JIM DOWDALLS/PHOTO RESEARCHERS INC.	The promise will only be realized with more support By Ira Mellman Despite its obvious scientific appeal, immunotherapy as an approach to cancer has yet to live up to expectations. Initial attempts at using cytokines to stimulate anticancer T cells, or deploying toxin-conjugated monoclonal antibodies as "magic bullets," were never quite successful despite having attracted considerable attention.

Therapeutic vaccines for cancer have proven similarly disappointing. Steven Rosenberg, a noted cancer immunologist at the National Cancer Institute, reviewed progress to date in 2004 and concluded that the objective clinical response rate for roughly 1,000 patients fell below an unimpressive 4%.¹ Skepticism and a lack of support has impeded research in the area such that even a role for the immune system as a natural surveillance mechanism to detect and eliminate incipient cancers remains without wide acceptance, despite a large body of experimental and clinical evidence.²

Yet, as a treatment for diseases other than cancer, immunotherapy - defined broadly as modulation of the immune system for therapeutic benefit - has emerged as one of the most exciting, promising, and effective treatment strategies for chronic inflammatory disorders, diabetes, transplantation, and other debilitating conditions (see the "Success Stories" sidebar below). Is it really not a viable approach in cancer?

The fact is that most work to date has been conducted in the absence of sufficient knowledge of the human immune response, particularly the response to cancer. Thus it is dangerously premature to conclude that cancer vaccines will not work when they have yet to be adequately conceived, supported, and coordinated.

Despite dramatic advances in our understanding of cancer cell biology and continuous if incremental gains in cancer treatment efficacy, conventional therapy has remained fundamentally unchanged for decades. In general, treatment still involves surgery, where possible, followed by broad-spectrum cytotoxic chemotherapy in an attempt to kill the patient's cancer before killing the patient.

ﾩ ARTWORK: ERICA P. JOHNSON

THE IMPORTANCE OF MATURITY:
In the path from peripheral tissue to lymph node, dendritic cells mature to the point at which they can arm B cells, T cells, natural killer cells, and NKT cells to attack tumors.
Click to view larger

Two recent developments have begun to change this situation. First, targeted chemotherapeutic drugs such as Gleevec, Sutent, Tarceva, and Iressa zero in on specific tyrosine kinases, or kinase mutants, associated with a given cancer cell's survival or proliferation. Although promising, such agents have thus far proved most effective in treating relatively rare cancers.

? 2005 HYBRID MEDICAL ANIMATION

Second, there has been a veritable renaissance in monoclonal antibodies (mAbs). These can be considered immunotherapeutics, as their very production requires mobilizing a core feature of the immune system, and in many cases they initiate elements of the patient's immune system (e.g., natural killer cells and macrophages), either to kill antibody-coated tumor cells or possibly help stimulate anticancer immunity.

Nevertheless, huge gaps in the armamentarium remain, and filling them will require a change in the way we look at cancer. Although cancer is commonly considered as a disease that must be "cured," for many patients it may be more realistic to view cancer as a chronic condition that must be managed. If immunotherapy has been a success at managing chronic inflammatory disorders, why not cancer?

Whether or not one accepts that cancer is normally suppressed by continuous immunosurveillance, experimental and clinical evidence clearly supports the idea that we should be able to mobilize the immune system for meaningful, even dramatic therapeutic benefit. For example, when advanced melanoma patients resolve spontaneously or after vaccination with tumor antigens there is typically pronounced vitiligo, indicating the cytotoxic killing of normal melanocytes that share antigens with the tumor cells. Patients with cutaneous T-cell lymphoma that show infiltration by CD8+ T cells either before or after vaccine therapy have greatly improved prognosis. The challenge is to understand the mechanisms at work to optimize what we do to benefit a substantially larger fraction of patients.

Achieving this goal will require thorough study of a wide range of immunotherapeutic approaches and new insight into basic principles of human biology. It will also require that we substantially change how we organize and fund research in human cancer immunotherapy. We cannot rely on laboratory investigation alone, yet funding agencies, pharmaceutical companies, and academic institutions are not yet up to the task of efficiently enabling investigation in the only laboratory that really counts: the cancer clinic.³

Two fundamental, opposing forces control the immune system: immunity and tolerance. These forces are normally in balance, allowing vigorous and selective responses to invading microorganisms (immunity), while avoiding unwanted responses to our own proteins, cells, and tissues (tolerance).

Thus far, most efforts at cancer immunotherapy have focused only one side of this equation: enhancing immunity. This is understandable: Relatively little is known about tolerance. Moreover, injecting antigens for the purposes of immunization is a deceptively simple process, making it a popular approach. Vaccines to prevent infectious disease have saved millions of lives.

Eliciting immunity - the minimal prerequisite for achieving a protective or therapeutic vaccine - is no simple matter, however. Compared to the mouse, we know precious little about the human immune response. As a result, many vaccine trials have been constructed on a fragmentary or inaccurate understanding of how to immunize against cancer antigens, or any antigens for that matter. Often, assays performed to determine whether an immune response had been obtained were inadequate or not even performed.

The immune system evolved to protect against microbial infections, not cancer. Accordingly, it is instructive to consider how immunity to infection works. Immune response initiation revolves around a recently appreciated family of leukocytes known as dendritic cells (DCs). Upon encountering invading organisms or microbial products, DCs ingest the invaders and begin migrating via lymphatic vessels to lymphoid organs. They mature en route, activating their ability to convert antigens to 10- to 15-mer peptides bound to major histocompatibility complex (MHC) class I and class II molecules. Mature DCs also upregulate production of surface "costimulatory molecules" (e.g., CD80, CD86) and cytokines needed to stimulate the antigen-specific T cells they encounter in lymph nodes. Diverse populations of DCs in the blood and nodes can also initiate T-cell responses by directly capturing soluble antigens.

The Monoclonal Question

JAMES KING-HOLMES/AUTOMATION PARTNERSHIP/PHOTO RESEARCHERS, INC.

Antibodies, immunotherapies in their own right, have already been developed against cancer. Most popular among these are antibodies to important signaling receptors or receptor ligands such as HER2 for breast cancer (Herceptin), the EGF receptor in colon cancer and other tumors (Erbitux), or the angiogenic signal VEGF (Avastin). Although the clinical benefit due to these agents is often not markedly better than with conventional chemotherapy, mABs are generally easier to tolerate. in the case of Herceptin, however, a recent study has indicated a dramatic reduction in breast cancer recurrence, if used as a secondary treatment.^1,2

There is strong justification for further development in this area, although much effort should be devoted to elucidating the basic cell biology of antibody-tumor interactions, how antibodies (and antibody derivatives) are targeted in vivo, and how targeting and clinical response can best be quantified.

A number of therapeutic mABs have also been developed against cancers of the immune system, i.e., certain forms of leukemia. Rituxan is a widely used antibody to the CD20 molecule on B lymphocytes that has impressive utility in certain B-cell lymphomas, although the cancer cells can develop resistance by losing CD20 expression. Campath binds CD52 and is highly effective for chronic lymphocytic leukemia. Other agents include Mylotarg (anti-CD33) and Bl22 (anti-CD22), both of which are used as toxic conjugates to cytotoxic agents or bacterial toxins.

Nonsolid tumors should in principle be more accessible to injected mABs. But it remains a question as to whether any one mAB can be identified that will reliably target more than relatively minor individual indications.

Targeting the still elusive population of cancer stem cells using mABs - for both leukemias and solid tumors - is another exceedingly promising avenue that should be considered for new antibody and T cell-based immune therapies. But this will require far more information regarding the nature of stem cells that might be responsible for cancer recurrence, even after therapy.

1. M.J. Piccart-Gebhart et al., ?Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer,? N Engl J Med, 353:1659-72, Oct. 20, 2005.

2. E.H. Romond et al., ?Trastuzumab plus adjuvant chemotherapy for operable HER2-positive breast cancer,? N Engl J Med, 353:1673-84, Oct. 20, 2005.

DC maturation is fundamental for a number of reasons. Most importantly, it links antigen uptake and processing to the detection of microbial products (via the family of Toll-like receptors) or the sequelae of trauma or inflammation. These microbial or inflammatory signals act to trigger maturation, a function mimicked by artificial adjuvants (e.g., Freund's adjuvant) used for years to produce antibodies in animals for experimental purposes. Interestingly, different microbial stimuli elicit qualitatively different immune responses. On the other hand, if DCs present antigen without upregulating costimulatory molecule and cytokine production, they can instead induce T-cell tolerance; this seems to make sense, since only self-antigens would be present in the absence of microbial stimuli. Other, nonmicrobial means also exist to initiate DC maturation, including alterations in cell adhesion and interaction with various innate lymphocytes such as natural killer (NK) cells or NKT cells. Because such cell types have the ability to recognize and kill tumor targets, their relationship to DCs should be considered carefully, as they may work synergistically with the DCs they stimulate to further enhance the immune response.

So, when contemplating how to vaccinate against cancer, one has to consider what DCs should be targeted, how antigens should be delivered, to what intracellular compartments they should be taken, and what maturation stimuli should be used. Both CD4 and CD8 responses are likely required to provide immunity to cancer, meaning that cells must load the antigens not only onto MHC class II molecules (CD4), but also onto MHC class I molecules. The latter provides a significant challenge: Exogenous antigens must escape from endocytic vesicles into the cytosol and then into the endoplasmic reticulum to find the waiting MHC class I molecules.⁴

The choice of antigens is similarly complex. Tumor-associated antigens may be proteins affected by cancer-specific mutations, proteins that are expressed only in cancer cells and germ-line cells (so-called cancer-testis antigens), or proteins specific to the lineage from which a given cancer is derived (differentiation antigens). Since a given tumor may not be associated with a known antigen, or antigen expression in a tumor may drift or be heterogeneous in a tumor, there has also been considerable interest in using apoptotic tumor cells or tumorcell homogenates.

Dosage and time course are other important considerations. There are many variables and, as mentioned earlier, remarkably little information concerning the human immune response. Many difficulties are associated with obtaining the desired antigens or adjuvants under conditions approved for human use. To paraphrase US Secretary of Defense Donald Rumsfeld: You perform your trial with the reagents you have, not the reagents you may wish for. This is no way to run clinical research.

Due in part to the challenges associated with reagent availability, studies conducted thus far have not been able to optimize these variables. Without providing definitive clinical benefits, they have contributed to the negative impressions of vaccine therapy. When combined with advanced immunological monitoring procedures, however, this work has begun to provide essential information concerning the human immune response.

A number of investigators (and some commercial ventures) have devised cell-based strategies that make use of DCs in the most direct fashion possible. DCs are isolated from a patient with cancer, loaded ex vivo with a preparation of antigen, and then infused back into the patient. While some immunity and perhaps some clinical response has been noted from such approaches, fundamental challenges remain. In many of the commercial attempts at DC-based therapies, little care is taken to characterize the populations of cells used for vaccination or the resulting immune response. Only clinical endpoints are monitored, a scientifically useless measure if not correlated with a detailed analysis of the type of immunity (if any) that has been elicited.

Michel Nussenzweig, Ralph Steinman, and colleagues have pioneered an appealing strategy, targeting DCs in vivo using antigens coupled to DC-specific mAbs.⁵ This approach has yielded impressive results in mice, enhancing the efficiency and the kinetics of the immune response relative to soluble antigen. The strategy is inherently satisfying because it mimics how DCs normally use receptors to detect and internalize microbial antigens. Using high-affinity anti-DC mAbs as vaccine delivery vehicles might give administered tumor antigens a greater targeting advantage, given that the approach seeks to generate immunity against proteins that are closer to "self " than to "foreign."

There is another, potentially even more important reason why vaccine trials have not yet proved successful: Efforts thus far have been devoted entirely to enhancing immunity without paying attention to overcoming tolerance. By the time a tumor is detected, it is possible that a patient's immune system has become tolerant to its major antigenic components. Generating an immune response to cancer may require manipulations that overcome tolerance, in addition to those that enhance and target immunity.

Although DCs contribute considerably to tolerance, there is much interest in a population of T cells called regulatory T cells, or T^regs. These poorly understood lymphocytes have the property of opposing antigen-specific immune responses in vitro as well as in vivo. In mice, manipulations that deplete their numbers or interfere with their function are already being shown to enhance immunity, increasing chances for effective vaccination.⁶ In principle, approaches could be developed in humans that would do much the same, although substantial advances will require considerably more basic investigation of human T^regs. For example, it appears likely that approaches to manipulating T^regs will have to take into account their antigen specificity, as is the case for any other T cell-oriented approach. In the interim, admittedly crude nonselective approaches are possible. Ablating a patient's immune system prior to bone marrow transplant might reduce the frequency of T^regs. If immunization is attempted following bone marrow repopulation, responses may prove more robust. Indeed, the therapeutic benefit seen after bone marrow transplants in patients with cancer may reflect this phenomenon.

Targeting T cells, in general, might also prove effective at boosting immunity. Treatment with inhibitory mAbs to the T-cell surface protein CTLA4, whose function is to suppress Tcell responsiveness, for instance, has proven promising in early human trials.⁹ Further, adoptive transfer of antigen-specific T cells - possibly harboring genetically engineered receptors for known tumor-associated antigens - may increase the frequency of cytotoxic cells that would enhance clinical benefit. Indeed, such studies are underway. the roles played by regulatory t cells will help.

It is not only plausible but also highly likely that one can modulate the immune response to focus and enhance anticancer responses. Doing so may impart an equilibrium between continued growth of tumor cells and the immune system's ability to recognize and eliminate them. But the responses have to be sufficiently robust and sustainable.

Learning how to accomplish this goal will require an open mind concerning the mechanisms and strategies to be tested. It will require that the twin problems of reagent inaccessibility and regulatory restrictions be solved. It is difficult, time consuming, and expensive to produce agents for human use in the academic environment. Unless the pharmaceutical and biotechnology industries can make their reagents available for early-stage, investigational trials, academia, the NIH, and other organizations will have to take up the slack. Finally, the scientific community, and the agencies that fund our work, must recognize the task at hand as an exciting and important scientific challenge. Clearly, it is a challenge that will benefit from the same type of systematic, reductionist approaches we have so effectively applied to model systems in the laboratory.