Comparison of the efficacy of hematopoietic stem cell mobilization regimens: a systematic review and network meta-analysis of preclinical studies

Mobilization failure may occur when the conventional hematopoietic stem cells (HSCs) mobilization agent granulocyte colony-stimulating factor (G-CSF) is used alone, new regimens were developed to improve mobilization efficacy. Multiple studies have been performed to investigate the efficacy of these regimens via animal models, but the results are inconsistent. We aim to compare the efficacy of different HSC mobilization regimens and identify new promising regimens with a network meta-analysis of preclinical studies. We searched Medline and Embase databases for the eligible animal studies that compared the efficacy of different HSC mobilization regimens. Primary outcome is the number of total colony-forming cells (CFCs) in per milliliter of peripheral blood (/ml PB), and the secondary outcome is the number of Lin− Sca1+ Kit+ (LSK) cells/ml PB. Bayesian network meta-analyses were performed following the guidelines of the National Institute for Health and Care Excellence Decision Support Unit (NICE DSU) with WinBUGS version 1.4.3. G-CSF-based regimens were classified into the SD (standard dose, 200–250 μg/kg/day) group and the LD (low dose, 100–150 μg/kg/day) group based on doses, and were classified into the short-term (2–3 days) group and the long-term (4–5 days) group based on administration duration. Long-term SD G-CSF was chosen as the reference treatment. Results are presented as the mean differences (MD) with the associated 95% credibility interval (95% CrI) for each regimen. We included 95 eligible studies and reviewed the efficacy of 94 mobilization agents. Then 21 studies using the poor mobilizer mice model (C57BL/6 mice) to investigate the efficacy of different mobilization regimens were included for network meta-analysis. Network meta-analyses indicated that compared with long-term SD G-CSF alone, 14 regimens including long-term SD G-CSF + Me6, long-term SD G-CSF + AMD3100 + EP80031, long-term SD G-CSF + AMD3100 + FG-4497, long-term SD G-CSF + ML141, long-term SD G-CSF + desipramine, AMD3100 + meloxicam, long-term SD G-CSF + reboxetine, AMD3100 + VPC01091, long-term SD G-CSF + FG-4497, Me6, long-term SD G-CSF + EP80031, POL5551, long-term SD G-CSF + AMD3100, AMD1300 + EP80031 and long-term LD G-CSF + meloxicam significantly increased the collections of total CFCs. G-CSF + Me6 ranked first among these regimens in consideration of the number of harvested CFCs/ml PB (MD 2168.0, 95% CrI 2062.0−2272.0). In addition, 7 regimens including long-term SD G-CSF + AMD3100, AMD3100 + EP80031, long-term SD G-CSF + EP80031, short-term SD G-CSF + AMD3100 + IL-33, long-term SD G-CSF + ML141, short-term LD G-CSF + ARL67156, and long-term LD G-CSF + meloxicam significantly increased the collections of LSK cells compared with G-CSF alone. Long-term SD G-CSF + AMD3100 ranked first among these regimens in consideration of the number of harvested LSK cells/ml PB (MD 2577.0, 95% CrI 2422.0–2733.0). Considering the number of CFC and LSK cells in PB as outcomes, G-CSF plus AMD3100, Me6, EP80031, ML141, FG-4497, IL-33, ARL67156, meloxicam, desipramine, and reboxetine are all promising mobilizing regimens for future investigation.


Background
Hematopoietic stem cell transplantation (HSCT) is a lifesaving strategy for variety of severe disorders, including bone marrow failure after high-dose radiation and various hematological malignancies [1]. Peripheral blood stem cells (PBSCs) have been gradually replaced bone marrow (BM) as the predominant source of stem cell for transplantation in clinical practice [2,3]. It has been demonstrated that PBSCs transplantation is associated with more convenient and safer harvest procedure, faster hematological recovery, lower risk of graft failure, and comparable disease-free survival (DFS) and overall survival (OS) in comparison with BM transplantation [4][5][6]. However, hematopoietic stem cells (HSCs) mainly reside in specialized BM niches during steady state; the number of HSCs circulating in peripheral blood (PB) is very low and not sufficient for harvest [6]. Administration of exogenous cytokines or chemokines could induce the egress of HSCs from BM into PB in a process termed mobilization. Successful mobilization allows for efficient collection of HSCs sufficient for transplantation, and increment in the dose of harvested HSCs could improve transplantation efficiency via promoting hematopoietic reconstitution, as well as reducing the need for supportive blood transfusion and the risks of infections [7,8]. Therefore, efficient mobilization is the key to successful HSCT and sustained hematopoietic recovery.
Granulocyte colony-stimulating factor (G-CSF) is the most commonly used steady-state HSC mobilization agent in clinical practice. However, mobilization failure may occur when G-CSF is used alone [8]. In addition, mobilization using G-CSF alone requires multiple doses beginning at least 4 days before first apheresis and a median of 2-5 apheresis sessions to collect sufficient PBSCs, which increased the risk of adverse events [7]. The incidences of bone pain induced by G-CSF is higher than 80% at day 4, in addition, other G-CSF-related severe adverse events including myocardial infarctions, pulmonary embolism, and splenic rupture also have been reported [9][10][11]. To improve mobilization efficacy and attenuate toxicity, novel mobilization regimens are developed and investigated in a variety of animal studies before applied in clinical practice, but the results are inconsistent. This study aims to review and compare the efficacy of different HSC mobilization regimens and identify new promising regimens with a network metaanalysis of preclinical studies, which may be helpful for guiding future clinical trials.

Literature search and study selection
We searched Medline and Embase from inception to February 23, 2021, with the search term "stem cell mobilization" and a filter of "animals". The titles and abstracts of retrieved citations were independently screened by two investigators (CXL and XL) for eligibility. Disagreements were resolved by full-text review and discussion with a third investigator (SNX). Preclinical studies that met the following criteria were included for review: (1) compared the efficacy of two or more different regimens in the mobilization of hematopoietic stem and progenitor cells (HSPCs) and (2) using any species of mice as experimental animals. As for network metaanalysis, the inclusion criteria were (1) using the poor mobilizer mice model-C57BL/6 mice as experimental animals [12] and (2) reporting data for at least one of the outcomes of mobilization efficacy, including the number of total colony-forming cells (CFCs) and Lin − Sca1 + Kit + (LSK) cells per milliliter of peripheral blood (/ml PB). Since aged mice were reported to have better mobilization efficiency compared with young mice and no significant difference was reported among mice younger than 3 months, we excluded studies using mice older than 12 weeks in meta-analysis to reduce heterogeneity [13]. In addition, we only included studies that administrated G-CSF via subcutaneously injection. Furthermore, G-CSF-based regimens were classified into the SD (standard dose, 200-250 μg/kg/day) group and the LD (low dose, 100-150 μg/kg/day) group based on G-CSF doses and were classified into the short-term (2-3 days) group and the long-term (4-5 days) group based on administration duration of G-CSF. Studies with significant heterogeneity in dosage and injection route of G-CSF were excluded in meta-analysis.

Data extraction and quality assessment
Full text of all eligible studies was reviewed, and two investigators (CXL and XL) independently extracted data using predesigned data collection forms. Data was extracted on studies characteristics, animal's characteristics, dosage of mobilization regimens, and efficacy outcomes. We chose the number of total CFCs per milliliter PB as primary outcome, and the number of LSK cells per milliliter PB as secondary outcome. The mean, standard deviation (SD) or standard error (SE) of each outcome are extracted directly from published text or from related graphs with Adobe Photoshop version CS3 via previously validated methods [14]. The methodological quality of included studies was assessed using the SYstematic Review Centre for Laboratory animal Experimentation (SYRCLE) risk of bias tool, which contains 10 items, including random sequence generation, similar baseline characteristics, allocation concealment, random housing, blinding of caregivers and investigators, random selection for outcome assessment, blinding of outcome assessor, adequate addressing of incomplete outcome data, free from selective outcome reporting, and free from other bias [15]. For each item, judgment of "yes", "no", and "unclear" respectively indicate low, high and unclear risk of bias.

Statistical analyses
We conducted network meta-analyses to compare the efficacy of multiple mobilization regimens simultaneously. Network plot for each outcome was obtained using Stata version 12.0. Bayesian network meta-analyses were performed with WinBUGS version 1.4.3 (MRC Biostatistics Unit, Cambridge, UK), employing the Markov Chain Monte Carlo (MCMC) approach and following the guidelines of the National Institute for Health and Care Excellence Decision Support Unit (NICE DSU) [16]. We used the WinBUGS code previously established by Dias et al., which could handle trials with multiple arms and rank treatments with additional code [16]. Three chains were run to yield 150,000 iterations, and the initial 5000 burn-ins were discarded. The convergence of models was assessed with trace plots and Brooks-Gelman-Rubin statistic. Model fit of fixed-effect model and random-effect model were compared with the Deviance Information Criterion (DIC), and model with lower DIC was adopted. Long-term SD G-CSF monotherapy was chosen as the common comparator. Estimates of treatment effects were reported as mean differences (MD) with the associated 95% credibility interval (95% CrI). The 95% CrI calculated in Bayesian meta-analysis can be interpreted like the 95% confidence intervals (95% CI) in traditional meta-analysis [17]. The probability of each regimen to be the best was calculated by ranking the relative effects of all treatments in each iteration and defined as the proportion of times a regimen ranked first. This work is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) extension statement for network meta-analyses [18].

Characteristics of included trials
We identified 3826 records from database searches. After removing 835 duplicates, 2991 records were screened on title and abstract, and 2749 clearly irrelevant records were excluded. We retrieved the full text of the remaining 242 records for further assessment. We excluded 147 records for the reasons listed in the flow diagram ( Fig. 1). Ultimately, 95 eligible studies were included for review . The characteristics of the 95 included studies are summarized in Supplementary  Table 1. The efficacy of 94 HSC mobilization agents were investigated, including cytokines, agents targeting the CXCR4 (C-X-C chemokine receptor type 4)/ CXCL12 (SDF1, stromal cell-derived factor-1) axis, agents targeting the VLA-4(very late antigen-4)/VCAM-1 (vascular cell adhesion molecule-1) axis, chemotherapeutic agents, nonsteroidal anti-inflammatory drugs (NSAIDs), and other agents. Most of these agents not only can induce the mobilization of hematopoietic stem and progenitor cells (HSPCs) alone, but also can enhanced the mobilization mediated by G-CSF or AMD3100 synergistically or additively. The detailed information and mobilization efficacy of these agents are reviewed in Table 1. Compared with the conventional G-CSF, modified G-CSF including SD/0 (an engineered pegylated G-CSF), IMG-CSF (G-CSF immobilized on polyethylenoxide by nanotechnology), and PEGLip-G-CSF (pegylated liposome formulated G-CSF) exhibited enhanced mobilization efficacy. Among other cytokines, IL-33 showed superior mobilization potential than G-CSF and AMD3100, and tGROβ (a truncated form of chemokine GROβ) showed superior mobilization AMD3100. AMD3100-a CXCR4 antagonist was the most commonly used agents in combined regimens, which can significantly increase mobilization with a single dose when in combination with G-CSF. There are 7 new CXCR4 antagonist investigated and compared with AMD 3100, among which T-140, POL5551, and CX0714 showed significant superior mobilization AMD3100.
After excluding studies using non-C57BL/6 mice model, studies including mice older than 12 weeks, studies that did not reported data about the number of total CFCs or LSK cell per milliliter PB, studies with significant heterogeneity in G-CSF dosage, and studies that were unrelated to the main network, 21 eligible studies were included in meta-analysis [51, 56, 72, 74, 77, 78, 81-86, 91, 94, 96, 104-106, 108, 109, 112]. All of the 21 included studies are controlled studies, and the most widely used controls are phosphate-buffered saline (PBS), saline, and G-CSF. There are 40 mobilization agents and 57 regimens investigated. The characteristics of these 21 studies are summarized in Table 2. The results of methodological quality evaluation are listed in Supplementary Table 2. Risk of bias regarding random allocation and blinding in all included studies are unclear since the lack of relevant information. The baseline characteristics including mice strain and gender are unified among groups in 10 studies, the other 11 studies did not report the animal gender and age. There are 2 studies that only reported representative data for mobilization outcomes; the other studies are all free from bias caused by incomplete outcome data, selective outcome reporting, and other reasons.

Total CFCs
The number of total CFCs (also known as colonyforming units, CFUs) per milliliter of PB was reported as primary outcome in 17 studies and involved 43 mobilization regimens. The network graph of all comparisons in these 17 studies is shown in Fig. 2. The results of Bayesian network meta-analysis indicate that compared with long-term SD G-CSF alone, 14 mobilization regimens significantly increased the number of total CFCs/ml PB, including long-term SD G-CSF + Me6 (   Mauch et al. [21] IL-11 Interleukin-11 Multi-day IL-11 alone can induce mobilization; IL-11 plus SCF or G-CSF synergistically enhanced mobilization.
Vater et al. [80] Me6 ab An alkaloid analog that can disrupt the SDF-1α/CXCR4 interaction Single-dose Me6 was more effective in mobilization than AMD3100 or G-CSF alone; Me6 synergized with G-CSF in mobilization with higher efficiency than G-CSF + AMD3100.

Desipramine, reboxetine
Norepinephrine reuptake inhibitors Desipramine alone did not induce mobilization; desipramine and reboxetine enhanced G-CSFinduced mobilization but did not affect AMD3100-induced mobilization.
Lucas et al. [74] Adrenaline Catecholaminergic neurotransmitter Adrenaline alone did not induce mobilization; adrenaline enhanced the mobilization efficiency of G-CSF.
Ogle et al. [96] THI An inhibitor of sphingosine phosphate lyase THI enhanced mobilization induced by G-CSF and AMD3100.
Adamiak et al. [93] Anti-CD69 Ab An antibody of CD69 that can increase S1PR1 expression Anti-CD69 Ab induced mobilization of the same magnitude as AMD3100 but did not synergize with AMD3100. ProGP-mobilized cells exhibited greater spleen colony-forming activity and competitive repopulating activity than that of G-CSF.
Liu et al. [45] s-kit A soluble form of c-kit receptor s-kit alone can induce mobilization; s-kit increased G-CSF-induced mobilization.
Wang et al. [98] PTH Parathyroid hormone 6-day PTH alone can induce mobilization; a combination of PTH and G-CSF showed slight additional effects.
Albanese et al. [60] Im-HD Immobilized hyaluronidase The native hyaluronidase and Im-HD alone did not induce significant mobilization; Im-HD enhanced G-CSF-induced mobilization.
Dygai et al. [76] Meloxicam, indomethacin  (Fig. 3). Long-term SD G-CSF + Me6 ranked first among these regimens in regard to the ability to mobilize CFCs. AMD1300 + desipramine, Cdc42 activity-specific inhibitor (CASIN) alone, AMD3100 alone, EP80031 alone, and meloxicam alone are inferior to long-term SD G-CSF. No significant differences are identified between the other regimens and long-term SD G-CSF.

LSK cells
The number of LSK cells/ml PB was reported as primary outcome in 11 studies, which have evaluated the efficacy of 34 mobilization regimens with mice models. The network graph of all comparisons in these 11 studies is shown in Fig. 4. The results of Bayesian network metaanalysis indicate that in comparison with long-term SD G-CSF alone, 7 mobilization regimens significantly increased the number of LSK cells collected from peripheral blood, including long-term SD G-CSF + AMD3100 ( The MD and 95% CrI of all included regimens are presented in forest plot in the order of median rank (Fig. 5). Long-term SD G-CSF + AMD3100 ranked first among these regimens considering this parameter since it is associated with most favorable MD and ranked first in all simulations. AMD3100 + LECT2, long-term LD G-CSF, short-term SD G-CSF, AMD3100 + IL-33, meloxicam alone, LECT2 alone, short-term LD G-CSF, and EP80031 alone are inferior to Xing et al. [89] Dexamethasone Glucocorticoid Dexamethasone enhanced AMD3100-induced mobilization.
Yan et al. [92] Viagra Sildenafil citrate Single-dose Viagra did not induce mobilization, but significantly improved AMD3100-induced mobilization.
Fang et al. [112] C57BL/6 HF51116; AMD 3100; long-term SD G-CSF; G-CSF + HF51116; long-term SD G-CSF + AMD 3100 G-CSF, 100 μg/kg every 12h s.c. for 4 days; AMD 3100, 5 mg/kg s.c.; HF51116, 5 mg/kg s.c.   Table 4. In summary, lethally irradiated recipient mice received mobilized PB cells from donor mice with or without competitive cells, and the long-term repopulating ability are assessed by the survival of recipients and the long-term reconstitution donor-derived cells at different time point (usually in at months after transplantation). Furthermore, serial transplantation analysis was performed via transplanting BM cells from primary recipients to secondary or tertiary recipients to assess the long-term repopulating and self-renewing capacity of mobilized cells in 20 studies. Results indicate that the combination of new mobilization agents (including FLT-3L, MIP-1α, IL-8, PEG-rHuMGDF, SB-251353, s-kit, AMD3100, T-140, tGROβ, VTP195183, SCA, erlotinib, EP80031, meloxicam, UDP-G, Anti-VCAM-1 Ab, heparin, Me6, HF51116, and CSF1-Fc) significantly increased the mobilization of long-term repopulating HSCs compared with G-CSF alone. In addition, the combination of BOP, BIO5192, SEW2871, VPC01091, LGB321, or Viagra with AMD3100 enhanced the mobilization of long-term repopulating HSCs compared with AMD3100 alone. Moreover, cells mobilized by LECT2, POL5551, UDP-G, or CoPP alone showed Fig. 3 Forest plots for total CFCs. Forest plot of the Bayesian network meta-analysis results about the number of harvested total colony-forming cells (CFCs) per milliliter of peripheral blood (/ml PB). Estimate of treatment effect for each mobilization regimen was reported as mean differences (MD) with the associated 95% credibility interval (95% CrI). Granulocyte colony-stimulating factor monotherapy (G-CSF) is a common comparator. All included regimens are described in the supplementary materials superior long-term repopulating capacity than those mobilized by G-CSF, whereas cells mobilized by Me6, CasNa, or CASIN alone showed superior long-term repopulating capacity than those mobilized by AMD3100.
To our best of knowledge, this study is the first network meta-analysis that compared the efficacy of different HSC mobilization regimens with data from preclinical studies. We provide a comprehensive summary of new mobilization agents that have been investigated in mice models. The efficacy of these agents alone or in combination with other agents was indirectly compared via network meta-analysis. Moreover, we ranked all of the investigated regimens based on their ability to mobilize HSCs into blood stream. We identified several promising agents and regimens that have the most potent mobilizing capacity. The majority of mobilization regimens that show great improvements over G-CSF are combined regimens containing both G-CSF and new mobilization agents. Although these regimens would be unlikely to reduce severe adverse events, they provide a perspective that the incorporation of new agents could reduce the incidences of G-CSF-related adverse events through reducing the doses of G-CSF that required to mobilization sufficient HSCs since they can synergistically enhance the G-CSF-mediated mobilization. In addition, we identified several agents showed superior mobilization potential than G-CSF even when used alone, such as Me6 and POL5551. It is worth further investigation that whether these agents could reduce mobilizationrelated toxicity compared with G-CSF.
Among the new agents, EP80031, Me6, FG-4497, and ML141 significantly improved the efficiency of G-CSFinduced HSC mobilization. EP80031 is a synthetic octosaccharide mimicking the structure of heparan sulfate. A single dose of EP80031 (15mg/kg, intravenously injection) could lead to rapid and prominent mobilization of hematopoietic stem and progenitor cells (HSPCs), and the combination of EP80031 with G-CSF and AMD3100 resulted in 3-fold increase in the number of LSK cells and total CFCs [72]. In addition, HSCs mobilized with the regimen G-CSF + AMD3100 + EP80031 are associated with enhanced hematopoietic reconstitution [72]. Me6 is a small molecule that was screened from a group of chemicals by Zhang et al. and has been proved to have robust ability of mobilizing HSPCs [85]. The combination of Me6 and G-CSF (G-CSF + Me6) resulted in remarkable increase in the number of total CFUs, moreover, it is suggested that Me6-mobilized HSCs are associated with greater long-term repopulating capacity and more efficient engraftment [85]. FG-4497 is a prolyl hydroxylase inhibitor that could enhance HSC mobilization through stabilizing the hypoxia-inducible transcription factor-1α (HIF-1α) protein [106]. The addition of FG-4497 significantly increased the mobilization of HSPCs induced by G-CSF [86,106]. In addition, FG-4497 exerts protective effects in ischemia-induced kidney injury and high-dose irradiation-induced BM failure [115]. ML141 is an inhibitor of cell division control protein 42 (Cdc42). The mobilization effect of ML141 is modest, but ML141 could synergistically enhance G-CSF-mediated mobilization of LSK cells and CFCs in mice model [81]. Taking our results of meta-analysis together into consideration, G-CSF + AMD3100 + EP80031, G-CSF + Me6, G-CSF + FG-4497, and G-CSF + ML141 are new promising mobilization regimens that could significantly increase the quantity of HSCs in PB without interfering their functions. However, the safety profiles of these new agents remain unclear. Further studies are required to determine the efficacy and safety of these potential regimens in human before applied in clinical practice.
In addition, we established the favorable efficacy of G-CSF and AMD3100 in HSC mobilization, which has been verified by clinical trials. AMD3100, also known as plerixafor, is an antagonist of the chemokine receptor CXCR4 that could rapidly induce the mobilization of stem cells through antagonizing the interaction of CXCR4 and stromal cell-derived factor-1α (SDF-1α) [116]. Multiple studies have demonstrated that AMD3100 alone mobilized lower numbers of HSCs compared with G-CSF, but the addition of AMD3100 dramatically increased the G-CSF-induced mobilization of HSCs both in mice models and non-human primates' model [117,118]. Our results from network meta- Estimate of treatment effect for each mobilization regimen was reported as mean differences (MD) with the associated 95% credibility interval (95% CrI). Granulocyte colony-stimulating factor monotherapy (G-CSF) is a common comparator. All included regimens are described in the supplementary materials analyses indicated that G-CSF in combination of AMD3100 not only significantly increased the number of LSK cells, but also increased total CFCs. Despite we only pooled data from murine models, which are different from human in regard to physiological conditions, our conclusions are consistent with that obtained from clinical studies in human beings. A group of randomized controlled trials (RCTs) have demonstrated that G-CSF in combination of AMD3100 led to higher rates of successful mobilization and increased the total collection of HSCs without increasing the risk of severe adverse events in patients with non-Hodgkin's lymphoma (NHL) and multiple myeloma (MM) [119][120][121]. Moreover, it is suggested that AMD3100-mobilized cell products are associated with greater capacity to repopulate the marrow and potential of protecting against graft-versus-host disease due to an enrichment of regulatory T cells (GVHD) [118,122]. AMD3100 has been approved for HSC mobilization and subsequent autologous transplantation in patients with NHL and MM [123]. Therefore, before the efficacy and safety of new regimens in human were well established, G-CSF in combination with AMD3100 remains the most efficient and safe regimens in patients with high risk of mobilization failure. Although G-CSF plus AMD3100 significantly improved mobilization efficiency compared with G-CSF alone, two well-designed RCTs indicated that successful rate of achieving optimal target with G-CSF plus AMD3100 is only 59.3% in NHL patients and 75.7% in MM patients [119,120]. Therefore, we speculate that almost 25-40% of patients with high risk of mobilization failure would still benefit from new mobilization regimens.
Nevertheless, there are some limitations in this study. Firstly, we integrated evidences from animal models. It is suggested that HSC mobilization is evolutionarily conserved from mice to humans, so mice models also represent a valuable experimental system for investigating the efficacy and mechanisms of mobilization regimens [67,81]. Even so, animal model could not completely simulate the physiological condition of human; hence, the translation of our results integrated from preclinical studies to human should be in cautions. Future clinical trials are needed for validation these regimens in human. Secondly, our meta-analysis did not include safety outcomes. Most of the included studies did not provide information about toxicity, and the toxicity data collected from animal experiments are hard to be pooled with metaanalysis. Further studies are required to compare the safety of these new mobilization regimens. Thirdly, the results of meta-analysis may be confounded by the heterogeneity in mice gender since it was reported that male mice have better mobilization outcome compared with female mice [114]. It is impractical to perform subgroup analysis based on animal gender since most of the studies did not report the gender of mice and some studies included both male and female mice. However, since the network meta-analyses were performed with wellestablished methods and the most efficacious regimens are associated with robust MD values, we believe that the effects of these differences are minimal. Last but not least, there is a big gap between our results and translational medicine since the lack of data from human systems, but we think this study may contribute to the translation of basic research results into clinical investigations through providing comprehensive review of new promising mobilization regimens and related mechanisms.

Conclusions
In summary, this study identified several promising mobilization agents and regimens that significantly increased the mobilization of HSCs compared with the conventional agent G-CSF alone. We think that our results can provide important perspectives for future researches.