Sage Journals: Discover world-class research

Abstract

BACKGROUND:

Acute myeloid leukemia (AML) is a malignant disorder of hematopoietic stem and progenitor cells, characterized by accumulation of immature blasts in the bone marrow and peripheral blood of affected patients. Response to chemotherapy treatment in patients with AML is wide-ranging, and to date there are no adequate molecular biomarkers used to predict clinical outcome.

OBJECTIVE:

The aim of this study was to identify potential protein biomarkers which could help predict response to induction treatment in AML patients.

METHODS:

Peripheral blood samples were obtained from 15 AML patients both before and after treatment. A comparative proteomic analysis was performed using 2D gel electrophoresis followed by Mass Spectrometry.

RESULTS:

CONCLUSIONS:

This study gives an insight into a panel of protein biomarkers with prognostic potential that should be further investigated.

Keywords

Acute myeloid leukemia induction therapy prognosis proteomics two-dimensional gel electrophoresis

1. Introduction

Acute myeloid leukemia (AML) is a malignant disorder of hematopoietic stem and progenitor cells, which results in the build-up of immature blasts in the bone marrow and eventually in the peripheral blood of affected patients. There is enormous genetic heterogeneity within defined AML subsets, with many gene mutations identified [1], which means AML is still a disease with a very variable prognosis. Current clinical guidelines in AML recognize three prognostic groups: favorable, intermediate, and poor risk, based on cytogenetic abnormalities and/or gene mutations [2]. Patients in the favorable-risk group have a complete remission rate of over 90%, and overall survival of 60%, while those in the poor-risk group have a disease-free survival and overall survival ranging between 5 and 15%, which is largely due to a high treatment resistance rate during induction chemotherapy. Most AML patients fall within the intermediate risk group (those with a normal karyotype); however, for these patients the outcome of treatment is very heterogeneous.

Accurately assessing patient’s prognosis is very important for clinical management of the disease, which is why there are several prognostic factors such as age, performance status at diagnosis, platelet count, serum creatinine or albumin that are considered by the clinician when deciding the course of treatment [3]. However, the patient’s cytogenetic profile is currently the strongest prognostic factor; for example, a complex karyotype, monosomy 5 or 7, t(6;9), inv(3) or 11q changes other than t(9;11) have all been associated with a significantly lower response to treatment and overall survival (OS) [4].

It is clear that genetic studies are very valuable; however, isolated from a context in which thousands of proteins mediate cellular function, this prognostic model is not complete. In this study we compared the protein profiles of peripheral blood mononuclear cells (PBMC) taken from patients with AML at time of diagnosis and after induction treatment, with the aim of identifying potential prognostic protein biomarkers in terms of response to treatment. As there are currently very few proteomic studies in the literature regarding AML, and less so that focus on prognosis, this study gives an insight into a panel of protein biomarkers with prognostic potential that could be further investigated.

2. Materials and methods

2.1 Experimental subjects

Peripheral blood was obtained from 15 newly diagnosed patients with de novo AML, at Hospital Manuel Uribe Angel, Colombia, between 2015–2017. The disease was diagnosed based on morphology, cytochemical staining, and immunophenotypic analysis. Two blood samples were taken from each patient: at time of diagnosis (before the start of chemotherapy), and once again after the completion of the first round of induction therapy, which was typically 2–3 weeks after induction or when medullary aplasia was resolved. Patients included in the study did not have neutropenic fever at the time of the first blood draw. This study was reviewed and approved both by the Instituto Tecnologico Metropolitano’s Ethics Committee (6 June 2014) and the Hospital Manuel Uribe Angel Ethics Committee (17 April 2015). Written informed consent was appropriately obtained from all patients, and international as well as institutional guidelines were complied with, specifically The Code of Ethics of the World Medical Association (Declaration of Helsinki), printed in the British Medical Journal (18 July 1964).

Complete remission (CR) was defined by the presence of $<$ 5% blasts in bone marrow after induction therapy, absence of extramedullary leukemia, and peripheral blood count recovery, with a neutrophil count of at least 1 $\times$ 10 ${}^{9}$ /L and a platelet count of at least 100 $\times$ 10 ${}^{9}$ /L. Partial response (PR) was defined by a decrease in the percentage of blasts after treatment to 5–20% blasts. Resistance to therapy (Res) or treatment failure was defined by $>$ 20% blasts after induction therapy. For PR and Res there was no neutrophil and platelet threshold.

2.2 Isolation of PBMCs and protein extraction

All blood samples were processed the same day of collection. PBMCs were isolated from peripheral blood by standard density gradient centrifugation with Ficoll Histopaque ${}^{\text{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}}$ -1077 (Sigma-Aldrich). To extract proteins from PBMCs, the cells were lysed (0.5% Triton x-100, 50 mM Tris-HCL pH 8.0, 150 mM NaCl, 1 mM EDTA, protease inhibitors) and the proteins precipitated in a 20% (v/v) final concentration of trichloroacetic acid. The protein pellet was resuspended in rehydration buffer (7 M urea, 2% CHAPS, 0.5% carrier ampholytes) and stored at $-$ 20 ${}^{\circ}$ C.

2.3 2D electrophoresis

Proteins (50 $\mu$ g) were loaded by passive rehydration onto 7cm ZOOM ${}^{\text{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}}$ IPG (Immobilized pH gradient) strips pH 3–10 NL (ThermoFisher Scientific) at room temperature. Isoelectric focusing was carried out using the following voltage ramp: 100 V for 1 h, 150 V for 1 h, 200 V for 5 min, 450 V for 5 min, 600 V for 5 min, 750 V for 5 min, 950 V for 5 min, 1200 V for 10 min, 1400 V for 10 min, 1600 V for 10 min, and 2000 V for 45 min. The IPG strips were then reduced with 100 mM DTT and alkylated with 2.5% iodoacetamide according to the manufacturer’s recommended protocol. After this, the IPG strips were loaded onto SDS-PAGE NuPAGE™ Novex™ 4–12% Bis-Tris Protein Gels, 1,5 mm (ThermoFisher Scientific) and run at 200V for 45 minutes. After electrophoresis these were stained with SYPRO ${}^{\text{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}}$ Ruby (Invitrogen™, ThermoFisher Scientific) and the gel images were acquired using the ChemiDoc™ MP System (Biorad). All samples were run in duplicate.

2.4 Image analysis and 2DE gel comparison

Gel images were analyzed and compared using the PDQuest Advanced 2-D Software (BioRad). The protein spots in each gel image were detected and then the images within each group of patients were matched to create synthetic images containing the spot data from all the gels in the match set, so synthetic images were obtained for CR before treatment, CR after treatment, PR before treatment, PR after treatment, Res before treatment, Res after treatment, and Death. After automated matching, spot alignment was improved by manual spot detection and matching, using some common spots as landmarks to align the images. The spot intensities were normalized to equalize the total densities of each gel image using the Local Regression Model. Normalized protein spot volumes in the proteomes were then compared among experimental groups, according to the response to induction therapy. The 2DE proteomes before and after treatment were clustered in subgroups to identify differentially expressed spots: (a) comparison of protein expression between CR before and after treatment; (b) comparison between PR (c) Res and (d) death during induction.

Parametric test analyses were performed with bootstrapping, which is a resampling method that provides a critical adjusted p-value, controlling for a type I error (false positive), reducing over-fit bias, and validating the accuracy estimates. We performed all the analyses to identify differentially expressed proteins based on 1000 bootstrap samples. In all analyses, the confidence interval (CI) was 95% and p values less than 0.05 were considered as significant. For protein identification by mass spectrometry, we filtered out significant spots with a fold change below 1.5-fold in density.

To analyze intragroup variability, the Correlation Coefficient and Coefficient of Variation were determined for each gel in the match set. Lastly, scatter plot analysis was performed to calculate the linear dependency between the spot values of one gel in comparison to another.

2.5 Mass spectrometry

For mass spectrometry (MS), protein samples were pooled, and 300 $\mu$ g loaded onto preparative gels. The spots of interest were excised from the gels, de-stained and subjected to tryptic in-gel digestion with 10 ng/ $\mu$ l trypsin in 50 mM Ambic at 37 ${}^{\circ}$ C for 18 h. Mass spectra for the spots were acquired by nanoUPLC-QTOF-MS (nanoAcquity UPLC, Waters Co.) using data-independent acquisition (DIA). Spectra were then processed using the ProteinLynx™Global Server version 2.2 (Waters Co.) and MASCOT ${}^{\text{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}}$ (www.matrixscience.org) to identify each protein. Parameters used: Minimum fragment ions per peptide, 3; Minimum fragment ions per protein, 7; minimum number of peptides per protein, 2; missed cleavages, 1; fixed modifications, carbamidomethylation; variable modifications, methionine oxidation; peptide charge 1 $+$ ; monoisotopic mass.

2.6 Protein network analysis

A protein-protein interaction (PPI) network was constructed using the 6 proteins identified by MS as associated with unfavorable prognosis, using databases STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) [5] and BioGRID (Biological General Repository) for Interaction Datasets [6]. Cytoscape 3.2.1 [7] was used to perform the topological analysis of the PPI network under default parameters. The three main topological parameters analyzed were: connectivity degree ( $k$ ), betweenness centrality (BC) and closeness centrality (CC).

3. Results

3.1 Clinical characteristics of patients

For this study, 15 patients diagnosed with de novo AML were recruited. Of these patients, six achieved complete remission, three had a partial response, three were resistant to therapy, and three died during induction. For the last group of patients, it was only possible to analyze samples taken before treatment started. There were four cases of Acute Promyelocytic Leukemia (APL) (patients 1, 2, 13 and 14), two cases treated with the PETHEMA protocol (patients 1 and 2), one case treated with the FLUGA protocol (patient 10), and the remaining patients were treated with the 7 $+$ 3 induction protocol. The clinical characteristics of these patients are presented in Table 1.

Table 1
Clinical characteristics of patients included in the study

Patient	Age	Sex	FAB	Karyotype	Leukocyte count	% Blasts(i) % Blasts(f)	Induction protocol	Treatement response
1	35	M	M3	Normal	33000	86%–3%	PETHEMA	Complete response
2	42	F	M3	Neg PML-RARA	28720	85%–0%	PETHEMA
3	56	F	M4	Normal	6090	65%–4.5%	7 $+$ 3
4	18	F	M1/M2	–	2280	91%–0.54%	7 $+$ 3
5	53	F	M2	46XX t(8;21) (q22;q22)	13680	19%–1%	7 $+$ 3
6	67	M	M0	45XY t(8;21) (q22;q22)	129520	91%–1%	7 $+$ 3
7	28	F	M2	Normal	2090	31.82%–7.6%	7 $+$ 3	Partial response
8	62	F	M1	–	1780	65%–18%	7 $+$ 3
9	50	M	M1/M2	46XY del(5), del(15)	5040	47%–8.41%	7 $+$ 3
10	67	M	M4	Normal	2530	14%–35%	FLUGA	Resistance
11	65	F	M4	Normal	9870	42%–22%	7 $+$ 3
12	46	M	–	Normal	7110	68%–64%	7 $+$ 3
13	62	M	M3	–	6390	60%	–	Death
14	30	F	M3	PML-RARA	12170	–	–
15	64	F	–	–	1780	84%	7 $+$ 3

% Blasts(i): Percentage of initial blasts – before treatment. % Blasts(f): Percentage of final blasts – after treatment. “–”: no data available.

Table 2

Spots detected in PBMC proteomes before and after treatment

Treatment response	Patient	No. spots		Match		CV		CC
		Before	After	Before	After	Before	After	Before	After
Complete response	1	96	68	28	6	77.33	81.56	1	1
	2	256	251					0.45	0.25
	3	601	337					0.27	0.13
	4	337	403					0.59	0.20
	5	621	408					0.65	0.66
	6	642	151					0.32	0.77
	Average	425.5	269.6
Partial response	7	138	69	55	36	75.4	111.5	0.16	0.72
	8	118	38					0.2	1
	9	170	146					1	0.22
	Average	142	84.3
Resistance	10	183	132	59	74	74.7	55.5	0.7	0.67
	11	154	85					0.6	1
	12	73	163					1	0.71
	Average	136.6	126.6
Death	13	153	–	27	–	54.33	–	0.42	–
	14	160	–					0.36
	15	71	–					1
	Average	128	–

CV: coefficient of variation. CC: coefficient of correlation.

3.2 Proteomic profile of PMBC isolated from AML patients before and after treatment

We analyzed the proteome of 15 AML samples: 6 with complete response, 3 with partial response, 3 with resistance and 3 who died during induction. Analysis of 2DE images revealed an average of 201,8 spots before treatment and 160,1 spots after treatment, distributed within a pI range of 3–10 and molecular weight between 10–110 kDa, as seen in Fig. 1 and Table 2. Samples from patients with CR exhibited the greatest number of spots (both before and after treatment) compared with PR, Res and death, which could be due to the high number of blasts present in these patients. In all groups we observed high intragroup variability, due in part to normal interindividual variability within the population, as well as to the different clinical characteristics of patients within a group (see Table 1) and the small sample size.

Figure 1.

Representative gel images.

The proteomic profiles of the four patient groups were subsequently analyzed for differentially expressed protein spots.

3.3 Differential protein expression according to treatment response

A total of 27 2DE images were analyzed with PDQuest: 15 images before treatment and 12 after treatment. The comparison of spots between groups, both before and after treatment, by paired T-test and Mann-Whitney U test, revealed 37 spots that were significantly ( $p\leqslant$ 0.05) altered with more than a 1.5-fold change. Particularly, there was a notable difference in the expression of spots present in CR (favorable response) compared to spots in PR, Res, and death (unfavorable response), indicating an association of these proteins with favorable response to treatment. There was no statistically significant difference between clinical data of patients (age, sex, FAB, karyotype and leukocyte count) and protein spots ( $R^{2}<$ 0.25).

These 37 spots were analyzed by nanoHPLC-QTOF mass spectrometry to determine their identity and following a Mascot database search using the acquired MS data, 30 spots were identified; however, some of these proteins were identified more than once and we excluded extracellular (fibrinogen) and contaminant proteins (apoliprotein AI, hemoglobin subunit beta and keratin type II) so in total 17 proteins of interest were identified. The following 12 proteins identified were significantly associated with favorable prognosis: ATP synthase $\beta$ , $\alpha$ 1-antitrypsin, leukocyte elastase inhibitor, triosephosphate isomerase, immunoglobulin kappa constant, malate dehydrogenase, 6-phosphogluconolactonase, Rho GDP-dissociation inhibitor 2, phosphoglycerate kinase 1, protein disulfide-isomerase, heat shock protein beta-1, and glyceraldehyde-3-phosphate dehydrogenase. On the other hand, the following 6 proteins identified were significantly associated with unfavorable prognosis: glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (identified 4 times), superoxide dismutase (SOD1), glutathione S-transferase P (GSTP1), Cofilin-1 (CFL1), annexin A1 (ANXA1) (identified 3 times) and putative elongation factor 1-alpha-like 3 (eEF1A1). Details of the proteins identified with their theoretical pI value, molecular weight, protein score, sequence coverage, number of matched peptides, as well as average relative change and p-values are shown in Supplementary Table S1.

Table 3
Proteins in the backbone network of unfavorable prognosis

No.	Protein	BC	$k$	CC	No.	Protein	BC	$k$	CC
1	GAPDH	0.466	0.759	15	13	TRIM25	0.022	0.489	3
2	EGFR	0.186	0.629	10	14	HNRPA1	0.008	0.489	4
3	HSP90AA1	0.044	0.579	7	15	EEF1A1	0.001	0.478	3
4	PTK2	0.037	0.564	6	16	ANXA1	0.023	0.468	2
5	APP	0.059	0.550	4	17	FN1	0.015	0.458	2
6	CFL1	0.039	0.550	7	18	SOD1	0.000	0.458	2
7	ESR1	0.056	0.537	7	19	RHOA	0.020	0.440	4
8	HSP90AB1	0.002	0.524	4	20	ARHGDB	0.019	0.415	3
9	BRCA1	0.054	0.512	4	21	GSTP1	0.005	0.407	2
10	TP53	0.019	0.512	5	22	IGKC	0.003	0.407	2
11	YWHAQ	0.007	0.512	4	23	UBE3A	0.017	0.386	3
12	MDM2	0.018	0.500	5

BC: betweenness centrality. $k$ : connectivity degree. CC: closeness centrality.

Figure 2.

Backbone network of unfavorable prognosis. BC: betweenness centrality. $k$ : connectivity degree.

This study had two cases treated with PETHEMA, one treated with FLUGA, and the remaining patients were treated with the 7 $+$ 3 induction protocol (Table 1). Therefore, we performed statistical analysis of GAPDH, eEF1A1, ANXA1, CFL1 and GSTP1 with these 9 patients (excluding patient 15 who died in induction) and found that expression at time of diagnosis of GSTP was related with favorable prognosis, while GAPDH (isoforms 40 kDa, pI 6.8, 6.9 and 7.8) was related with unfavorable prognosis PR and Res ( $p$ value $<$ 0.05).

3.4 Protein interaction network

The PPI network was constructed with 6 seed proteins associated with unfavorable prognosis in AML patients. This network was found to have 458 nodes and connected by 1.899 edges (Supplementary Fig. S1). The proteins in the network were classified into four categories according to their $k$ and BC values: 256 nodes were nonhub-nonbottleneck (low $k$ and low BC), 152 nodes were hub-nonbottlenecks (high $k$ and low BC), 17 nodes were nonhub-bottlenecks (low $k$ and high BC), and 33 nodes were hub-bottlenecks (high $k$ and high BC), the latter being the most interesting as they are the most central and well-connected nodes in the network. To identify the most central node in the network, we compared these $k$ and BC values and found two proteins of interest with the highest $k$ and BC: GAPDH (BC 0.531; $k$ 146) and eEF1A1 (BC 0.241; $k$ 91). As GAPDH had the highest CC value (0.491) it was selected as the central node in the network.

Of this main network we retrieved the proteins within the top 5% highest BC (23 protein nodes and 54 edges) to construct a backbone network. Of the 23 nodes comprising the backbone network, GAPDH is the hub-bottleneck node and has 15 primary nodes, including 4 seed proteins (eEF1A1, CFL1, SOD1, ANXA1) (see Fig. 2). The nodes with the highest $k$ , BC and CA were: GAPDH, EGFR and HSP90AA1 (see Table 3).

4. Discussion

In this study, the comparative proteomic analysis of AML patients with differing response to treatment, led to the identification of 12 proteins associated with favorable prognosis and 6 proteins associated with unfavorable prognosis. As protein-protein interaction networks help in understanding the functional relevance of expressed proteins, we implemented this strategy to further analyze the proteins related to poor treatment response. Both in the main and backbone networks constructed, two proteins of interest stood out: GAPDH and eEF1A1; while the backbone network also highlighted proteins ANXA1, CFL1, and GSTP1.

GAPDH was identified as the central node in both the main and the backbone networks. Interestingly, in the proteomic analysis, GAPDH was identified in both favorable (isoforms 25 kDa pI 7.1 and 7.8) and unfavorable prognosis (isoforms 40 kDa pI 6.8, 6.9 and 7.8) patient groups suggesting that the different isoforms of GAPDH have a multifunctional role favoring both mechanisms. In a study published in 2009, GAPD was purified from normal leukocytes of healthy subjects and leukocytes of patients with chronic myeloid leukemia (CML) showing that GAPDH from CML patients was distinctly different from GAPDH in healthy volunteers [8]. The most notable difference was in terms of molecular weight and subunit composition, as GAPDH from healthy sources was found to be a homotetramer with a molecular mass of 140 kDa, while GAPDH from leukemic leukocytes was a heterodimer of approximately 87 kDa. This could be an explanation as to why we identified several GAPDH isoforms in both favorable and unfavorable prognosis.

GAPDH is an essential enzyme of the glycolytic pathway. One of the well-known features of cancer cells is the increase in glucose metabolism, which actively contributes to an increase in cell proliferation. Therefore, our finding of increased expression of GAPDH in patients with unfavorable diagnosis is logical. A study using the transgenic mouse model of non-Hodgkin’s lymphoma revealed that an increase in GAPDH expression, but not in other glycolytic enzymes, was associated to tumor aggressiveness (increased growth and vascularization) [9]. This was confirmed in patients with diffuse large B-cell lymphoma, where 7 out of 13 patients presented elevated GAPDH expression at time of diagnosis. GAPDH was found to enhance the transcription of hypoxia-inducing factor-1 $\alpha$ (HIF-1 $\alpha)$ which allows cells to adapt to hypoxic environments and is often upregulated in cancer [10].

It is worth noting that GAPDH has other intracellular functions and locations within the cell. A key study reported that GAPDH accumulation in the mitochondria contributes to mitochondrial membrane permeabilization prior to apoptosis [11], which could also explain why we found increased levels of two isoforms of GAPDH in patients with a favorable response to treatment.

In our analysis, at time of diagnosis high levels of expression of eEF1A1 were found in the three patients who died and only in one patient of the RES group; however, in both protein networks that were constructed eEF1A1 was one of the central nodes. This protein is involved in the GTP-dependent binding of aminoacyl-tRNA to the ribosomes during protein biosynthesis. Until recently, there were not any reports into the prognostic significance of this protein in AML. However, a study published in 2020 [12] showed that in AML cell lines eEF1A1 promotes proliferation and migration, and inhibits apoptosis, which are well-known features of malignancy. Another study, published in 2021, which investigated prognostic markers for pediatric AML patients reported that eEF1A1 is associated with poor prognosis [13], and as such it could be considered as a novel strategy in AML treatment. These reports support our findings.

Annexin A1(ANXA1), was another of the proteins of interest in the protein network analysis, and we found it to be overexpressed in patients with unfavorable prognosis. The role of ANXA1 in cancer prognosis is rather unclear, in some cases, it has been reported that ANXA1 overexpression increases cell proliferation, migration and invasion [14, 15] due to the constitutive activation of the NF $\kappa$ B pathway; while other studies have reported that ANXA1 reduces cell viability, colony formation, and induces apoptosis [16]. It has also been reported that this protein is downregulated in bone marrow of patients with myelodysplastic syndromes (MDS) and thus may be useful for predicting risk of leukemia transformation [17].

Another protein of interest is cofilin 1, which we found to be notably increased in patients with an unfavorable prognosis. CFL1 is an actin regulatory protein that in its active form (unphosphorylated) causes depolymerization of F-actin to the G-actin form which translocates to mitochondria promoting the release of cytochrome c, and the activation of caspases, culminating with the activation of apoptosis [18, 19]. Overexpression of cofilin has been reported in multiple cancer types and is associated with a poor prognosis (malignant phenotype and metastasis) [20]. This is logical for solid tumors, as cofilin regulates cytoskeletal reorganization promoting tumor cell migration; however, recently cofilin has also been flagged as a protein of interested in leukemia. In a human leukemia cell line, it was shown that silencing cofilin 1 notably reduces cell proliferation [21], while gene expression analysis of 180 patients with chronic lymphocytic leukemia showed upregulation of cofilin1 expression in these patients, compared to healthy controls.

The last protein of interest that shall be discussed is GSTP1. Glutathione S-transferases (GSTs) are a family of enzymes which play a role in processes of detoxification, and as such has been reported to be of interest in several types of cancer, such as breast and prostate cancer. In terms of treatment response, in breast cancer cells it has been reported that overexpression of GSTP1 causes chemoresistance, while silencing this protein has the opposite effect (reduces chemoresistance) [22]. In patients with chronic myeloid leukemia this is also the case [23], thus supporting our result of increased GSTP1 expression in AML patients who did not respond to treatment.

This study is one of the first to report how protein expression changes before and after induction treatment in different AML patient groups; however, it is not without limitations.

This study is limited by the high interindividual variability within the AML population due to different AML subtypes, karyotypes and gene mutations. The French-American British (FAB) classification system has classified AML patients into subtypes (M0–M7) based on the type of cell the leukemia develops from and how mature the cells are. Particularly, Acute Promyelocytic Leukemia (APL) is a unique subtype of AML, classified as the M3 subtype, of which there were four cases included in this study. The genetic hallmark of APL is the fusion PML-RAR $\alpha$ gene caused by the reciprocal translocation t(15;17) (q22;q12). This limitation is inherent to the genetic variability of this population; however, it can be somewhat overcome with a larger sample size.

The enormous genetic heterogeneity within defined AML subsets means that AML is still a disease with a very variable prognosis. A large percentage of AML patients have a normal karyotype, and as such fall within the intermediate risk group. For these patients the outcome of treatment is very heterogeneous; however, molecular markers such as FLT3, NPM1 and CEBPA have been found to improve risk stratification in this patient group and have already been incorporated into clinical practice in many countries. However, this is not the case in Colombia, so details of these molecular markers were absent from the study as they were not included in the routine clinical workup.

Lastly, it is important to note that due to the lack of a validation cohort, the proposed biomarkers are prognostic in nature rather than predictive.

5. Conclusions

This comparative proteomic study, combined with a protein network analysis, revealed several proteins that could be considered potential biomarkers of poor prognosis in AML: GAPDH which favors increased glucose metabolism; eEF1A1 and Annexin A1 that promote proliferation and migration, cofilin 1 which plays a role in the activation of apoptosis; and GSTP1 which is involved in the processes of detoxification and chemoresistance. This study gives an insight into a panel of protein biomarkers with prognostic potential that bares further investigation, as there are not many studies in the literature that focus on prognosis in AML.

Author contributions

Author contributions are as follows.

Conception: JPA, SR.

Interpretation or analysis of data: LRR, JPA, EC, SR.

Preparation of the manuscript: LRR, JPA, SR.

Revision for important intellectual content: LRR, JPA, EC, SR.

Supervision: SR.

Supplementary data

The supplementary files are available to download from http://dx.doi.org/10.3233/CBM-210540.

sj-docx-1-cbm-10.3233_CBM-210540.docx - Supplemental material

Supplemental material, sj-docx-1-cbm-10.3233_CBM-210540.docx

Footnotes

Acknowledgments

The authors gratefully acknowledge the support received by the Instituto Tecnológico Metropolitano (Project Code: P17215) and by Minciencias (Project Code: 115065743926). The authors would also like to thank the E.S.E. Hospital Manuel Uribe Angel, Colombia.

References

Prada-Arismendy

Arroyave

J.C.J.C.

and Röthlisberger

, Molecular biomarkers in acute myeloid leukemia, Blood Rev 31 (2017), 63–76. doi: 10.1016/j.blre.2016.08.005.

Pollyea

D.A.

Bixby

Perl

Bhatt

V.R.

Altman

J.K.

Appelbaum

F.R.

de Lima

Fathi

A.T.

Foran

J.M.

Gojo

Hall

A.C.

Jacoby

Lancet

Mannis

Marcucci

Martin

M.G.

Mims

Neff

Nejati

Olin

Percival

M.E.

Prebet

Przespolewski

Rao

Ravandi-Kashani

Shami

P.J.

Stone

R.M.

Strickland

S.A.

Sweet

Vachhani

Wieduwilt

Gregory

K.M.

Ogba

and Tallman

M.S.

, NCCN Guidelines Insights: Acute Myeloid Leukemia, Version 2.2021: Featured Updates to the NCCN Guidelines, Journal of the National Comprehensive Cancer Network 19 (2021), 16–27. doi: 10.6004/JNCCN.2021.0002.

Walter

R.B.

Othus

Borthakur

Ravandi

Cortes

J.E.

Pierce

S.A.

Appelbaum

F.R.

Kantarjian

H.A.

and Estey

E.H.

, Prediction of early death after induction therapy for newly diagnosed acute myeloid leukemia with pretreatment risk scores: A novel paradigm for treatment assignment, Journal of Clinical Oncology 29 (2011), 4417–4423. doi: 10.1200/JCO.2011.35.7525.

Mroźek

Marcucci

Niclet

Maharry

K.S.

Becker

Whitman

S.P.

Metzeler

K.H.

Schwind

Y.Z.

Kohlschmidt

Pettenati

M.J.

Heerema

N.A.

Block

A.W.

Patil

S.R.

Baer

M.R.

Kolitz

J.E.

Moore

J.O.

Carroll

A.J.

Stone

R.M.

Larson

R.A.

Bloomfield

C.D.

Walter

R.B.

Othus

Borthakur

Ravandi

Cortes

J.E.

Pierce

S.A.

Appelbaum

F.R.

Kantarjian

H.A.

and Estey

E.H.

, Prognostic significance of the European LeukemiaNet standardized system for reporting cytogenetic and molecular alterations in adults with acute myeloid leukemia, Journal of Clinical Oncology 30 (2012), 4515–4523. doi: 10.1200/JCO.2012.43.4738.

Szklarczyk

Morris

J.H.

Cook

Kuhn

Wyder

Simonovic

Santos

Doncheva

N.T.

Roth

Bork

Jensen

L.J.

and von Mering

, The STRING database in 2017: quality-controlled protein-protein association networks, made broadly accessible, Nucleic Acids Res 45 (2017), D362–D368. doi: 10.1093/nar/gkw937.

Chatr-Aryamontri

Oughtred

Boucher

Rust

Chang

Kolas

N.K.

O’Donnell

Oster

Theesfeld

Sellam

Stark

Breitkreutz

B.J.

Dolinski

and Tyers

, The BioGRID interaction database: 2017 update, Nucleic Acids Res 45 (2017), D369–D379. doi: 10.1093/nar/gkw1102.

Shannon

Markiel

Ozier

Baliga

N.S.

Wang

J.T.

Ramage

Amin

Schwikowski

and Ideker

, Cytoscape: a software environment for integrated models of biomolecular interaction networks, Genome Res 13 (2003), 2498–2504. doi: 10.1101/gr.1239303.

Patra

Ghosh

Bera

Roy

Ray

and Ray

, Molecular characterization of tumor associated glyceraldehyde-3-phosphate dehydrogenase, Biochemistry (Mosc) 74 (2009), 717–727. doi: 10.1134/S0006297909070037.

Chiche

Pommier

Beneteau

Mondragón

Meynet

Zunino

Mouchotte

Verhoeyen

Guyot

Pagès

Mounier

Imbert

Colosetti

Goncalvès

Marchetti

Brière

Carles

Thieblemont

and Ricci

J.E.

, GAPDH enhances the aggressiveness and the vascularization of non-Hodgkin’s B lymphomas via NF-κB-dependent induction of HIF-1α, Leukemia 29 (2015), 1163–1176. doi: 10.1038/LEU.2014.324.

10.

Jun

J.C.

Rathore

Younas

Gilkes

and Polotsky

V.Y.

, Hypoxia-Inducible Factors and Cancer, Curr Sleep Med Rep 3 (2017), 1. doi: 10.1007/S40675-017-0062-7.

11.

Tarze

Deniaud

le Bras

Maillier

Molle

Larochette

Zamzami

Jan

Kroemer

and Brenner

, GAPDH, a novel regulator of the pro-apoptotic mitochondrial membrane permeabilization, Oncogene 26 (2007), 2606–2620. doi: 10.1038/sj.onc.1210074.

12.

Xiao

Wang

Liu

Tang

Deng

and Fang

, Dimethylation of eEF1A at Lysine 55 Plays a Key Role in the Regulation of eEF1A2 on Malignant Cell Functions of Acute Myeloid Leukemia, Technol Cancer Res Treat 19 (2020). doi: 10.1177/1533033820914295.

13.

Cai

Zhang

and Wu

, A three-gene signature and clinical outcome in pediatric acute myeloid leukemia, Clin Transl Oncol 23 (2021), 866–873. doi: 10.1007/S12094-020-02480-X.

14.

Han

Huang

Dai

and Zhang

, Effect of Annexin A1 gene on the proliferation and invasion of esophageal squamous cell carcinoma cells and its regulatory mechanisms, Int J Mol Med 39 (2017), 357–363. doi: 10.3892/ijmm.2016.2840.

15.

Bist

Phua

Q.H.

Shu

Anbalagan

Lee

L.H.

Sethi

Low

B.C.

and Lim

L.H.K.

, Annexin-A1 controls an ERK-RhoA-NκB activation loop in breast cancer cells, Biochem Biophys Res Commun 461 (2015), 47–53. doi: 10.1016/j.bbrc.2015.03.166.

16.

Hsiang

C.H.

Tunoda

Whang

Y.E.

Tyson

D.R.

and Ornstein

D.K.

, The impact of altered annexin I protein levels on apoptosis and signal transduction pathways in prostate cancer cells, Prostate 66 (2006), 1413–1424. doi: 10.1002/pros.20457.

17.

Wan

Xie

Liang

Wen

and Liu

, Expression and clinical significance of ANXA1 and DICER1 in myelodysplastic syndromes, Journal of the College of Physicians and Surgeons Pakistan 30 (2020). doi: 10.29271.

18.

Zhang

Liu

Zhang

Chen

Liu

and Gao

, Dephosphorylation and mitochondrial translocation of cofilin sensitizes human leukemia cells to cerulenin-induced apoptosis via the ROCK1/Akt/JNK signaling pathway, Oncotarget 7 (2016), 20655–20668. doi: 10.18632/oncotarget.7994.

19.

Cheng

Liu

Zhou

Shan

Zhou

Liu

and Gao

, Mitochondrial translocation of cofilin is required for allyl isothiocyanate-mediated cell death via ROCK1/PTEN/PI3K signaling pathway, Cell Communication and Signaling 11 (2013), 50. doi: 10.1186/1478-811X-11-50.

20.

Huang

Zhao

Liu

and Wu

, Cofilin: A Promising Protein Implicated in Cancer Metastasis and Apoptosis, Front Cell Dev Biol 9 (2021). doi: 10.3389/fcell.2021.599065.

21.

Ling

Lei

Jia

Liu

Xia

Tan

Zeng

and Su

, Diallyl disulfide induces downregulation and inactivation of cofilin 1 differentiation via the Rac1/ROCK1/LIMK1 pathway in leukemia cells, Int J Oncol 56 (2020), 772–782. doi: 10.3892/ijo.2020.4968.

22.

Yang

Shen

Ruan

Jin

Song

Guo

Zhang

Shao

and Quan

, CLDN6 promotes chemoresistance through GSTP1 in human breast cancer, Journal of Experimental and Clinical Cancer Research 36 (2017). doi: 10.1186/s13046-017-0627-9.

23.

Rostami

Assad

Ghadyani

Hamid

Karami

Jalaeikhooand

and Kalahroodi

R.A.

, Influence of glutathione S-transferases (GSTM1, GSTT1 and GSTP1) genetic polymorphisms and smoking on susceptibility risk of chronic myeloid leukemia and treatment response, Mol Genet Genomic Med 7 (2019). doi: 10.1002/mgg3.717.

Comparative proteomic analysis of patients with acute myeloid leukemia before and after induction therapy

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1. Introduction

2. Materials and methods

2.1 Experimental subjects

2.2 Isolation of PBMCs and protein extraction

2.3 2D electrophoresis

2.4 Image analysis and 2DE gel comparison

2.5 Mass spectrometry

2.6 Protein network analysis

3. Results

3.1 Clinical characteristics of patients

Table 1 Clinical characteristics of patients included in the study

Table 3 Proteins in the backbone network of unfavorable prognosis

4. Discussion

5. Conclusions

Author contributions

Supplementary data

sj-docx-1-cbm-10.3233_CBM-210540.docx - Supplemental material

Footnotes

Acknowledgments

References

Table 1
Clinical characteristics of patients included in the study

Table 3
Proteins in the backbone network of unfavorable prognosis