Leaving Tumors No Way Out

Time after time, Steven A. Rosenberg, chief of the surgery branch in the National Cancer Institute's (NCI) Clinical Oncology Program, has seen cancer immunotherapy destroy melanomas that conventional therapies leave unmolested. When immunotherapy works, even bulky metastatic tumors are destroyed. The frustration and tragedy, he says, speaking recently to a Grand Rounds audience at the National Institutes of Health, "is that it happens in only a small percentage of patients." For nearly 20 years Rosenberg's research at the NCI has probed how the immune system responds to cancer and set the pace for developing this new form of cancer therapy. He and his colleagues pioneered the isolation of tumor antigens that activate cytotoxic T cells against cancers, especially melanomas, and are working out new therapies based on those antigens. Rosenberg's reward is seeing therapy response rates slowly climb to about one-third of his patients. Steven A. Rosenberg Now researchers stand at what may be the final barrier holding them from more widespread success. The central question of his field now is how cancers escape therapy; that is, "what are the mechanisms limiting cancer regression, despite in vivo generation of large numbers of T cells against tumor antigens?" With answers, better therapies are sure to follow. To uncover the escape routes that cancer cells use to flee immune destruction, and then tailor therapies that grant no stay of execution, Rosenberg relies on what has already carried him such a far journey: a detailed molecular analysis of his patients' responses to therapy. "Studying one patient can uncover an immunologic principle involved in a clinical phenomenon," he comments. "Virtually every patient, provides us with the opportunity to answer important questions, if we only work hard enough to determine the right questions to ask." On being a Patient and Tutor One of those patients was a 26-year-old patient (named MG) with metastatic melanoma who was referred to Rosenberg in 1989. MG's clinical history illustrates how the back-and-forth effort between the lab and clinic yielded deeper principles and subsequent improved therapy. At the time Rosenberg was having a small amount of success treating melanoma patients with high doses of IL-2, the cytotoxic (CD8+) T cell regulator produced by helper (CD4+) T cells. He was trying to pin down the cells responsible for tumor regression. Candidates for the honor were cells discovered within melanoma nodules called tumor-infiltrating lymphocytes (TIL). MG became one of the stars of a clinical trial testing a combined treatment of TIL and IL-2. TIL capable of recognizing her tumor in vitro was cultured from one of her soft tissue masses. When she was infused with autologous TIL and IL-2, MG experienced a complete regression. The trial, which studied 73 people, was a milestone. On IL-2 alone, only about 15 percent of patients had complete or partial responses. Adding TIL doubled that response rate. When researchers later began cloning the antigens in tumors that TIL recognized, tumor antigens of spectacular responders like MG were of special interest. Rosenberg colleague Paul Robbins found that MG's antigen was tyrosinase, a normal, nonmutated enzyme involved in a melanin synthesis. Several years later MG suffered a recurrence and needed retreatment. After receiving a second TIL infusion from a recently removed pelvic mass, MG once again joined the lucky few to enjoy a complete remission. Then lab analysis came in with a surprise: Her tumor antigen had changed. Now it was a mutant version of beta-catenin, a cell adhesion protein. The second TIL recognized a novel epitope on beta-catenin created by a point mutation replacing serine with phenylalanine; the slim tether that kept MG among the living had been the change of a single base. Turning Principles into Treatments Applying cloning techniques to tumor DNA, Rosenberg's lab has identified a host of class I cell surface tumor antigens recognized by CD8+ cytotoxic T cells. The antigens come from a variety of venues, including normal melanocyte differentiation proteins, intronic sequences, alternative open reading frames, single base mutations, and chromosomal rearrangements. They prefer not to work alone, Rosenberg observes, "Each time we study a patient in detail, we find that they recognize more than one antigen." MG's TIL recognized eight epitopes on seven cloned tumor antigens. And some tumor antigens are shared; one, NY-ESO-1, occurs frequently in melanomas, breast, and lung cancers. With isolated tumor antigens Rosenberg could theorize that patients could be immunized against cancer by synthetic peptides derived from those antigens. In 1998 his group reported that peptides representing CD8+ T-cell epitopes immunized patients against melanoma.1 According to the paper, an element of design exists in this work because most peptides are not immunogenic partly because of poor binding to HLA molecules that present peptides as antigens to T cells. So the peptides were modified at what are called anchor positions, residues that can be changed to increase the strength of binding. Immunized T cells recognize not only the modified peptide, but also the original tumor antigen.2 A pilot trial testing synthetic peptides plus IL-2 found that nine out of 10 patients were successfully immunized, with more than 40 percent showing objective responses. Pleased with how well patients respond to peptide vaccination, Rosenberg calls their levels of immunity "surprisingly high;" often up to 1 percent of a patient's CD8+ T cells recognize the tumor antigen. He is especially impressed that "in three patients we saw the reduction of brain metastases, something we never saw with IL-2 alone." Rosenberg cautions, however, that a randomized trial to confirm the power of peptide vaccination is still ongoing. That power is being enhanced by studies of class II tumor antigens, made possible by a cloning technique invented by Rong-fu Wang at Rosenberg's lab.3 These antigens are recognized by CD4+ T cells. The list of class II antigens found so far shows they have much in common with their class I counterparts, including a variety of mutational events from which they form, and that patients can have more than one (two were found in MG's tumors). Tumor peptide antigens also provide the opportunity to study T-cell clones raised from immunized patients. Colleague Mark Dudley screens clones isolated from TIL or peripheral blood to identify those with the highest peptide avidity, grow them into large numbers, activate them by peptide exposure, and then infuse them into patients. "This allows us to give patients what we think are the best cells, and to understand the cellular mechanism of tumor regression when we give patients cells with a single specificity," explains Rosenberg. CD8+ clone properties are widely diverse:4 avidity for antigen can span three orders of magnitude, clones differ by lytic ability and are unalike in the amounts and kinds of cytokines they secrete. Pitting their various attributes against melanoma will gradually reveal which are most deadly. Escape-proof Therapies The combination of tumor peptide antigens and T-cell clones enable Rosenberg to design therapies that cancer cells will find harder to escape. This of course depends on knowing more about how they escape, as MG's most recent troubles make plain. Last September MG developed a subcutaneous mass on her flank that was removed by surgery; presently, she has no evidence of disease. That is fortunate, as study of the flank mass revealed that remission through a third round of TIL would have been unlikely--A cell line grown from the melanoma proved unable to express beta-2 microglobulin, a cell surface protein indispensable for presenting class I antigens. According to Rosenberg, this approach is a rare means of moving beyond the immune system's reach; his studies estimate fewer than five percent of melanomas lose all ability to express class I antigens. A more likely means of escape would be down regulating expression of a tumor antigen targeted by an immunizing peptide.5 To counter this possibility, a multi-peptide therapy is proposed. Currently Rosenberg's group is organizing a clinical trial to test simultaneous immunizations with four different tumor antigen peptides. And in a pilot trial, the focus will be on what happens to melanoma patients simultaneously immunized with both classes of tumor antigen peptide. "It's probably not enough to just immunize for CD8+ cells," he says. "One has to immunize against class II determinants. And now in the last year-and-a-half we have many of these for the first time." Another pilot trial recently tested a method to close off three possible escape routes: too few antitumor T cells, avidity too low, and lack of regulatory factors for optimal activity in vivo. The protocol treated metastatic melanoma patients with an infusion of high-avidity cloned T cells following chemotherapy to ablate their lymphocytes. Depletion of the native lymphocytes prior to infusion may increase survival of transferred T cells. The patients also received injections of high-dose IL-2. After one important adjustment, the protocol gave dramatic remission for a 58-year-old woman with advanced melanoma. In addition to her CD8+ T cell clone, she was infused with her heterogeneous TIL from whence the clone came. The heterogeneous TIL "must have contained something that was vital for mediating her regression," Rosenberg says, possibly CD4+ cells, which is mostly what her TIL contained. A teenage boy who received a supplement of bulk TIL that also was high in CD4+ cells had 80 percent regression and was able to return to college. Another therapeutic idea in development is to give CD8+ cells within TIL the ability to make their own IL-2. A new technique developed in Rosenberg's lab puts the IL-2 gene into CD8+ TIL by viral transduction. The transduced cells grow in the absence of externally supplied IL-2 for well over a month. There are many other possible ways for tumors to escape immune destruction. To learn more about their survival tactics, researchers are using gene array techniques to compare gene expression patterns between tumors that respond to immunotherapy and tumors that do not. The results may turn up clues about which genes may be inhibiting the immune response. By using all of these means, Rosenberg hopes to see big strides forward, helping his nascent branch of medicine complement, and maybe one day outshine the other cancer therapy he understands so thoroughly, surgery. He knows well what surgery can do for cancer, and what it cannot. Which may be a spur to his research. Above his desk hangs this sign: "Surgery is like an armed savage which attempts to get by force that which a civilized man would get by stratagem." Tom Hollon (thollon@starpower.net) is a freelance writer in Rockville, Md. References 1. S.A. Rosenberg et al., "Immunologic and therapeutic evaluation of a synthetic peptide vaccine for the treatment of patients with metastatic melanoma," Nature Medicine 4:321-7, 1998. 2. M.R. Parkhurst et al., "Improved induction of melanoma-reactive CTL with peptides from the melanoma antigen gp100 modified at HLA-A*0201-binding residues," Journal of Immunology, 157:2539-48, 1996. 3. R.F. Wang et al., "Cloning genes encoding MHC class II-restricted antigens: mutated CDC27 as a tumor antigen," Science, 284:1351-4, 1999. 4. M.E. Dudley et al., "T-cell clones from melanoma patients immunized against an anchor-modified gp100 peptide display discordant effector phenotypes," Cancer Journal, 6:69-77, 2000. 5. M.C. Panelli et al., "A tumor-infiltrating lymphocyte from a melanoma metastasis with decreased expression of melanoma differentiation antigens recognizes MAGE-12," Journal of Immunology, 164:4382-92, 2000.

Leaving Tumors No Way Out

Interested in reading more?

Become a Member of