Invariant NKT Cell Reconstitution in Pediatric Leukemia Patients Given HLA-Haploidentical Stem Cell Transplantation Defines Distinct CD4⁺ and CD4⁻ Subset Dynamics and Correlates with Remission State

Allogeneic hematopoietic stem cell transplantation (HSCT) is an established approach to treat pediatric patients with relapsed or high-risk hematological malignancies (1–3). The therapeutic effect of HSCT relies on tumor reduction by the conditioning regimen and eradication of residual leukemia cells via the graft-versus-leukemia (GVL) effect (4). GVL is mainly mediated by donor-derived mature T lymphocytes transferred with the graft (5). About 60% of patients that would benefit from HSCT, however, lack a HLA-identical sibling donor. T cell-depleted HSCT from HLA-haploidentical family donors is thus increasingly used to overcome this limitation (6). This strategy offers the advantage of using donors that are promptly available for almost any patient. The presence of allogeneic MHC barriers between donor and recipient of haploidentical HSCT (hHSCT) requires almost complete elimination of donor T cells from the graft, to prevent graft-versus-host disease (GVHD). This results in a state of profound post-hHSCT immunodeficiency of the recipients, which lasts until the immune system is newly generated from donor hematopoietic stem cells (HSCs), and predisposes patients both to life-threatening infections and to leukemia relapse, the leading causes of death in this approach (7, 8). A deeper understanding of the reconstitution dynamics of all the components of the immune system after hHSCT is hence instrumental to optimize the control of infections, leukemia recurrence, and GVHD. In pediatric hHSCT, NK cells reconstitute earlier (i.e., within 1–3 mo post-hHSCT) than T lymphocytes, which need up to 1 y to reach normal values (9). The rapid NK cell recovery can partially compensate for the prolonged T cell deficiency in providing a GVL effect, which occurs when donor NK cells are alloreactive toward the HLA-disparate patient (10–12). T lymphocytes account for the long-lasting protection against infectious pathogens and leukemia cells, and reconstitute in pediatric hHSCT recipients essentially via a thymus-dependent pathway, which recapitulates T cell ontogeny and is very efficient in children, where the thymus is still functional (8, 13).

Invariant NKT (iNKT) cells are a conserved subset of T lymphocytes that in humans express an invariant Vα24-Jα18 TCR chain, which is preferentially paired with Vβ11 (14, 15). This semi-invariant TCR is specific both for microbial and endogenous lipid Ags, and for the strong synthetic agonist α-galactosyl ceramide (αGalCer), presented by CD1d (16). iNKT cells display constitutive effector functions that are acquired on development independently of exogenous Ag encounter (17, 18). Mouse iNKT cells are equally divided in two CD4⁺ and CD4⁻CD8⁻ subsets, which originate in the thymus and undergo a maturation process conventionally defined by the orderly acquisition of IFN-γ expression and the upregulation of NK receptor CD161 (NK1.1 or NKR-P1C in mice, NKR-P1A in humans) (16). The phenotypic and functional maturation dynamics of human iNKT cells is less known, because of the difficulty to follow longitudinally their development. Cross-sectional studies have identified only the initial and final steps of the maturation program of human iNKT cells. Human thymic iNKT cells display an immature CD4⁺CD161⁻ phenotype, which dominates also the neonatal peripheral compartment (19–22). In adults, peripheral iNKT cells increase their frequency and are divided into two main subsets expressing CD161: the CD4⁺ helper/regulatory and the CD4⁻ (mainly CD4⁻CD8⁻) effector/inflammatory, present on average with a CD4⁺<CD4⁻ ratio (23, 24).

iNKT cell activation results in the swift production of copious amounts of different cytokines and in the contact-dependent activation of APCs, which together help jump-start both adaptive and innate immune responses (16). As a consequence of these helper functions, iNKT cells are implicated in the control of infections and cancer. Evidence in mouse models and in cancer patients indeed suggests an active role of iNKT cells in the immune surveillance of different solid tumors, lymphoma, and multiple myeloma (25–27). Whether iNKT cell might also play a role in controlling other relevant hematological malignancies, such as leukemia, is not known.

To combine the possibility of investigating the maturation dynamics of human iNKT cells and evaluating the role for these cells in the control of leukemia, we took advantage of the clinical setting provided by T cell-depleted hHSCT in pediatric patients affected by hematological malignancies. The circulating T lymphocytes that appear at engraftment are very likely generated, via thymic differentiation, from the CD34⁺ precursors cells present in the graft, allowing to investigate the emergence of immature iNKT cell and their peripheral maturation and expansion. We undertook a thorough longitudinal analysis of iNKT cell reconstitution in 22 consecutive pediatric patients undergoing T cell-depleted hHSCT after a myeloablative regimen for the treatment of acute leukemia and myelodysplastic syndromes. Frequencies, number, phenotype, and effector functions of iNKT cells were assessed, in direct comparison with mainstream T cells, in the peripheral blood of recipients at specified predefined intervals up to 18 mo post-hHSCT, correlating their emergence with disease relapse. The reconstitution and maturation kinetics of the two iNKT cell subsets were analyzed in statistical terms using mathematical models. The findings on reconstitution, expansion, and maturation of iNKT cells were further corroborated by a cross-sectional analysis in a second group of 11 patients, who were in stable clinical remission of their hematological malignancies for 2–6 y post-hHSCT.

The study includes a first group of 22 consecutive patients (patients 1–22, median age: 9 y, range: 4–22 y) affected by hematological malignancies, which were analyzed longitudinally up to 18 mo post-hHSCT, and a second group of 11 patients (patients 23–33, median age: 6 y, range: 3–20 y) in stable remission, analyzed after 2–6 y post-hHSCT. All patients received a T cell-depleted hHSCT at the Pediatric Onco-hematology Unit of San Matteo Hospital Pavia between October 2003 and March 2009 (detailed in Table I). All patients were transplanted after a fully myeloablative preparative regimen including total body irradiation and/or chemotherapy and antithymocyte globulin. HSCs were mobilized in peripheral blood of the donor after administration of G-CSF. CD34⁺ cells were positively selected using the one-step CliniMacs device (Miltenyi Biotech, Bergisch Gladbach, Germany). The median number of CD34⁺ and CD3⁺ cells infused per kilogram of recipient body weight was 23 × 10⁶ (range: 10.0–41.2) and 5 × 10³ (range: 0.5–35), respectively. No patient received posttransplantation pharmacologic immune suppression. Patients with evidence of donor engraftment surviving more than 14 and 90 d post-hHSCT were evaluated for the occurrence of acute and chronic GVHD, respectively. These complications were diagnosed and scored according to established criteria (28, 29). Donor/recipient human CMV serology was determined by ELISA (30). Donor/recipient HLA-A, HLA-B, HLA-C, HLA-DRB1, and HLA-DQB1 alleles were identified by PCR single-strand polymorphism and sequence-based typing (31, 32).

The Institutional Review Board of Fondazione IRCCS Policlinico San Matteo, Pavia, Italy, approved the design of this study. Written informed consent in accordance with the Declaration of Helsinki was obtained from patients' parents to collect bone marrow (BM) and peripheral blood.

PBMCs and BM cells were isolated by Ficoll-Hypaque density gradient and immediately analyzed. Cells were cultured as previously described (22).

Patient PBMCs were analyzed by flow cytometry at defined time points post-hHSCT. iNKT (Vα24⁺Vβ11⁺CD3⁺) and T cells (CD3⁺) were stained with anti–Vα24-PE or anti–Vα24-biotin (Coulter, Fullerton, CA), anti–Vβ11-FITC or anti–β11-PE (Coulter), anti–CD3-allophycocyanin (BD Biosciences, San Jose, CA) or anti–CD3-Pacific Blue (BioLegend, San Diego, CA), anti–CD4-Pacific Blue (eBioscences, San Diego, CA) or anti–CD4-PE (BD Biosciences), anti–CD161-biotin (Serotec, Oxford, U.K.) or anti–CD161-PerCP-Cy5.5 (eBioscences), anti–IFN-γ–allophycocyanin, anti–IL-4–PE (BD Biosciences) mAbs, and streptavidin-Quantum Dot 800 (Invitrogen UK) or PeCy5.5 (BD Biosciences). Intracellular cytokine production was detected as described previously (22). Thawed PBMCs from patients were incubated at 37°C with PMA + ionomycin for 1 h + 2 h with brefeldin A (10 μg/ml; Sigma, St. Louis, MO). Thawed PBMCs from patients (when sufficient cells were available) or healthy adult donors were also cocultured at 37°C for 16 h + 2 h with brefeldin A (10 μg/ml) with C1RCD1d cells, preloaded for 1 h at 37°C with 100 ng/ml αGalCer. A minimal number of 2 × 10⁶ T cells/sample was acquired on a BD Canto II (BD Biosciences) and analyzed with FlowJo software (TreeStar, Ashland, OR). Dead cells and cell aggregates were excluded from the analysis. Absolute numbers of iNKT and T cells in each patient were determined by multiplying the absolute lymphocyte number by the iNKT or T cell percentage among lymphocytes.

To describe kinetics of expansion and phenotypic maturation for each iNKT cell subset (depicted in Figs. 2B, 3B), we used mixed-effects models (33). Mixed-effects models include parameters associated with the entire population, called fixed effects, and subject-specific parameters, called random effects. For this reason, mixed-effects models are able to describe the common dynamics of the phenomenon under investigation, even in presence of high variability between subjects, providing both estimates for the entire population’s model and for single subjects.

To evaluate differences in iNKT, CD4⁺ T, and CD8⁺ T cell reconstitution between patients who did or did not experience leukemia relapse within 18 mo post-hHSCT (Fig. 5A, 5B), we considered for each patient the piecewise linear curve interpolating the observations of the quantity under study and computed their area under the curve (AUC). The differences between the distributions of normalized AUC in patients who did or did not experience disease recurrence were tested by using the nonparametric Wilcoxon rank sum test. A p value ≤0.05 was considered to be statistically significant. All statistical analyses were carried out using the R language and environment for statistical computing (34).

To obtain a dynamic view of iNKT cell reconstitution, 22 consecutive patients with various hematological malignancies (patients 1–22; Table I) were studied longitudinally from the day of transplantation up to 18 mo post-hHSCT. In this group, eight patients had leukemia relapse (36%) at a median time of 5 mo (range: 3–18) post-hHSCT and six of them died because of disease progression. None of the potentially relevant variables, including donor NK alloreactivity, predicted the risk for disease recurrence. In 11 patients, fungal/bacterial/viral infections (50%) were recorded and 2 children experienced development of grade II acute GVHD (10%) (Table I). iNKT cell reconstitution was also investigated by cross-sectional analysis post-hHSCT in 11 additional patients (patients 23–33, Table I) maintaining stable remission for 2–6 y after the allograft.

iNKT- and T cell reconstitution was investigated by flow cytometry on PBMCs from patients 1–22 (Table I) at predefined time points post-hHSCT: +30, +60, +90, +180, +360 d (±10 d, depending on the schedule of clinical controls). In 10 patients, the analysis continued monthly or every 2 mo until 18 mo post-hHSCT. Monitoring was discontinued at time of disease relapse. iNKT cells emerged in periphery as early as 3 mo post-hHSCT, with an estimated delay of 2 mo compared with T cells. Fig. 1 shows the temporal curves depicting the absolute numbers and frequencies of iNKT, CD4⁺ T, and CD8⁺ T cells, obtained for each patient included in the longitudinal analysis performed in the first 18 mo post-hHSCT. Both iNKT and T cell compartments progressively expanded after their appearance in the periphery. At 18 mo post-hHSCT, the median numbers of cells that we found were 0.549 iNKT cells/μl (range, 0.08–0.8), 913 CD4⁺ T cells/μl (range, 273–2522), and 608 CD8⁺ T cells/μl (range, 371–1021), whereas the median cell frequencies among total T cells were 0.029% (range, 0.008–0.06) iNKT, 62.5% (range, 27–66) CD4⁺ T, and 37% (range, 28–54) CD8⁺ T cells. The numbers and frequencies recorded for iNKT and T cells were in the range of those found in age-matched healthy children (21, 35), suggesting that iNKT and T cells reached normal reference values by 18 mo post-hHSCT.

The longitudinal analysis of patients belonging to the group of patients 1–22 (Table I) allowed us to investigate the posttransplantation reconstitution dynamics of the CD4⁺ and CD4⁻ iNKT cell subsets. As shown for one representative patient (Fig. 2A), during the early post-hHSCT period, iNKT cells exhibited a nearly homogeneous CD4⁺ phenotype, resembling that previously reported in the thymus and in umbilical cord blood of healthy or nonleukemic subjects (19–22). CD4⁻ iNKT cells became detectable 2–4 mo later than the CD4⁺ ones, corresponding to, on average, 6 mo post-hHSCT. The numbers of both CD4⁺ and CD4⁻ iNKT cells increased significantly over the 18-mo period (Fig. 2A), suggesting that both subsets progressively expanded in periphery, and that the reduction of the CD4⁺ iNKT cell frequency was due to the expansion of CD4⁻ subset.

For a statistical evaluation of the overall reconstitution kinetics of the two iNKT cell subsets in the patients selected earlier, the mixed-effects models (33) determined the following estimates for the first 18 mo post-hHSCT: 1) the CD4⁺ iNKT cell frequency declined from 100% to an asymptote of 70%, whereas the CD4⁻ iNKT cells increased to an asymptote of 30% (Fig. 2B); and 2) the number of CD4⁺ iNKT cells grew at a rate twice as high as that of the CD4⁻ cells (0.025 versus 0.013 cells/month; Fig. 2B), a difference in the expansion rates between the two iNKT cell subsets that was statistically significant. By 18 mo post-hHSCT, the frequencies of the two iNKT cell subsets were comparable with those found in age-matched nonleukemic children, with the CD4⁺ subset dominating over the CD4⁻ one (21). Yet, these frequencies were still different from those reported for adult healthy subjects, which bear an inverted CD4⁺<CD4⁻ iNKT cell ratio, suggesting that these cells had not yet reached equilibrium by 18 mo post-hHSCT.

To determine the time needed for the recovery of CD4⁺ and CD4⁻ iNKT cells to values found in adult healthy subjects, we investigated cross sectionally the frequencies and numbers of the two iNKT cell subsets in the second cohort of patients, which received hHSCT for hematological malignancies and were maintaining remission up to 6 y after the allograft (patients 23–33, Table I). As shown in Fig. 2C, CD4⁺ iNKT cells remained the dominant subset up to 4 y post-hHSCT, although their frequency tended to progressively decline compared with the first 18 mo posttransplantation. Between 4 and 6 y post-hHSCT, the frequency of CD4⁺ iNKT cells dramatically diminished, whereas that of the CD4⁻ cells increased, resulting in the inverted CD4⁺<CD4⁻ iNKT cell ratio observed in adults. The number of CD4⁺ iNKT cells remained essentially constant between 18 mo and 6 y post-hHSCT, whereas the number of CD4⁻ iNKT cells expanded slowly up to 4 y post-hHSCT and then underwent a brisk expansion between 4 and 6 y, making the CD4⁻ cells the predominant subset (data not shown).

Altogether, these results showed that the CD4⁺ and CD4⁻ iNKT cell subsets reached adult-like values by 4–6 y post-hHSCT, following two remarkably independent expansion dynamics.

The expression of CD161 conventionally distinguishes mature iNKT cells (16). We hence determined the post-hHSCT maturation dynamics of CD4⁺ and CD4⁻ iNKT cells by investigating longitudinally their CD161 acquisition rate in patients 1–22 (Table I). As shown for one representative patient (Fig. 3A), both iNKT cell subsets emerged in the periphery exhibiting an immature CD161⁻ phenotype, demonstrating the presence of this rather elusive precursor also in the CD4⁻ subset.

After emergence, both subsets underwent progressive phenotypic maturation by showing acquisition of CD161 expression (Fig. 3A). To assess in statistical terms the possible differences in the maturation kinetics between the two iNKT cell subsets, we again used the mixed-effects models (33). This analysis estimated that, by 18 mo post-hHSCT, CD161 was expressed by 57 and 93% of CD4⁺ and CD4⁻ iNKT cells, respectively (Fig. 3B). The values obtained for CD4⁻ cells approached the percentage of CD161 expression reported by iNKT cells of adult healthy donors, suggesting that these cells were almost completely phenotypically mature by 18 mo post-hHSCT. Interestingly, the model estimated that the velocity of CD161 acquisition was significantly faster in CD4⁻ than in CD4⁺ cells, as indicated by the slope of the tangents to the kinetic curves of CD161 expression shown in Fig. 3B, which were 0.13 and 0.37 for CD4⁺ and CD4⁻ cells, respectively.

Collectively, these results showed that both CD4⁺ and CD4⁻ iNKT cells emerge in the periphery displaying an immature CD161⁻ phenotype, and that the CD4⁻ subset matured with significantly faster kinetics than the CD4⁺ one.

To assess the kinetics of effector differentiation by reconstituting iNKT cells, we investigated their capacity to produce IFN-γ and IL-4 ex vivo on a brief stimulation. Previous studies showed that the frequency of human iNKT cells producing IFN-γ increases from 10–20% at birth to 50–60% in adults, suggesting that these cells acquire full effector functions only late after birth (20, 22). Considering the paucity of circulating iNKT cells in the early posttransplantation period, we investigated cytokine production by iNKT cells at 6 mo after their first detection post-hHSCT in the group of patients 1–22 (Table I). At this time, the estimated numbers of circulating iNKT cells were sufficient to perform the analysis. The production of IFN-γ and IL-4 by iNKT cells from hHSCT patients and adult healthy donors were first compared after a very short stimulation time with PMA/ionomycin, to unveil their immediate effector potential (22). As indicated in Fig. 4A (upper panel), at 6 mo after their first detection post-hHSCT, 40% of CD4⁻ iNKT cells from the patients were already capable of producing IFN-γ, whereas only 20% of the CD4⁺ ones made this cytokine. In adult healthy donors, ∼50% of the cells in both iNKT cell subsets could make IFN-γ on the short stimulation. The same experimental conditions did not reveal functional differences in terms of IL-4 production between the CD4⁺ and CD4⁻ iNKT cell subsets from both patients and adult healthy donors (mean 10% of cells producing IL-4, data not shown).

In three patients, we had enough PBMCs to assess the responsiveness of their iNKT cell subsets, on activation in vitro with CD1d-expressing lymphoblastoid cell lines loaded with αGalCer. At 6 mo after their first detection post-hHSCT, the percentage of IFN-γ–producing CD4⁻ iNKT cells was greater than that of the CD4⁺ subset (mean 25 and 8%, respectively), also in response to Ag-specific stimulation (Fig. 4A, lower panel). The frequency of CD4⁻ iNKT cells that reached IFN-γ expression competence was in the range of adult cell values, although in adults, the fraction of cytokine-expressing cells was again comparable in both subsets.

Because mouse iNKT cells produce IFN-γ before CD161 acquisition (16), we investigated the IFN-γ production of iNKT cells in relation with their CD161 expression. We found that human iNKT cells that produced IFN-γ were all CD161⁺, regardless of their CD4 coreceptor expression (Fig. 4B). This finding suggested a linear maturation pathway that, unlike mice, proceeded from an immature CD161⁻IFN-γ⁻ stage, through an intermediate CD161⁺IFN-γ⁻ stage, to a final mature CD161⁺IFN-γ⁺ stage.

Altogether, these results suggested that iNKT cells acquire rapidly effector competence post-hHSCT, contributing to the early provision of IFN-γ. Moreover, they indicate that, consistent with their markedly faster phenotypic maturation, the CD4⁻ cells acquire IFN-γ production significantly more rapidly than the CD4⁺ ones and approached adult-like values already by 6 mo after their emergence in the periphery.

We finally sought to determine whether iNKT cell reconstitution post-hHSCT correlated with remission maintenance in patients 1–22 (Table I) who were investigated longitudinally. As shown in Fig. 5A, iNKT cells remained undetectable during the whole follow-up in the eight children of the group of patients 1–22 (Table I) in whom leukemia relapsed post-hHSCT. In seven of eight patients (patients 1, 5, 6, 7, 15, 16, and 20, Table I), the relapse occurred from 4–18 mo post-hHSCT—that is, at a time when iNKT cells had already emerged and were clearly detectable in all patients who maintained remission. Only one patient (patient 14, Table I) experienced disease relapse at a time (3 mo post-hHSCT) when iNKT cells had just begun to emerge and were not yet clearly detectable in all patients.

The difference in iNKT cell number between patients in remission and those experiencing disease relapse was statistically significant. Although circulating CD4⁺ and CD8⁺ T cells were always clearly detectable in the relapsed patients, their absolute numbers were also statistically lower than those found in patients maintaining remission (Fig. 5A). Interestingly, the analysis of the frequencies of iNKT, CD4⁺ T, and CD8⁺ T cells demonstrated a statistically significant reduction only of iNKT cells in the relapsed cohort compared with the nonrelapsing patients (Fig. 5B). By contrast, the frequencies of CD4⁺ and CD8⁺ T cells were comparable in the two groups of patients. This result suggested that the frequency of iNKT cells, but not that of either CD4⁺ T or CD8⁺ T cells, significantly correlates with a remission state after hHSCT.

Because leukemia recurs usually in the BM, we investigated the presence of iNKT and T cells in this compartment using serial BM aspirates collected post-hHSCT from the following patients: 1) a patient in stable remission for the whole follow-up; and 2) a second patient who experienced leukemia relapse (Fig. 5C). Both T and iNKT cells were normally present in the BM of the patient in stable remission. By contrast, T but not iNKT cells could be detected in the BM of the patient before relapse, suggesting that the lack of circulating iNKT cells in this patient did not reflect a selective recruitment of these cells into the BM. Together, these findings suggest that a defective iNKT cell reconstitution might contribute to an impaired immune surveillance of leukemia.

The results of this study define the expansion and maturation dynamics of human CD4⁺ and CD4⁻ iNKT cell subsets, and suggest that these cells in hHSCT recipients might contribute to the maintenance of the remission state, possibly through the early provision of antitumor cytokine IFN-γ.

In our patients, total T lymphocytes emerged by 2 mo post-hHSCT and displayed reconstitution kinetics comparable with previously published data (8). iNKT cells appeared in the periphery of hHSCT recipients with a slower kinetics than T cells, though they approached normal age-matched reference values within 18 mo posttransplantation. Only two studies so far have addressed iNKT cell reconstitution after allogeneic HSCT and demonstrated that iNKT cells emerged and reconstituted to normal values before T cells. The possible correlation between iNKT cell reconstitution and remission was not addressed, although one study suggested that a low number of iNKT cells could be related to the development of GVHD (36). Both studies were performed in adult patients receiving either T cell-replete HLA-matched allogeneic HSC or umbilical cord blood (36, 37), and thus required posttransplantation treatment with immune-suppressive drugs to prevent GVHD. This faster iNKT cell reconstitution dynamics could depend, at least in patients given HLA-matched HSC, on the adoptive transfer of donor-derived mature lymphocytes present in the graft and/or the different transplantation protocols used in these studies, as compared with our approach of T cell depletion of the graft.

The immune reconstitution in T cell-depleted hHSCT patients depends on the functional recovery of the thymus, which is efficient in pediatric patients and recapitulates the physiological development of T lymphocytes (13). Consistent with their recent thymic emigration, the first iNKT cells emerging in our patients after hHSCT displayed the immature CD4⁺CD161⁻ phenotype. We could also detect immature CD4⁻CD161⁻ precursors, emerging, on average, by 6 mo post-hHSCT. This elusive CD4⁻ immature stage has not been detected either in fetal or postnatal thymus (19–22), suggesting the possibility that, in humans, the bifurcation between CD4⁺ and CD4⁻ iNKT cells may occur in the periphery, at variance with mouse CD4⁺ and CD4⁻ iNKT cells, which separate already at the level of thymic precursors (18). However, we cannot exclude that the human immature CD4⁻CD161⁻ iNKT cells emerge already in the thymus at frequencies below detection.

The distinct peripheral expansion and maturation dynamics of human CD4⁺ and CD4⁻ iNKT cell subsets demonstrated by our study is remarkable and could not be anticipated from the published data from mouse CD4⁺ and CD4⁻ iNKT cells, which simultaneously expand and reach full maturation within the first 2–4 mo of age (38). The generation of the human CD4⁺ and CD4⁻ iNKT cell repertoires could be controlled by two distinct mechanisms, in line with the hypothesis of Baev et al. (20). A thymic output in the early posttransplantation period would account for a progressive peripheral accumulation of immature CD4⁺ iNKT cells, which then slowly divide and mature. In contrast, the peripheral CD4⁻ iNKT cell compartment would be originated from rare precursors that rapidly mature and subsequently undergo an expansion burst, becoming the predominant subset by 4–6 y after their peripheral appearance. The robust expansion burst of CD4⁻ iNKT cells that we document by 4–6 y post-hHSCT, leading to the characteristic CD4⁺<CD4⁻ iNKT cell ratio found in healthy adults, is consistent with the lower quantity of TCR excision circles found in adult CD4⁻ compared with CD4⁺ iNKT cells (20), suggesting a greater cell division rate in the former subset. What mechanisms control such different and asynchronous rates of expansion and maturation exhibited by the two human iNKT cell subsets remains an open question. It can be hypothesized that the two human iNKT cell subsets are differentially responsive to specific environmental cues generated at various stages post-hHSCT, which may include the homeostatic cytokines IL-7 and IL-15 (20, 22), and could differently impact on their cell division and maturation programs.

In our study, all the patients who relapsed post-hHSCT failed to reconstitute the iNKT cell repertoire, suggesting that the absence of these cells might be a factor contributing to leukemia regrowth. Because mainstream T cells were also significantly reduced in relapsing patients, our findings underscore the relevance of reconstituting a peripheral T cell repertoire containing a correct proportion of iNKT and T cells for the maintenance of leukemia control after hHSCT. This might be a general requirement for tumor immune surveillance, because Molling et al. (39) recently showed that undetectable to low levels of circulating iNKT cells (<48 iNKT/10⁶ T cells), present in patients with head and neck squamous cell carcinoma before radiotherapy, was significantly associated with locoregional recurrence and poor survival. This iNKT/T cell ratio is similar to that found in the small group of our relapsing patients at 5 mo post-hHSCT (19.6 iNKT cells/10⁶ T cells). The fact that a rapid iNKT cell reconstitution post-hHSCT might be beneficial in terms of leukemia recurrence is also indirectly supported by the evidences obtained from both animal studies and patients, showing the iNKT cell capacity to control hematological or solid tumors via a variety of effector mechanisms, which rely on the following: 1) direct CD1d-dependent recognition of malignant cells (40–42); and 2) indirect antitumor effects, such as enhancement of leukemia-specific CD8⁺ T cell cytotoxicity and GVL via APC licensing (43, 44), killing of CD1d-expressing tumor-infiltrating macrophages (45), or the production of the critical antitumor cytokine IFN-γ (46, 47). In the cohort of patients included in this study, we could not correlate leukemia remission or relapse with the expression of CD1d on leukemia blasts (C.d.L., data not shown); however, as discussed earlier, direct CD1d-dependent recognition of blasts by iNKT cells would not be a prerequisite for their antitumor effect.

In addition to their GVL activity, iNKT cells exert beneficial effects in patients given HSCT also by suppressing GVHD via provision of Th2 cytokines, which polarize pathogenic antihost T cell responses toward less harmful Th2 responses and/or promote the expansion of CD4⁺CD25⁺Foxp3⁺ T regulatory cells (48–50).

Although obtained in a cohort including a limited number of patients, our results could have interesting clinical implications for T cell-depleted pediatric hHSCT (51). Screening for the lack of iNKT cell reconstitution in blood samples might provide a noninvasive prognostic parameter useful to identify patients at risk for leukemia recurrence post-hHSCT. Furthermore, we can envisage that the adoptive transfer of donor-derived iNKT cells into hHSCT recipients, who do not properly reconstitute the iNKT cell compartment within the first 4–6 mo after the graft, might represent a possible therapeutic option for preventing leukemia relapse. The lack of CD1d polymorphisms overcomes the problem of histocompatibility barriers between haploidentical donors and recipients, making the transfer of mature iNKT cells after hHSCT a safer and more applicable approach than that of conventional T lymphocytes (51).

We thank Dr. Chiara Bonini for critical reading of the manuscript and insightful suggestions.

The authors have no financial conflicts of interest.