IL-17-producing T cells contribute to acute graft-versus-host disease in patients undergoing unmanipulated blood and marrow transplantation

Authors


Abstract

The aim of this study was to investigate the effects of IL-17-producing T cells, including Th17 and Tc17 cells, on acute graft-versus-host disease (aGVHD) in patients who had undergone granulocyte colony-stimulating factor (G-CSF)-mobilised peripheral blood progenitor cell (PBPC) and G-CSF-primed bone marrow (G-BM) transplantation. Allografts from forty-one patients were analysed for IL-17-producing T cells with respect to aGVHD. Furthermore, ten patients with aGVHD onset were monitored for the presence of Th17 cells in the peripheral blood by flow cytometry. Patients who received a higher dose of Th17 cells in the G-BM (>8.5×104/kg, p=0.005) or a higher dose of Tc17 cells in PBPC (>16.8×104/kg, p=0.001) exhibited a higher incidence of aGVHD. An increased Th17 population (up to 4.99% CD4+ T lymphocytes) was observed in patients with aGVHD onset. In contrast, the percentage of Th17 population decreased drastically in aGVHD patients following treatment to achieve partial and complete remission (p=0.013 and p=0.008, respectively). All percentages of Th17 and Tc17 cells were significantly reduced after in vivo G-CSF application. Our results suggested that IL-17-producing T cells contributed to aGVHD. The application of G-CSF in vivo aided in reducing the occurrence of aGVHD through a decrease in IL-17 secretion by T cells.

Introduction

Allogeneic haematopoietic stem cell transplantation (allo-HSCT) provides curative therapy for patients with various haematological malignancies 1. However, the incidence of acute graft-versus-host disease (aGVHD) limits the application of allo-HSCT 2. aGVHD is a proinflammatory process that is mediated in part by mature donor T cells present in the allograft, which are polarised towards a Th1 phenotype and recognise minor or major histocompatibility disparities between the donor and the host 3–5. Previous work has demonstrated that the cytokine profiles of T cells, such as IFN-γ and IL-4, are important in the pathogenesis of aGVHD 6–8. The available data have clearly shown that a newly identified subset of IL-17-producing CD4+ T lymphocytes named Th17 cells play a crucial role in triggering inflammation and tissue injury in various autoimmune diseases 9–11. The role of Th17 cells in aGVHD has been controversial in recent mouse and human transplantation studies 12–20. To date, four mouse models have been used to investigate the role of Th17 cells in the occurrence of aGVHD. The first study to report the use of IL-17A−/− donors suggested that IL-17A attenuated severe aGVHD due to the suppression of Th1 differentiation and that neutralising IL-17 might augment aGVHD 20. A subsequent report using the same mouse strain combinations suggested that IL-17 production by donor T cells was not required for aGVHD in recipients of non-selected donor T cells; however, IL-17 production contributed to the development of aGVHD in recipients of CD4+ T cells by increasing the production of proinflammatory cytokines and recruiting or priming Th1 cells in lymphoid organs during the early phase after transplantation 16. The third study demonstrated that a highly purified population of Th17 cells was capable of inducing extensive pathologic coetaneous and pulmonary lesions, which are hallmarks of lethal aGVHD. These results were not model specific; the lesions occurred in both a haploidentical and a minor mismatch model 12. A fourth report suggested that wild-type T cells develop by default into Th1 cells that primarily mediate gut and liver aGVHD. In comparison, Th1/Th2 depletion predisposes naive T cells to Th17 cell outgrowth, which manifests primarily as skin GVHD. The association of Th2 cells with lung GVHD has been reported previously, and this association occurs following Th1/Th17 depletion 19. Currently, there are no data available concerning the role of the dose of Th17 or Tc17 cells in the grafts and the occurrence of aGVHD in humans after unmanipulated transplantation. Therefore, we hypothesised that the number of IL-17-producing T cells, including Th17 and Tc17 cells, in allografts infused into recipients could be associated with aGVHD. To test our hypothesis, we investigated the dosage of Th1, Th17, Th2, Tc1, Tc2 and Tc17 cells in the allograft, and we countered the numbers of the reconstituted Th1, Th17, Th2, Tc1, Tc2 and Tc17 cells in the peripheral blood at day 30 post-transplantation. Furthermore, we examined the proportions of Th17 and Tc17 cells in patients with aGVHD after transplantation. Moreover, we studied the effects of in vivo G-CSF application on the proportions of IL-17-producing T cells in G-CSF-primed bone marrow (G-BM) and G-CSF-mobilised peripheral blood progenitor cell (PBPC) allografts.

Results

Patient characteristics

Table 1 shows the characteristics of the 41 patients. Three subgroups were designated according to the median of the Th17 dose in G-BM (8.5×104/kg) and the median of the Tc17 dose in PBPC (16.8×104/kg). The characteristics of 10 patients evaluated for the association of aGVHD and Th17 expression are shown in Table 2. All the patients achieved engraftment and complete donor chimerism after transplantation. As of May 1, 2010, 50 patients had survived longer than 100 days without relapse or death, and one patient relapsed on day 80 after transplantation. Grades I, II and III aGVHD occurred in 10, 8 and 2 patients, respectively, among the 41 patients.

Table 1. Patient characteristics in the low- and high-dose groups
CharacteristicsLow doseOther doseHigh dosep-Value
  1. a

    Patients were designated as the “high dose” (Th17 dose in the G-BM >8.5×104/kg and Tc17 dose in PBPCs >16.8×104/kg, n=12), the “low dose” (Th17 dose in the G-BM <8.5×104/kg and Tc17 dose in PBPCs <16.8×104/kg, n=12) or the “Other dose” (Th17 dose in the G-BM >8.5×104/kg and Tc17 dose in PBPCs <16.8×104/kg or Th17 dose in the G-BM <8.5×104/kg and Tc17 dose in PBPCs >16.8×104/kg, n=17) according to the median of the Th17 dose in the G-BM (8.5×104/kg) and the median of the Tc17 dose in PBPCs (16.8×104/kg). AML, acute myeloblastic leukaemia; ALL, acute lymphoblastic leukaemia; CML, chronic myelocytic leukaemia; MDS, Myelodysplastic syndrome; NHL, non-Hodgkin's lymphoma; AA, aplastic anemia. High risk: patients with MDS, NHL, or a more-advanced stage of ALL/AML or CML beyond first complete remission or first chronic phase.

N121712 
Patient age, median (range)25.5 (9–46)35.5 (18–57)28 (6–35)NS
Patient sex, male, n (%)3 (25%)5 (29.4%)5 (41.7%)NS
Donor sex, male, n (%)3 (25%)8 (47.1%)9 (75%)0.049
High risk, n (%)3 (25%)7 (41.2%)5 (41.7%)NS
Diagnosis   NS
AML, n (%)5 (41.7%)4 (23.5%)7 (58.3%) 
ALL, n (%)4 (33.3%)8 (47.1%)3 (25%) 
CML, n (%)03 (17.6%)1 (8.3%) 
MDS, n (%)1 (8.3%)1 (5.9%)0 
NHL, n (%)1 (8.3%)1 (5.9%)1 (8.3%) 
AA, n (%)1(8.3%)00 
Patient/donor HLA incompatibility   NS
No loci (%)8 (66.7%)8 (47.1%)2 (16.7%) 
Single locus (%)01 (5.9%)1 (8.3%) 
Two loci (%)2 (16.7%)5 (29.4%)3 (25%) 
Three loci (%)2 (16.7%)3 (17.6%)6 (50%) 
WBC engraftment, median (range)14.5 (10–19)13 (10–17)13 (12–16)NS
PLT engraftment, median (range)14 (8–23)14 (11–34)17 (8–38)NS
Table 2. Clinical characteristics of ten patients with aGVHDa)
Patient number/sexAgeDiagnosisConditioning regimensSex disparityDonorPatient/donor HLA compatibilityGrade of GVHDGVHD treatment
  • a)

    a) F, female; M, male; AML, acute myeloblastic leukaemia; ALL: acute lymphoblastic leukaemia; Cy, cyclophosphamide; Bu, busulfan; ATG, anti-human thymocyte immunoglobulin; MTX, methotrexate; CSA, cyclosporine A.

1/F13ALLBu/Cy+ATGYesRelatedTwo lociIMethylprednisolone+anti-CD25
2/M37AMLBu/CyYesSiblingNo lociIIIAnti-CD25
3/F19ALLBu/Cy+ATGNoRelatedTwo lociII/IMethylprednisolone+MTX
4/M20ALLBu/Cy+ATGNoRelatedTwo lociIIAnti-CD25
5/M6AMLBu/Cy+ATGYesRelatedOne locusIIIAnti-CD25
6/M25ALLBu/Cy+ATGYesRelatedThree lociISwitch from CSA to FK506
7/M21ALLBu/Cy+ATGYesRelatedTwo lociIIIAnti-CD25
8/M18ALLBu/Cy+ATGNoRelatedThree lociIIMethylprednisolone
9/M23ALLBu/Cy+ATGNoRelatedThree lociIIMethylprednisolone
10/F37AMLBu/CyYesSiblingNo locusIIIMethylprednisolone+MTX+anti-CD25

Differences in the proportions and doses of Th1, Th17, Tc1 and Tc17 cells infused in BMT patients

First, we compared the percentages and the doses of type 17 (Th17 and Tc17), type 1 (Th1 and Tc1) and type 2 (Th2 and Tc2) effectors cells in CD4+ and CD8+ T cells in the G-BM or PBPC allografts infused in patients with and without aGVHD. Patients who subsequently developed aGVHD demonstrated greater proportions and doses of Th17 and Tc17 cells infused in the G-BM (Fig. 1A and B) or PBPC allografts (Fig. 1C and D). The representative FACS plot of Th17 and Tc17 cells infused in the G-BM and in PBPCs allografts are shown in patients without GVHD and with aGVHD (Supporting Information Fig. 1). Furthermore, we found significantly higher proportions and doses of Th1 and Tc1 cells in G-BM or PBPC allografts in patients with aGVHD (data not shown). When we analysed the correlations between type 17 T-cell (Fig. 2A and B) and type 1 T-cell doses (Fig. 2C and D) infused in the G-BM and PBPC allografts, statistically significant differences in the occurrence of aGVHD were discovered. However, no association between type 2 T-cells and the occurrence of aGVHD were found (data not shown).

Figure 1.

Differences in the proportions and doses of Th1, Th17, Tc1 and Tc17 cells infused in patients with and without aGVHD. The proportions and doses levels of Th17 and Tc17 cells infused in the G-BM (A and B) and in PBPC (C and D) allografts are shown among patients without GVHD (n=21) and with aGVHD (n=20). Th17 (%) or Tc17 (%) represent the percentages of CD4+ or CD8+ T cells expressing IL-17. Data represent the mean±SD (Mann–Whitney U test, two-tailed).

Figure 2.

Differences in the overall doses of Th1, Th17, Tc1 and Tc17 cells infused in patients with and without aGVHD. The overall doses of Th1 (A), Th17 (B), Tc1 (C) or Tc17 cells (D) infused in the G-BM and PBPC allografts are shown among patients without GVHD (n=21) and with aGVHD (n=20). Data represent the mean±SD (Mann–Whitney U test, two-tailed).

Acute GVHD occurred in 20 patients, including 1 patient with gastrointestinal GVHD, 5 patients with skin plus gastrointestinal GVHD and 14 patients with simple skin GVHD. A significant association between the dose of IL-17-producing T cells and the occurrence of simple skin GVHD was also found (data not shown). Furthermore, we hypothesised that the pattern of GVHD organ involvement observed after transplantation was likely to be more complex than the dominance of a single cytokine and was most likely due to the balance of type 1, type 2 and type 17 effectors in both CD4+ and CD8+ T cells. Therefore, we investigated the type 1, type 2 and type 17 effectors in the allografts simultaneously. However, we failed to find an association between a specific balanced ratio of type 1, type 2 and type 17 effectors and the pattern of GVHD organ involvement (data not shown).

Predictive value of Th1, Th17, Tc1 and Tc17 cells in aGVHD allografts

To confirm the outcomes and to adjust for potential confounding factors, we constructed multivariate Cox models to test the proportional hazards assumption and interaction terms with covariates. The variables included in the Cox models had p-values that were smaller than 0.1 after the univariate analyses (Supporting Information Table 1). The final multivariate models were constructed using a forward stepwise selection approach. As shown in Table 3, the dose of Th17 cells in the G-BM, the dose of Tc17 cells in PBPC and the number of HLA locus mismatches emerged as independent factors that influenced the occurrence of aGVHD. Therefore, according to the median of Th17 cells in the G-BM (8.5×104/kg) or Tc17 cells in PBPC (16.8×104/kg), the patients were grouped in the “high Th17 group” and the “low Th17 group” or the “high Tc17 group” and the “low Tc17 group”. Patients in the “high Th17 group” or “high Tc17 group” demonstrated a higher cumulative incidence of aGVHD compared with those in the “low Th17 group” (70±10.14% versus 28.57±10.82%; p=0.005; Fig. 3A) or “low Tc17 group” (76.19±9.88% versus 20±9.2%; p=0.001; Fig. 3B). Furthermore, we found a significant and striking difference in the cumulative incidence of aGVHD between the patients who received a high dose of Th17 and of Tc17 cells (n=12) and those who received a low dose of Th17 and of Tc17 cells (n=12) (91.67±10.40% versus 8.33±8.0%; p<0.0001; Fig. 3C). In addition, an intermediate incidence of aGVHD was observed in patients who received other concentrations of Th17 and Tc17 cells (n=14) compared with those patients who received high concentrations of Th17 and Tc17 cells concomitantly (47.06±12.62 versus 91.67±10.40%; p=0.037), or those who received low concentrations of Th17 and Tc17 cells concomitantly (47.06±12.62 versus 8.33±8.0%; p=0.010; Fig. 3C). The characteristics of the patients among these three groups are compared in Table 1 and Supporting Information Table 2. No significant differences were found, except for the dose of CD3+ T cells, the dose of Th1 cells in the allografts and the gender of the donor. The numbers of HLA locus mismatches were associated with a high risk for aGVHD, and there was a strong bias towards a three-antigen mismatch (50%) in the high Th17 and Tc17 groups compared with the low Th17 and Tc17 groups (17%). Therefore, to deplete the confounding effects of HLA mismatch, or even the three-antigen mismatch, we further analysed the predictive roles of the Th17 dose in G-BM and the Tc17 dose in PBPCs among patients with an identical HLA transplantation (n=18), a haploidentical HLA transplantation (n=23) or a three-antigen mismatch transplantation (n=11). As shown in Supporting Information Figs. 1–3, the cumulative incidence of acute aGVHD was significantly increased in the high Th17 and Tc17 groups than that in the low Th17 and Tc17 groups, irrespective of whether the patients had undergone an identical HLA transplantation (50 versus 0%, p=0.046, Supporting Information Fig. 1), a haploidentical HLA transplantation (100 versus 25%, p=0.019, Supporting Information Fig. 2) or even a three-antigen mismatch transplantation (100 versus 0%, p=0.032, Supporting Information Fig. 3).

Figure 3.

Predictive value of Th17 cells and Tc17 cells in allografts in the occurrence of aGVHD. The cumulative incidence estimates of aGVHD for patients in the “low” and “high” Th17 cell groups were separated according to the median of the Th17 cell dose in the G-BM (8.5×104/kg, A), and the estimates for patients in the “low” and “high” Tc17 cell groups were separated according to the median of the Tc17 cell dose in PBPCs (16.8×104/kg, B). The cumulative incidence of aGVHD were shown among the patients (C) in the “high dose of Th17 and Tc17 group” (n=12), the “low dose of Th17 and Tc17 group” (n=12) and the “other dose of Th17 and Tc17 group” (n=17) (Gray's test were used to determine p-value).

Table 3. Multivariate analysis: factors that affect aGVHDa)
FactorsHR (95% CI)p-Value
  • a)

    a) To avoid potential confounding factors, multivariate Cox proportional hazards models were assessed for interaction terms with covariates. All concentrations of immune cells analysed in the Cox model were continuous variables. The final multivariate models were constructed using a forward stepwise selection approach.

Th17 infused in the G-BM (×104/kg)1.095 (1.032–1.162)0.003
Tc17 infused in the PBPC (×104/kg)1.063 (1.017–1.112)0.008
HLA-locus mismatch (from 0/6 to 3/6)1.84 (1.18–2.87)0.008

Correlation between Th17 and Th1 in allografts

Because there were significant differences in the dose of Th1 infused in the allografts among the three groups, we hypothesised that there might be some correlation between the type 1 and type 17 cells. In the PBPC and G-BM allografts, significant correlations were observed between the doses of Th17 and Th1 cells (Fig. 4A and B), respectively. However, no correlations were detected between the doses of Tc17 and Tc1 cells in the G-BM or PBPC allografts.

Figure 4.

Correlations between the doses of Th17 and Th1 cells in the G-BM (A) or in the PBPC (B) allograft.

Reconstitution levels of Th1, Th17, Tc1 and Tc17 cells on day 30 post-transplantation

The quantitative evaluation of circulating type 1 and type 17 T-cells was performed in the last 32 patients among the 41 patients included in the study according to the transplantation time on day 30 after transplantation and in 12 age- and sex-matched healthy donors (HD). Because the occurrence of aGVHD and the associated therapy affected immune recovery, only 15 of the 30 patients survived without leukaemia and were exempt from aGVHD; we analysed the reconstitution levels of type 1 and type 17 T-cells in patients with or without aGVHD, respectively. The reconstitution of type 1 T-cells was rapid, and the levels of type 1 T-cells recovered to the levels observed in HD during the first month post-transplantation (Fig. 5A and B). However, on day 30 post-transplantation, the reconstitution levels of type 17 T-cells were significantly lower than those determined in the HD (Fig. 5C and D).

Figure 5.

Reconstitution levels of type 1 and type 17 cells on day 30 after transplantation. The reconstitution levels of Th1 (A), Tc1 (B), Th17 (C) and Tc17 cells (D) on day 30 after transplantation are shown among patients without GVHD (n=15) and with aGVHD, including skin aGVHD only (n=12) or skin and gastrointestinal aGVHD (n=5). All of the reconstituted levels were compared to their respective healthy donors (HD, n=12 in type 17 and n=10 in type 1 cells). Th1 (%) or Th17 (%) represent the percentage of CD4+ T cells expressing IFN-γ or IL-17. Tc1 (%) or Tc17 (%) represent the percentage of CD8+ T cells expressing IFN-γ or IL-17. Data represent the mean±SD (Mann–Whitney U test, two-tailed).

The initial appearance of clinical signs of GVHD occurred at a median of 26.5 days post-transplantation (range, 14–90 days). Patients who developed aGVHD demonstrated a greater proportion of type 1 T-cells (p=0.003 for Th1 and p=0.027 for Tc1) on day 30 post-transplantation (data not shown). However, we failed to detect any significant differences between patients with and without aGVHD with respect to the reconstitution levels of type 17 T cells on day 30 post-allo-HSCT (Fig. 5C and D).

Moreover, upon further comparison of the patients according to the aGVHD target organ, greater numbers of Th1 and Tc1 cells were found in patients with skin plus gastrointestinal aGVHD compared with those determined in patients without aGVHD (p=0.0001 for Th1 and p=0.002 for Tc1) and in patients with simple skin aGVHD (p=0.037 for Th1 and p=0.048 for Tc1) (Fig. 5A and B).

Expression of IL-17-producing T cell in the peripheral blood of patients with aGVHD

To detect an association between the frequencies of type17; T-cells and the onset of aGVHD, we collected the peripheral blood of patients at the onset of GVHD, and of whom had achieved partial remission (PR) or complete remission (CR) following treatment. Peripheral blood was also collected from a HD. Our unpublished data suggested that the reconstitution dynamics of IL-17-producing T cells was significantly different at different time points post-transplantation in patients without GVHD. Therefore, we used the HD as a control rather than patients without GVHD. The Th17 populations in patients at the onset of aGVHD showed a tendency towards an increase compared with those determined in the HD (Fig. 6A). In contrast, when patients achieved PR or CR after treatment, the proportion of Th17 cells was significantly reduced compared with that determined at the onset of aGVHD and was lower than that in the HD (Fig. 6A and B). No significant differences between the status of CR and PR were found in the proportions of Th17 cells.

Figure 6.

Analysis of IL-17-producing T cells in the peripheral blood of patients with aGVHD. The phenotype (A) and the individual percentages (B, C) of Th17 and Tc17 cells were evaluated in patients with aGVHD (aGVHD onset, n=11) after treatment and achievement of PR (n=11) and CR (n=11) and in healthy donors (HD, n=12) by intracellular staining for CD3+CD8 T cells (Th17) and CD3+CD8+ T cells (Tc17). Data represent the mean±SD. A Wilcoxon signed-rank test was used to compare the patients between aGVHD onset and PR or CR. The Mann–Whitney U test (two-tailed) was used to compare the aGVHD patients and HD.

The kinetics of the Tc17 cells during the aGVHD treatment phase was the same as those observed for the Th17 cells (Fig. 6A and C). Furthermore, a significant difference in the proportion of Tc17 cells was observed between patients in CR or PR (Fig. 6C).

Effect of G-CSF on the IL-17-producing-T-cell phenotype

To analyse the effect of G-CSF on IL-17-producing T cells, we collected samples from 12 of the donors for the 41 patients who had undergone allo-HSCT and evaluated the percentages of type 17 T-cells in the peripheral blood before and after in vivo G-CSF application (NG-PB and G-PB). Samples were collected from another 7 donors for the 41 patients to analyse the percentages of type 17 T-cells in the bone marrow before and after in vivo G-CSF application (NG-BM and G-BM). Following the in vivo G-CSF application, the percentages of Th17 cells among CD4+ T cells were significantly reduced in the G-BM (p=0.018) and G-PB (p=0.006) compared with those determined prior to the application. Furthermore, the proportions of Tc17 cells in CD8+ T cells were significantly decreased in the G-BM (p=0.028) and G-PB (p=0.007) compared to those determined before the G-CSF treatment (Fig. 7).

Figure 7.

Effect of G-CSF on Th17 and Tc17 expression. Differences were detected in the proportions of Th17 (A) and Tc17 (B) cells in unprimed peripheral blood versus G-CSF-mobilised PBPC products (NGPB versus GPB) or in unprimed BM versus G-BM (NG-BM versus. G-BM). Data represent the mean±SD (percentage) of Th17 cells among the CD4+ T cells or the percentage of Tc17 cells among the CD8+ T cells. The Wilcoxon signed-rank test was used.

Discussion

The important findings of the current study include the following: (i) the dose of IL-17-producing T cells infused in the allograft is associated with the occurrence of aGVHD following transplantation; (ii) IL-17-producing T cells are involved in the clinical course of GVHD after transplantation; (iii) G-CSF could aid in reducing the occurrence of aGVHD through a reduction of IL-17 secretion by T cells.

Previous reports have demonstrated contradictory results concerning the ability of Th17 cells to induce aGVHD 12, 16, 19, 20, which suggests a protective or an exacerbating effect on aGVHD depending on the mouse model tested. However, the ability of Th17 cells infused in allografts to induce aGVHD has not been explored in human transplantation. In the present study, donor-derived Th17 cells in the allografts emerged as the independent risk factor during the course of aGVHD, which suggests that IL-17-producing cells might be a type of initiating cell in aGVHD. These results support previous observations demonstrating an association between the presence of IL-17–producing cells and aGVHD, in particular simple skin GVHD, in a mouse model. In accordance with the study reported by Dander et al. 14, the peripheral blood of patients presenting active GVHD showed an increased number of IL-17 lymphocytes after transplantation, and this increase was associated with the clinical course of aGVHD. In contrast, Philippe Ratajczak et al. 17 demonstrated that Th17 cells in human GVHD were not associated with any evidence of severe tissue damage at the onset of disease, but in situ quantification of the Th17/regulatory T-cell (Treg) ratio was a sensitive and specific marker of human GVHD. The effector molecules responsible for Th17-mediated GVHD are not entirely known. Most likely, IL-17 serves as an important effector cytokine in the development of GVHD because it has been shown to cause direct tissue damage, and neutralising IL-17 markedly decreases skin and lung GVHD 12, 14. Furthermore, polarised Th17 cells produce significant levels of TNF-α after restimulation in vitro and in vivo in allogeneic recipients 16. The contribution of TNF-α in Th17-mediated GVHD is further supported by the observation that the administration of TNF-α antagonists significantly reduces the GVHD induced by Th17 cells 16. Moreover, the positive correlations between Th17 cells and Th1 cells in allografts and the association between Th1 and GVHD by day 30 after transplantation suggested that Th17 and Th1 cells play a cooperative or synergetic function in the pathogenesis of aGVHD. The biological effects of Th17 cells promote a cytokine imbalance towards a Th1-type immune response, which may induce aGVHD. Therefore, it could be postulated that the grafts that go on to develop GVHD contain increased heterogeneity in the populations of primed CD4+ T cells that can secrete multiple cytokines in addition to IFN-α and IL-17. In addition, studies have suggested that a high donor FOXP3-positive Treg content is associated with a low risk for GVHD following HLA-matched allo-HSCT 21. Moreover, experimental data suggest a reciprocal relationship between a Th17-induced pathology and the regulatory role of Tregs. Philippe Ratajczak et al. 17 demonstrated that the Th17/Treg ratio could be a sensitive and specific pathological in situ biomarker for aGVHD. Therefore, the imbalance of the Th17/Treg content in allografts must be clarified.

To date, the effects of Tc17 cells on the occurrence of GVHD have been minimally explored. CD8+ T cells have been shown to be activated in the presence of the cytokines IL-6 or IL-21 plus TGF-β, and they may develop into IL-17-producing (Tc17) cells 22. These cells display a great suppression of cytotoxic function together with low levels of cytotoxic T lymphocyte markers. Rather, these cells express hallmark molecules of Th17 programming, including retinoic acid receptor-related orphan receptor (ROR)γt, RORα, IL-21 and IL-23R 23. In the present study, compared with Th17 cells, Tc17 cells in the allograft were also part of the dominant T-cell subtype involved in the occurrence of aGVHD, which suggested the presence of a significant collaboration between the dose of Th17 cells in the G-BM and that of Tc17 cells in PBPC in inducing aGVHD. Moreover, the percentages of Tc17 cells were also correlated with the clinical course of GVHD. However, in contrast to the dose of Th17 cells, the dose of Tc17 cells in the allograft did not show any correlation with the other cytokines produced by T cells. Therefore, the mechanism by which Tc17 cells induce GVHD might differ somewhat from that associated with Th17 cells. Further exploration will be crucial in determining these pathways.

However, with improvements in immunosuppressive therapy and supportive care, decreasing numbers of patients develop severe aGVHD; only 2/41 patients developed grade III or IV GVHD in the present clinical study. Meanwhile, although our supporting data demonstrated that the dose of Th17 and Tc17 played predicative roles irrespective of a matched or haploidentical HLA, the risks for aGVHD (donor types ranging from matched to haplo) among the patients were heterogeneous. Therefore, in our subsequent studies we should study more patients to capture a sufficient range of aGVHD severity and to develop firm conclusions based on a more homogeneous patient population demonstrating clinical aGVHD. IFN-γ and IL-17 cytokine secretion are now recognised to be unrestricted to Th1 and Th17 subsets; that is, Th1 cells can produce IL-17 and Th17 cells can secrete IFN-γ 24. An improved operational definition for a Th1 or Th17 cell could be determined using multi-colour flow cytometry to examine multiple cytokines and transcription factors including T-β and ROR-γ.

In a previous study, we have shown that G-CSF can induce bone marrow and peripheral blood T-cell hyporesponsiveness and can polarise T cells from a Th1 phenotype to a Th2 phenotype 25–28. However, the effect of G-CSF on IL-17-producing T cells remains somewhat controversial 29–31. Two groups demonstrated that G-CSF improved IL-17 production by T cells 29, 31. However, one group showed that G-CSF could reduce the levels of T cells with a Th17 phenotype (CD4+IL-17A+CCR6+IL-23R+); this reduction may be more than three times the level detected in normal controls 30. Our results demonstrated that the effect of G-CSF on Th17 and Tc17 expression is inconsistent with the results reported by Geoffrey R Hill et al. 31. In accordance with Han C. Toh et al. 30, the immunophenotype study of Th17 and Tc17 cells demonstrated significant decreases in the bone marrow and peripheral blood following G-CSF application in vivo. Taken together, the effects of G-CSF on the expression of IL-17-producing T cells are not entirely clear. They might be dependent on the different types of T cells (human versus mouse), the different doses of G-CSF used or even the different culture conditions employed in vitro, among other conditions. Therefore, G-CSF-induced immune tolerance could be provided in part through the downregulation of IL-17-producing T cells. Further studies are needed to investigate the possible mechanisms responsible for this phenomenon.

In summary, we demonstrated for the first time that the dose of IL-17-producing T cells in allografts was associated with an increased risk for aGVHD. Moreover, the present results suggested that IL-17-producing T cells were associated with the active status of aGVHD and correlated with the clinical course of this condition. Therefore, careful monitoring of the number of IL-17-producing T cells in the allograft and after transplantation might aid in predicting the occurrence of aGVHD. In the future, the specific targeting of Th17 and Tc17 cell differentiation or function in allografts may provide a new approach to prevent aGVHD and to improve transplant outcomes in transplantation settings with T-cell repletion.

Materials and methods

Patients

We first enrolled ten patients in the study who developed aGVHD during November 2009 to confirm the changes in the dynamics of Th17 and Tc17 cells during GVHD treatment. Next, we investigated the predictive roles of the Th1, Th17, Th2, Tc1, Tc2 and Tc17 doses infused in the allografts on the occurrence of aGVHD. Two different cohorts of patients (total n=51) were studied sequentially.

  • (i)Ten patients with aGVHD onset after haploidentical (n=8) and HLA-identical sibling (n=2) allo-HSCT in November 2009 were monitored for the presence of Th17 and Tc17 cells by flow cytometry of the peripheral blood. Because one patient developed aGVHD twice after transplantation, we documented 11 onsets of GVHD among the 10 patients. Blood samples were collected on the day of GVHD onset/flare and before the beginning of GVHD treatment (n=11), and subsequently, the patients who achieved PR (n=11) or CR (n=11) after GVHD treatment were monitored.
  • (ii)Forty-one patients with haematological malignancies who were undergoing haploidentical (n=23) and HLA-identical sibling (n=18) allo-HSCT between November 2009 and February 2010 were enrolled consecutively in this study. The concentrations of Th1, Th2, Th17, Tc1, Tc2 and Tc17 cells in the allografts were analysed by flow cytometry. Among these 41 patients, the reconstituted levels of the Th1, Tc1, Th17 and Tc17 cells were monitored in the last 32 patients on day 30 after allo-HSCT. All of patients and donors provided written informed consent, and the Institutional Review Board of Peking University Institute of Haematology approved the study.

Transplant procedure

The conditioning, mobilisation, collection of stem cells and the prevention of GVHD have been described previously 32, 33.

G-CSF treatment of healthy donors and sample collection

NG-PB (n=12) or NG-BM (n=7) was obtained from normal donors before treatment with human recombinant G-CSF (filgrastim; Kirin Brewery, Tokyo, Japan). G-BM (n=7) were obtained on the fourth day of treatment by aspiration prior to harvest to avoid contamination by peripheral blood. G-PB (n=12) collections were performed on the fifth day of the G-CSF treatment as described previously.

Detection of IL-17 production by intracellular flow cytometry

Intracellular staining was performed using the Pharmingen Intracellular Staining Kit (BD Pharmingen, San Diego, CA, USA). The cells were incubated for 5 h with phorbol myristate acetate (PMA) (50 ng/mL) plus ionomycin (2.5 μg/mL, all reagents from Sigma Chemical) to stimulate maximal IL-17 production; Golgistop (0.7 μL/mL) was added to the sample during the last 4 h to trap the protein in the cytoplasm. The monoclonal antibodies CD3-Percp, CD8-APC, IL-17A-PE or IL-4-PE and IFN-γ-FITC (BD. Bioscience Mountain View, CA) were used to stain cell surface markers and intracellular cytokines. Th17, Th2 and Th1 cells were identified as CD3+CD8IL-17A+, CD3+CD8IL-4+ and CD3+CD8IFN-γ+, respectively, and Tc17, Tc2 and Tc1 cells were CD3+CD8+IL-17A+, CD3+CD8+IL-4+ and CD3+CD8+IFN-γ+, respectively.

Statistical analysis

A two-sided Mann–Whitney U test was applied to compare the differences in cytokine secretion in T cells between the patients with and without aGVHD. To test the differences in the cytokine secretion in T cells between donors who were treated before and after G-CSF and between patients with aGVHD onset and after treatment to achieve PR or CR status, a Wilcoxon signed-rank test was used. The associations between Th1, Th17, Tc1 and Tc17 cells in the allografts and aGVHD were calculated using cumulative incidence curves to accommodate competing risks. Gray's test was used in the cumulative incidence analyses. The risk factors calculated for the univariate analysis according to the aGVHD status included the recipient and donor age, sex, diagnosis, numbers of HLA locus mismatches between the donor and recipients (0, 1, 2, and 3 locus mismatches), the pre-transplantation risk category, the concentrations of CD3+, CD4+ and CD8+ T cells, CD34+ cells, and the dose of Th1, Th2, Th17, Tc1, Tc2 and Tc17 infused in the G-BM and PBPC allografts. All the concentrations of immune cells and CD34+ cells analysed in the Cox model were continuous variables. To confirm the outcomes and adjust for potential confounding factors, multivariate Cox models were assessed for the proportional hazards assumption, and interaction terms with covariates were tested. The factors included in the Cox model had p-values less than 0.1 in the above univariate analysis. The final multivariate models were constructed using a forward stepwise selection approach. The characteristics of the patients among the groups were compared using the χ2 test for categorical variables and the Mann–Whitney U test for continuous variables. The calculations were performed using the SPSS 13.0 statistical software, and R software was used to calculate the cumulative incidence when considering the presence of competing risks.

Acknowledgements

This work was supported by the National Outstanding Young Scientist's Foundation of China (grant no. 30725038), the National Natural Science Foundation of China (grant nos. 30971292 and 30800485) and the Beijing Novel Program (grant no. 2008B05). American Journal Experts (www.journalexperts.com) provided editorial assistance to the authors during the preparation of this manuscript. The authors thank Xu-Hua Wang for excellent technical support and thank all of the core facilities at the Peking University Institute of Haematology for sample collection.

Conflict of Interest: The authors declare no financial or commercial conflict of interest.

Ancillary