Full correspondence: Prof. Xiao-Jun Huang, Peking University People's Hospital & Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation, No. 11 Xizhimen South Street, Beijing 100044, China
The roles of Th17 cells and IL-21 in the pathogenesis of chronic graft-versus-host disease (cGVHD) in patients who undergo allogeneic transplantation are still unknown. Here, we examined this question by monitoring eight patients with new-onset cGVHD for the presence of Th17 cells, Th1 cells, and IL-21. Allografts from an additional 41 patients were also analyzed for Th17 and Th1 cells. Out of these 41 patients, the last 32 enrolled patients were further analyzed for Th17 cells, Th1 cells, and plasma IL-21 levels at day 30 post-transplantation regarding cGVHD. Th17 cells and IL-21 plasma levels were significantly increased at cGVHD onset and drastically decreased after complete remission. Patients who received a higher number of Th17 cells in their allograft had a higher incidence of cGVHD compared with patients who received a lower number of Th17 cells. Meanwhile, positive plasma IL-21 levels at day 30 post-transplantation predicted cGVHD occurrence. These results suggest that Th17 cells and IL-21 may contribute to the development of cGVHD.
Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is considered an effective treatment method for hematological malignancies, but graft-versus-host disease (GVHD) is a potentially lethal complication after allo-HSCT. Chronic GVHD (cGVHD) is not only a major cause of nonrelapse mortality but also induces substantial morbidity, which can severely affect the quality of life [1, 2]. Despite improvements in the practice of allo-HSCT over the last 30 years, there has been little change in the incidence, morbidity, and mortality of cGVHD. One of the difficulties in combating cGVHD is the poor understanding of its pathogenesis .
Chronic GVHD differs from acute GVHD (aGVHD) in many aspects. cGVHD often presents with clinical manifestations that resemble those observed in autoimmune diseases, such as systemic lupus erythematosus, Sjögren's syndrome, lichen planus, and scleroderma. However, in contrast to aGVHD, it is unclear how autoimmune responses develop in cGVHD [3, 4].
IL-17-producing CD4+ T lymphocytes, named Th17 cells, play a crucial role in triggering inflammation and tissue injury in several autoimmune diseases . Increasingly, the Th17 differentiation pathway has been shown to play important roles in aGVHD [6, 7]. We previously demonstrated that IL-17-producing T cells contribute to aGVHD after allogeneic blood marrow transplantation . However, the role of Th17 cells in cGVHD is controversial [7, 9-11]. Using the B10.D23→BALB/c cGVHD model, Nishimori et al.  show that cGVHD is significantly ameliorated in recipients of IL-17−/− B10. D2 donor T cells compared with recipients of wild-type T cells, suggesting that Th17 cells contribute to cGVHD in this model. However, the (B6→BALB/c)→B6 Rag cGVHD model of Chen et al.  shows that cGVHD is not significantly ameliorated in recipients transplanted with IL-17−/− B6 donor T cells, suggesting that IL-17 is not required for autoimmune-mediated pathologic damage during cGVHD. Furthermore, recent clinical data have also demonstrated contradictory roles for IL-17 as a central mediator of cGVHD pathology [7, 11]. The work of Dander et al.  demonstrates an increased Th17-cell population in patients with cGVHD in addition to an inflammatory process. However, Ratajczak et al.  found that although skin-infiltrating Th17 cells can be identified by immunohistochemistry through IL-17A expression, the Th17/Treg-cell ratio is significantly lower than that observed in non-GVHD controls, which argues against a pathogenic role for the Th17 subset. Therefore, the role of Th17 cells in the pathogenesis of cGVHD must be further explored.
Moreover, a recent study has shown that IL-21, one of the main cytokines produced by Th17 cells , is expressed in aGVHD target organs . Inhibiting IL-21 could decrease disease severity in murine B6→BALB/c aGVHD models, which suggests that anti-human IL-21 mAb may be an effective treatment for suppressing aGVHD in the clinic [14, 15]. However, there are no related reports concerning the relationship between IL-21 and cGVHD in either mouse models or clinical transplantation.
In our study, we first enrolled patients with new-onset cGVHD to determine the presence of Th17 cells, Th1 cells, and plasma IL-21 levels. Next, we analyzed the levels of Th17, Th1, and Th2 cells in allografts, the reconstituted levels of Th17 and Th1 cells, and the plasma level of IL-21 in peripheral blood (PB) on day 30 post-transplantation to explore the association between Th17 cells, Th1 cells, IL-21, and cGVHD. We found that high levels of both Th17 cells and IL-21 correlated with the occurrence of cGVHD. In addition, the number of Th17 cells transferred in an allograft was correlated with the reconstituted Th17 cell numbers on day 30 post-transplantation. In contrast, the number of Th1 cells infused in an allograft and reconstituted levels of Th1 cells on day 30 post-transplantation had no relationship with the occurrence of cGVHD. Therefore, our studies demonstrated that both Th17 cells and IL-21 might contribute to cGVHD induction in this transplantation model.
Eight patients with new-onset cGVHD were first enrolled to monitor the presence of Th17 cells and Th1 cells, plasma IL-17 and IL-21 levels, and the expression of lineage-specific transcription factors (Foxp3, ROR-γ, T-bet, and STAT5). The patient characteristics were as follows: three men and five women, four haplo-identical and four identical transplantations, four extensive and four local cGVHD cases, and a median age of 35 years (range 19–47 years). An additional ten patients without cGVHD who were matched with the previously mentioned eight patients by age, donor type, transplantation type, and time after transplantation were enrolled as controls. Table 1 details the characteristics of the other 41 patients. All patients achieved engraftment and complete donor chimerism after transplantation. Grade I, II, and III aGVHD occurred in ten, eight, and two patients, respectively. As of June 1, 2012, the median follow-up time was 874 days (range 851–935 days) after transplantation. According to the Seattle criteria , cGVHD occurred in 22 patients, including six patients with simple skin cGVHD involving the eye/mouth, four patients with only liver cGVHD, and 12 patients with a combination of skin and liver cGVHD involving the eye/mouth/gastrointestinal tract/lung. In total, ten patients were classified as local cGVHD, and 12 patients were classified as extensive cGVHD. Three patients died from transplantation-related mortality (TRM) because of severe infection, and eight patients relapsed. Of the relapsed patients, four were treated with modified donor lymphocyte infusion (mDLI) and chemotherapy, survived, and re-achieved complete remission (CR), and the remaining four patients died of relapse. The median 2.5 years overall survival (OS) rate was 82.9 ± 5.9%, and the event-free survival (EFS) rate was 63.4 ± 7.5%.
Table 1. Patient characteristics in the low- and high-dose groupsa
High Th17 group
Low Th17 group
Patients were designated as the “high dose” (Th17 dose in the G-BM plus PBSCs >1.67 × 106/kg, n = 20), the “low dose” (Th17 dose in the G-BM plus PBSCs <1.67 × 106/kg, n = 21) according to the median of the Th17 dose in the G-BM plus PBSCs (1.67 × 106/kg).
High risk: patients with MDS, NHL, or a more advanced stage of ALL/AML or CML beyond first complete remission or first chronic phase.
A total of 17 patients received mDLI treatment after transplantation. Five patients were administered prophylactic mDLI due to either nonremission status or because the second complete remission prior to transplantation implied a high risk of leukemia relapse in the absence of GVHD. Therapeutic mDLI was performed in 11 patients as a treatment for minimal residual disease positive (MRD+) or relapse after transplantation. One patient received mDLI because of delayed poor engraftment after transplantation. Among the 17 patients treated with mDLI, ten patients developed cGVHD after transplantation (10/17), which was comparable with patients not treated with mDLI (12/24, p = 0.577).
Correlation between Th17 and the clinical status of cGVHD patients
To detect an association between the frequencies of Th17 cells and the onset of cGVHD, we collected the PB of patients at cGVHD onset who had achieved partial CR following treatment. Patients without cGVHD onset post-transplantation and healthy donors (HDs) were also enrolled as controls. When cGVHD was completely controlled, the percentage of Th17 cells (p = 0.003, Fig. 1A) and Th1 cells (p = 0.036, Fig. 1B) was significantly reduced compared with that observed at cGVHD onset. Meanwhile, the percentage of Th17 cells (p = 0.002) was higher in patients at cGVHD onset than patients without cGVHD post-transplantation but comparable to donor levels (Fig. 1A). However, the percentage of Th1 cells was comparable among patients with active cGVHD, patients without cGVHD and HDs (Fig. 1B).
The kinetics of the plasma IL-17 (p = 0.012, Fig. 1C) and IL-21 levels (p = 0.043, Fig. 1D) during the cGVHD treatment phase were the same as the pattern observed for Th17 cells (Fig. 1C and D). Meanwhile, the plasma IL-17 (p = 0.007, Fig. 1C) and IL-21 (p = 0.043, Fig. 1D) levels were also higher in patients at cGVHD onset than patients without cGVHD post-transplantation but comparable to donor levels (Fig. 1C and D).
Moreover, we evaluated the expression of lineage-specific transcription factors (Foxp3, ROR-γ, T-bet, and STAT5) and presented data as the threshold cycle (Ct). There were no significant differences in the transcript levels of FoxP3, ROR-γ, T-bet, or STAT5 between patients with cGVHD onset and those who achieved CR after treatment (Fig. 1E–I). However, the Ct levels of FoxP3, T-bet, and STAT5 (Fig. 1E, and I) and the ratio of FoxP3 and ROR-γ Ct levels (Fig. 1G) were higher in patients with cGVHD and after cGVHD was controlled than HDs, except for the Ct levels of ROR-γ (Fig. 1F). Because the numerical value of the Ct is inversely related to the amount of amplicon in the reaction (i.e. the lower the Ct, the greater the amount of amplicon) , the transcription levels of FoxP3, T-bet, or STAT5 were lower in patients with cGVHD onset or after cGVHD was controlled than in HDs. Considering that RNA was extracted from PB mononuclear cells without controlling for T-cell numbers, we further analyzed CD4+ and CD8+ T-cell subsets and found that the percentage of CD4+ T cells was lower in patients with cGVHD onset (p = o.oo5) or after cGVHD was controlled (p = o.oo5) than in HDs (Fig. 1J), but the CD8+ T-cell levels were comparable (Fig. 1K).
Correlation between Th17 cells in allograft and post-transplantation with cGVHD occurrence
We compared the percentages and absolute numbers of Th17, Th1, Th2, Tc17, Tc1, and Tc2 cells within CD4+ and CD8+ T-cell populations in the granulocyte colony-stimulating factor (G-CSF) primed bone marrow (GBM) and peripheral blood stem cell (PBSC) allografts of patients with and without cGVHD. Patients with cGVHD had greater proportions of Th17 cells in their GBM and PBSC allografts (Fig. 2A), especially patients with local cGVHD (Fig. 2B) and simple skin cGVHD (Fig. 2C). Therefore, we found significantly higher numbers of Th17 cells in the GBM plus PBSC allografts of patients with cGVHD (Fig. 3A), especially in patients with local cGVHD (Fig. 3C) and simple skin cGVHD (Fig. 3D). However, the numbers of Th1 cells in the GBM plus PBSC allografts of patients with cGVHD were comparable to patients without cGVHD (Fig. 3B). No association was found between Th2, Tc1, Tc2, or Tc17 cells and cGVHD occurrence (data not shown). Furthermore, we failed to find any association between a specific balanced ratio of type 1, type 2, and type 17 effectors and the pattern of cGVHD organ involvement (data not shown).
Of the 41 patients included in the study, the last 32 enrolled patients were quantitatively evaluated for circulating Th1, Th2, Tc1, and Tc17-cell reconstitution and plasma IL-17 and IL-21 levels on day 30 post-transplantation. Patients with cGVHD demonstrated a greater proportion of Th17 cells on day 30 post-transplantation compared with those without cGVHD (p = 0.038, Fig. 4A). Moreover, upon further comparison of the patients according to the cGVHD target organ, greater proportions of Th17 cells were found in patients with local cGVHD (p = 0.009, Fig. 4B), especially skin cGVHD (p = 0.042, Fig. 4C), compared with patients without cGVHD. Meanwhile, patients with skin cGVHD had higher absolute numbers of Th17 cells on day 30 post-transplantation compared with those without cGVHD (p = 0.040, Fig. 4C). No significant differences were observed in the percentages or the numbers of Th17 cells on day 30 post-allo-HSCT between patients with extensive cGVHD and without cGVHD (Fig. 4A–C). The plasma IL17 levels on day 30 post-transplantation were not correlated with the occurrence of total cGVHD (Fig. 4A); however, further comparison showed that plasma IL-17 levels were correlated with limited cGVHD (p = 0.0003, Fig. 4B). No significant differences were observed in the percentages of Th1, Tc1, or Tc17 cells between patients with and without cGVHD (data not shown).
We also retrospectively diagnosed cGVHD patients according to the NIH criteria ; however, four cases who could not be classified as classic cGVHD using NIH criteria because involvement was confined to the liver alone, were removed from the study. Accordingly, two patients were classified as mild cGVHD, six patients were classified as moderate cGVHD, and ten patients were classified as severe cGVHD. As shown in the Supporting Information, the number of Th17 cells in the allograft (Supporting Information Fig. 1A) and at day 30 post-transplantation (Supporting Information Fig. 1B and C) was more associated with mild/moderate cGVHD rather than severe cGVHD.
Higher plasma IL-21 levels on day 30 post-transplantation positively predict cGVHD occurrence
As shown in Figure 4A and B, there were significant differences in serum IL-21 levels on day 30 post-transplantation between patients with and without cGVHD (p = 0.010), mainly between patients without cGVHD versus extensive cGVHD (p = 0.024). Meanwhile, according to cGVHD NIH criteria, day 30 plasma IL-21 levels were particularly associated with severe cGVHD (p = 0.007, Supporting Information Fig. 1D). Based on the lowest plasma IL-21 level detectable by our ELISA kit (48 pg/mL), patients were subgrouped into “IL-21-positive” (≥ 48 pg/mL, n = 16) and “IL-21-negative” (<48 pg/mL, n = 16) groups. The cumulative incidence of overall cGVHD (p = 0.0017), especially extensive cGVHD (p = 0.0074), was significantly higher in the IL-21-positive group than the IL-21-negative group (Fig. 4D). However, there were no differences in the relapse, TRM, OS, or EFS rates between patients in the IL-21-positive and IL-21-negative groups (data not shown). Furthermore, the number of Th17 cells infused in the allograft was significantly correlated with the reconstituted Th17 cell numbers on day 30 post-transplantation (p = 0.040, Fig. 5). However, there was no relationship between the proportions of Th17 cells in allografts or the reconstituted levels of Th17 cells (regardless of the proportion or absolute number of Th17 cells) and plasma IL-21 levels on day 30 post-transplantation (data not shown).
Predictive value of the Th17 cell dose in allografts in cGVHD
To confirm the predictive value of Th17 in clinical outcomes and adjust for potential confounding factors, we constructed multivariate Cox models. The final multivariate models were constructed using a forward stepwise selection approach. The parameter information considered is shown in the statistical analyses section. The variables entered in the Cox models had p-values less than 0.1 in the univariate analyses. As shown in Table 2, the Th17-cell dose in the allografts and IL-21 groups (positive versus negative) emerged as independent factors that influenced cGVHD occurrence. In addition, IL-21 grouping (positive versus negative) was a single independent risk factor for extensive cGVHD.
Table 2. Multivariate analysis: factors that affect cGVHDa
Clinical outcomes and factor
HR (95% CI)
To avoid potential confounding factors, multivariate Cox proportional hazards models were assessed for interaction terms with covariates. All concentrations of immune cells analyzed in the Cox model were continuous variables. The final multivariate models were constructed using a forward stepwise selection approach. Factors included in the Cox model include Th17 group, pretransplantation disease risk category, II-IV aGVHD, Th17 levels on day 30 post-transplantation and IL-21 positive versus negative in plasma at day 30 after transplantation.
Th17 group (high versus low)
Pretransplantation disease risk (high risk versus low risk)
IL-21 (positive versus negative)
IL-21 (positive versus negative)
Based on the median Th17-cell numbers in the allograft (1.67 × 106/kg), patients were subgrouped into either the “high Th17” group or “low Th17” group. Patients in the high Th17 group demonstrated a higher cumulative incidence of cGVHD compared with patients in the low Th17 group (75.0 ± 10.31% versus 33.33 ± 10.61%; p = 0.018; Fig. 6A). However, there were no differences in the cumulative incidence of extensive cGVHD between the high Th17 and low Th17 groups (p = 0.94; data not shown). The patient characteristics of these two groups are compared in Table 1. No significant differences were found, with the exceptions of the CD3+ T-cell dose in the allografts, sex of the donor, and occurrence of II-IV aGVHD.
Moreover, we analyzed the association between the Th17 numbers in allografts and the relapse, TRM, OS, and EFS rates. As shown in Figure 6B, patients in the high Th17 group tended to have a relatively lower cumulative incidence of relapse and TRM post-transplantation. Therefore, the OS and EFS rates tended to be higher in patients in the high Th17 group than patients in the low Th17 group (Fig. 6C and D).
In this study, we observed the following important findings: (i) both the absolute number of Th17 cells in allografts and the percentage of Th17 cells at day 30 post-transplantation are correlated with the development of cGVHD after transplantation; and (ii) plasma IL-21 levels at day 30 post-transplantation could predict subsequent cGVHD development. Our results suggest that both Th17 and IL-21 might contribute to cGVHD development in this transplantation model.
Recent preclinical and clinical data support a role for IL-17A as a central mediator of cGVHD pathology, particularly within the skin . In this study, our results showed that Th17 cells and Th1 cells are associated with the clinical course of cGVHD, which is in accordance with the study reported by Dander et al. , which showed that the PB of patients presenting with active cGVHD showed an increased number of IL-17 lymphocytes after transplantation and that this increase was associated with the clinical course of aGVHD. In another recent study, Nishimori et al.  showed that the infusion of IL-17-deficient T cells could attenuate cGVHD in the skin and salivary glands in a cGVHD model, which suggests that Th17 cells contribute to cGVHD development. In contrast, Ratajczak et al.  demonstrated that the percentage of Th17 cells in the PB of 31 patients at a mean of 3 months post-transplantation was not associated with any evidence of severe tissue damage at cGVHD onset. However, in situ quantification of the Th17/regulatory T (Treg) cell ratio showed that a low Th17/Treg cell ratio was correlated with severe clinical and pathologic GVHD, which argued against a pathogenic role for the Th17 subset. In addition, Chen et al.  showed in a different cGVHD model that donor IL-17-deficient T cells did not abrogate cGVHD pathology because Th1 responses were preserved in the absence of donor IL-17. Therefore, the authors suggested that IL-17 is not required for autoimmune-mediated pathological damage during cGVHD. The reasons for these apparently discordant results are currently unclear, but they may reflect differences in the (i) cGVHD mouse model/transplantation and GVHD prophylaxis; (ii) sample type and time-points post-transplantation; (iii) Th1-cell response; or (iv) involvement of Treg cells. The substantial differences between our study and that of Ratajczak et al.  are the relatively greater number of cGVHD patients (22 cGVHD patients here compared with 3 cGVHD patients in ) and the different time-points assayed (allograft and day 30 post-transplantation here compared with 3 months post-transplantation in ) to analyze the association between Th17 cells and cGVHD development. Litjens et al.  demonstrated in vitro that the repertoire of alloreactive CD4+ T cells is biased to a Th17-cell response, with an average of 24% of alloreactive CD154+ CD4+ memory T cells producing IL-17 after polyclonal stimulation. Thus, the Th17-cell subset might play a prominent role in the early pathogenetic mechanisms of human cGVHD. In addition, the difference in Th1-cell responses between studies may produce contradictory results concerning the ability of Th17 cells to induce cGVHD. Given the plasticity of the Th17 lineage and the ability of these cells to convert to conventional Th1 cells [5, 21], Th17 pathways may play more important roles in cGVHD . Furthermore, considering that Treg cells and Th17 cells were verified to be inversely correlated , the conflicting roles of Th17 cells in cGVHD development in the previous study might be explained by the involvement of Treg cells in cGVHD. The mouse model used by Chen et al.  demonstrated that the absence of Treg cell control in Th1 and Th17 cells is responsible for autoimmune-mediated pathology in cGVHD. Considering that the ratio of Treg cells to Th17 cells has been implicated in GVHD pathology , careful monitoring of the reconstitution kinetics of Th17, Treg, and Th1 cells and the routine evaluation of FoxP3 expression post-transplantation might aid in predicting cGVHD occurrence. Therefore, a prospective study to routinely monitor FoxP3/ROR-γexpression by RT-PCR and Treg/Th17 cells levels by flow cytometry is necessary. Finally, in accordance with our previous work [23, 24], the incidence of cGVHD was not increased after mDLI compared with the non-mDLI group. Therefore, we ruled out an effect of mDLI on cGVHD. Moreover, mDLI treatment was performed after 1 month post-transplantation, and therefore, mDLI had no influence on the expression of Th17 cells in the allograft and PB by day 30 post-transplantation. Furthermore, the use of G-CSF-mobilized DLI combined with short-term immunosuppressants for GVHD prophylaxis may play a role in reducing cGVHD occurrence . The effect of in vivo G-CSF application on T-cell hyporesponsiveness has been extensively explored [25-28]. In addition, our previous data also indicate that in vivo G-CSF application could induce a decrease in Th17 cells .
Although the kinetics of Th17 cells, Th1 cells and plasma IL-17, and IL-21 levels coincided with the activity of cGVHD after transplantation, only Th17 cells and plasma IL-21 levels could predict the occurrence of cGVHD. While Th17 cells predicted only the occurrence of local cGVHD, plasma IL-21 levels were particularly associated with the occurrence of extensive cGVHD. This phenomenon might suggest that (i) the pathogenesis of local and extensive cGVHD is different; (ii) as shown in the cGVHD mouse model, Th17 cells may be involved in skin and lung infiltration and are therefore mainly associated with local cGVHD ; and (iii) IL-21 is not only generated by Th17 cells  but could also be secreted by other memory/effector T cells or NK T cells . IL-21 signals through the IL-2Rc and IL-21R complex [30, 31] to promote the activation, differentiation, maturation, and expansion of NK cells, B cells, CD8+, and CD4+ T cells, dendritic cells, and macrophages and could therefore initiate extensive cGVHD. Thus, Th17 is associated with limited or mild/moderate cGVHD, and local cGVHD had been demonstrated to induce a graft-versus-leukemia effect rather than increasing TRM after transplantation [32, 33], therefore translating into a better outcome. In contrast, higher levels of IL-21 are mainly associated with the development of extensive cGVHD without any benefit in terms of final outcome. Care should therefore be taken in the selective targeting of Th17 cells. Specifically targeting IL-21 might be a safer and more effective method to prevent extensive cGVHD after allogeneic HSCT.
To consider the impact of factors that have previously been shown to influence cGVHD, we considered the potential influence of prophylaxis on GVHD incidence (with or without using in vivo T-cell depletion) and the occurrence of aGVHD. No difference was found when these groups were analyzed, except for the occurrence of aGVHD. Previous studies have demonstrated that a history of aGVHD is the greatest risk factor for cGVHD, and strategies to prevent aGVHD may help to prevent cGVHD [34, 35]. However, scleroderma, the primary manifestation of cGVHD, can also develop de novo late after BMT in the absence of prior clinical aGVHD . After demonstrating that IL-17-producing T cells in allografts were associated with an increased risk for aGVHD, our study further confirmed that Th17 cells were also involved in the pathogenesis of cGVHD after over 2 years of post-transplantation follow-up. However, because aGVHD and cGVHD are two different conditions, the threshold to predict aGVHD could be different from cGVHD. A multivariate analysis in our previous study  showed that the median number of Th17 cells in the GBM (8.5 × 104/kg) or Tc17 cells in PBSCs (16.8 × 104/kg) could better predict the risk of acute GVHD. However, the above-mentioned threshold was not associated with the occurrence of chronic GVHD, but the threshold of Th17 cells in the present data (1.67 × 106/kg) was not only correlated with chronic GVHD but also associated with acute GVHD (Supporting information Table 1 – from our previously published paper ). Therefore, Th17 cells might play a prominent role in the transition from acute to chronic GVHD in human transplantation.
In summary, we demonstrate that both Th17 cells and IL-21 are associated with an increased risk of cGVHD. Th17 cells and plasma levels of IL-21 are associated with active cGVHD and correlate with the clinical course of this condition. Therefore, Th17 cells and IL-21 may contribute to the development of cGVHD.
Materials and methods
Three cohorts of patients (total n = 59) were studied sequentially. We first enrolled eight patients who developed cGVHD between March and November 2011 to confirm the dynamics of changes in Th17 and Th1 cells by flow cytometry and changes in plasma IL-17 and IL-21 levels by ELISA in PB during the course of cGVHD. Blood samples were collected on the day of cGVHD onset/flare before beginning cGVHD treatment (n = 8) and when the patient achieved CR (n = 8) after cGVHD treatment. Subsequently, an additional 41 patients undergoing haploidentical (n = 23) and HLA-identical sibling (n = 18) allo-HSCT between November 2009 and February 2010 were enrolled in this analysis after being originally selected using a protocol exploring the contribution of Th17 cells to aGVHD . After over 2 years of follow-up, we further investigated the predictive role of the Th17 dose in allografts in the occurrence of cGVHD in 41 patients undergoing transplantation. The characteristics of the 41 patients are detailed in Table 1. Th1, Th2, and Th17 concentrations in the allografts were analyzed by flow cytometry. Among these 41 patients, the reconstituted levels of Th1 and Th17 cells and plasma levels of IL-17 and IL-21 on day 30 after allo-HSCT were further monitored by flow cytometry or ELISA in the final 32 patients. Meanwhile, an additional ten patients without cGVHD onset post-transplantation matched by age, sex, disease, transplantation type, and post-transplantation time were enrolled as without cGVHD controls. In addition, age- and sex-matched HSC donors were included as healthy donor controls. All patients and donors provided written informed consent, and the Institutional Review Board of the Peking University Institute of Hematology approved the study.
Transplant procedure and definition
The conditioning, mobilization, and collection of stem cells as well as GVHD prevention were conducted as previously described [37-40]. All patients received a myeloablative regimen, and conditioning was performed as follows. In HLA-matched sibling transplants , patients received a regimen consisting of 80 mg/kg hydroxyurea orally on day 10, 2 g/m2/d cytarabine intravenously on day 9, 3.2 mg/kg/d busulfan intravenously on days 8 to 6 pretransplantation, 1.8 g/m2/d cyclophosphamide intravenously on days 5 to 4 pretransplantation, and 250 mg/m2 of methyl-N-(2-chloroethyl)-N′-cyclohexyl-N-nitrosourea orally on day 3. In HLA-haploidentical donor transplants, patients received a regimen similar to HLA-matched patients, except for the addition of 4 g/m2/d cytarabine on days 10 to 9 and 2.5 mg/kg/d antithymocyte globulin (SangStat) intravenously on days 5 to 2 pretransplantation [37, 38].
Donors received 5 μg/kg rhG-CSF (Filgrastim) daily for 5 to 6 days. On the fourth day, bone marrow cells (G-BMs) were harvested. The target total nucleated cell count was 3.0 × 108 (median, 3.0 × 108; range, 0.98–5.78 × 108) cells/kg recipient weight. On the fifth and sixth days, peripheral blood progenitor cells (PBSCs) were collected. The target mononuclear cell count was 3.0 × 108 (median, 6.51 × 108; range, 1.7–14.35 × 108) cells/kg recipient weight. The fresh and unmanipulated GBMs and PBSCs were infused into the recipients on the day of collection.
GVHD prophylaxis included cyclosporine A and short-term methotrexate with mycophenolate mofetil (MMF) [37, 38]. Cyclosporine was started intravenously on day 9 at a dosage of 2.5 mg/kg and switched to an oral formulation as soon as the patient was able to take medication after the graft. The dosage was adjusted for blood levels. A 0.5 g of MMF was administered orally every 12 h from 9 days before transplantation until day 30 after transplantation. MMF was then given at 0.25 g twice daily for 1 to 2 months. The dosage of methotrexate was 15 mg/m2 administered intravenously on day 1 and 10 mg/m2 administered on days 3, 6, and 11 after transplantation.
Bone marrow aspirations were performed 1, 2, 3, 4.5, 6, 9, and 12 months after transplantation and at 6-month intervals thereafter to assess the graft. To detect donor chimerism, HLA genotyping and DNA fingerprinting (short tandem repeat) were performed. All subjects were monitored for MRD after transplantation using LAIPs, WT1, BCR/ABL, and ETO as previously described [40-42]. Modified DLI was administered when patients without aGVHD showed evidence of recipient chimerism or MRD detected by molecular, cytogenetic, or hematologic methods following a trial of immunosuppressant withdrawal. The mDLI regimen consisted of G-CSF-primed PBSCs instead of harvested unprimed donor lymphocytes and short-term immunosuppressive agents [23, 43, 44].
The diagnosis and grading of GVHD were performed according to published criteria [16, 45]. Leukemia relapse was scored as BM, extramedullary, or both according to common morphological criteria. EFS was defined as the time elapsed from the initiation of treatment (with imatinib, mDLI, chemotherapy, or allo-HSCT) to the appearance of one of the following events: the absence of hematologic response at 3 months; loss of previously obtained CHR, MCR, or CCR; post-transplantation MRD+; relapse in AP or BP; or death from any cause.
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 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-allophycocyanin, IL-17A-PE, IL4-PE, and IFN-γ-FITC (BD Bioscience, Mountain View, CA, USA) were used to stain cell surface markers and intracellular cytokines. Th17, Th2, and Th1 cells were identified as CD3+CD8−IL-17A+, CD3+CD8−IL-4+, and CD3+CD8–IFN-γ+, respectively, and Tc17, Tc2, and Tc1 cells were CD3+CD8+IL-17A+, CD3+CD8+IL-4+, and CD3+CD8+ IFN-γ+, respectively. The percentage of Th17 cells was classified as the percentage of Th17 cells of CD4+ T cells, and the quantity of Th17 cells was classified as the absolute number of Th17 cells in PB (cells/μL).
IL-17 and IL-21 ELISA
To quantify human IL-17 and IL-21 levels in PB, plasma samples (stored at −40°C) were collected from 32 patients on day 30 post-transplantation (n = 32), eight additional patients pairs with new-onset cGVHD and after cGVHD was completely controlled, ten patients without cGVHD and the 24 healthy donor controls. Plasma IL-17 and IL-21 levels were detected in a total of 82 patients by ELISA using the human IL-17 and IL-21 ELISA Development Kit from PeproTech (PeproTech, Rocky Hill, NJ, USA). The lowest and highest quantification limits of IL-21 were 48 pg/mL and 355 pg/mL, respectively, when the assay was performed according to the manufacturer's instructions.
ROR-γ, T-bet, FoxP3, and STAT5 RT-PCR
Because STAT5, ROR-γ, T-bet, FoxP3 are key transcription factors that regulate Th17, Th1, and FoxP3+ Treg cells, we assayed their transcript levels during cGVHD treatment. Total RNA was isolated from mononuclear cells from the eight patients at cGVHD onset and after cGVHD was completely controlled using standard Trizol protocols (Invitrogen). RNA was amplified with the GoldScript cDNA Amplification Kit (Invitrogen) according to the manufacturer's protocol. RNA was subjected to quantitative real-time PCR using the TaqMan One-Step RT-PCR Master Mix reagent (Applied Biosystems). Relative transcript levels were analyzed by real-time PCR in a 20 μL reaction volume in 96-well plates using an ABI 7500 real-time PCR system (Applied Biosystems). Transcript levels were normalized to GAPDH transcript levels. All experiments were performed in duplicate and repeated at least three times.
A two-sided Mann–Whitney U test was applied to compare the percentage or dose of Th1, Th17, and Th2 cells and IL-21 levels between patients with and without cGVHD. To test the differences in the percentage of Th17 and Th1 cells and plasma IL-21 levels between cGVHD onset and cGVHD CR, a Wilcoxon signed-rank test was used. Associations between Th17-cell number or the percentage or number of Th17 cells and plasma IL-21 levels at day 30 post-transplantation and cGVHD were calculated using cumulative incidence curves to accommodate competing risks. Gray's test was used in the cumulative incidence analyses. Statistical associations between Th17 numbers in allografts and Th17 recovery or IL-21/IL-17 plasma levels after transplantation were investigated using linear regression analysis. The risk factors included in the univariate analysis according to cGVHD status were recipient and donor age; sex; diagnosis; number of HLA locus mismatches between the donor and recipients (zero, one, two, and three locus mismatches); pretransplantation risk category; dose of CD3+, CD4+, CD8+, CD34+, Th1, Th2, and Th17 cells infused in the allografts (GBM+ PBSCs); prophylaxis on GVHD incidence (with or without using in vivo T-cell depletion); the occurrence of acute GVHD; the percentage of Th17 cells; the plasma levels of IL-21 by day 30 post-transplantation; and the initiation of mDLI treatment. To confirm outcomes and adjust for potentially confounding factors, multivariate Cox models were used to assess the proportional hazards assumption, and interaction terms with covariates were tested. The factors included in the Cox model had p-values of less than 0.1 in the above univariate analysis. The final multivariate models were constructed using a forward stepwise selection approach. All of the immune cell doses and CD34+ cell doses analyzed in the Cox model were continuous variables. The patient characteristics were compared among the groups using the χ2 test for categorical variables and Mann–Whitney U test for continuous variables. The calculations were performed using SPSS 13.0 statistical software, and R software was used to calculate the cumulative incidence considering the presence of competing risks.
This work was supported by the National Natural Science Foundation of China (grant numbers 81270644 and 81230013, the names of the principal funding recipients were Xiao-Jun Huang and Xiang-Yu Zhao, respectively), the Beijing Novel Program (grant number 2008B05, the name of the principal funding recipient was Xiang-Yu Zhao), the Beijing Natural Science Foundation (grant number 7113171, the name of the principal funding recipient was Xiang-Yu Zhao), doctoral funding of The Ministry of Education of China (grant numbers 200800011087 and 20110001110039, the names of the principal funding recipients were Xiang-Yu Zhao and Xiao-Jun Huang, respectively) and the Major State Basic Research Development Program of China (973 Program Number 2013CB733700, the name of the principal funding recipient was Kai-Yan Liu). The authors thank all the core facilities at the Peking University Institute of Hematology for sample collection.
Conflict of interest
The authors declare no financial or commercial conflict of interest.