Pathophysiology of the Graft-Versus-Tumor Effect
The GVT effect occurs due to a predominantly T-cell-mediated immunologic attack on tumor cells. In HLA-identical transplants, GVT effect is mediated by naïve T cells; for development of effector function, these must first be primed by host APCs. This requires the following: presentation of MiHAs or tumor-specific antigens on HLA; appropriate costimulatory molecules, including CD28, OX40, CD40L, and 41BB; and an appropriate “third signal,” provided by IL-12, interferon gamma (IFN-γ) or adjuvant (4). Restraining influences limiting the degree of immune activation are present to protect the host from an excessive immune response and include expression of CTLA4 (which competes with CD28 for binding to CD80/86) and programmed cell death protein 1 (PD1) and its interactions with its lig and PDL1, which limit T-cell activation and expansion during normal, pathogen-directed immune responses.
Soon after transplant, there is activation and expansion of MiHA-reactive T cells, followed by a decline, similar to that seen in pathogen-directed immune reactions. This may in part relate to the development of peripheral tolerance/anergy and also to replacement of host hematopoiesis, with resulting loss of host APCs. For tumor-associated antigen presentation to continue, there must be cross presentation on donor APCs. In addition, the initial alloresponse to MiHAs results in recruitment of T cells targeting either tumor-associated antigens or nonpolymorphic genes, which are either overexpressed or aberrantly expressed by the tumor (4).
Some tumor-associated antigens of importance include pathogenesis related protein (PR1), an epitope shared by proteinase-3 and elastase proteins, which is expressed in normal neutrophils and overexpressed in myeloid leukemias; and PR1-specific CD8+ T-cell responses, which can be detected in a range of myeloid and nonmyeloid malignancies post-HCT and correlate with outcome in chronic myelogenous leukemia (CML) (5). CD8+ T-cell responses to WT1, which is frequently overexpressed in acute myelogenous leukemia (AML), can be induced with vaccination strategies in both the autologous and allogeneic settings (6,7,8). Clinical responses from tumor vaccines have thus far been suboptimal.
Natural killer (NK) cells can also mediate antitumor effects; this is discussed in further detail in this chapter.
Tumor Escape From Immunologic Destruction
Tumors utilize numerous mechanisms to escape immunological destruction, including the following:
Induction of regulatory T cells
Production of inhibitory cytokines
Downregulation of costimulatory molecules (9) and HLA I (10)
Induction of coinhibitory molecules
Induction of myeloid-derived suppressor cells in the microenvironment that inhibit immune responses through multiple mechanisms.
Invasion of immunologically privileged sites (11).
Cellular therapeutic approaches are designed with the intent to abrogate these escape mechanisms.
Cellular Therapy to Induce Graft-Versus-Tumor Effect
Donor Lymphocyte Infusions
Donor lymphocyte infusion (DLI) may induce remissions in patients with molecular or overt relapse of their malignancy and can reverse CD8+ T-cell exhaustion. (12). However, the likelihood of success varies significantly according to the underlying disease. Chronic myelogenous leukemia is most sensitive; follicular lymphoma (FL) and Hodgkin lymphoma (HL) are also highly responsive (13,14). Responses to DLI in AML or myelodysplastic syndrome (MDS) are less frequent, and durability is often poor. Acute lymphoblastic leukemia (ALL) is the least responsive to DLI. Reported response rates are 60% to 73% in CML, 15% to 29% in AML, and 0% to 18% in ALL (15,16).
Problems and Challenges Associated With Donor Lymphocyte Infusion
Development of Graft-Versus-Host Disease
Graft-versus-host disease occurs in 40% to 60% of patients with HCT (15,16) and is more likely to occur in unrelated donor recipients (13). Lympho-depleting chemotherapy given prior to DLI enhances alloreactive T-cell proliferation, potentiating the GVT effect but increasing GVHD (17). Approaches to reduce the GVHD incidence include the following:
Reducing T-cell dose. A dose-response relationship exists for both GVT and GVHD effects. In CML, no increased response is seen with cell doses greater than 4.5 × 108 CD3+ cells/kg. In AML, response rates plateau beyond 1.5 × 108/kg; higher doses increase GVHD. Gradual dose escalation schedules have been successfully utilized in relapses of indolent diseases. Follicular lymphoma appears highly sensitive to DLI after T-cell-depleted HCT, responding to low-dose DLI at doses from 1 to 10 × 106/kg with less than 20% incidence of clinically significant acute GVHD (aGVHD) and chronic GVHD (cGVHD) (14). A similar strategy has shown success in HL, mainly in the setting of low-volume disease detected on surveillance positron emission tomographic (PET) imaging (18). More sensitive surveillance techniques, such as deep sequencing, to detect low-volume disease may allow earlier institution of DLI and maximize efficacy; utilization of these techniques remains experimental.
Transduction of donor lymphocytes with a suicide gene (19).
Selection of T cells to target tumor-associated antigens/antigens with restricted or differential expression (analogous to the use of viral-specific T cells [VSTs]). Infusion of MiHA-specific T cells was effective at eradicating tumor in mouse models (20). In addition, tumor-infiltrating lymphocytes (TILs) recognizing tumor-associated antigens have been successfully used for selected metastatic solid tumors (21). However, the technology to isolate TILs remains restricted to a few specialized centers and is yet to be applied in large-scale clinical studies.
If there is insufficient residual donor hematopoiesis prior to DLI, eradication of host hematopoiesis by the infused lymphocytes can result in marrow aplasia; chimerism studies should therefore be performed prior to DLI to ensure adequate donor hematopoiesis (22).
Responses to DLI may not be seen for up to 2 months (11). In indolent diseases (eg, FL), this may not be problematic, but in overt relapse of aggressive diseases (eg, AML), chemotherapy may be required first to achieve disease control (23).
Prophylactic Donor Lymphocyte Infusion in High-Risk Patients
The delayed onset of action of DLI has led to the use of preemptive DLI to prevent relapse in high-risk patients. Mixed donor/recipient chimerism within the T-cell lineage is frequently seen in T-cell-depleted transplants and is associated with higher rates of relapse in CML (24), FL (14), and HL (18), likely due to development of bidirectional tolerance with resulting tumor escape from immunologic control. Donor lymphocyte infusion can induce full donor chimerism in both FL and HL, and subsequent relapse rates are low; no formal comparison to similar groups not receiving DLI has been performed.
Bi-Specific T-Cell-Engaging Antibodies
The topic of bi-specific T-cell-engaging antibodies has been discussed in detail elsewhere. Bi-specific T-cell-engaging antibodies are single-chain antibodies that engage T cells via CD3 and direct them to an antigenic target present on tumor cells, typically CD19, resulting in T-cell redistribution, activation, expansion, and perforin-mediated killing of target cells (25). Blinatumomab is highly efficacious in ALL in the setting of persistent MRD (25), overt relapse, or refractory disease (26).
Chimeric Antigen Receptor-Modified T Cells
The ideal cellular therapy for a malignant disease should expand and persist in vivo and selectively target cancer cells. This can be achieved by modifying autologous T cells with a chimeric antigen receptor (CAR). The CAR consists of a single-chain monoclonal antibody (scFv) targeted to a tumor-associated antigen, which is thus recognized in a major histocompatibility complex (MHC)–independent fashion (unlike unmodified T-cell-mediated GVT effect); the scFv is coupled via an extracellular hinge domain and transmembrane domain to an intracellular signaling domain, typically the CD3ζ chain (Fig. 16-2) (27). Autologous T cells are collected from peripheral blood (PB) via a steady-state blood draw or apheresis procedure and transduced with the CAR construct via a lentiviral or retroviral vector or using electroporation and a transposon/transposase system (28). Cells are cultured and expanded ex vivo using either CD3/28 beads (29) or artificial APCs (30) and specific cytokines prior to infusion. The persistence and clinical activity of CAR T cells (CAR-T) in vivo can be enhanced by the addition of a costimulatory molecule to the CAR construct (31), usually CD28 (31,32). An initial report in a patient with highly refractory CLL treated with an anti-CD19 CAR-T utilizing CD137 (4-1BB) as the costimulatory domain generated great excitement (33): Infused T cells expanded more than 3 log, the patient developed a cytokine-release syndrome (CRS) and tumor lysis syndrome (TLS), and achieved complete remission (CR); long-term persistence of CAR-T and persistent normal B-cell aplasia (a predictable, on-target effect when targeting CD19) were demonstrated.
Schematic of basic chimeric antigen receptor (CAR) construct. The CAR consists of a single-chain monoclonal antibody (scFv) targeted to a tumor-associated antigen, and the scFv is then coupled via an extracellular hinge domain and transmembrane domain to an intracytoplasmic signaling domain, typically the CD3ζ chain.
Subsequent results in CLL have been heterogeneous; updated results from the University of Pennsylvania showed that 5 of 24 patients treated achieved durable CRs, 7 had partial responses (PRs), and 12 had no response (34). The variables underlying response to treatment are not well understood, but in vivo CAR-T expansion is a prerequisite and appears to be more important than the dose of infused cells (30). In ALL, results have been particularly impressive, with a CR rate of 86% in children treated for relapsed/refractory disease; patients were MRD negative even when tested with highly sensitive deep-sequencing techniques (34). Long-term survival rates are not yet known, particularly as in many cases the treatment has been used as a “bridge to transplant.” At least two patients with ALL have relapsed with CD19-negative disease (34). Similarly impressive CR rates have been reported by the groups at the National Institutes of Health (35) and Memorial Sloane Kettering Cancer Center (36). All three groups reported a similar toxicity profile (see Table 16-1). It is unclear whether this treatment can replace HCT in a proportion of patients; long-term survival outcomes in patients ineligible for HCT will be important in answering this question. The CAR-T targeting CD19 have now been used in a range of B-cell malignancies, with responses seen in both aggressive and indolent lymphomas, CLL, and ALL (Table 16-1).
Table 16-1Summary of Reported Studies Using Chimeric Antigen Receptor (CAR) T Cells Directed Against CD19 in Hematologic Malignancies ||Download (.pdf) Table 16-1 Summary of Reported Studies Using Chimeric Antigen Receptor (CAR) T Cells Directed Against CD19 in Hematologic Malignancies
|Reference ||Cancer ||CAR Endodomains ||Number of Patients ||Clinical Outcome ||Toxicities |
|Kochenderfer et al, 2014 (91) ||CLL, DLBCL, NHL, PMBCL, SMZL ||CD28 & CD3ζ ||15 ||8 CR, 4 PR, 1 SD, 2 NE ||Fever, hypotension, renal failure, confusion, aphasia |
|Maude et al, 2014 (35) ||ALL ||CD28 & CD3ζ ||30 ||90% CR (15 prior HCT); 67% EFS, 78% OS at 6 mo. ||SIRS 27%, B-cell aplasia 73% |
|Lee et al, 2014 (36) ||ALL ||CD28 & CD3ζ ||21 ||70% CR ||33% severe SIRS |
|Davila et al, 2014 (35) ||ALL ||CD28 & CD3ζ ||16 ||88% CR ||43% severe SIRS |
|Kochenderfer et al, 2013 (92) ||CLL, Lymphoma ||CD28 & CD3ζ ||10 ||1 CR, 1 PR, 2 PD, 6 SD ||Fever, SIRS, TLS |
|Brentjens et al, 2013 (93) ||ALL ||CD28 & CD3ζ ||5 ||5 CR ||SIRS |
|Grupp et al, 2013 (94) ||ALL ||4-1BB & CD3ζ ||2 ||2 CR ||SIRS, central nervous system toxicity |
|Kochenderfer et al, 2012 (38) ||CLL, Lymphoma ||CD28 & CD3ζ ||8 ||1 CR, 5 PR, 1 SD, 1 died (influenza) ||Mild SIRS |
|Brentjens et al, 2011 (32) ||CLL, ALL ||CD28 & CD3ζ ||8 ||1 PR, 2 SD, 3 NR, 1 PD, 1 died (sepsis-like disease) ||Fever, death |
|Savoldo et al, 2011 (31) ||NHL ||CD28 & CD3ζ versus CD3ζ ||6 ||2 SD, 4 NR ||None |
|Porter et al, 2011 (33) ||CLL ||4-1BB & CD3ζ ||1 ||CR ||TLS, SIRS |
|Kalos et al, 2011 (95) ||CLL ||4-1BB & CD3ζ ||3 ||2 CR, 1 PR ||Fever, rigors, dyspnea, cardiac dysfunction, febrile syndrome, hypotension |
|Kochenderfer et al, 2010 (96) ||Lymphoma ||CD28 & CD3ζ ||1 ||PR ||None |
|Jensen et al, 2010 (27) ||Lymphoma ||CD3ζ ||2 ||2 NR ||None |
While this therapy shows great promise, many aspects require optimization. Due to the heterogeneity in technique for CAR-T production, it is difficult to compare across trials to determine optimal manufacturing methods. The controversies are numerous: First, while it has been shown that addition of a costimulatory domain to the CAR enhances expansion and persistence (31), it is unclear whether CD28 or 4-1BB, or both, is superior. Second, lympho-depleting chemotherapy may enhance CAR-T expansion and persistence (32), but the optimal drugs and schedule are unknown. Third, the epitope targeted by the antibody fragment is likely important, in terms of both the spatial location of epitope binding and binding affinity. Fourth, the hinge and transmembrane domains are important in determining interaction with antigen and formation of the immunologic synapse, but little is known about optimizing this aspect of the CAR design. Fifth, the method of T-cell transduction (viral vs transposon) may be important in determining efficacy. Sixth, the ex vivo culture technique and duration (eg, CD3/28 beads vs artificial APCs and which supplemental cytokines to provide) may be important; for example, culture after transduction using the Sleeping Beauty system is relatively prolonged, which is problematic in kinetically active diseases such as ALL (37). The dose and composition (unselected vs specific ratios of CD4/8 cells) of T-cell product infused are also variables requiring consideration. Finally, the bulk of tumor present at the time of CAR-T infusion may potentially affect the in vivo proliferation of the infused CAR-T, and the ideal time of infusion (eg, in an MRD state vs overt relapse) has not been elucidated.
Adverse Events and Optimizing Safety
All responding patients have had some degree of CRS and have developed B-cell aplasia (38). Cytokine-release syndrome is characterized by fever with variable systemic symptoms, including hypotension, and high levels of inflammatory cytokines, of which IL-6 appears particularly important (37). Macrophage activation syndrome (MAS) may accompany CRS. Major neurological symptoms, including seizures, have occurred. The mechanism of neurological events is unclear; it may be cytokine mediated, associated with MAS, or due to direct CAR-T infiltration. Unexpectedly, CD19 CAR-T have been found in the cerebrospinal fluid of some patients without central nervous system disease (37). Cytokine-release syndrome can be managed with the anti-IL-6 antibody tocilizumab, with prompt responses in the majority of patients (37). Corticosteroids, while potentially efficacious in managing CRS/MAS, are toxic to the infused cells and may limit efficacy.
Unanticipated on-target toxicity may occur; for example, a toxic death occurred in a patient treated with a CART-T directed against Human Epidermal Growth Factor Receptor 2, or ErbB2, due to unanticipated low-level pulmonary epithelial expression (39). The inclusion of a suicide gene within the CAR, such as inducible caspase 9 (iCaspase9), which could be triggered in the event of severe toxicity, would provide an added safety measure.
Most human trials to date have focused on CD19 as a target in B-cell diseases, but a number of novel targets show potential in different diseases.
Adoptive Transfer of Natural Killer Cells to Enhance Antitumor Effect
In contrast to T and B lymphocytes, NK cells do not express rearranged, antigen-specific receptors; rather, NK effector function is dictated by the integration of signals received through germ-line-encoded receptors that can recognize ligands on their cellular targets. Functionally, NK cell receptors are classified as activating or inhibitory. Natural killer cell function, including cytotoxicity and cytokine release, is governed by a balance between inhibitory receptors, notably the killer immunoglobulin-like receptors (KIRs) and the heterodimeric C-type lectin receptor (NKG2A), and activating receptors, in particular the natural cytotoxicity receptors (NCRs) NKp46, NKp30, NKp44, and the membrane protein NKG2D (40). Inhibitory receptors bind to HLA class I molecules, expressed on the surface of normal cells, resulting in signals that block NK cell triggering and inhibit killing. In the setting of malignancy or viral infection, HLA class I is often downregulated or has altered peptide expression, resulting in failure of KIR-mediated recognition by the NK cell and resultant cell killing (40). Activating receptors, such as NKG2D, bind ligands that are induced by cellular stress (eg, viral infection and malignant transformation); binding results in NK cell activation and target lysis (41) (Fig. 16-3).
Selective killing of transformed cells by NK cells: In normal cells, the inhibitory signals triggered by KIR-HLA-I molecule engagement overrides activating signals. In the context of cancer, expression of stress ligands for activating receptors, in conjunction with low expression of HLA-I molecules, attenuates the triggering of inhibitory receptors and results in an activating signal.
Early NK cell recovery (within 30 days) postallogeneic stem cell transplant has been associated with reduced rates of both relapse and aGVHD, with resultant improved survival (42). This dual benefit makes allogeneic NK cells an attractive option for adoptive cellular therapy peritransplant. Adoptive transfer of NK cells has previously been limited by the small numbers of circulating NK cells (5%-15% of the total lymphocytes) and consequently the low numbers obtained in an apheresis procedure (43).
Use of allogeneic NK cells may be more efficacious than autologous NK cells due to inhibition of autologous NK cell activity by recognition of host HLA. Adoptive transfer of ex vivo–expanded haploidentical NK cells after lympho-depleting chemotherapy is safe. High-dose, but not low-dose, chemotherapy facilitates in vivo NK cell expansion, likely due to both prevention of host T-cell-mediated rejection and reduction in competition for cytokines, particularly IL-15. Persistence for at least 4 weeks has been achieved in some patients and responses have been observed in high-risk AML without inducing GVHD (44). The NK cell expansion ex vivo has traditionally included culture with cytokines (IL-2 or IL-15) and cell selection (CD3 depletion) (44); the use of “feeder cells” (Epstein-Barr virus [EBV]–transformed lymphoblastic cell lines or gene-modified, irradiated K562 cells), and large-scale expansion flasks have dramatically increased NK cell yield and activation status. Clinically relevant NK cell numbers can be obtained from both cord blood (CB) and adult donors (43). The CB-derived NK cells show a similar phenotype and are similarly active against leukemic targets as PB-derived NK cells (45).
There are several potential limitations of NK cell adoptive transfer, particularly limited persistence and the potential for passenger lymphocyte-related complications. The NK cells may rapidly develop exhaustion in vivo after adoptive transfer, despite initial expansion and activity (46). In part, this may relate to NK cells’ exquisite sensitivity to cytokines such as IL-2 and IL-15. In vivo use of IL-2 (which can expand NK cell numbers) can lead to severe toxicity and to T-regulatory (T-reg) expansion, which limits NK cell activation (47). In contrast, IL-15/IL-15Rα complexes promoted NK cell activation and enhanced function without the detrimental effects of IL-2 (48). Whether in vivo use of IL-15/IL-15Rα will rescue NK cells from this phenomenon is not known. Lymphocyte contamination of the infused product can be avoided by proper selection techniques. T-cell contamination should be limited to less than 1 to 5 × 105/kg (49) to minimize the risk of GVHD; this can be achieved by CD3 depletion (44). Addition of CD56+ selection reduces B-cell contamination to less than 1%, minimizing passenger B-lymphocyte-mediated EBV-posttransplant lymphoproliferative disorder (PTLD) and acute hemolytic anemia (50).
Optimizing Natural Killer Cell Efficacy
The NK cells have a range of highly polymorphic KIRs, which are divided into inhibitory and activating subtypes. The KIRs are inherited as haplotypes (KIR-A and KIR-B). The KIR-A haplotypes, found in one-third of adult Caucasians, have one activating receptor, while the KIR-B haplotypes have 2 or more. Transplantation in AML from a KIR-B haplotype donor is associated with lower relapse rates and superior survival (51). Donor KIR2DS1 (an activating KIR) and recipient HLA-C type influence relapse risk. The KIR2DS1-associated reduction in the rate of AML relapse is restricted to donors with HLA-C1/C1 or C1/C2, in whom KIR2DS1-expressing NK cells are presumed to be “educated,” and the benefit was eliminated in transplants from donors with HLA-C2/C2, where KIR2DS1-expressing NK cells are expected to be tolerized in the setting of self HLA-C2 (52). Selection of adult or CB donors for ex vivo NK cell expansion based on KIR genotype may therefore enhance NK cell efficacy.
Other future strategies to enhance NK cell tumor killing include the use of immunomodulatory drugs such as lenalidomide and the use of bi-specific killer engagers (BiKEs), which consist of a single-chain Fv against CD16 and a tumor-associated antigen. A CD16x33 BiKE has been shown to have activity in refractory AML (53).
Interaction of PD1/PDL1 induces T-cell dysfunction in CLL (54). Anti-PD1 antibodies are efficacious in a subset of patients with metastatic solid tumors as monotherapy and in combination with the anti-CTLA4 antibody ipilimumab (55). They have also shown remarkable activity in relapsed/refractory HL (56).
The potential importance of immune checkpoints has been demonstrated in AML. CD8+ T-cell responses directed against the MiHA Liver receptor homolog-1 (LRH-1) have resulted in remission post-DLI. Despite persistence of CD8+ T-memory cells specific for LRH-1, subsequent relapse with LRH-1-positive blasts occurs, unaccompanied by LRH-1-specific T-cell expansion, suggesting anergy/functional impairment. LRH-1-specific T cells from patients with relapsed AML have elevated levels of PD-1. The addition of anti-PD1 antibody to a coculture system resulted in marked LRH-1-specific T-cell expansion, IFN-γ production and cytotoxicity, suggesting a specific inhibitory effect induced by the PD-L1/PD1 interaction (57). Blockade of the PD-L1/PD1 axis may therefore represent an immunomodulatory target for patients with persistent/relapsed AML post-alloHCT.