridization, leaving only those that were tissue-specific, and then make a library out of those. We constructed the Lymphochip using cDNA clones isolated from subtractive libraries made from normal germinal center B cells. Following our initial publication (2), we started printing the Lymphochip at the NCI and used it to profile DLBCL and other blood cancers, including chronic lymphocytic leukemia, mantle cell lymphoma, Burkitt lymphoma, primary mediastinal B-cell lymphoma, and follicular lymphoma. Each one of those efforts was highly illuminating regarding the nature of heterogeneity in those cancer types, especially how that heterogeneity influenced survival following therapy. Sometimes that had to do with signatures of gene activity in the malignant cells, and sometimes it had to do with gene signatures in the immune microenvironment. All of the discoveries we made have stood the test of time, although they have been elaborated on and refined.What role do you see for new technology or tools versus new ideas in driving discoveries?Tools have always driven scientific discovery. Just consider the story of Galileo Galilei, who did not invent the telescope. The first telescope was invented by Dutch scientists, but it was only at 4× magnification. Galileo improved the optics and raised the magnification to 16×, and with that he said, “Well, maybe I will find something interesting if I look up.” And so he pointed it into the night sky and saw the moons of Jupiter and rings of Saturn, and the rest is history.I have always been one of the first adopters of new technologies. As a graduate student, I jumped on monoclonal antibodies, a new technology at that time—but that is not just because it was a shiny new object. It was the chance to discover something that had not been seen before. The research I like best is what can be loosely termed as discovery research, where there might be something to find in a certain area of interest, but you are not sure. You take a look, and that is the beginning of making a discovery that leads to more hypothesis-driven research.The use of new tools is powerful when you have an inherent idea about what it is you are looking for. Profiling for its own sake, or sequencing for its own sake, will not get you very far. I like to discover things after thinking about what I might discover. As Louis Pasteur said, “chance favors the prepared mind.” You want to be expecting certain things and then look for them.Many consider you a pioneer in using gene expression profiling to molecularly distinguish subtypes of DLBCL. How did this work develop into our current understanding of genetic and molecular classifications of cancer?Microarrays distinguished various cancers that pathologists could already distinguish. Our first paper on DLBCL subtypes—activated B cell–like (ABC) and germinal center B cell–like (GCB)—was one of the first descriptions of novel molecular cancer subtypes (2). As people applied microarray technology to various cancers, that became the rule rather than the exception.The second major transformation was sequencing technology and resequencing cancer genes. I was involved peripherally in the TCGA project, a group effort to sequence 500 tumors from particular types of cancer. Several different cancers were chosen, but a diffuse lymphoma initiative never got off the ground. Therefore, my own laboratory generated a TCGA-like high-quality discovery data set derived from 574 diffuse lymphomas, which we shared with the research community (3).TCGA often used methylation status, copy-number changes, and some gene expression changes to identify cancer subtypes. I thought that oncogenic mutations that implicated critical pathways could be used to identify subtypes and the pathways relevant to each subtype. I challenged my longtime colleague and friend, George Wright, a computational mathematical statistician, to develop a new algorithm to use constellations of co-occurring genetic abnormalities to define genetic subtypes. With his novel algorithm, we eventually identified seven subtypes of DLBCL.We published our findings (3), and almost simultaneously, a group led by Margaret Shipp and Gad Getz published a similar paper (4). Their findings were so in sync with ours that it was uncanny. The world had coalesced. Our current thinking is that there is a phenotype that a cancer cell is selected to have, and that the initiation of lymphomagenesis by a particular mutation or translocation may dictate selection for a second, third, or fourth mutation. Thus, lymphomas develop that can look like each other under the microscope and have related gene expression profiles but have taken a different evolutionary path. Therefore, the concept of genetic subtypes is not at all in opposition with the previous identification of gene expression subtypes.The important question is how closely the genetic subtypes have aligned with therapeutic outcomes. An obvious example of this came from a large phase III clinical trial that my colleague Wyndham Wilson and I co-led in diffuse lymphomas (NCT01855750; ref. 5). We aimed to show that, when added to chemotherapy, the BTK inhibitor ibrutinib can benefit patients with the ABC type of DLBCL because it targets B-cell receptor signaling. We showed that for two genetic subtypes, treatment with ibrutinib plus chemotherapy yielded a 100% 3-year progression-free survival rate versus 50% or less with chemotherapy alone. This is an excellent example of precision medicine.Was precision medicine considered a feasible concept prior to gene expression profiling or genetic subtyping?It definitely was, but the question was how generalizable the concept was. Chronic myelogenous leukemia with BCR-ABL translocation, for example, was one of the first cancers to be treated with a targeted agent developed against an oncogenic kinase, which led to remarkable lifelong remissions. And in acute promyelocytic leukemia with the PML–RARA fusion, treatment with retinoic acid was remarkably effective. Therefore, you may say: just find your pathway, get a drug, and cure cancer. But in both of those situations, that seemed too easy—and it was. In ABC DLBCL, we discovered that the B-cell receptor signaling pathway is essential for the survival of the lymphoma cells, and we showed that BTK inhibitors could block this pathway, killing the lymphoma cells (6). However, it became clear that lymphomas have multiple redundant survival pathways (7), explaining why treatment with BTK inhibitors alone shrink tumors for a few months but do not typically induce durable remissions. I think precision medicine took a little bit of a hit when single-agent targeted therapies did not meet expectations. But that is because cancer is complicated. You have to embrace its complexity to find the right regimen of targeted agents that will leave the cancer with no way out.Considering your recent article on the ViPOR regimen (8), how do you view combinations of targeted therapies compared with more conventional chemotherapy-heavy regimens, such as R-CHOP?We are excited about ViPOR because we believe that, in the future, we may be able to cure most patients with aggressive lymphomas with nonchemotherapy-based regimens. This is in part due to the success of the ViPOR regimen, which includes five agents targeting oncogenic signaling pathways in cancer cells (venetoclax, ibrutinib, prednisone, obinutuzumab, and lenalidomide), as well as an anti-CD20 antibody, to enlist the immune system (8).Even in tumors that have resisted three or four therapies, including chimeric antigen receptor (CAR) T-cell therapy, we have seen durable complete remissions with ViPOR. Importantly, we realized early on that giving these drugs long term as a maintenance therapy simply added more toxicity. My colleague Wyndham Wilson and I wondered: “Could you deliver these drugs in combination over several cycles in a short period of time, and then stop altogether?” Incredibly, without any ongoing therapy, tumors treated with ViPOR have stayed in complete remission in about 40% of patients with particular lymphoma subtypes. This is in line with CAR T-cell therapy survival rates, but ViPOR has advantages. In patients with tumors refractory to chemotherapy, CAR T-cell therapy works, but there is a caveat. To precondition the bone marrow to accept the CAR T-cell product, oncologists use fludarabine and cyclophosphamide, which are fairly toxic. I am glad we have CAR T-cell therapy, but ViPOR has fewer side effects, and it is certainly less costly.How do you see lymphoma classifications evolving and informing therapies in 10 to 15 years?The goal is to continue to define [molecular] cancer subtypes because they are not going away, they are a part of nature. However, the question you are really asking is what will precision medicine look like? Do I need one particular therapy for subtype A and a different one for subtype B or subtype C? We do not believe so. One reason for developing a five-drug regimen was to create a regimen that might be a “one-size-fits-all” treatment. ViPOR likely works better in some subtypes than in others, but it still has activity across many subtypes, and we believe there is a commonality of the survival pathways that cancer cells use. First, ViPOR targets B cell receptor signaling, which is an essential survival pathway shared by most forms of B-cell lymphoma. Second, ViPOR lowers the apoptotic threshold by targeting a Bcl-2 family member. Third, ViPOR targets transcription factors that are lymphoid-specific, which can be done using lenalidomide, for example. And fourth, ViPOR engages the immune system. After you put all of that together in a combination therapy, tumor cells may have very few escape routes. Just because a therapy can treat multiple molecular subtypes does not mean the subtypes do not exist. We just want to generate a regimen that can overcome the differences between subtypes and that is safe for patients.The novelty with ViPOR is not the concept of drug combinations per se, but rather the fact that ViPOR targets multiple survival pathways simultaneously. A common approach to developing drug combinations is to first try drug doublets, find a doublet with activity, and then maybe add one more drug, but this incremental approach is not time-efficient. ViPOR skips a few steps and combines multiple drugs at once. Our careers are too short to wait for testing single combinations; we want good results for our patients sooner. This puts the scientist and laboratory investigators really front and center in the development of new therapies, because it is no longer something empirical. It is understanding the survival pathways in the malignant cell and then devising a regimen that might hit multiple pathways simultaneously.In all, we do not think we are done yet. Our ViPOR platform is well tolerated, with virtually all recipients capable of completing all six cycles. We are looking forward to adding immunotherapies to ViPOR, especially bispecific T-cell engagers. The real win would be to knock chemotherapy off the pedestal for patients newly diagnosed with diffuse lymphoma.