Prognostic significance of programmed death ligand 1 expression in adult patients with de-novo acute myeloid leukemia

Nevine N Mostafa; Emad A Abdelmohsen; Amro M.S El-Ghammaz; Alia M Saeed; Mohammed T Hamza

doi:10.4103/ejh.ejh_27_18

ORIGINAL ARTICLE

Year : 2018 | Volume : 43 | Issue : 4 | Page : 158-165

Prognostic significance of programmed death ligand 1 expression in adult patients with de-novo acute myeloid leukemia

Nevine N Mostafa¹, Emad A Abdelmohsen¹, Amro M.S El-Ghammaz¹, Alia M Saeed¹, Mohammed T Hamza²
¹ Department of Internal Medicine, Clinical Hematology and Oncology Division, Faculty of Medicine, Ain Shams University, Cairo, Egypt
² Department of Clinical Pathology, Faculty of Medicine, Ain Shams University, Cairo, Egypt

Date of Submission	05-Aug-2019
Date of Acceptance	06-Aug-2018
Date of Web Publication	10-Apr-2019

Correspondence Address:
Alia M Saeed
Department of Internal Medicine, Clinical Hematology and Oncology Division, Faculty of Medicine, Ain Shams University, Cairo, 11711
Egypt

Source of Support: None, Conflict of Interest: None

Check

DOI: 10.4103/ejh.ejh_27_18

Abstract

Context Acute myeloid leukemia (AML) exhibits one of the therapeutic challenges to the clinician owing to unsatisfactory outcomes obtained by conventional chemotherapy protocols. Immune checkpoints have gained attention in the recent years in the field of oncology as a presumable mechanism of cancer to evade immunity, but their status in AML has yet to be investigated.
Aims The aim was to measure programmed death ligand-1 (PDL-1) expression on the blast cells in patients with de novo AML at time of diagnosis, followed by investigating its relationship to different patients’ characteristics as well as disease prognostic variables and therapy outcomes.
Setting and design A total number of 40 adult patients with de-novo AML were recruited.
Materials and methods Surface expression of PDL-1 on the blast cells was evaluated by multicolor flow cytometry. The collected data were revised, coded, tabulated, and introduced to a PC using IBM SPSS version 20.0.
Results PDL-1 has been expressed amongst the study cohort with a mean expression of 43.01±24.72. PDL-1 expression was not different among different risk categories of the disease and did not influence the therapeutic response. Despite a higher PDL-1 expression in refractory cases in comparison with responders, being 68.9 and 43.4%, respectively, this did not reach a statistical significance.
Conclusions PDL-1 expression did not show a discernible relationship with any patients’ or disease parameters. Moreover, it did not influence patients’ response to treatment or survival. Refractory cases displayed higher expression, but they were too few to draw statistical inferences, with the need of a more ample sample size.

Keywords: de-novo acute myeloid leukemia, outcome, programmed death ligand 1

How to cite this article:
Mostafa NN, Abdelmohsen EA, El-Ghammaz AM, Saeed AM, Hamza MT. Prognostic significance of programmed death ligand 1 expression in adult patients with de-novo acute myeloid leukemia. Egypt J Haematol 2018;43:158-65

How to cite this URL:
Mostafa NN, Abdelmohsen EA, El-Ghammaz AM, Saeed AM, Hamza MT. Prognostic significance of programmed death ligand 1 expression in adult patients with de-novo acute myeloid leukemia. Egypt J Haematol [serial online] 2018 [cited 2023 Jun 5];43:158-65. Available from: http://www.ehj.eg.net/text.asp?2018/43/4/158/255873

Introduction

Acute myeloid leukemia (AML) exhibits a great deal of morphologic, clinical, immunophenotypic, cytogenetic, molecular as well as epigenetic heterogeneity, which affects the responsiveness to chemotherapeutics and the disease outcome. This fact leaves the door open to novel therapeutic options to join the AML therapeutic armamentarium for the hope of optimizing the long-term outlook of the disease [1].

Leukemic cells harbor many dysregulated genes either genetically or epigenetically whose products may serve as plausible targets to immune attack. In spite of that, no efficient immune response is mounted against these cells so as to clear them, which suggests a potential alteration of the tumor microenvironment facilitating immune evasion [2].

Programmed death-1 (PD-1) represents one of the immune cell intrinsic checkpoints important for regulation of T-cell activation. It becomes physiologically induced on the surface of T cells in the setting of inflammation, creating a negative feed-back loop to halt down the unnecessary tissue damage and guard against autoimmunity. However, its chronic over-expression has been demonstrated in many solid cancers and hematopoietic malignancies facilitating tumor hijacking of the immune system [3].

Immune checkpoint blockade has been postulated to be a potential therapeutic option in many malignancies, and this is applied in many solid tumors as malignant melanoma and hematologic malignancies such as Hodgkin’s disease. This has not been clinically established yet in AML in spite of the encouraging experimental studies [4].

Patients and methods

Patients

This prospective study was conducted in the Clinical Hematology and Oncology Unit, Internal Medicine Department, Ain Shams University, Cairo, Egypt. It was held between June 2016 and December 2017. A total of 40 adult cases of de-novo AML were recruited. An informed written consent was obtained from all the study participants. The study was approved by the Ethical Committee Board of Ain Shams University and is in accordance with the Declaration of Helsinki. Data were extracted and recorded from the patients’ files.

Inclusion criteria

The following were the inclusion criteria:

Age range from 18 to 65 years old.
Patients with de-novo AML.
Patients who were deemed fit for intensive chemotherapy of curative intent.

Exclusion criteria

The following were the exclusion criteria:

Age less than 18 years.
Relapsed patients with AML.
Patients with AML with chemotherapy given before enrolment in the study.
Secondary AML with preceding hematologic disorder.
Other types of acute leukemia other than AML.
Past history of autoimmunity.
Past history of other solid tumors.
Patients who were considered ineligible to receive intensified treatment.

Plan of treatment

Induction phase

All patients received standard chemotherapy in accordance with NCCN guidelines 2016 [5]. Patients with APL received induction by PETHEMA protocol which consisted of idarubicin 12 mg/m² intravenous (i.v.) bolus on days 2, 4, 6, and 8 as well as all-trans retinoic acid (ATRA) 45 mg/m²/day in two divided doses (ATRA is started on day 1 through day 45), whereas patients without APL received 3+7 protocol formed of cytosine arabinoside 200 mg/m²/day by continuous i.v. infusion for 7 days plus doxorubicin 25 mg/m²/day i.v. for 3 consecutive days.

Consolidation phase

The patients who had achieved complete remission (CR) following induction chemotherapy were subjected to consolidation with either allogeneic stem cell transplantation (SCT) if they had intermediate or poor risk cytogenetics along with the availability of matched sibling donor, whereas those lacking a donor or who had been categorized as good-risk patients from the start were given high-dose chemotherapy as their consolidation of remission as follows: patients with APL received first cycle made up of idarubicin 5 mg/m² i.v. bolus on days 1–4, second cycle in the form of mitoxantrone 10 mg/m² i.v. bolus on days 1–3, whereas third cycle consisted of idarubicin 12 mg/m² i.v. bolus for 1 day. ATRA was administered concurrently at a dose of 45 mg/m²/day in two divided doses for 15 days with each 4-week cycle of consolidation chemotherapy. Patients without APL were consolidated with high-dose chemotherapy using high-dose cytosine arabinoside 2 gm/m² i.v. over 3 h every 12 h on days 1, 3, and 5. A total of four doses repeated every 28 days were planned to be given for patients with no possibility of transplantation, whereas patients who had been transplant eligible and had had matched sibling donor received consolidation chemotherapy cycles till their access to their transplantation procedure with a maximum number of four cycles.

For those who were refractory to the first-line therapy, they received re-induction by one of the salvage protocols. In our cohort, refractory cases received FLAG-IDA as the salvage protocol aiming at achieving remission. It consisted of fludarabine 30 mg/m²/day i.v. over 30 min on days 1 through day 5, cytosine arabinoside 2 gm/m²/day i.v. over 4 h and given 4 h after fludarabine on days 1 through day 5, idarubicin 10 mg/m²/day i.v. on days 1 through day 3, and filgrastim 5 μg/kg/day SC to begin on day 6 until neutrophil recovery.

Response assessment

On day 28, all the patients who survived the induction were evaluated regarding their responsiveness to chemotherapy. Assessment included full blood picture as well as bone marrow (BM) examination. Consequently, they were classified into responders, who attained CR, and nonresponders, who were refractory to chemotherapy; this classification was based on response criteria mentioned by Cheson et al. [6].

CR was defined as an absolute neutrophilic count greater than 1000/μl, platelet count greater than or equal to 100 000/μl, and less than 5% BM blasts with no evidence of extramedullary disease.

Refractory disease was defined as survival greater than or equal to 7 days following completion of initial treatment course with persistent leukemia in the last peripheral blood smear or BM, and/or persistent extramedullary disease [6].

Methods

Sampling

Approximately 3–4 ml of BM aspirate was obtained and divided as follows: 0.5–1 ml of BM aspirate was obtained, from which smears were prepared and stained with Leishman and MPO stains for morphological and cytochemical examination. Overall, 3 ml of BM aspirate was obtained on K2-EDTA for immunophenotyping and cytogenetics. Samples were sent to each laboratory respectively on the same day of collection.

Flow cytometric immunophenotyping

In the current study, flow cytometric analysis of PDL-1 expression was performed on blast cells (supplied by Beckman Coulter, Hialeah, Florida, USA). BM samples were processed on the same day of sample collection. They were counted using Coulter LH750 cell counter (Beckman Coulter), and the total leucocytic count was adjusted to be around 5.0×10⁹/l using PBS, 120 mmol/l NaCl, 2.7 mmol/l KCl, and 10 mmol/l phosphate buffer, with PH 7.4 (commercially available from Sigma, St Louis, Missouri, USA). Overall, 50 μl of adjusted samples were aliquoted in the control as well as in each of the sample tubes, and then 5 μl of each monoclonal antibody were added. After incubation for 15 min at room temperature protected from light, 1–2 ml of ammonium chloride-based erythrocyte lysing solution were added to every tube [8.29 g (0.15) NH₄Cl, 1 g (10 mmol/l) KHCO₃, 0.037 g (0.1 mmol/l) EDTA, and 1 l distilled water, adjusted to pH 7.3]. Tubes were vortexed and then analyzed using Coulter Navies flow cytometer.

Statistical methods

The collected data were revised, coded, tabulated, and introduced to a PC using statistical package for social science (Released 2011, IBM SPSS Statistics for Windows, version 20.0; IBM Corp., Armonk, New York, USA). Data were presented, and suitable analyses were done according to the type of data obtained for each parameter. Numerical data were expressed as means and standard deviations or medians and ranges as appropriate. Qualitative data were expressed as frequencies and percentages. Student’s t-test was used to assess the statistical significance of the difference between two study group means, whereas Mann–Whitney test was used to assess the statistical significance of the difference of a nonparametric variable between two study groups. Spearman’s method was used to assess the strength of association between two quantitative variables. The correlation coefficient denoted symbolically as ‘r‘ defines the strength and direction of the linear relationship between two variables. Kruskal–Wallis test was used to assess the statistical significance of the difference between more than two study groups’ nonparametric variables. χ²-Test was used to compare qualitative variables between groups. The confidence interval was set to 95%, and the margin of error accepted was set to 5%.

P value defined the level of significance as follows:

P value greater than 0.05: nonsignificant.
P value less than or equal to 0.05: significant.

Results

This prospective study was conducted on 40 adult newly diagnosed patients with de-novo AML. [Table 1] displays the demographic data and the clinical characteristics of the study cohort.

Table 1 Demographic data and patients’ characteristics

Click here to view

The study participants exhibited a median age of 43.4 years, ranging from 19 to 65 years, and comprised 22 (55%) females and 18 (45%) males.

BM studies were performed at the time of diagnosis including cytomorphology, flow cytometric, and cytogenetic studies. BM aspirate was infiltrated by a median blast percentage of 67.5% ranging from 20 to 98%. Immunophenotyping divided cases by their French American British (FAB) subtypes. Most cases belonged to myeloblastic phenotype with different stages of differentiation, with either no (M0), minimal (M1) or evident differentiation (M2), making ∼75% of the entire cohort. Three patients were AML (M0), 10 cases were AML (M1), and 17 cases were AML (M2). Moreover, AML with monocytoid differentiation made up 15% of the study cohort, that is, six cases, of which five cases were AML (M4) and one case was AML (M5). APL contributed to the study group by three (7.5%) patients, whereas megakaryoblastic leukemia had the lowest frequency of presentation with only one case.

Cytogenetic studies were performed on fresh BM samples using both Giemsa banding so as to assess for the presence of any karyotypic abnormalities and FISH using specific probes to assess for the presence of recurrent cytogenetic abnormalities. Patients were categorized as follows: intermediate risk encountered in 32 patients (i.e. 80%), of which 29 patients exhibited normal karyotype with no distinct recurrent cytogenetic abnormalities distinguished by FISH, that is, cytogenetically normal AML (CN-AML), whereas three had trisomy 8. Good risk was evident in five patients accounting for 12.5% of the cohort. Of these five patients, three patients had positive t(15;17), whereas core binding factor mutations affecting the alpha or beta subunits were there in two patients: one with inv(16) and another with t(8;21). Poor risk cytogenetics had the lowest contribution with only three cases (7.5%); all of them had t(9; 22).

Treatment outcome

After the proper assessment of the patients recruited to determine their eligibility to chemotherapy, all of them received induction chemotherapy of curative intent. By the end of induction on day 28, 14 (35%) patients of the cohort succumbed to their illness, of which, 10 patients died owing to septicemia and septic shock. Adult respiratory distress syndrome was incriminated to be the cause of death in three patients who had clinical, laboratory, and radiologically proven invasive pulmonary aspergillosis, and one patient died owing to complicated central venous catheter insertion causing hemopneumothorax. Survivors underwent re-evaluation of their disease response to chemotherapy, and they were further categorized into 21 (80.7%) responders and five (19.3%) with refractory disease. Refractory cases were subjected to salvage chemotherapy so as to attain remission. The five refractory cases died at the nadir stage owing to septicemia.

Follow-up of the cases was done for 6 months. Of 40 patients recruited, 18 (45%) were still alive by the end of the study, as illustrated in [Table 2].

Table 2 Different treatment outcomes on day 28 as well as 6-month survival among cases

Click here to view

Programmed death ligand 1 expression

Using monoclonal antibodies and multi-color flow cytometry, assessment of PDL-1 expression on myeloblasts has been performed. It is conspicuous that PDL-1 had a median expression of 42.95% ranging from 1.52 to 88.1%.

PDL-1 expression was related to various patients’ demographic and clinical data with no statistically significant association between PDL-1 expression from one side and age or sex from the other side, with P values of 0.66 and 0.35, respectively. Patients with extramedullary disease, which is considered as one of the aggressiveness criteria denoting hostile tumor behavior, exhibited almost similar median PDL-1 expression to that displayed by free patients with no extramedullary infiltration; their median expressions were at 46. 1 and 41.2% in turn, with a P value of 0.86.

Correlating PDL-1 expression to different laboratory quantitative variables, including total leukocyte count, hemoglobin (HGB), platelet, peripheral blood blast percentages as well as their absolute numbers, and BM blast percentages, failed to prove an association between any of these parameters and PDL-1 expression, with P values of 0.22, 0.09, 0.21, 0.49, 0.21, and 0.2, respectively. [Table 3] shows the correlations between different laboratory findings and PDL-1 expression.

Table 3 Correlations between clinical and laboratory data and percent of programmed death ligand 1 expression

Click here to view

[Table 4] illustrates the variability of PDL-1 expression amongst various FAB categories of AML. It is prominent that the lowest expression was for the one case of AML (M5) with 10.2%, whereas the highest expression was for the single case of AML (M7) at 69.3%. When a comparison was held amongst the remaining cases regarding their median PDL-1 expression level, there was variability in the expression amongst the various FAB subtypes, but it was distinct that this variability did not bear a statistical significance, with a P value of 0.57.

Table 4 Comparison amongst different acute myeloid leukemia FAB subtypes in terms of their programmed death ligand 1 expression

Click here to view

Furthermore, there was no association that could be delineated between PDL-1 expression from one side and status of different cluster of differentiation from the other side, with P values of greater than 0.05. The comparison was not done amongst CD13 and CD33 negative and positive cases, as only one patient lacked the expression of each. Focusing more, it is prominent that CD34-negative cases enjoyed a median PDL-1 expression, which was higher than CD34-positive cases, with an expression at 61.60 and 38.50%, respectively. This difference had a tendency towards statistical significance, with a P value of 0.058. Moreover, the cases that had an aberrant expression of one of the lymphoid or NK-cell antigens constituted 37.5% of the cohort. They did not show statistically remarkable difference as for their PDL-1 levels in comparison with cases not exhibiting immunophenotypic aberrations. The median expression for cases with aberrant expression was 46.1%, whereas it was 41.2% for cases without aberrant expression, with a P value of 0.67. [Table 5] and [Table 6] illustrates a comparison of PDL-1 expression in positive and negative cases in reference to different CD markers.

Table 5 Comparison of programmed death ligand 1 expression among different cases in relation to the expression of different cluster of differentiation markers.

Click here to view

Table 6 Comparison of different clinical and cytogenetic risk categories in terms of programmed death ligand 1 expression

Click here to view

Comparing median PDL-1 expressions amongst variable cytogenetic risk strata failed to prove a difference between various categories that might be statistically discernible, with a P value of 0.82. Intermediate-risk subgroup showed a median PDL-1 expression at 40.85%, whereas good-risk patients had a median value of 44.7%. Poor-risk category had a median PDL-1 level at 45.5%, which confirms almost comparable levels of expression of PDL-1 amongst the different groups.

Comparison of the study participants regarding their D28 treatment outcome and PDL-1 expression was done as illustrated by [Table 2]. It was discernible that higher PDL-1 expression did not confer higher mortality risk on day 28, with a P value 0.59. Survivors enjoyed a median PDL-1 of 43.4% whereas the dead cases had a median at 40%. Among patients who survived the induction chemotherapy regimen, 21 cases responded to chemotherapy attaining remission, whereas five cases were refractory. Comparison of responders versus non-responders in terms of PDL-1 expression showed that there was a discrepancy between both groups with median PDL-1 expression at 41.2 and 68.9%, respectively. In spite of that difference, it was not that substantial enough to reach a statistical significance, with a P value of 0.42. By the end of the study, 18 cases were alive whereas 22 cases were dead. A comparison was held between survivors and dead cases in terms of their PDL-1 expressions. It was noticed that survivors exhibited a median PDL-1 level at 43.4% whereas dead cases exhibited PDL-1 of 41%, with no statistically significant difference, with a P value of 0.89.

Discussion

PDL-1 is one of the immune cell intrinsic checkpoints that are responsible for regulation of the process of T-cell activation. It is one of the incriminated pathways that are upregulated by cancer cells, so as to halt down antitumor effector T cells’ activity and allow the tumor to evade the immune system [7].

Despite the sensitivity of AML to the immune attack, the microenvironment in AML is immunosuppressive, facilitating the immune tolerance of leukemia cells. In vitro studies have demonstrated that the factors secreted by primary AML cells can prevent T-cell activation and expansion [8].

The main purpose of this prospective study was to assess PDL-1 expression status in adult patients with de-novo AML and to relate it to patients and disease characteristics and prognostic factors. Moreover, the study aimed at identifying a relationship between this marker expression on one side and the clinical outcome of the patients presented with their responsiveness to chemotherapy and overall survival on the other side.

The present study included 40 adult patients who had de-novo AML. Our patients’ cohort demographics showed a median age of 43.4 years. Of the 40 patients included in the study, 18 (45%) were males and 22 (55%) were females.

Upon the assessment of PDL-1 expression in our cohort, the study participants exhibited a median PDL-1 expression at 42.95% ranging from 1.52 to 88.1%. The majority of them (75%) expressed the marker on more than or equal to 20% blasts, whereas 25% expressed it on less than 20% blasts.

Assessment of the relationship of PDL-1 expression and the different demographic as well as disease parameters has been performed. It is clear that there was no obvious association between PDL-1 expression on one side and age or sex on the other side, with P values of greater than 0.05.

Upon correlating PDL-1 expression with the various laboratory quantitative variables, PDL-1 showed no significant correlation with total leukocyte count, HGB, platelet, peripheral blood blast percentage or absolute number and BM blast percentage.

Moreover, comparison of PDL-1 expressions in different AML immunophenotypes or cytogenetic risk categories failed to prove any association of certain FAB subtype or cytogenetic risk group with higher PDL-1 expression. Different FAB subtypes exhibited variable PDL-1 expression with the lowest of all being for the only case of AML (M5) at 10.2%, whereas the highest expression was for the single case of AML (M7) at 69.3%. However, these discrepancies were not of statistical significance, with a P value of 0.53. Poor risk group had mildly higher PDL-1 expression in relation to intermediate or good risk groups, being at 45.5, 40.85, and 44.7% respectively, although this was not significantly remarkable, with a P value of 0.81.

These results show concordance with the findings highlighted in the study conducted by Berthon and colleagues [7] including 79 patients with AML who were tested for their PDL-1 expressions using both PCR and flow cytometry. Upon comparing different patients and disease variables with respect to PDL-1 expression, they found no relationship between PDL-1 expression on one side and age, FAB subtype, karyotype, leukocyte count, or molecular markers on the other side.

On the contrary, Brodská et al. [9] contend that PDL-1 expression is usually low in patients with AML upon diagnosis. However, they observed a very high incidence of PDL-1 positivity among patients with leucocytosis. This made them assume that high diagnostic leukocyte count may be associated with immune response failure in AML [9].

Moreover, our findings are in sharp contrast to the work done by Chen et al. [10]. Our study claimed that the single case with AML(M5) immunophenotype had the lowest expression of PDL-1, whereas their study reported that PDL-1 was highly expressed on leukemia cells of patients with AML (M5) in particular and it could significantly affect the immunotherapeutic strategies directed against that immunophenotype and proposed it as a novel prognostic marker for AML (M5) [10]. This considerable conflict cannot be validated, as we had only a single case of that FAB subtype in our cohort upon which we cannot draw solid conclusions.

Another comparison was held between different disease parameters on one side and PDL-1 expression on the other side. Patients with extramedullary infiltration had slightly higher PDL-1 expression in relation to those free from extramedullary disease at 46.1 and 41.2%, respectively, but this was not statistically substantiated, with a P value of 0.86.

This finding is not in line with an experimental study done by Zhou et al. [11] in which they infused mice with C1498 leukemia cell line to create an animal model of AML. Then, they assessed PD-1 expression in liver and spleen, which highlights PD-1/PDL-1 pathway activation. It was obvious that PD-1 expression was significantly higher in the liver in comparison with the spleen. Moreover, liver was considered as a major site for tumoral dissemination underscoring the importance of that mechanism for cancer cell progression and metastasis [11]. Whether the difference in our findings is related to the difference of the type of study, being experimental versus prospective human study, or relating to the different marker pursued; PD-1 versus PDL-1, is a subject of further investigation.On dividing our cohort in accordance with their response to induction protocol, it was obvious that of 26 patients who were alive by the end of chemotherapy, 21 were responders and only five cases were refractory. Refractory cases enjoyed a higher median expression of PDL-1 in comparison with responders at 68.9 and 41.2%, respectively. Despite this clear discrepancy, it was not of statistical importance, with a P value of 0.42. This might be related to the very low number of refractory cases. Moreover, there was no clear distinction between PDL-1 expressions amongst those who succumbed and those who survived the induction period, with comparable PDL-1 expressions at 40 and 43.4% in turn, with a P value of 0.59.

The experimental study of Zhou et al. [11] on PD-1 knockout and wild-type mice shows contradictory results with our study. Our study failed to prove a relationship between PDL-1 expression and different survival time points, either median or six-month survival. Zhou et al. [11] could assert a significant inverse relationship of the activation of PD-1/PDL-1 pathway with survival. They demonstrated higher mean survival time in PD-1 knockout mice (70 days) in relation to their wild-type counterparts (20 days). This may be attributed to the nature of research, being experimental versus prospective human study. In addition, they did not compare cases that express PD-1 or PDL-1 on the surface of leukemia cells; instead of that, they compared mice with non-functional versus functional PD-1 gene. Different methodologies might explain in part the different outcomes.

As for overall survival, our study did not examine this variable because of the short period.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

References

1.	Prokocimer M, Molchadsky A, Rotter V. Dysfunctional diversity of p53 proteins in adult acute myeloid leukemia: projections on diagnostic workup and therapy. Blood 2017; 130:699–712.
2.	Yang H, Bueso-Ramos C, DiNardo C, Estecio MR, Davanlou M, Geng QR et al. Expression of PD-L1, PD-L2, PD-1 and CTLA4 in myelodysplastic syndromes is enhanced by treatment with hypomethylating agents. Leukemia 2014; 28:1280–1288.
3.	Sehgal A, Whiteside TL, Boyiadzis M. PD-1 checkpoint blockade in acute myeloid leukemia. Expert Opin Biol Ther 2015; 15:1191–1203.
4.	Kim H, Hwang H, Han A, Lee J, Min W. Modulating the immunologic microenvironment in AML by blocking both CXCR4 and PD-L1. Blood 2017; 130:1364.
5.	National Comprehensive Cancer Network. Acute myeloid leukemia (Version 2. 2016). Available at: http://www.nccn.org/professionals/physician_gls/pdf/aml.pdf. [Accessed 27 October 2016].
6.	Cheson BD, Bennett JM, Kopecky KJ, Büchner T, Willman C, Estey EH et al. International Working Group for Diagnosis, Standardization of Response Criteria, Treatment Outcomes, and Reporting Standards for Therapeutic Trials in Acute Myeloid Leukemia. J Clin Oncol 2003; 21:4642–4649.
7.	Berthon C, Driss V, Liu J, Kuranda K, Leleu X, Jouy N et al. In acute myeloid leukemia, B7-H1 (PD-L1) protection of blasts from cytotoxic T cells is induced by TLR ligands and interferon-gamma and can be reversed using MEK inhibitors. Cancer Immunol Immunother 2010; 59:1839–1849.
8.	Sehgal A, Whiteside TL, Boyiadzis M. Programmed death-1 checkpoint blockade in acute myeloid leukemia. Expert Opin Biol Ther 2015; 15:1191–1203.
9.	Brodská B, Fuchs O, Otevřelová P, Salek C, Kuželová K. PD-L1 is frequently expressed in acute myeloid leukemia patients with leukocytosis. Blood 2016; 128:5229.
10.	Chen X, Liu S, Wang L, Zhang W, Ji Y, Ma X. Clinical significance of B7-H1(PD-L1)expression in human acute leukemia. Cancer Biol Ther 2008; 7:622–627.
11.	Zhou Q, Munger ME, Highfill SL, Tolar J, Weigel BJ, Riddle M et al. Program death-1 signalling and regulatory T cells collaborate to resist the function of adoptively transferred cytotoxic T lymphocytes in advanced acute myeloid leukemia. Blood 2010; 116:2484–2493.