Prognostic Impact of Serum Growth Differentiation Factor 15 Level in Acute Myeloid Leukemia Patients

Abstract

Growth differentiation factor 15 (GDF15) plays an important role in cancer pathophysiology and prognosis. However, limited studies analyzed its level and prognostic value in acute myeloid leukemia (AML) patients. This study included 56 adult AML patients. Serum GDF15 level was measured at diagnosis in all patients by enzyme-linked immunosorbent assay. Remission and survival statuses were assessed at 90 days following treatment. GDF15 level was significantly higher in patients than in controls (P < 0.001). GDF15 level correlated positively with age (P < 0.001), hemoglobin level (P = 0.027), and platelet count (P = 0.024). High GDF15 above the median level was associated with inferior OS (P = 0.044) together with high platelet count (P = 0.006) and high bone marrow blast percent (P = 0.038). There was no statistically significant difference between patients with GDF15 above and below the median level regarding DFS (P = 0.881). On multivariate analysis for OS, GDF15 level was an independent risk factor (P = 0.047). In conclusion, serum GDF15 level is significantly elevated in AML patients and high GDF15 level is associated with inferior OS.

Introduction

Growth differentiation factor 15 (GDF15), also known as macrophage inhibitory cytokine-1, is a divergent transforming growth factor-β superfamily member [1]. It is a 40-kDa secretory propeptide that is cleaved in the endoplasmic reticulum to produce a 25-kDa circulating protein [1]. GDF15 is a stress-responsive cytokine that is present in different cells (vascular smooth muscle cells, cardiomyocytes, adipocytes, fibroblasts, macrophages, endothelial cells), tissues (vessels, adipose tissue, tissues of central and peripheral nervous system) and organs (liver, heart, placenta, brain) [2]. Also, it was found to be highly expressed in many types of cancer, e.g. colorectal, oral, esophageal, gastric, pancreatic cancers and others [3]. It has been suggested that GDF15 can regulate some cellular processes, e.g. growth inhibition, induction of apoptosis, and tumor invasiveness [4]. GDF15 has been shown to be a powerful prognostic protein in some diseases and conditions such as cardiopulmonary diseases, end-stage renal disease, liver injury, chronic inflammatory diseases, sepsis and acute respiratory distress syndrome [5].

Acute myeloid leukemia (AML) is a heterogeneous disorder characterized by clonal expansion of myeloid progenitors (blasts) in the bone marrow (BM) and peripheral blood [6]. It is characterized by the accumulation of somatically acquired genetic changes in hematopoietic progenitor cells that alter normal mechanisms of self-renewal, proliferation, and differentiation [7]. Outcome is affected by different factors including patient features, e.g. age, comorbidities, and performance status, and disease characteristics of which the genetic profile of the disease is the most important [8]. Serum GDF15 levels were found to be increased in some hematological malignancies including AML, polycythemia rubra vera, essential thrombocythemia, primary myelofibrosis, multiple myeloma and others [9]. However, the prognostic value of GDF15 in AML was not studied previously. Hence, we measured the serum level of GDF15 in AML patients and evaluated its prognostic impact on their outcome.

Materials and Methods

Patients

This study included 56 newly diagnosed adult AML patients and 30 age- and sex-matched healthy controls. Exclusion criteria were patients with acute promyelocytic leukemia, patients older than 60 years, patients with chronic diseases especially cardiac diseases, rheumatoid arthritis and renal diseases, patients with history of dyserythropoietic disorders (e.g. thalassemia and myelodysplastic syndromes), patients with past or present history of solid tumors and recent/current pregnancy. Patients were classified according to BM morphological and cytochemical characteristics into M0–M7 according to the French-American-British (FAB) classification [10]. The Southwest Oncology Group/Eastern Cooperative Oncology Group cytogenetic risk classification was used to categorize patients into favorable, intermediate, and poor risk groups [11]. Recruitment of patients started on April 2018 and ended on August 2018.

Treatment Plan and Follow Up

All patients received treatment by the 7 + 3 induction protocol consisting of cytarabine 100 mg/m² as continuous IV infusion for 7 days and Adriamycin 30 mg/m² IV for 3 days [12]. Remission status was evaluated through performing BM examination on day 28 after induction cycle start. Patients who achieved complete remission (CR) received high dose Ara-C (HIDAC) consolidation chemotherapy cycles consisting of cytarabine 3 g/m² bid IV on days 1, 3 and 5 [13]. Patients with refractory/relapsed disease received reinduction with FLAG-Ida protocol consisting of granulocyte-colony stimulating factor from day 1 to day 5 SC, fludarabine 30 mg/m² from day 1 to day 5 IV, cytarabine 2 g/m² from day 1 to day 5 IV and idarubicin 8 mg/m² days 3, 4, and 5 IV [14]. Evaluation of remission status thereafter was performed on day 28 of each post-induction cycles. Patients were followed up for 90 days and were monitored for occurrence of death, refractoriness or relapse.

GDF15 Assessment

Sera were obtained through peripheral blood sampling from all included subjects at diagnosis before starting treatment and were stored at −80 °C for a maximum duration of two months. At time of analysis, samples were thawed and serum GDF 15 level was determined using Human GDF15 Enzyme-Linked Immunosorbant Assay (ELISA) kit (Wuhan Fine Biological Technology Co., Hubei, China). Briefly, the microtiter plate provided had been precoated with an anti-GDF15 antibody. The biotin conjugated anti-GDF15 antibody was used as detection antibody. The standards, test samples and biotin conjugated detection antibody were added to the wells subsequently, and then washed with wash buffer. Horseradish peroxidase (HRP)-streptavidin was added and unbound conjugates were washed away with wash buffer. Tetramethyl benzidine (TMB) substrates were used to visualize HRP enzymatic reaction. TMB was catalyzed by HRP to produce a blue color product that changed into yellow after adding acidic stop solution. The density of yellow is proportional to the GDF15 amount of sample captured in plate. The color change was measured spectrophotometrically at a wavelength of 450 nm. The concentration of GDF15 in the samples was then determined by plotting the optical density of the samples against concentrations on the standard curve. The detection range was 23.438–1,500 pg/ml.

Definitions and Statistical Methods

CR was defined as absence of extramedullary leukemia; normal BM by light microscopy with < 5% blasts, platelets ≥ 100 ×  10⁹/l, absolute neutrophil count ≥ 1.0 ×  10⁹/l and absence of circulating blasts [15]. Overall survival (OS) was defined as the time from study entry to death from any cause [15]. Disease free survival (DFS) was defined as time from CR to time of disease progression or death due to any cause whichever came first [15]. Patients with no reported event at the time of analysis were censored at the most recent assessment date. Descriptive statistical analysis of variables was carried out (mean, standard deviation, median, range, number and percentage). Variables in two or more groups were compared using the unpaired t test, one-way analysis of variance, or χ²-test as appropriate. Spearman correlation coefficients were used to assess the relation between two quantitative parameters in the same group. The linear regression model was used to calculate R² which reflects the amount of variance in GDF15 level explained by significantly correlating variables. Survival probabilities were estimated using the Kaplan–Meier method and survival curves were compared by the log-rank test. The median values of GDF15 and other continuous variables were used to dichotomize patients into high and low subgroups in survival analyses. Multivariate analysis of variables was performed using Cox proportional hazard model. Statistical significance was determined at the 0.05 level. All P values were two-sided. Standard computer program SPSS for Windows, release 17.0 (SPSS Inc., Chicago, IL, USA) was used for data entry and analysis.

Results

Baseline Patient Characteristics

Patient characteristics at diagnosis are summarized in Table 1. Aberrantly expressed cluster of differentiation (CD) markers were CD19 in 3 patients, CD7 in 2 patients, CD11c in two patients, CD5 in one patient and CD64 in one patient. Favorable risk cytogenetics included t(8;21) in nine patients and inv (16) in three patients; intermediate risk cytogenetics included normal karyotype in 35 patients and + 8 in four patients; poor risk cytogenetics included complex karyotype in two patients and t(9;22) in one patient.

Table 1.

Baseline patient characteristics

Characteristics	Mean ± SD	Median (range)
Age (years)	40.7 ± 13.2	41 (19–58)
TLC (cells × 109/l)	34.8 ± 57.3	15.4 (1–270)
Hemoglobin (g/dL)	7.0 ± 1.8	7.1 (4.2–13)
Platelets (cells × 109/l)	43.6 ± 40.8	29.5 (7–203)
LDH (U/l)	111.6 ± 14.5	110 (90–140)
ESR (mm in 1st h)	1338 ± 951.3	1072.5 (212–5000)
Bone marrow blasts (%)	69.3 ± 17.3	74.5 (30–90)

	N (%)
Sex
Male	26 (46.4%)
Female	30 (53.6%)
Extramedullary disease
+ve	0 (0%)
−ve	56 (100%)
Antecedent MPN
+ ve	2 (3.6%)
−ve	54 (96.4%)
FAB subtype	0 (0%)
M0
M1	12 (21.4%)
M2	34 (60.7%)
M4	4 (7.1%)
M5	6 (10.7%)
M6	0 (0%)
M7	0 (0%)
Aberrant CD expression by flowcytometry
+ve	9 (16.1%)
−ve	47 (83.9%)
Cytogenetic risk
Favorable	14 (25%)
Intermediate	39 (69.6%)
Poor	3 (5.4%)

TLC total leukocytic count, ESR erythrocyte sedimentation rate, LDH lactate dehydrogenase, MPN myeloproliferative neoplasm, FAB French-American-British, CD cluster of differentiation, SD standard deviation, N number

GDF15 Level in Patients and Controls and Its Correlation with Other Variables

There was significant difference between patients and controls regarding GDF15 level (mean level = 550.4 pg/ml (range  275–865) versus 64.9 pg/ml (range 3–95) respectively; P < 0.001). GDF15 level did not differ between male and female patients (mean level = 563.5 pg/ml vs. 538.9 pg/ml respectively; P = 0.637), between patients with and without aberrant CD expression by flowcytometry (mean level = 614.2 pg/ml vs. 538.1 pg/ml respectively; P = 0.280), among different FAB subtypes (mean level in M1 = 454.5 pg/ml, M2 = 576.2 pg/ml, M4 = 603.5 pg/ml, M5 = 560.3 pg/ml; P = 0.273) or among cytogenetic risk categories (mean level in favorable risk category = 526.8 pg/ml, intermediate risk category = 548.7 pg/ml, poor risk category = 681.3 pg/ml; P = 0.455). GDF15 level had significant positive correlation with age (r = 0.475; P < 0.001), hemoglobin level (r = 0.295; P = 0.027), and platelet count (r = 0.302; P = 0.024). Age, hemoglobin level and platelet count in combination were associated with a variance in GDF15 level of 41.3% (R² = 0.413; P < 0.001). On the other hand, GDF15 level had no significant correlation with total leukocytic count (TLC) (r = 0.221; P = 0.102), erythrocyte sedimentation rate (ESR) (r = 0.008; P = 0.954), lactate dehydrogenase (LDH) (r = -−0.047, P = 0.729) and BM blast percent (r = −0.068; P = 0.617).

Outcome of Patients

The median duration of follow up was 56 days (range  9–90). At day 28, 42 patients (75%) were alive and ready for evaluation of remission status whereas 14 (25%) died before evaluation. Of the former living group, 36 patients (85.7%) were in CR whereas 6 (14.3%) were resistant and received FLAG-Ida salvage protocol. At study end, 34 patients (60.7%) out of the whole cohort remained alive whereas additional 8 patients died resulting in a total number of death events equaling 22 (39.3%). Of the 42 patients evaluated at day 28, 30 (71.4%) were alive and in CR at the end of the study, 2 (4.8%) were dead and in CR, 4 (9.5%) were alive and having leukemia progression/relapse and 6 (14.3%) were dead and having leukemia progression/relapse. Causes of death before day 28 were septicemia (9 patients), tumour lysis syndrome (2 patients), intracranial hemorrhage (1 patient), mucor mycosis (1 patient) and doxorubicin-induced cardiotoxicity (1 patient). Causes of death after day 28 were leukemia progression/relapse (6 patients) and septicemia (2 patients).

Impact of GDF15 on OS and DFS

Patients with GDF15 > 554.5 pg/ml (i.e. median GDF15 level for the whole patient population) had significantly inferior OS in comparison to those with GDF15 < 554.5 pg/ml (OS = 50 vs. 71.4%; median survival = 41 days vs. not reached (NR) respectively; P = 0.044) (Fig. 1). On the other hand, there was no statistically significant difference between patients with GDF15 > 554.5 pg/ml and those with GDF15 < 554.5 pg/ml regarding DFS (DFS = 85.7 vs. 81.8%; median survival = 56 days vs. NR respectively; P = 0.881) (Fig. 2).

Fig. 1.

Comparison of OS of AML patients with GDF15 > 554.5 pg/ml (dashed line) and those with GDF15 < 554.5 pg/ml (continuous line) using Kaplan Meier curves

Fig. 2.

Comparison of DFS of AML patients with GDF15 > 554.5 pg/ml (dashed line) and those with GDF15 < 554.5 pg/ml (continuous line) using Kaplan Meier curves

Univariate and Multivariate Analyses for OS

On Univariate analysis, platelet count > 29.5 cell × 10⁹/l, bone marrow blast percent > 74.5% and GDF15 level > 554.5 pg/ml were associated with inferior OS (P = 0.006, P = 0.038 and P = 0.044 respectively) (Table 2). On multivariate analysis of these variables, only GDF15 level > 554.5 pg/ml was an independent risk factor for inferior OS (P = 0.047) (Table 2).

Table 2.

Univariate and multivariate analysis of GDF15 level and other variables regarding OS

Variables	Univariate analysis for OS			Multivariate analysis for OS
	Alive/total N (%)	Mean/median survival (days)	P	HR	95% CI	P
Age (years)
> 41	14/28 (50%)	56.1/41	0.066
< 41	20/28 (71.4%	72.3/NR
TLC (cells × 109/l)
> 15.4	18/28 (64.3%)	66.9/NR	0.526
< 15.4	16/28 (57.1%)	61.4/NR
Hemoglobin (g/dL)
> 7.1	16/28 (57.1%)	61.2/NR	0.516
< 7.1	18/28 (64.3%)	67.1/NR
Platelet count (cells × 109/l)
> 29.5	12/28 (42.9%)	52/35	0.006*	2.577	0.843–7.755	0.097
< 29.5	22/28 (78.6%)	76.4/NR
ESR (mm in 1st h)
> 110	16/28 (57.1%)	62.6/NR	0.717
< 110	18/28 (64.3%)	65.7/NR
LDH (U/l)
> 1072.5	15/28 (53.6%)	69.5/NR	0.247
< 1072.5	19/28 (67.9%)	58.9/NR
Bone marrow blast percent
> 74.5	13/28 (46.4%)	55.4/36	0.038*	1.663	0.575–4.808	0.347
< 74.5	21/28 (75%)	73/NR
GDF15 level (pg/ml)
> 554.5	14/28 (50%)	55.5/41	0.044*	2.429	1.012–5.827	0.047*
< 554.5	20/28 (71.4%)	72.8/NR
Sex
Male	16/26 (61.5%)	65.2/NR	0.852
Female	18/30 (60%)	63.3/NR
Antecedent MPN
+ve	1/2 (50%)	46.5/33	0.907
−ve	33/54 (51.1%)	64.3/NR
FAB subtypea
M1	8/12 (66.7%)		0.052
M2	16/34 (47.1%)
M4	4/4 (100%)
M5	6/6 (100%)
Aberrant CD expression
+ve	4/9 (44.4%)	50.6/35	0.183
−ve	30/47 (63.8%)	66.8/NR
Cytogenetic risk
Favorable	10/14 (71.4%)	71.2/NR	0.598
Intermediate	22/39 (56.4%)	61.1/NR
Poor	2/3 (66.7%)	61/NR

TLC total leukocytic count, ESR erythrocyte sedimentation rate, LDH lactate dehydrogenase, GDF15 growth differentiation factor 15, MPN myeloproliferative neoplasm, FAB French-American-British, CD cluster of differentiation, OS overall survival, N number, NR not reached, HR hazard ratio, 95% CI 95% confidence interval

*Significant

^aMean/median survival were not calculated because all FAB M4 and M5 cases were censored

Discussion

The role of GDF15 in neoplastic disorders has not been fully elucidated. Several Authors linked GDF15 with pro-apoptotic effects on cancer cells [16–18]. Others associated GDF15 with pro-tumorigenic effects [4]. In this study we analyzed serum GDF15 level in newly diagnosed adult AML patients and assessed its impact on their outcome. We excluded patients with conditions known to be associated with altered GDF15 level, i.e. patients with chronic diseases especially cardiac diseases, rheumatoid arthritis and renal diseases [19–22], patients with history of dyserythropoietic disorders (e.g. thalassemia and myelodysplastic syndromes) [23–25], patients with past or present history of solid tumours [26–29], and recent/current pregnancy [30, 31]. Patients received the 7 + 3 induction protocol. Adriamycin was used instead of the traditionally used daunorubicin and idarubicin because of the unavailability of the latter two agents in our center during the study period. Our regimen was studied by the Cancer and Leukemia Group B researchers who compared impacts of daunorubicin at dose of 45 mg/m², daunorubicin at dose of 30 mg/m² and Adriamycin at dose of 30 mg/m² on outcome. Despite an inferior response rate in patients younger than 60 years in the Adriamycin arm in comparison to the daunorubicin arm at dose of 45 mg/m², Adriamycin was associated with comparable response rate to daunorubicin at dose of 30 mg/m². Also, Adriamycin had similar toxicity profile and treatment-related mortality rate in comparison to both doses of daunorubicin. However, Adriamycin was generally more toxic to the gastrointestinal tract than daunorubicin [12].

We found that there is a statistically significant higher mean level of GDF15 in AML patients when compared to the control group. This agrees with what has been reported by a Japanese study where serum level of GDF15 has been measured in patients with various hematological malignancies including AML, myelodysplastic syndromes, chronic myeloid leukemia, primary myelofibrosis, polycythemia rubra vera, essential thrombocytosis, myeloma, lymphoma, acute lymphoblastic leukemia and adult T-cell leukemia. The median serum GDF15 levels in all these disease categories were significantly elevated relative to the levels in healthy controls [9]. A Chinese study reported that GDF15 was highly expressed by leukemic cell lines and that these cells also secreted GDF15. This secreted GDF15 was involved in the morphological remodeling of BM adipocytes, which can in turn help leukemic cell growth, indicating that GDF15 may be a possible target for treatment of AML patients [32]. Moreover, it has been demonstrated in another study that GDF15 is also released from cancer associated fibroblasts that are present in the BM of AML patients and that it acts as a chemo-protective factor for AML cells [33].

GDF15 level in our study was not influenced by gender, CD expression, FAB subtype or cytogenetic risk category. Up to our knowledge no other study analyzed the correlation between level of GDF15 and types of cytogenetic abnormalities. Of note, GDF15 levels in the two AML patients with antecedent MPN were higher than the mean GDF15 level in AML patients without (842 pg/ml and 885 pg/ml vs. 538.8 pg/ml) but no statistical comparison has been performed because of the small number of patients in the former group. It has been reported that GDF15 levels are higher in patients with MPN, especially those characterized by increased erythropoiesis, and that this effect is more pronounced particularly in individuals with JAK2-V617F mutation [34]. GDF15 level had no correlation with TLC, ESR, LDH or BM blast percent. Nevertheless, GDF15 level had significant positive correlations with age, hemoglobin level and platelet count. It has been suggested that GDF15 plays a major role in biological processes associated with ageing which may explain its positive correlation with age [3]. In contrast to our finding, a negative correlation between GDF15 serum concentration and hemoglobin level and platelet count has been demonstrated in patients with multiple myeloma [35]. Interestingly, it has been revealed that the use of recombinant human GDF15 had a slight inhibition effect on hematopoiesis in vitro and not only on erythropoiesis [36]. Hence, it is important to elucidate the mechanism of anemia in AML, especially GDF15-related anemia, in further studies in order to provide therapeutic options to correct this problem.

GDF15 level above the median for the whole patient population was associated with inferior OS but not DFS in our study. Also, GDF15 level was found to be an independent prognostic marker for OS in AML patients in multivariate analysis. The impact of GDF15 level on prognosis of AML patients was not addressed in other studies. In agreement with our result, a study reported that myeloma patients with high levels of plasma GDF15 levels had inferior event-free survival and OS rates 30 months after diagnosis compared to those with low plasma GDF15 levels [37]. Conversely, another study reported that GDF15 level do not have an impact on survival of myeloma patients [38]. It has also been demonstrated that GDF15 expression level could be a prognostic biomarker in some solid tumors, e.g. colorectal cancer [39], prostate cancer [40], and ovarian cancer [41]. Moreover, it has been reported that elevated plasma GDF15 levels are associated with increased cancer-related weight loss and decreased survival in cancer patients [42]. Notably, although cytogenetic risk category is a well-documented prognostic factor in AML patients, it did not influence OS in univariate analysis of variable in our study [6]. Such finding can be attributed to the low numbers of patients in our cohort and to the short duration of follow up.

Our study holds significance because it is the first one analyzing the impact of GDF15 serum level on the prognosis of AML patients. The limitations of our study are the small number of patients and the short duration of follow up. We conclude that serum GDF15 level is significantly elevated in AML patients. Also, GDF15 level can be used as predictor for inferior OS in AML patients regardless of their cytogenetic risk category. However, we recommend performing other studies on a larger scale of patients and for a longer duration of follow up to confirm our results. Also, further studies concerned with the role of GDF15 in the pathophysiology of AML are required in order to assess a possible therapeutic role for targeting GDF15 in AML patients. Finally, we suggest conducting studies concerned with the expression level of GDF15 on BM cells of AML patients determined by immunohistochemistry and analyzing its prognostic significance.

Compliance with Ethical Standards

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the ethical committee of Faculty of Medicine—Ain Shams University and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

Informed consent was obtained from all individual participants included in the study.