Predictive values of plasma TNFα and IL-8 for intracranial hemorrhage in patients with acute promyelocytic leukemia

Abstract

In patients with acute promyelocytic leukemia (APL), intracranial hemorrhage (ICH), if not identified promptly, could be fatal. It is the leading cause of failure of induction and early death. Thus, biomarkers that could promptly predict severe complications are critical. Here, cytokine differences between patients with APL with and without ICH were investigated to develop predictive models for this complication. The initial cytokine profiling using plasma samples from 39 patients and 18 healthy donors found a series of cytokines that were remarkedly different between patients with APL and healthy controls. The APL patients were subsequently divided into high and low white blood cell count groups. Results showed that tumor necrosis factor α and interleukin 8 (IL-8) were vital in distinguishing patients with APL who did or did not develop ICH. In addition, verification in 81 patients with APL demonstrated that the two cytokines were positively correlated with the cumulative incidence of ICH. Finally, in-vitro and in-vivo experimental evidence were provided to show that IL-8 influenced the migration of APL-derived NB4 cells and impaired the blood–brain barrier in PML/RARα positive blast-transplanted FVB/NJ mice. These assessments may facilitate the early warning of ICH and reduce future mortality levels in APL.

Keywords

acute promyelocytic leukemia / intracranial hemorrhage / cytokines / biomarker

1 Introduction

Acute promyelocytic leukemia (APL) is a hematological malignancy driven by the promyelocytic leukemia/retinoic acid receptor-α (PML/RARα) fusion protein. Although great improvement has been shown in prognosis on the account of all-trans retinoic acid (ATRA) and related therapeutic regimens, APL is often characterized by bleeding diathesis, which results in considerable mortality rate attributed to the occurrence of severe hemorrhage before and during induction therapy. The hemorrhaging could be fatal if not identified and intervened promptly [1,2]. Intracranial hemorrhage (ICH), which is often catastrophic and accounts for the highest proportion of fatal bleeding events in APL [3], remains a major challenge for the management of APL.

The potential biomarkers for ICH, especially fatal hemorrhage, include leukocytosis, increased peripheral blast count, decreased platelet count and fibrinogen (Fg) level, prolonged prothrombin time (PT), and increased dehydrogenase and creatinine level. Among these biomarkers, leukocytosis is considered a robust predictor [4,5]. However, many patients without these characteristics still develop ICH, indicating that additional factors may affect the ICH occurrence, especially in patients without leukocytosis. Furthermore, distinguishing patients vulnerable to ICH in a high white blood cell (WBC) group is also challenging. Defective microenvironments are well known to be frequently observed in patients with AML, and they could contribute to cancer initiation and progression [6–8]. Importantly, plasma elements associated with leukemic microenvironments have potential application as translatable biomarkers.

Cytokines form a family of extracellular proteins that stimulate biological responses in hematopoietic cells by activating their cognate receptors, and they play crucial roles in the tight control of hematopoiesis [9–11]. Abnormal cytokine expression is an essential feature of AML, including APL. For example, interleukin 6 (IL-6) secreted by mesenchymal cells contributes to the differentiation and apoptosis of APL cells [12]. Tumor necrosis factor α (TNFα) and IL-6 facilitate the cell death pathway of APL cells by stimulating DNA trap’s release [13]. In addition, wild-type promyelocytic leukemia protein (PML) is a crucial regulator of cytokine signaling [14], and the disruption of PML signaling by PML/RARα results in abnormal cytokine expression in patients with APL. Furthermore, TNFα and IL-1β produced by APL cells could contribute to endothelial and monocyte activation and cause clotting dysfunctions [15]. However, comprehensive cytokine profiling remains under-explored in APL, and investigations for plasma cytokine makers that may predict ICH in APL have not been undertaken.

In this study, to identify the cytokine biomarkers for APL patients with ICH, the clinical information and cytokine profiling of plasma samples in a cohort of newly diagnosed APL patients, accompanied with age- and sex-matched healthy donors, were investigated. Plasma testing and bioinformatic analysis were performed on low and high WBC count groups, respectively, to identify the predictive cytokines for patients with APL-ICH.

2 Materials and methods

2.1 Diagnosis of patients

This study was approved by the Ethics Committee of Ruijin Hospital affiliated to Shanghai Jiao Tong University School of Medicine following the Declaration of Helsinki. Demographic, clinical, and laboratory information of patients with APL were collected at the first referral to the hospital from 2014 to 2017. A total of 120 patients with APL without comorbidities, such as other cancers, chronic hepatitis, severe infection, and autoimmune diseases, were included in the study. Majority of the patients (n = 109) were from the APL2012 trial (NCT01987297), while the remaining 11 patients were not due to suffering from early death in the emergency department. The written informed consent was obtained from all subjects or their legal guardians. The diagnosis for each patient was made on the basis of morphological features and confirmed by the presence of the PML/RARa fusion gene and/or the t(15;17) translocation in bone marrow (BM) samples. Risk classification was performed in accordance with the Sanz criteria [16]. Patients with high WBC count were defined as the same as high-risk patients (WBC count at diagnosis ≥ 10 × 10⁹/L), while patients with low WBC count characterized by WBC count at diagnosis < 10 × 10 ⁹/L were denoted as low- and intermediate-risk patients. ICH was diagnosed using computed tomography scans or by expert neurologists. In total, 80 age- and sex-matched healthy donors were also included.

2.2 Sample collection and analysis

Plasma samples from 120 patients with APL were collected at diagnosis before starting medication. Among them, the samples of 39 patients were examined using Luminex testing for 41 cytokines, chemokines, and growth factors (Table S1). Fifteen patients were randomly selected from all the patients with ICH to avoid the influence of sample size on the authenticity of the results. The remaining 24 patients without ICH were also chosen randomly from those without ICH. The concentrations of the tested elements were measured in duplicates. Assay was implemented using Bio-Rad Pro Human Cytokine multiplex kits (Bio-Rad, Hercules, CA, USA) in accordance with the standard procedures. The results were evaluated using Bio-Plex Manager 6.1 (Luminex, Austin, TX, USA). Samples from 18 healthy donors were used as controls. Afterwards, the two important cytokines were validated in the remaining 81 patients with APL by using chemiluminescent immunoassay (IMMULITE/IMMULITE 1000, Siemens Healthineers AG, Erlangen, Germany).

2.3 Treatment strategy

ATRA and arsenic trioxide (ATO), the current first-line treatment for APL, were taken as induction treatment (ATRA 25 mg/m² per day orally and ATO 0.16 mg/kg intravenously daily till complete remission, with weekends not excluded). High-risk patients (WBC count at diagnosis ≥ 10 × 10⁹/L) and intermediate-risk patients suffering from leukocytosis later (WBC count > 10 × 10 ⁹/L) were additionally given anthracycline during the induction course. Patients who suffered from ICH were treated with blood components, such as platelets, cryoprecipitate, and mannitol and/or glycerol fructose, to reduce intracranial hypertension and edema.

2.4 Cell culture and transwell assay

NB4 cells were cultured in RPMI 1640 (Gibco, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS, Gibco) and maintained in a humidified and 5% CO₂ atmosphere at 37 °C. These cells were washed and suspended in RPMI 1640 medium, followed by serum starvation for 12 h. Human IL-8 (8.9 kDa protein containing 77 amino acids; Perrotech, NJ, USA) was plated into the lower chamber of a 24-well plate transwell cabin with 0.8 mm pores (Corning Costar, NY, USA) in RPMI 1640 medium with 20% FBS. The NB4 cells (2 × 10⁶/mL) were added into the upper chamber. After 24 h of incubation, the cell counts in the lower chamber were calculated using Counterstar BioTech Automated Cell Counter (ALIT Life Science Co., Ltd., Shanghai, China).

2.5 Blood–brain barrier leakage tests

The animal studies were approved by the Animal Care and Use Committee of Ruijin Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. PML/RARα positive blasts from APL transplantable mice were transduced into FVB/NJ mice by tail intravenous injection. Premorbid FVB mice were then treated with 100 ng/mL (100 µL) of IL-8 after being anesthetized with isoflurane (2%; Yuyan Instrument Co., Ltd., Shanghai, China). A total of 200 μL of Evans blue staining (2%; Aicon Biotech, Tigard, USA) was injected into the mice via cardiac perfusion. The mice were homogenized in 1000 μL of PBS, and the brains were harvested and observed using a Nikon SMZ18 microscope (Nikon Corp., Tokyo, Japan).

2.6 Statistical analysis

Continuous variable distribution was analyzed using Mann–Whitney U test. Groups with nominal variables were compared using either Chi-square or Fisher’s exact test. The random forest technique (“randomForest package,” version 4.6-14 in R) [17,18] was used to estimate the weights of the candidate cytokines in each tree. By using the four measures of variable importance from the random forest, which factors have the most classified power could be determined. Normally, the factors with higher value for importance measure has more effect on the classification of random forest. All codes are available upon reasonable request. Cox proportional hazard regression was performed for multivariable analysis. P value < 0.2 was taken as the threshold to be included in multivariate analysis. Cumulative incidence was considered to estimate the incidence of ICH within 40 days of diagnosis. P value < 0.05 was considered significant. Statistical analyses were conducted using SPSS 22.0 statistic software (IBM, Armonk, NY, USA).

3 Results

3.1 Patient characteristics

Cytokine profiling was performed using plasma samples from 39 patients with APL evaluated with Luminex assay. The median age was 36 years (range, 16–70), within which 15 cases (38.46%) suffered from ICH, containing six (15.38%) patients who developed hemorrhagic death. The 15 patients with ICH comprised one low-risk, six intermediate-risk, and eight high-risk patients. The remaining 24 patients (five, 10, and nine in the low-, intermediate-, and high-risk groups, respectively) did not endure ICH. Their detailed clinical information is provided in Tab.1. In a baseline assessment of ICH-related clinical factors, the intergroup comparison between patients with APL-ICH (n = 15) and those with APL but without ICH (APL-non-ICH, n = 24) subsets showed that patients with lower platelet counts (P = 0.028) and Fg levels (P = 0.038), increased D-dimer (P = 0.035), and high BM blast counts (P = 0.003) tended to suffer from ICH. Patients with higher WBC counts (P = 0.058) and elevated international normalized ratios (INRs, P = 0.058) showed tendency towards ICH. Only higher lactic dehydrogenase (LDH; hazard ratio (HR) = 3.62, 95% confidence interval (CI) = 1.26–10.42, P = 0.017) showed significance in the multivariable analysis (Tab.1).

3.2 Identification of cytokines as predictive biomarkers for patients with ICH

Luminex assays were conducted on 41 cytokines, chemokines, and growth factors (Table S1). In 18 healthy donors, the concentrations of nine factors, namely, IL-1α, IL-1β, IL-10, IL-2, IL-5, LIF, M-CSF, TNFα, and TNFβ, were below the detection limits. In the patients with APL, 35 of 41 (approximately 85%) pro-inflammatory and proliferation-associated cytokines, chemokines, and growth factors were significantly different compared with those in healthy controls, with 30 elevated and five depressed (Tab.2).

Several factors were significantly differentially expressed between the APL-ICH and APL-non-ICH groups. The circulation levels of the angiogenesis-promoting chemokine SDF-1α were in line with those of the apoptosis-promoting cytokine TRAIL, and both were lower in patients with APL-ICH. Pro-inflammatory factors, such as IL-2Rα and IL-8, and angiogenesis-promoting cytokine HGF were increased in patients with APL-non-ICH (Fig.1).

3.3 TNFα and IL-8 were the critical cytokines associated with ICH in low and high WBC groups, respectively

As an influencing factor for ICH, leukocytosis is an important confounding factor that affects the expression of inflammatory factors. Thus, the 39 APL patients were further divided into low WBC count (n = 22, seven with ICH and 15 without ICH) and the high WBC count (n = 17, eight with ICH and nine without ICH) groups. Random forest permutation-based variable importance measures were performed to determine the potential biomarkers to predict ICH at diagnosis. In the low WBC count group, three of the four variable importance measures indicated that TNFα was the most critical cytokine for distinguishing patients having ICH. Meanwhile, MIP-1β was nominated by three measures as the second important factor. The following factors were IL-2Rα, RANTES, and GROα (Tab.3). Therefore, TNFα was determined to be the most powerful predictive cytokine in the low WBC count group. For the patients with high WBC count, IL-8 was confirmed as the most important factor by measures 1, 2, and 3, followed by monocyte chemotactic protein-1 (MCP-1), which was recognized as second most important (ranked the first by measure 4 and second by measures 1 and 3). The remaining factors among the best were basic FGF (ranked relatively high by all of the measures) and SDF-1α (ranked relatively high by measures 1 and 3) in addition to TNFα (ranked relatively high by measures 1 and 4). The comparison between TNFα and IL-8 expression in the ICH and non-ICH groups is shown in Table S2.

3.4 Cumulative incidence are related to plasma cytokines

As indicated by the strengths of predictive cytokine values, abnormal plasma concentrations of TNFα and IL-8 may also predict ICH-related incidence. Considering that ICH mainly occurs in the early stage of disease onset [19,20], a 40-day near-term follow-up in 81 patients with APL (28 high-risk and 53 low-to-intermediate risk patients, Table S3) was conducted to determine the correlations between cytokines and prognosis. In terms of the high WBC count group, the incidence of ICH was 50% (95% CI = 16.3%–76.8%) in patients with increased IL-8 levels, significantly higher than that in patients with lower IL-8 level (5.6%, 95% CI = 0.3%–23.1%, P = 0.008; Fig.2). In the low WBC count group, patients with increased TNFα had a significantly higher ICH cumulative incidence (69.2%, 95% CI = 34.2%–88.1%, P < 0.001; Fig.2) than those with lower TNFα level (2.5%, 95% CI = 0.2%–11.4%). In total, only one (1.89%) of the patients with ICH had a lower TNFα level. In addition, four of the remaining 43 patients had higher TNFα levels (9.3%). TNFα and IL-8 also showed higher expression in patients with ICH in the two groups (Table S2).

3.5 Effects of IL-8 on inducing the migration of NB4 cells

Given that this study is the first report of IL-8 serving as an APL-related ICH-associated cytokine, whether IL-8 contributes to vessel infringement through the migration of APL blasts was explored. Transwell experiments were performed to explore the role of IL-8 in the chemotaxis on patients-with-APL-derived NB4 cells. As shown in Fig.3 and 3B, the number of NB4 cells migrating to the lower chamber significantly increased along with the IL-8 concentration in the lower chamber compared with the control group (P < 0.001). The difference between the 100 and 200 ng/mL IL-8 groups indicated that the mobility of NB4 cells increased along with the IL-8 concentration.

3.6 Effects of IL-8 to the injury of the blood–brain barrier

An APL mouse model was utilized to examine the effect of IL-8 on vessels. PML/RARα positive blasts were transplanted into FVB/NJ mice, and the development of APL was mimicked. The mice were treated with IL-8 through intracardiac injections before APL onset. As shown in Fig.3, Evans blue staining showed an increased permeability in IL-8-treated FVB mice. The data indicated that IL-8 was capable of destroying the blood–brain barrier of APL mice.

4 Discussion

An abnormal cytokine milieux located in myeloid neoplasms contributes to the proliferation and survival of leukemic cells [21–26]. The present study comprehensively investigated the constitutive abnormalities in the plasma surroundings of patients with APL and established translatable biomarkers for APL-related ICH.

A series of cytokines, chemokines, and growth factors was found to be elevated in patients with APL compared with healthy donors. Inflammatory mediator synthesis could impede the differentiation of APL cells and contribute to cancer initiation and progression [27,28]. For instance, IL-13 suppresses group 2 innate lymphoid cells and monocytic myeloid-derived suppressor cells in APL, resulting in the increased survival rate of leukemia cells [29]. The abnormal immune-inflammatory state is involved in the migration, invasion, and metastasis of malignant cells [30], as exemplified by the migration-promoting function of MCP-1 and IL-8 in APL cells [31,32]. Moreover, many cytokines, chemokines, and growth factors, especially pro-inflammatory factors IL-2Rα and IL-8, and proliferation-related factor HGF, in the APL-ICH group were elevated compared with those in the APL-non-ICH group. On the contrary, angiogenesis- and apoptosis-promoting factors were abnormal in the APL-ICH group, suggesting that immune inflammation, vascular endothelium disorder, and excessive leukocyte proliferation are highly correlated with ICH in APL. TNFα and IL-8 were determined to be critical cytokines associated with ICH in the low and high WBC count groups, respectively.

TNFα has been reported to be produced by APL cells and associated with severe toxicity in the pro-inflammatory processes [33]. TNFα also enhances the effects of antiangiogenic and pro-inflammatory function and inhibits the angiogenic function of pericytes; it forms an important cellular component of the blood–brain barrier [34]. TNFα contributes to the loss of the endothelial barrier integrity in a model of blood–brain barrier composed of primary human brain-derived pericytes and endothelia [35]. Besides, leukemia cells overexpress anti-apoptotic genes in response to TNFα via the TNF/JNK-AP1 signal [36]. In addition, TNFα has been used to mimic leaky blood–brain barrier by modulating endothelial permeability [37]. The predictive approach in the present study may ultimately serve as a powerful guide for clinical decisions.

Plasma IL-8 levels had a strong ability to predict ICH in the high WBC groups of patients with APL. With a relatively ideal sensitivity, IL-8 could recognize the risk of ICH among patients under hyperleukocytosis state and predict ICH more accurately than the increased WBC count that is frequently used in present clinical practice. IL-8 is generally secreted by activated microglia, astrocytes, endothelial cells, and macrophages. The expression of IL-8 is correlated with brain endothelial injury in infectious disease in early reports [38]. The CXCL8 (IL-8)/CXCR2 network aggravates brain parenchymal damage caused by neutrophil infiltration [39], suggesting that plasma IL-8 promotes the deterioration of the microenvironment through the positive feedback triggered by the endothelial stimulation. In the present study, the effects of IL-8 were confirmed using in-vitro and in-vivo experiments. The results showed that IL-8 effectively increased the migration of APL cells and the permeability of the blood–brain barrier, consistent with the functions of IL-8 reported in other neoplasms [40,41].

Recent studies have explored the risk factors of ICH in APL, including low albumin, high creatinine, uric acid, aspartic acid transferase, and body mass index [42–44]. However, these indicators always appear in the terminal stage of the disease, sacrificing the best time to predict and not providing accurate representativeness for leukemia-related pathophysiological processes. A practical biomarker for anticipation is also scanty in terms of APL-ICH, especially for patients with low WBC count. Besides, specific therapeutic targets for ICH in APL have not been provided hitherto. Cytokine biomarkers could probably act as therapeutic targets regarding APL-ICH by facilitating the practicability of predictive cytokine models. Moreover, previous studies on hemorrhage predictors were mainly based on clinical trial cohorts, in which the vast majority of patients with ICH were composed of individuals with high WBC due to screening procedures. Therefore, patients who died in an emergency before incorporation were excluded. In addition, detailed analysis for ICH in patients with low WBC has been ignored due to insufficient patients samples. In the present study, efforts were made to avoid these shortcomings by gathering the individuals at first referral, whether emergency or outpatient. Although the predictive value was demonstrated, the sample size was limited, with the WBC count not reaching a significant level of difference.

As noted above, intracranial bleeding events occupy most early death and remain the problem that needs to be solved urgently. This study initially highlighted the predictive value of cytokine profiling for the issue in patients with APL. In the future, more sufficient and balanced sampling across diverse WBC level groups and mechanism studies are necessary to determine the biological role of the three cytokines in the ICH course and potentially attribute them to targeted therapy to prevent ICH in APL.

References

[1] dela Serna J, MontesinosP, VellengaE, RayónC, ParodyR, LeónA, EsteveJ, BerguaJM, MiloneG, DebénG, RivasC, GonzálezM, TormoM, Díaz-MediavillaJ, GonzálezJD, NegriS, AmutioE, BrunetS, LowenbergB, SanzMA. Causes and prognostic factors of remission induction failure in patients with acute promyelocytic leukemia treated with all-trans retinoic acid and idarubicin. Blood 2008; 111( 7): 3395– 3402

[2] YanadaM, MatsushitaT, AsouN, KishimotoY, TsuzukiM, MaedaY, HorikawaK, OkadaM, OhtakeS, YagasakiF, MatsumotoT, KimuraY, ShinagawaK, IwanagaM, MiyazakiY, OhnoR, NaoeT. Severe hemorrhagic complications during remission induction therapy for acute promyelocytic leukemia: incidence, risk factors, and influence on outcome. Eur J Haematol 2007; 78( 3): 213– 219

[3] JillellaAP, KotaVK. The global problem of early deaths in acute promyelocytic leukemia: a strategy to decrease induction mortality in the most curable leukemia. Blood Rev 2018; 32( 2): 89– 95

[4] ManthaS, GoldmanDA, DevlinSM, LeeJW, ZanninoD, CollinsM, DouerD, IlandHJ, LitzowMR, SteinEM, AppelbaumFR, LarsonRA, StoneR, PowellBL, GeyerS, LaumannK, RoweJM, ErbaH, CoutreS, OthusM, ParkJH, WiernikPH, TallmanMS. Determinants of fatal bleeding during induction therapy for acute promyelocytic leukemia in the ATRA era. Blood 2017; 129( 13): 1763– 1767

[5] Mantha S, Tallman MS, Devlin SM, Soff GA. Predictive factors of fatal bleeding in acute promyelocytic leukemia. Thromb Res 2018; 164(Supp 1): S98–S102

[6] De BoerB, ShevelevaS, ApeltK, VellengaE, MulderAB, SchuringaGH, JacobJ. The IL1–IL1RAP axis plays an important role in the inflammatory leukemic niche that favors acute myeloid leukemia proliferation over normal hematopoiesis. Haematologica 2021; 106( 12): 3067– 3078

[7] ForteD, García-FernándezM, Sánchez-Aguilera A, StavropoulouV, FieldingC, Martín-PérezD, LópezJA, CostaASH, TronciL, NikitopoulouE, BarberM, GallipoliP, MarandoL, Fernándezde Castillejo CL, TzankovA, DietmannS, CavoM, CataniL, CurtiA, VázquezJ, FrezzaC, HuntlyBJ, SchwallerJ, Méndez-FerrerS. Bone marrow mesenchymal stem cells support acute myeloid leukemia bioenergetics and enhance antioxidant defense and escape from chemotherapy. Cell Metab 2020; 32( 5): 829– 843.e9

[8] StuaniL, SarryJE. Microenvironmental aspartate preserves leukemic cells from therapy-induced metabolic collapse. Cell Metab 2020; 32( 3): 321– 323

[9] BulaevaE, PellacaniD, NakamichiN, HammondCA, BeerPA, LorzadehA, MoksaM, CarlesA, BilenkyM, LefortS, ShuJ, WilhelmBT, WengAP, HirstM, EavesCJ. MYC-induced human acute myeloid leukemia requires a continuing IL-3/GM-CSF costimulus. Blood 2020; 136( 24): 2764– 2773

[10] YamashitaM, PasseguéE. TNF-α coordinates hematopoietic stem cell survival and myeloid regeneration. Cell Stem Cell 2019; 25( 3): 357– 372.e7

[11] SinclairA, ParkL, ShahM, DrotarM, CalaminusS, HopcroftLE, KinstrieR, GuitartAV, DunnK, AbrahamSA, SansomO, MichieAM, MacheskyL, KrancKR, GrahamGJ, PellicanoF, HolyoakeTL. CXCR2 and CXCL4 regulate survival and self-renewal of hematopoietic stem/progenitor cells. Blood 2016; 128( 3): 371– 383

[12] ChenF, ZhouK, ZhangL, MaF, ChenD, CuiJ, FengX, YangS, ChiY, HanZ, XueF, RongL, GeM, WanL, XuS, DuW, LuS, RenH, HanZ. Mesenchymal stem cells induce granulocytic differentiation of acute promyelocytic leukemic cells via IL-6 and MEK/ERK pathways. Stem Cells Dev 2013; 22( 13): 1955– 1967

[13] MaR, LiT, CaoM, SiY, WuX, ZhaoL, YaoZ, ZhangY, FangS, DengR, NovakovicVA, BiY, KouJ, YuB, YangS, WangJ, ZhouJ, ShiJ. Extracellular DNA traps released by acute promyelocytic leukemia cells through autophagy. Cell Death Dis 2016; 7( 6): e2283

[14] PalibrkV, SuganthanR, SchefflerK, WangW, BjøråsM, BøeSO. PML regulates neuroprotective innate immunity and neuroblast commitment in a hypoxic-ischemic encephalopathy model. Cell Death Dis 2016; 7( 7): e2320

[15] BreenKA, GrimwadeD, HuntBJ. The pathogenesis and management of the coagulopathy of acute promyelocytic leukaemia. Br J Haematol 2012; 156( 1): 24– 36

[16] SanzMA, LoCoco F, MartínG, AvvisatiG, RayónC, BarbuiT, Díaz-MediavillaJ, FioritoniG, GonzálezJD, LisoV, EsteveJ, FerraraF, BoluferP, BernasconiC, GonzalezM, RodeghieroF, ColomerD, PettiMC, RiberaJM, MandelliF. Definition of relapse risk and role of nonanthracycline drugs for consolidation in patients with acute promyelocytic leukemia: a joint study of the PETHEMA and GIMEMA cooperative groups. Blood 2000; 96( 4): 1247– 1253

[17] BreimanL. Random Forests. Mach Learn 2001; 45( 1): 5– 32

[18] LiawA, WienerM. Classification and regression by randomForest. R News 2002; 2 : 18– 22

[19] LehmannS, RavnA, CarlssonL, AntunovicP, DenebergS, MöllgårdL, DerolfAR, StockelbergD, TidefeltU, WahlinA, WennströmL, HöglundM, JuliussonG. Continuing high early death rate in acute promyelocytic leukemia: a population-based report from the Swedish Adult Acute Leukemia Registry. Leukemia 2011; 25( 7): 1128– 1134

[20] GurnariC, BrecciaM, Di GiulianoF, ScalzulliE, DivonaM, PiciocchiA, CicconiL, De BellisE, VendittiA, Del PrincipeMI, ArceseW, Lo-CocoF, GaraciF, VosoMT. Early intracranial haemorrhages in acute promyelocytic leukaemia: analysis of neuroradiological and clinico-biological parameters. Br J Haematol 2021; 193( 1): 129– 132

[21] Sanchez-CorreaB, BerguaJM, CamposC, GayosoI, ArcosMJ, BañasH, MorgadoS, CasadoJG, SolanaR, TarazonaR. Cytokine profiles in acute myeloid leukemia patients at diagnosis: survival is inversely correlated with IL-6 and directly correlated with IL-10 levels. Cytokine 2013; 61( 3): 885– 891

[22] CareyA, EdwardsDK 5th, EideCA, NewellL, TraerE, MedeirosBC, PollyeaDA, DeiningerMW, CollinsRH, TynerJW, DrukerBJ, BagbyGC, McWeeneySK, AgarwalA. Identification of interleukin-1 by functional screening as a key mediator of cellular expansion and disease progression in acute myeloid leukemia. Cell Rep 2017; 18( 13): 3204– 3218

[23] KatsumuraKR, OngIM, DeVilbissAW, SanalkumarR, BresnickEH. GATA factor-dependent positive-feedback circuit in acute myeloid leukemia cells. Cell Rep 2016; 16( 9): 2428– 2441

[24] BinderS, LucianoM, Horejs-HoeckJ. The cytokine network in acute myeloid leukemia (AML): a focus on pro- and anti-inflammatory mediators. Cytokine Growth Factor Rev 2018; 43 : 8– 15

[25] BiKH, JiangGS. Relationship between cytokines and leukocytosis in patients with APL induced by all-trans retinoic acid or arsenic trioxide. Cell Mol Immunol 2006; 3( 6): 421– 427

[26] ChoudhryA, DeLougheryTG. Bleeding and thrombosis in acute promyelocytic leukemia. Am J Hematol 2012; 87( 6): 596– 603

[27] JambrovicsK, UrayIP, KeresztessyZ, KeillorJW, FésüsL, BalajthyZ. Transglutaminase 2 programs differentiating acute promyelocytic leukemia cells in all-trans retinoic acid treatment to inflammatory stage through NF-κB activation. Haematologica 2019; 104( 3): 505– 515

[28] HanahanD, WeinbergRA. Hallmarks of cancer: the next generation. Cell 2011; 144( 5): 646– 674

[29] TrabanelliS, ChevalierMF, Martinez-UsatorreA, Gomez-CadenaA, SaloméB, LeccisoM, SalvestriniV, VerdeilG, RacleJ, PapayannidisC, MoritaH, PizzitolaI, GrandclémentC, BohnerP, BruniE, GirotraM, PallaviR, FalvoP, LeibundgutEO, BaerlocherGM, Carlo-StellaC, TaurinoD, SantoroA, SpinelliO, RambaldiA, GiarinE, BassoG, TresoldiC, CiceriF, GfellerD, AkdisCA, MazzarellaL, MinucciS, PelicciPG, MarcenaroE, McKenzieANJ, VanheckeD, CoukosG, MavilioD, CurtiA, DerréL, JandusC. Tumour-derived PGD2 and NKp30-B7H6 engagement drives an immunosuppressive ILC2-MDSC axis. Nat Commun 2017; 8( 1): 593

[30] MantovaniA, AllavenaP, SicaA, BalkwillF. Cancer-related inflammation. Nature 2008; 454( 7203): 436– 444

[31] TsaiWH, HsuHC, LinCC, HoCK, KouYR. Role of interleukin-8 and growth-regulated oncogene-alpha in the chemotactic migration of all-trans retinoic acid-treated promyelocytic leukemic cells toward alveolar epithelial cells. Crit Care Med 2007; 35( 3): 879– 885

[32] TsaiWH, ShihCH, LinCC, HoCK, HsuFC, HsuHC. Monocyte chemotactic protein-1 in the migration of differentiated leukaemic cells toward alveolar epithelial cells. Eur Respir J 2008; 31( 5): 957– 962

[33] Ataca AtillaP, McKennaMK, TashiroH, SrinivasanM, MoF, WatanabeN, SimonsBW, McLean StevensA, RedellMS, HeslopHE, MamonkinM, BrennerMK, AtillaE. Modulating TNFα activity allows transgenic IL15-Expressing CLL-1 CAR T cells to safely eliminate acute myeloid leukemia. J Immunother Cancer 2020; 8( 2): e001229

[34] KangTY, BocciF, JollyMK, LevineH, OnuchicJN, LevchenkoA. Pericytes enable effective angiogenesis in the presence of proinflammatory signals. Proc Natl Acad Sci USA 2019; 116( 47): 23551– 23561

[35] SmythLCD, RustenhovenJ, ParkTI, SchwederP, JanssonD, HeppnerPA, O’CarrollSJ, MeeEW, FaullRLM, CurtisM, DragunowM. Unique and shared inflammatory profiles of human brain endothelia and pericytes. J Neuroinflammation 2018; 15( 1): 138

[36] VolkA, LiJ, XinJ, YouD, ZhangJ, LiuX, XiaoY, BreslinP, LiZ, WeiW, SchmidtR, LiX, ZhangZ, KuoPC, NandS, ZhangJ, ChenJ, ZhangJ. Co-inhibition of NF-κB and JNK is synergistic in TNF-expressing human AML. J Exp Med 2014; 211( 6): 1093– 1108

[37] LeeTH, HsiehST, ChiangHY. Fibronectin inhibitor pUR4 attenuates tumor necrosis factor α-induced endothelial hyperpermeability by modulating β1 integrin activation. J Biomed Sci 2019; 26( 1): 37

[38] DoranKS, LiuGY, NizetV. Group B streptococcal beta-hemolysin/cytolysin activates neutrophil signaling pathways in brain endothelium and contributes to development of meningitis. J Clin Invest 2003; 112( 5): 736– 744

[39] SempleBD, KossmannT, Morganti-KossmannMC. Role of chemokines in CNS health and pathology: a focus on the CCL2/CCR2 and CXCL8/CXCR2 networks. J Cereb Blood Flow Metab 2010; 30( 3): 459– 473

[40] MengL, ZhaoY, BuW, LiX, LiuX, ZhouD, ChenY, ZhengS, LinQ, LiuQ, SunH. Bone mesenchymal stem cells are recruited via CXCL8-CXCR2 and promote EMT through TGF-β signal pathways in oral squamous carcinoma. Cell Prolif 2020; 53( 8): e12859

[41] JayatilakaH, TyleP, ChenJJ, KwakM, JuJ, KimHJ, LeeJSH, WuPH, GilkesDM, FanR, WirtzD. Synergistic IL-6 and IL-8 paracrine signalling pathway infers a strategy to inhibit tumour cell migration. Nat Commun 2017; 8( 1): 15584

[42] CaiP, WuQ, WangY, YangX, ZhangX, ChenS. An effective early death scoring system for predicting early death risk in de novo acute promyelocytic leukemia. Leuk Lymphoma 2020; 61( 8): 1989– 1995

[43] JinB, ZhangY, HouW, CaoF, LuM, YangH, TianX, WangY, HouJ, FuJ, LiH, ZhouJ. Comparative analysis of causes and predictors of early death in elderly and young patients with acute promyelocytic leukemia treated with arsenic trioxide. J Cancer Res Clin Oncol 2020; 146( 2): 485– 492

[44] AblaO, RibeiroRC, TestiAM, MontesinosP, CreutzigU, SungL, Di GiuseppeG, StephensD, FeusnerJH, PowellBL, HasleH, KaspersGJL, Dalla-PozzaL, LassalettaA, TallmanMS, LocatelliF, ReinhardtD, Lo-CocoF, HitzlerJ, SanzMA. Predictors of thrombohemorrhagic early death in children and adolescents with t(15;17)-positive acute promyelocytic leukemia treated with ATRA and chemotherapy. Ann Hematol 2017; 96( 9): 1449– 1456

RIGHTS & PERMISSIONS

Higher Education Press