Discussion
We found MRD test results and GvHD-guided multiple consolidation chemotherapy and DLIs reduced CIR and improved LFS and survival compared with historical controls in patients with relapsed acute leukemia after allo-hematopoietic stem cell transplantation (HSCT). This was probably due to that the use of multiple consolidation chemotherapy and DLIs after induction chemotherapy and DLI could make more patients achieve and maintain a persistent negative MRD test. Although induction chemotherapy and DLI could make patients achieve complete remission in patients with relapsed acute leukemia after transplant, only 55 % of patients achieved a negative MRD test, and 54 % of these patients had subsequent positive MRD tests. But, after multiple consolidation chemotherapy and DLIs, 81 % of patients finally maintained a negative MRD test, compared with a 9 % of negative MRD test rate in historical control (
P < 0.0001). Many studies already report a positive MRD test is associated with an increased relapse risk post-transplant [
5‐
7,
12,
18,
19]. As well, based on the results of MRD tests, preemptive use of DLI could make patients achieve a negative MRD test and prevent relapse post-transplant in patients with standard risk acute leukemia [
8]. Besides, multiple consolidation chemotherapy and DLIs also induced chronic GvHD and stronger graft-vs.-leukemia (GvL) effects. Our present study suggested that the incidence of chronic GvHD in study group was significantly higher than that in historical control (52 vs. 33 %,
P = 0.039), as well as, the incidence of chronic GvHD post-DLI increased gradually with increased numbers of courses of consolidation chemotherapy and DLI (
P = 0.002). Many researchers have already demonstrated that the development of chronic GvHD post-DLI was a favorable factor of CIR and survival in patients with relapsed acute leukemia after allo-HSCT [
3,
4,
20,
21]. Therefore, all of these data suggested that the application of multiple consolidation chemotherapy and DLIs after induction chemotherapy and DLI could prevent the second relapse in patients with relapsed acute leukemia post-transplant, by inducing chronic GvHD and maintaining the negative MRD test.
Severe GvHD is a major risk of multiple consolidation chemotherapy and DLIs, which is usually correlated with higher TRM. However, we found that the incidences of ≥grade 2 acute GvHD and ≥grade 3 acute GvHD in study group were all similar to that in historical control (
P = 0.149 and
P = 0.366). And ultimately, there was no significant difference in the incidence of TRM between two groups (8.8 vs. 6.4 %,
P = 0.064). A probable reason is due to the application of immunosuppressive agents for 2–4 weeks in patients receiving DLI from HLA-identical related donors and for 6 weeks in patients receiving DLI from haploidentical related donors or unrelated donors. Our previous study suggested that the duration of immunosuppressive agents used after DLI was the only risk factor for the development of ≥grade 3 acute GvHD after DLI (
P = 0.018) and the cumulative incidence of ≥grade-3 acute GvHD in patients receiving immunosuppressive agents for ≥6 weeks was only 9.3 % [
22]. Besides, the application of immunosuppressive agents for ≥6 weeks after DLI from haploidentical related donors and 2–4 weeks after DLI from HLA-identical related donors could preserve GvL effects at the same time could reduce the incidence of ≥grade 3 acute GvHD after DLI [
10,
11]. As well, our present study also suggested that cumulative incidence of ≥grade 2 acute GvHD and ≥grade 3 acute GvHD were not significantly associated with numbers of courses of consolidation chemotherapy and DLI (
P = 0.17 and
P = 0.77), although cumulative incidences of chronic GvHD and ≥moderate chronic GvHD increased gradually with increased numbers of courses of consolidation chemotherapy and DLI (
P = 0.003). Another probable reason is due to the use of G-CSF mobilized peripheral blood cells instead of unstimulated lymphocytes. Huang et al. [
23,
24] reported that the application of G-CSF may modulate the polarization of T cells from a Th1 to a Th2 phenotype and indirectly induce T-cell hypo-responsiveness through the selective increase of DC2 cells and monocytes and the down-regulation of the CD28/B7 co-stimulatory signal. Moreover, Morris et al. [
25] confirmed that using G-CSF during blood cell mobilization augments NK-T-cell-dependent CD8
+ cytotoxicity and purportedly separates GvHD and GvL. Our previous study also suggested that compared with chemotherapy and DLI with unstimulated lymphocytes, chemotherapy and G-CSF mobilized peripheral blood cells infusion tended to be associated with a higher complete remission rate (7/9 vs. 3/5) and lower incidence of ≥grade 3 acute GvHD (0/9 vs. 1/5) in patients with relapsed hematological malignancies after HLA-identical related HSCT [
26]. Therefore, although trial subjects received multiple consolidation chemotherapy and DLIs, no subjects died of GvHD. Another risk of multiple consolidation chemotherapy and DLIs is pancytopenia, which usually leads to higher infection-related mortality. Raiola AM et al. [
21] found that chemotherapy and DLI with unstimulated lymphocytes was typically associated with a higher incidence of pancytopenia (18/100) and a longer duration of pancytopenia (median duration, 90 days). However, chemotherapy and G-CSF mobilized peripheral blood cells infusion was associated with a lower incidence of pancytopenia (2/57) [
27]. In this study, all trial subjects achieved neutrophil recovery with a median time of 17 day, 45 trial subjects achieved platelet recovery with a median time of 20 days, and only 3 subjects died of infection. All the results confirmed the safety of chemotherapy and DLI which used G-CSF mobilized peripheral blood cells followed by CSA or MTX to prevent GvHD. As well, due to the improvement of diagnosis and treatment of GvHD and infectious disease post-transplant, it also ensured the successful use of multiple consolidation chemotherapy and DLIs. Moreover, because multiple consolidation chemotherapy and DLIs were given based on the results of MRD test and GvHD post-DLI, therefore, it avoided unnecessary TRM while could prevent second relapse in patients with relapsed acute leukemia post-transplant.
Besides, although multiple consolidation chemotherapy and DLIs were given, nine patients were still in a positive MRD test and seven out of nine patients relapsed again post-DLI. Mo et al. [
9] also found that early-onset positive MRD test after DLI and persistent positive MRD test after DLI were usually associated with increased relapse risk (
P = 0.001) and impaired LFS (
P = 0.004). Therefore, for those patients with a positive MRD test post-DLI, more intensive therapy should be used so as to make them maintain a persistent negative MRD test. Mo et al. [
28] found that in patients who were in a positive MRD test post-transplant, compared with DLI, interferon-α (IFN-α) could provide comparable rate of negative MRD test, relapse rate (27.3 % vs. 35.6 %,
P = 0.514) and disease-free survival (68.2 vs. 60.0 %,
P = 0.517). Furthermore, even in patients who had unsatisfied response to DLI, IFN-α could also provide a decreased level of MRD or a negative MRD test [
29]. Besides, some articles have already suggested that second transplant could offered a chance of stable remission for patients with acute leukemia relapsing post-transplant and provided a 25–30 % of survival [
30,
31]. Moreover, our previous study found that in 24 patients receiving second transplant to treat relapsed acute leukemia post-transplant, although 12 patients had no response to chemotherapy and DLI, second transplant still provided 55.7 % of CIR, 35.1 % of non-relapse mortality and 30.9 % of survival [
32]. Therefore, for patients who had a persistent positive MRD test after two courses of consolidation chemotherapy and DLIs, DLI followed by IFN-α or second transplant were probably alternative methods. This needed to be investigated in future.
In addition, in this study, the complete remission rate (67 %, 47/60) is higher than that in article published before [
3]. This is maybe due to the application of chemotherapy before DLI. Collins RH et al. [
33] reported that complete remission rate in AML (
n = 39) and ALL (
n = 11) patients who had not received chemotherapy before DLI were only 15.4 and 18.2 %, respectively. But, in the article from Schmid C et al. [
3], 75 % AML patients received chemotherapy before DLI and 35 % patients achieved complete remission. Besides, our previous study also suggested that chemotherapy and DLI provided a 64 % of complete remission rate in subjects with relapsed acute leukemia post-transplant [
4]. Another reason is maybe due to the infusion of G-CSF mobilized peripherial blood cells infusion instead of unstimulate donor lymphocytes. Morris et al. [
25] confirmed that using G-CSF during blood cell mobilization augments NK-T-cell-dependent CD8
+ cytotoxicity. As well, our previous study suggested that compared with chemotherapy and DLI with unstimulated lymphocytes, chemotherapy and G-CSF mobilized peripherial blood cells infusion tended to be associated with a higher complete remission rate (7/9 vs. 3/5) in patients with relapsed acute leukemia after allo-HSCT [
26]. In addition, Levine JE et al. [
34] also reported that chemotherapy and DLI with G-CSF mobilized peripherial blood cells infusion could make 27 patients (27/57, 47 %) achieve complete remission.
Of course, there are some limitations to interpreting data from our study. The most important one is that historical cases were used as control, rather than randomized control. But, most variable were comparable between study group and historical control except for donor types (P = 0.037). In study group, more subjects receive a HLA-haplotype-matched transplant compared with historical control (66 % vs. 56 %, P = 0.037). However, the impact of this imbalance should be to increase rather than decrease the difficulties in interventions. As well, because multiple consolidation chemotherapy and DLIs were given based on the results of MRD tests and GvHD, therefore, it is difficult to perform randomized control study.