Science recently highlighted cancer immunotherapy as the “2013 Breakthrough of the Year”. More to the point, they stated "this year marks a turning point in cancer, as long-sought efforts to unleash the immune system against tumors are paying off - even if the future remains a question mark." (1) Most of the celebration involves the compelling results of targeted agents designed to reactivate the immune system by manipulating the PD-1/PD-L1 and CTLA-4 pathways. By blocking these recently identified suppressor molecules, these targeted therapies are designed to unleash the immune system either as monotherapies or in combination with traditional cytotoxic chemotherapy. The ultimate result of either strategy should improve the treatment of established, late stage disease, a patient population that has yet to be adequately addressed with modern modalities. While these investigations have provided a novel direction for enhancing cancer treatment, additional technologies still need to be developed to specifically identify tumor-associated antigens (TAAs) to harness the full power of the immune system. Active specific immunotherapy (ASI) has the potential to be that transformative technology by embracing the recently demonstrated genomic heterogeneity of tumor cells, through the use of live, metabolically active, autologous tumor cells which represent the entire antigenic diversity of each patient’s primary tumor.
ASI involves generating a robust, cell-mediated, cytotoxic immune reaction against tumor cells. This concept is rooted in the reality that patient-derived vaccines can induce a potent and lasting immune response against TAAs capable of halting or even eliminating tumors to prevent recurrence. If immunomodulatory agents are capable of rearming the immune system against cancer, then ASI serves as the guidance system. However, do we have physical evidence to suggest killer T-cells are capable of interacting with malignant cells in a therapeutic manner? There is a considerable amount of literature discussing the quantitative effect of in vitro T-cell mediated cytotoxicity, but there is a paucity of direct evidence. In a previous study, Bucana et al., (2,3), harvested immune lymphocytes and other white blood cells from the peritoneal cavity of guinea pigs that had previously been immunized and cured of a lethal, transplantable, syngeneic hepatocarcinoma. These primed immune cells, with proven in vivo cytotoxic capabilities, were then combined in vitro with the same tumor cell line and followed sequentially with time-lapse cinematography, scanning- and transmission-electron microscopy. In this manner, the authors were able to directly observe lymphocyte and macrophages interact with “familiar” cancer cells in vitro.
During co-culture, considerable migration of lymphocytes and other monocytes (later identified as macrophages), occurred on and around the tumor cells (Figure 1). Two important observations characterized the interaction of lymphocytes and activated macrophages with the tumor cell surface. First, active extension and contraction of monocyte cytoplasmic processes was observed around and into the tumor cell surface. This "probing" was sustained for long periods of time without significant lateral movement of the activated monocytes. Thus, considerable clasmatosis of the T-cells and macrophages was observed at the tumor cell surface. Secondly, exocytosis of osmiophilic organelles from the monocytes to the tumor cells was observed (Figure 2). These organelles had the morphologic and histochemical characteristics of primary and/or secondary lysosomes. Additionally, it was clear that some of the lysosomal organelles were still within segments of monocyte cytoplasm that had detached at the tumor cell surface. Clasmatosis of macrophage or monocyte cytoplasmic extensions with these organelles would allow for tumor cell internalization of toxic proteinases initiating the dramatic cytotoxic events that were observed during the study. Other experimental systems have also reached similar conclusions concerning the cytotoxic events occurring between the innate immune response and cancer cells (4,5). Following cell lysis, the existing macrophages further phagocytized the ruptured tumor cell cytoplasm enabling the identification of secondary or previously “hidden” TAAs. Based on the studies above, the innate immune system already possesses the tools to recognize and destroy malignant cells if appropriately directed.
The early claims of immunotherapy for cancer came from reports of infectious agents reducing or eliminating localized tumors both in animal models and man (for review see Hanna et al., 6). In fact, the first vaccine approved by the US Food and Drug Administration for the treatment of cancer was Bacillus Calmette-Guerin (BCG). In 1976, Morales et. al., (7) first reported the use of BCG for treatment of non-muscle invasive superficial bladder tumors. They reported a 12-fold reduction in recurrence rate of superficial bladder tumors following combined intravesical and intradermal administration of BCG. Subsequently, numerous prospective randomized clinical trials demonstrated the efficacy of intravesical BCG therapy for therapy of Carcinoma-in-situ (CIS) and later for preventing the progression and recurrence of superficial papillary bladder cancer. It seems that the immune system is triggered by the admixing of the BCG attaching to the tumor at the wall of the bladder and this is often considered to be more inflammation by the innate immune response, thus categorized as active nonspecific immunotherapy.
These results supported the enthusiasm for the specificity of ASI as a rational modality for cancer treatment and developing cancer vaccines as a means of achieving tumor-specific immune responses for disseminated disease. However, the majority of cancer vaccines have failed in practice (Figure 3). Over the last decade, the failure rate of these treatments in phase II/III clinical trials is over 70%. If we intend to make meaningful progress with vaccine-based cancer treatments, we need to resolve this glaring discrepancy between theory and practice.
First, almost all of these trials were conducted in patients with advanced, late stage disease as a primary or salvage treatment to improve overall survival. These patients are often heavily pretreated with extensive disseminated disease. However, we must understand these immune-based treatments are expected to be effective within a well-established, tumor microenvironment that is often immunosuppressive. As mentioned above, we now have considerable evidence that tumor-infiltrating lymphocytes (TILs) demonstrate an “exhausted” phenotype initiated by molecular interactions within the tumor cells. Specifically molecules, such as members of the PD-1/PD-L1 axis, negatively regulate the efficacy of these immune responses (8,9). This critical interaction prevents cytotoxic T-cell responses against cancer cells, essentially cloaking them from the immune system. Thus, even with a systemic, robust immune response, the functional immunocompetant cells are suppressed within the primary tumor.
The second issue complicating cancer vaccine effectiveness is the staggering degree of heterogeneity observed within established tumors and between patients of a given cancer type. In 2007, a major review of active cancer vaccines outlined the various disappointing results in the field (10). One of the first general considerations of this review highlighted the importance of antigen discovery: "select the most informative targets." The authors point out that ideal targets should be tumor-specific and "it is important to use the intended study population to assess the proportion of tumors that express the target of choice and the proportion of cells within each tumor that express it." Thus, it should be a common goal within the field to actively search for a convenient number of shared antigens that most effectively define a patient population of interest. However, this stipulation would require a disease with significant inter- and intra-patient homogeneity.
The first significant evidence of phenotypic heterogeneity in tumors was described by Fidler and Kripke, (11). They demonstrated that various clones of murine melanoma cells could be derived in vitro which varied greatly in their ability to produce lung metastases in syngeneic mice. This suggested that the parent tumor initially displayed a high degree of heterogeneity and clones with various metastatic potentials preexisted in the parental population. However, a competing view proposed by Peter Nowell (12) around this same time, posited that cancer is a disease arising from a surviving mutant clone which progresses into an established tumor with a high degree of homogeneity reflecting its clonal origin. Clearly, the latter hypothesis has been well-represented in the majority of cancer vaccine clinical trials, with the former only recently understood. Fortunately, improvements in DNA sequencing technology have been able to definitively address this debate.
An excellent example of the inter-tumoral heterogeneity inherent to cancer was provided by Wood and Vogelstein (13). To answer this question, Wood et. al, asked "how many genes are mutated in a human tumor?" Applying the latest DNA sequencing technology to a cohort of breast and colorectal tumors, the authors reported roughly 80 mutations that alter critical amino acids were evident in a typical tumor. About 95% of these mutations are single-base substitutions, whereas the remainder are deletions or insertions. By definition, the resulting altered proteins are unique from the perspective of the immune system and all are candidates for potent immunological markers or TAAs. However, when the sequencing results of individual tumors are visualized as mutational landscapes (Figure 5), a troubling view emerges. Despite sharing a similar number of mutations, breast and colorectal cancers demonstrated very different results with respect to the type of mutations and specific genes mutated. Of the ~80 mutations in an individual tumor, only about 3 of these mutations were shared between two different tumors. Additionally, many of the most common mutations are observed within intracellular signaling molecules (p53, PI3K, etc) that may not be effectively presented to the immune system. Consequently, a polyvalent cancer vaccine is technically limited from providing the diversity required to stimulate an appropriately robust and therapeutic immune response across a given patient population. Based on these results, antigen discovery for the development of “off the shelf” cancer vaccines takes on a new level of complexity and is fraught with logistical hurdles.
As our knowledge of inter-tumoral heterogeneity has expanded with improved DNA sequencing technology, we have simultaneously gained a greater appreciation for the troubling degree of intra-tumoral heterogeneity inherent to this disease. Recently, two definitive studies have proven that individual tumors are comprised of many clonal populations. Yachida et al, (14) were able to demonstrate this in pancreatic tumors (Figure 6) while Swanton et al. (15) found similar results in renal cancer samples. Undoubtedly, future studies will demonstrate this level of intra-tumoral heterogeneity is a general feature of cancer. While intertumoral heterogeneity calls into question the cancer vaccine trials of the past, intratumoral heterogeneity challenges the promise of “personalized medicine”. The major focus of cancer research today is profiling patient-specific mutations such that appropriate targeted agents can be used in a rational manner to clear primary disease. Given the degree of intratumoral heterogeneity, how can a randomly chosen biopsy be expected to adequately represent the complexity of the entire tumor? How many biopsies are required? What clones with known resistance lay undetected in the remaining tumor? This leads to the provocative yet critical question, is tumor heterogeneity of any practical value and how does one embrace heterogeneity in cancer treatment? With respect to cancer vaccines, the answer is employing a means of antigen discovery that is highly adaptable and exquisitely sensitive utilizing the entire array of parenchymal tumor cells as source material.
Autologous cancer vaccines, or the process of using a patient’s own tumor as source material for an individualized treatment, is not a new endeavor. However, given what we now know about tumor heterogeneity, we are primed to deploy these tools in the appropriate way. Using powerful, genomic sequencing technology and an updated understanding of tumor-immune system interactions, we now have the ability to design tools capable of addressing the biological realities of cancer. We are at the cusp of a renaissance for ASI, assuming we follow a basic set of guidelines:
- While antigen discovery platforms of the past emphasized the use of common antigens, based on tumor homogeneity, there is now indisputable evidence cancer is comprised of extreme genetic diversity from an inter- and intra-tumoral standpoint. It is now illogical to treat a heterogeneous disease with homogeneous tools.
- As immunologists, we are aware of one highly adaptable, exquisitely sensitive tool provided by evolution to address the magnitude of cancer diversity - the immune system.
- No longer can we use cancer vaccines to inappropriately treat established or advanced disease. We must be focused on preventing recurrence in the adjuvant setting by curing minimal residual disease (MRD). In this way, latent disease which has not yet established a tumor microenvironment, but is certainly capable of doing so later, would be the therapeutic target. This has the opportunity of significantly impacting cancer mortality as the majority of cancer patients (~80%) die due to recurrence.
- In the clinical setting described above, extending recurrence-free survival (RFS) should be the primary endpoint of autologous cancer vaccines. Overall survival will serve as a secondary clinical endpoint. A schematic which emphasizes this last point is provided in (Figure 4).
1. Frankel, JC. Breakthrough of the year 2013. (2013) Science, 342:1432-1434.
2. Bucana C, Hoyer LC, Hobbs B, Breesman M, McDaniel M, Hanna Jr MG (1976) Morphological evidence for the translocation of lysosomal organelles from cytotoxic macrophages into the cytoplasm of tumor target cells. Cancer Research 36 4444-4458.
3. Hanna Jr MG, Bucana C, Hobbs B, Fidler IJ (1976) Morphologic aspects of tumor cell cytotoxicity by effector cells of the macrophage-histiocyte compartment: In vitro and in vivo studies in BCG-mediated tumor regression. In Macrophage in Neoplasia. M Fink, ed. Academic Press, 113-133.
4. Cohn ZA (1968) The structure and function of monocytes and macrophages. Advan Immunol 9 163-214.
5. Cohn ZA (1975) Macrophage physiology. Federation Proc 34 1725-1729.
6. Morales A, Eidenger, D, Bruce, A (1976) Intracavitary bacillus Calmette-Guerin in the treatment of Superficial bladder Cancer. J Urol. 116:180-183.
7. Hanna Jr MG, DeJager, R, Guinan, P, Crispen, R, Lamm, A et al. (1992) Bacillus Calmette-Guerin (BCG) Vaccine for Tuberculosis: Antitumor Effect in Experimental Animals and Humans. Vaccine Research. 1:69-91.
8. Finke LH, Wentworth K, Blumenstein B, Rudolph NS, Levitsky H, Hoos A (2007) Lessons from randomized phase III studies with active cancer immunotherapies – outcomes from the 2006 meeting of the Cancer Vaccine Consortium (CVC). Vaccine 25 (Suppl 2): B97-109; PMID: 17916465.
9. Keir ME, Francisco LM, Sharpe AH (2007) PD-1 and its ligands in T-cell immunity. Curr Opin Immunol 19: 309-14.
10. Brahmer JR, Tyodi SS, Chwo LO et al (2012) Safety and activity of anti-PL1 antibody in patients with advanced cancer. N England J Med 366: 2443-54.
11. Fidler, IJ and Kripke ML (1977) Metastasis results from pre-existing variant cells within a malignant tumor. Science197: 4306 893-895.
12. Nowell PC (1976) The clonal evolution of tumor cell populations. Science. 194 (4260):2-8.
13. Woods DL, Parsons DW, Jones S, Lin, J, Sjoblom T, et al (2007) The genomic landscapes of human breast and colorectal cancers. Science 318: 1108-1113.
14. Yachida S, Jones S, Bozic I, Antal T, Leary R, et al (2010) Distant metastasis occurs late during the genetic evolution of pancreatic cancer. Nature 467: 1114-1117.
15. Swanton C (2012) Intratumor heterogeneity: Evolution through space and time. Cancer Res. 19:4875-4882.
Top left. Time lapse cinematography of transplantable, L 10 hepatocarcinoma of syngeneic Strain 2 guinea pigs with 3 peritoneal monocytes from L 10 immune animals. Note the extensions from the monocytes (clasmatosis) Top right. Implosion of the tumor cell. Bottom left and right. Cytoplasm from tumor cell being phagocytized by macrophages.
Top left. A lysosome trapped between a monocyte and tumor cell. Top right. Lysosome like organelle vacuoles in monocyte and organelle in tumor cell. Bottom left. An extension from the monocyte to and possible into the tumor cell. Intracytoplasmic organelles in the probing extension. Bottom right. Schematic of the several ways the lysomomal organelles can be transferred from activated or immune monocytes to the tumor cells.
In the two rectangles on the right, cancer vaccine candidates on the left and declared failed candidates, blue letters, on the right. Both of these used to treat patients with advanced disease. OncoVAX® an autologous tumor cell vaccine used to treat occult disease on the left.
Best Hope for Significant Progress with Solid Tumors is via Treatment of Minimal Residual Disease.
Genomic Landscape of Colorectal Cancer. Wood, L and B. Vogelstein, Science Vol. 318, November, 2007. A two-dimensional map of genes mutated in colorectal cancers, in which a few genes "mountains" are mutated in a large proportion of tumors while most are mutated infrequently. The mutations in two individual tumors are indicated in the lower map. Note that only 3 mutations (blue dots on bottom landscape) were common to both tumors indicating a potential for weak common immunogenicity
Heterogeneity within primary tumors. Metastatic subclones giving rise to liver and lung metastasis are non-randomly located within slice 3, indicated by blue circles. The clones are both geographically and genetically distinct from clones giving rise to peritoneal metastasis in this same patient, indicated in green. Yachida, et al., Nature 467: 114, 2010.