Introduction

Hematopoietic stem cell transplantation (HSCT) is the only curative option for a number of both malignant (e.g., acute and chronic leukemia, lymphoma, myelodysplastic syndrome) and non-malignant disorders (e.g., bone marrow failure syndromes, inborn errors of metabolism, immune deficiencies, and hemoglobinopathies) [1]. In all of these settings, both in children and adult patients, immune reconstitution is critical for a favorable outcome [2•], being associated with post-transplant infections [3], graft-versus-host disease (GVHD) [4] (both affecting transplant-related mortality (TRM) [5]), and, in the malignant setting, relapse incidence [6]. Correlations with clinical outcomes have been described for both specific (e.g., CD4+ T cells, regulatory T (Treg) cells) [5, 7] and non-specific (such as total lymphocyte count) [4] cell populations. Moreover, as expected, the interplay between immune reconstitution and transplantation outcomes has been demonstrated in all types of HSCT, from HLA-identical to partially matched HSCT [2•].

For all these reasons, strategies aimed at enhancing/accelerating immune reconstitution in order to (i) reinforce immune competence against pathogens and tumor cells and/or (ii) modulate donor T cell alloreactivity appear to be particularly desirable [8]. In this regard, while T cell immunotherapy currently represents the most attractive approach, the separation of GVHD from graft-versus-tumor/graft-versus-infection effect is particularly challenging. This review will provide an overview of the most widely used strategies aimed at enhancing post-transplant immune reconstitution by means of T lymphocyte administration, focusing on both the simplest and most widely used method, based on the infusion of unmodified donor T cells, and on more sophisticated approaches, based on selection/generation/expansion of specific T cell subsets.

Donor Lymphocyte Infusion

Donor lymphocyte infusion (DLI) represents the easiest (and thus most used) strategy to enhance immune reconstitution after HSCT. However, it is well known that the infusion of unmanipulated donor-derived lymphocytes carries the risk of provoking GVHD (or to worsening already present GVHD), thus increasing morbidity and mortality and compromising the beneficial effect of adoptive donor T cell transfer to the recipient. Several factors determine whether the infusion of donor T lymphocytes will cause GVHD [9], among others, (1) the intensity of the conditioning regimen, (2) the use of in vivo T cell depletion (TCD) with anti-thymocyte globulin or other approaches, (3) the timing of DLI after allogeneic HSCT, (4) the dose of T cells infused, (5) the degree of HLA disparity in the donor-recipient, pair, (6) the type of disease for which HSCT is performed, (7) the ongoing administration of immune suppressive drugs, and (8) the level of donor chimerism at the time of DLI (Fig. 1).

Fig. 1
figure 1

Relation between T cell dose, grade of donor-recipient HLA mismatch, graft manipulation, and risk of GVHD onset after DLIs

Full size image

In particular, since the more intensive the conditioning regimen, the more severe is the cytokine storm (and thus the higher the risk of GVHD), delaying the administration of donor lymphocytes after resolution of tissue damage (and the cytokine storm) may not only delay but also prevent/reduce acute GVHD after T-cell-depleted transplant [9].

Treatment of either overt or incipient disease relapse (or a condition of donor-recipient mixed chimerism) after HSCT represents the main indication for performing DLI [10]. In this regard, although DLIs have been used in almost all malignant diseases for which allogeneic HSCT is performed (specifically, chronic myelogenous leukemia (CML) [11], acute myeloid leukemia (AML) [12], juvenile myelomonocytic leukemia (JMML) [13], acute lymphoblastic leukemia (ALL) [14], Hodgkin lymphoma (HL) [15], non-Hodgkin lymphoma (NHL) [16], multiple myeloma (MM) [17]), CML represents the most consistently successful therapeutic model of this kind of approach. Several studies have reported, in large series of patients treated, response rates after CML relapse as high as 70 % [11, 18, 19]. The experience with DLI in CML has led to a better comprehension of the role of this approach. Although the optimal cell dose remains undefined, some concepts are well established, namely (i) the response is dose dependent; (ii) higher doses increase the risk of develo** GVHD; (iii) in recipients of HSCT from a matched unrelated donor (MUD), lower doses are sufficient to obtain the same therapeutic effect as in HLA-identical sibling HSCT; (iv) the best responses are obtained in patients with low tumor burden, such as either molecular or cytogenetic relapse of CML; and (v) there is a higher probability of response in patients relapsing late after transplantation.

DLI has also been used in other hematologic malignancies, such as AML, ALL, NHL, and MM [20]: however, less satisfactory results have been obtained in these diseases. In general, however, the more indolent disorders (such as MM and some type of NHL) seem to respond better to DLI than does acute leukemia.

In almost all of these studies, DLIs were used with a therapeutic purpose (tDLIs) aimed at treating disease relapse, either molecular, cytogenetic, or hematologic. A different approach is the prophylactic administration of donor T cells with a pre-emptive intent (pDLIs), namely before disease relapse, to prevent the occurrence of life-threatening infections. This strategy has been adopted mainly after myeloablative, in vivo, or ex vivo T-cell-depleted HSCT [21–26]. In the largest series reported to date by Montero et al., 112 patients with hematologic malignancies received DLIs containing 10 × 106 T cells/kg recipient body weight between day +45 and day +100 after T-cell-depleted peripheral blood stem cell transplantation (PBSCT) from an HLA-identical sibling donor [24]. With a median follow-up of 4 years, relapse incidence (RI), TRM, and disease-free survival (DFS) were 40, 20, and 46 %, respectively, with a cumulative incidence of grades III–IV acute and extensive chronic GVHD of 15 and 25 %, respectively. Of note, chronic GVHD was associated with lower overall mortality rate, due to a protective effect on disease recurrence [24].

In an interesting prospective study, Schaap and colleagues treated with DLI 31 patients with high-risk malignant disease at a median time of 22 weeks after partial T-cell-depleted bone marrow transplantation (BMT) from an HLA-identical sibling and compared their outcomes with those of 47 matched patients who did not receive DLI due to the previous onset of acute and/or chronic GVHD [25]. In comparison to the matched controls, patients receiving DLI experienced a statistically significant lower risk of relapse (18 versus 44 %, respectively) which resulted in an improved 3-year DFS (77 versus 45 %, respectively).

With the aim of optimizing T cell function and/or avoiding excessive alloreactivity, associated with “conventional” DLI, several approaches have been explored; they include the following:

  • CD8-depleted DLIs [27–30]; conversely, CD4-depletion does not prevent acute GVHD [31];

  • DLIs preventively depleted of the alloreactive component towards recipient tissues either by immunotoxins [32, 33] or by photodynamic purging [34];

  • Preventive incubation of donor T cells with rapamycin [35];

  • Transfection of T cells with suicide genes (see the following section).

Strategies aimed at enhancing T cell function include the following:

  • Treg-depleted DLIs [36];

  • Tumor-infiltrating leukocytes [37];

  • Ex vivo expanded and activated T cells [38] (e.g., pre-incubation of T cells contained in the DLI with cytokines able to augment their function [39]);

  • In vivo administration of IL-2 after DLI [40];

  • Association with monoclonal antibodies, such as anti-CD30 brentuximab vedotin in Hodgkin disease [41].

However, due to the small number of patients treated (and thus the lack of strong evidences of durable clinical effect) and/or the complexity of production methods, which may limit their use to few centers, most of those approaches have not been incorporated into routine clinical practice.

It is interesting to note that use of DLI has been recently reported also in the context of Cord Blood (CB) transplantation [42]. Berglund and coauthors reported nine infusions of ex vivo expanded CB-derived T cells in four patients because of mixed chimerism, minimal residual disease, or graft failure. The mean number of cells infused was 1 × 105/kg; only one patient developed acute GVHD, one patient reversed mixed chimerism, and, in another, control of positive minimal residual disease was achieved.

Genetic Manipulation of Donor T cells with Suicide Genes

In order to allow the infusion of higher numbers of donor-derived T cells also in a haploidentical HSCT setting, strategies based on the use of T cells engineered with safety switches have been developed [43]. Retro-viral mediated transduction of donor-derived T cells with suicide genes provides a tool to control the risk of acute GVHD in the presence of a high degree of HLA mismatch in the donor-recipient pair (Fig. 1). To date, two different approaches for creating cells with inducible suicide genes have been studied: the San Raffaele group in Milan tested the herpes simplex virus (HSV) thymidine kinase (HSV-TK) gene [44–46], which activated to kill the cell if gancyclovir is administered, and an investigator in Houston tested the use of inducible Caspase 9 (iCasp9) gene, which could be included by administration of the drug AP1503 [47, 48••, 49].

Ciceri and coworkers reported on 50 patients who received haploidentical HSCT after CD34+ cell selection for high-risk leukemia, 28 of whom were treated with HSV-TK-modified T cells starting from 22 days after transplantation at a dose ranging from 0.9 to 40 × 106/kg [46]. Ten patients developed grades I–IV acute GVHD and one developed extended chronic GVHD. In all of the patients, the GVHD was controlled by a 14-day course of ganciclovir (plus steroids and cyclosporine A in those patients who had developed the more severe stages of GVHD). Notably, patients obtaining improved immune reconstitution (defined as CD3+ in peripheral blood above 100/μl in two consecutive observations) had a statistically significantly lower probability of non-relapse mortality in comparison with those patients with delayed CD3+ recovery.

Since HSV-TK is potentially immunogenic [50] and exploiting its effect requires a drug, ganciclovir, which remains of significant utility for controlling post-transplantation human cytomegalovirus (CMV) reactivation, Brenner’s group in Houston developed a new suicide gene based on the fusion of human caspase 9 to a modified human FK-binding protein (inducible caspase 9, iCasp9). The construct encodes for a protein that dimerizes after exposure to a synthetic inert drug, named AP1903, thus becoming activated and leading to rapid cell death [47]. Zhou and colleagues reported long-term follow-up of 10 patients who received iCasp9 T cell infusion (from 1 × 106/kg up to 1 × 107/kg) 30 to 124 days after CD34+ cell-selected (and thus significantly T cell depleted) haploidentical HSCT [48••]. Investigators found long-term persistence of modified T cells up to 24 months after infusion, thus contributing, especially in the earlier phases after HSCT, to immune reconstitution; this was highlighted by evidences of immediate and sustained protection from pathogens like CMV and adenovirus. Four out of the 10 patients developed acute GVHD which was promptly controlled by AP1903 infusion which eliminated 85–95 % of circulating modified T cells within 30 minutes, with no recurrence of GVHD within 90 days. This kinetics of elimination of genetically modified T cells is much faster than that observed using HSV-TK-modified cells exposed to ganciclovir. It should be noted that virus-specific T cells recovered even after AP1903 administration and continued to provide sustained protection against infections. Noteworthy, both approaches improved immune reconstitution, not only because of the expansion of gene-modified T cells but also perhaps by accelerating recovery of endogenous (i.e., non-modified) T cells differentiating from donor hematopoietic stem cells [32, 46, 48••].

A phase I–II study evaluating the safety of iCAsp9 cell infusion in children receiving haploidentical HSCT both for malignant and for non-malignant disorders is currently ongoing at our institution.

Antigen-Specific Cytotoxic T Lymphocytes

Pathogen-Specific

After initial proof-of-principle studies on the restoration of immunity against CMV more than 20 years ago by the Seattle group [51, 52], the adoptive transfer of virus-specific T cells has been developed for many pathogens, especially viruses, including CMV [53], Epstein-Barr virus (EBV) [54], and adenovirus [55], ensuring good clinical results with an acceptable toxicity profile [56]. The techniques developed to obtain T lymphocytes specific for specified antigens are based on either repeated in vitro exposure of donor lymphocytes to relevant pathogen antigens/peptides followed by cell expansion to obtain larger number of cells to be infused [57], or on a rapid selection procedure based on magnetic-activated cells sorting (MACS) [58], including tetramers [59], streptamer isolation [60], and IFN-γ capture [61]. Each of these techniques is discussed below.

Culture Protocols

Successful ex vivo expansion of virus-specific T (VST) cells requires (i) the identification of the immune-dominant antigens/peptides, (ii) efficient APCs expressing HLA molecules that present virus-derived peptides, (iii) costimulatory signals ensuring T lymphocyte activation and expansion, and (iv) a prolonged period of culture (the process is time-consuming) [62]. To date, several groups have developed different protocols to improve T cell function and to reduce the time of the culture as well as the costs and complexity of the process (and thus to develop a widely accessible technique) [53, 63, 64].

The advantages of these strategies include the small quantity of cells required to start the process (especially compared to other approaches) and the possibility to obtain a broad polyclonal product containing both CD4+ and CD8+ T cells [63]. Moreover, the requirement that donors be seropositive donors to produce/expand specific T cells has been at least partly solved by the use of in vitro stimulation approaches for antigen-specific stimulation of naive (even CB-derived) T lymphocytes [65, 66].

Recent advances in this field include the generation of multi-specific VST cells and the possibility of creating third-party VST banks. In a seminal work, Leen and colleagues generated trivirus-specific (i.e., against CMV, EBV, and adenovirus) T cells from a single culture by preparing antigen-presenting cells (APCs) consisting of activated monocytes and EBV-lymphoblastoid cell lines (LCLs) transduced with an adenoviral vector encoding the immune-dominant CMV-pp65 antigen [67]. More recently, this strategy was further improved in terms of complexity, safety, and time required for generation of VSTs [68–70]. It is now possible to rapidly generate single-culture VSTs that recognize 12 immunogenic antigens from five viruses (EBV, adenovirus, CMV, BK virus, and human-herpes virus-6, HHV6) [71••]. Another potential breakthrough is the possibility of preparing banks of closely HLA-matched VSTs that can serve as “off the shelf” products, i.e., are available for immediate use. In fact, generating promptly specific VSTs for each individual patient is both impractical and impossible for widespread or urgent use. Despite concerns about safety, several studies have demonstrated the feasibility and efficacy of this approach, without a significant risk of GVHD [72, 73].

Selection by Multimers (Tetramers, Pentamers, and Streptamers)

These strategies are based on the binding of donor antigen-specific T cells via the T cell receptor (TCR) to an antigen-specific multimer (soluble peptide-HLA molecules) coupled with a system that subsequently allows cell selection (e.g., magnetic beads, fluorochrome-streptavidin complexes). This approach was first explored by Cobbold and colleagues who treated nine patients who had CMV reactivation with autologous pathogen-specific T cells selected by tetramers [59]. Although the median cells dose was 8.6 × 103/kg (and composed exclusively of CD8+ T cells), the cells expanded in vivo, leading to CMV viral load clearance in eight out of nine cases.

Since MHC multimer binding may interfere with the functional status of epitope-specific T cell populations in vivo (e.g., treatment with tetramers can induce epitope-specific tolerance in a dose-dependent manner) a different approach is that of streptamer technology. Streptamer technology involves low affinity Strep-tagged MHC molecules that are multimerized with a streptavidin derivative to generate multimers with high binding avidity. Conjugated fluorochromes or magnetic beads are used for cell staining/isolation. Upon addition of d-biotin, which competes with high affinity for the binding of Strep-tag to streptavidin, staining and isolation reagents dissociate rapidly from the cell surface. This reversibility enables multiple cell stainings/sequential positive cell selections. Streptamer technology allows reversible binding and thus does not alter T cell function or activate T cells through cross-linking TCRs [74].

The limitations of these techniques are the large starting number of cells needed (i.e., requiring an apheresis procedure), the restriction to HLA alleles for which antigen-specific viral peptides are available, the restriction to viruses for which the donor has detectable circulating T cells, and the risk of “immune escape” when only one peptide is targeted. For all of these reasons, multimer selection is actually available only for CMV [75], EBV [76], and adenovirus [77].

IFN-γ Capture

Another approach to rapidly selecting VST utilizes IFN-γ capture technique, which is based on the ability of Ag-specific T cells (both CD4+ and CD8+) to secrete cytokines (IFN-γ, in particular) after being challenged with antigens. Thus, after short-term (12–16 h) antigen exposure, IFN-γ catch reagent is attached to all leukocytes, and, in antigen-specific responding cells, IFN-γ remaining attached to the cell surface so that these cells can be subsequently isolated by magnetic selection [78].

The advantages of this strategy are that selected VST cells are both CD4+ and CD8+ and that it is not restricted to certain HLA alleles [61]. Again, although the number of VST cells that can be isolated is low, in vivo expansion has been reported to occur and can lead to viral clearance [79, 80].

Leukemia-Specific

Antigen-targeted immunotherapy with ex vivo expanded T cells is a promising approach to prevent or treat leukemia relapse by enhancing GVL effect, while avoiding GVHD, after allogeneic HSCT, especially in the setting of T-cell depleted HLA-haploidentical HSCT [81]. In this regard, donor-derived leukemia-reactive cytotoxic T lymphocytes (CTL) can be generated using as targets either specific antigens (namely tumor-associated antigens (TAAs)) or apoptotic leukemia blasts, in this case providing a broad antigen repertoire. The latter approach was used by Montagna et al. who generated and ex vivo expanded leukemia-reactive CTLs using donor-derived dendritic cells as APCs and apoptotic autologous leukemia blasts as the source of tumor antigen. Donor CTLs have been shown to selectively lyse leukemia blasts, with absent or low-level residual alloreactivity against non-malignant cells [82].

AML cells express several TAAs, which can be targets of ex vivo expanded and adoptively transferred CTLs. The most thoroughly studied leukemia-associated antigens are the Wilms tumor antigen (WT1), proteinase 3 (PR3), human neutrophil elastase (NE), and melanoma-associated antigen. Through in vitro experiments, Weber et al. showed that CTL lines, generated using 15mer peptide libraries of five TAAs, can target and kill AML cells. Moreover, they were multi-specific as assessed by IFN-γ enzyme-linked immunospot, regardless of their HLA type [83]. The same authors investigated a similar approach also in the setting of ALL, using WT1, Survivin, MAGE-A3, and PRAME as TAAs [84]. T cell lines were successfully expanded from all patients; moreover, tumor-specific responses were observed by reduction of autologous leukemia blasts in ELISpot, 51Cr-release assays, and coculture experiments.

Other possible targets are human minor histocompatibility antigens (mHAgs), which are T cell epitopes derived from polymorphic proteins and presented by various HLA class I and class II molecules. Some human mHAgs are preferentially expressed by leukemia cells, whereas non-hematopoietic tissues do not usually present them. Therefore, hematopoietic system-restricted mHAgs might be exploited to enhance immune responses in GVL, without increasing the risk of GVHD. A phase I clinical trial of adoptive immunotherapy with T cell clones specific for mHAgs has been reported by Warren and coworkers [85]. CTLs specific for tissue-restricted recipient mHAgs (assessed in in vitro assays) were infused into relapsed patients. Prior to infusion, T cell clones were expanded using culture methods to promote T cell proliferation and survival. This study showed that the adoptive transfer of mHAgs-specific T cells is able to mediate anti-leukemia activity in vivo, although in some patients treated with high T cell doses pulmonary toxicity has been observed [85].

Adoptive Transfer of Regulatory T Cells for Prevention of GVHD

Regulatory T cells restore tolerance in preclinical models of immune-mediated diseases. Among regulatory T cells, both CD4+ and CD8+ or double negative cells have been described [86]. Within the CD4+ regulatory T cell subsets, the best characterized are the CD4+ CD+ FOXP3+ T regulatory (Treg) cells and the type 1 regulatory T (Tr1) cells. Results from the first clinical trials exploring the adoptive transfer of Tregs in order to prevent GVHD and improve immune reconstitution after allogeneic HSCT are worthy of interest.

A prerequisite for the clinical use of naturally occurring Treg (nTreg) [86] cells is the ability to efficiently enrich and also expand this rare cell population, ensuring phenotypic homogeneity.

Several strategies have been investigated in order to purify and expand nTreg cells in vitro before infusion or to isolate nTreg cells in sufficient numbers for in vivo transfer. In the last years, the first trials with CD4+ CD25+ FOXP3+ Treg, either expanded in vitro or freshly isolated, have been reported. The group of Blazer from Minneapolis published the results of a trial of adoptive transfer of expanded third-party CB-derived polyclonal nTreg cells after non-myeloablative unrelated CBT. The adoptively transferred nTregs were present in the peripheral blood of patients up to 14 days after infusion of fresh cells and up to 4 days after the infusion of cryopreserved cells. Compared to identically treated historical controls, in these patients, a reduced incidence of grades II–IV acute GVHD was observed (43 versus 61 %) without increased risk of opportunistic infections, relapse, or early mortality [87]. However, more recently, the same group reported an increased cumulative incidence of viral reactivation in the early post-transplantation period in patients treated with adoptive transfer of Tregs [88].

In order to obtain a more homogeneous population of Treg cells for clinical application, investigators from Perugia, Italy, explored a different strategy [89, 90] in which they infused freshly, immunomagnetically enriched cell separation [91], donor CD4+ CD25+ nTreg cells to adults who were going to be recipients of T-cell-depleted HLA-haploidentical HSCT for hematological malignancies. The adoptive transfer of nTreg cells was followed, 3 days later, by the infusion of CD34+ cells together with a defined dose of donor mature T cells (conventional T cells, Tcon), in the absence of any pharmacologic immune suppression. Overall, the immune reconstitution was improved, and high frequencies of CD4+ and CD8+ T cells specific for opportunistic pathogens were detected. This approach demonstrated that the transfer of freshly purified nTreg cells is permissive for infusion of high doses of donor Tcon in the setting of HLA-haploidentical HSCT [89, 90].

Adaptive Tr1 cells are defined by their pattern of cytokine production, which include high levels of IL-10 in the absence of IL-4 and low IL-2. Roncarolo and coworkers conducted a phase I–II trial based on the administration of donor-derived IL-10-induced alloantigen-specific Tr1 cells (IL-10-DLI [92]), without immune suppression in patients with high-risk hematological malignancies transplanted with CD34+ cell-selected HLA-haploidentical HSCT. In fact, donor T cells primed ex vivo with host APCs and IL-10 are anergic towards host-HLA antigens but contain host-specific Tr1 cells and memory T cells able to respond to pathogens [93]. The five patients treated with this approach showed improved immune reconstitution, and four of them are alive with complete disease remission at 7.2 years (range 6.4–8.3) after HSCT [94, 95], thus providing the first proof-of-concept of feasibility and safety of Tr1 cell-based therapy and suggesting a clinical benefit of the use of Tr1 cells after HLA-haploidentical HSCT.

Novel approaches to increase Treg potency are actually under investigation. Very recently, the ex vivo fucosylation of third-party human Treg cells has been shown to improve Treg homing to the site of inflammation [96].

Conclusions

Apart from approaches of graft manipulation [97, 98], T cell therapy techniques represent the best option to enhance immune reconstitution (and thus transplant outcome) after HSCT, especially in high-risk patients (because of disease status or kind of HSCT). In this regard, new therapeutic approaches with gene-modified T cells, with the aim of enhancing effector functions (i.e., chimeric antigen receptors-modified T cells [99, 100], which, to date, have been rarely employed after HSCT) or of increasing safety through insertion of genes activating T cell apoptosis [47, 49], are particularly promising. Moreover, the possibility of rapid generation of multi-specific VST cells is attractive. The results of the different trials reported are highly encouraging but require confirmation in larger cohorts of patients, homogeneous in terms of disease and treatment, in order to obtain a truly effective comparison among different cell therapy approaches.

However, the real challenge for the future will be the standardization of the manufacturing of these products, together with the spread of these techniques in order to render them widely available, thus improving global transplant outcomes.