Blood Cells, Molecules and Diseases 83 (2020) 102436

Contents lists available at ScienceDirect

Blood Cells, Molecules and Diseases

journal homepage: www.elsevier.com/locate/bcmd

Biomarkers and genetic modulators of cerebral vasculopathy in sub-Saharan T

ancestry children with sickle cell anemia
Marisa Silvaa, Sofia Vargasa, Andreia Coelhoa, Emanuel Ferreiraa, Joana Mendonçaa,
Luís Vieiraa,b, Raquel Maiac, Alexandra Diasd, Teresa Ferreirad, Anabela Moraise,1,
Isabel Mota Soaresf, João Lavinhaa,g, Rita Silvah, Paula Kjöllerströmc, Paula Faustinoa,i,
⁎

a
Departamento de Genética Humana, Instituto Nacional de Saúde Dr. Ricardo Jorge, Lisbon, Portugal
b
ToxOmics, Faculdade de Ciências Médicas, Universidade Nova de Lisboa, Lisboa, Portugal
c
Unidade de Hematologia, Hospital de Dona Estefânia, Centro Hospitalar Universitário de Lisboa Central (CHULC), Lisbon, Portugal
d
Núcleo de Hematologia, Departamento de Pediatria, Hospital Prof. Doutor Fernando Fonseca, Amadora, Portugal
e
Departamento de Pediatria, Hospital de Santa Maria, Centro Hospitalar Universitário de Lisboa Norte, Lisbon, Portugal
f
Departamento de Pediatria, Hospital Garcia de Orta, Almada, Portugal
g
BioISI, Faculdade de Ciências, Universidade de Lisboa, Lisbon, Portugal
h
Unidade de Neuropediatria, Hospital de Dona Estefânia, CHULC, Lisbon, Portugal
i
Instituto de Saúde Ambiental (ISAMB), Faculdade de Medicina, Universidade de Lisboa, Lisbon, Portugal

ARTICLE INFO ABSTRACT

Editor: Mohandas Narla We investigated biomarkers and genetic modulators of the cerebral vasculopathy (CV) subphenotype in pediatric
Keywords: sickle cell anemia (SCA) patients of sub-Saharan African ancestry. We found that one VCAM1 promoter hap-
Sickle cell anemia lotype (haplotype 7) and VCAM1 single nucleotide variant rs1409419_T were associated with stroke events,
Cerebral vasculopathy stroke risk, as measured by time-averaged mean of maximum velocity in the middle cerebral artery, and with
Ischemic stroke high serum levels of the hemolysis biomarker lactate dehydrogenase. Furthermore, VCAM-1 ligand coding gene
Genetic modulators ITGA4 variants rs113276800_A and rs3770138_T showed a positive association with stroke events. An additional
Biomarkers positive relationship between a genetic variant and stroke risk was observed for ENPP1 rs1044498_A.
Conversely, NOS3 variants were negatively associated with silent cerebral infarct events (VNTR 4b_allele and
haplotype V) and CV globally (haplotype VII). The -alpha3.7kb–thal deletion did not show association with CV.
However, it was associated with higher red blood cell and neutrophil counts, and lower mean corpuscular
volume, mean corpuscular hemoglobin and red cell distribution width.
Our results underline the importance of genetic modulators of the CV sub-phenotype and their potential as
SCA therapeutic targets. We also propose that a biomarker panel comprising biochemical, hematological,
imaging and genetic data would be instrumental for CV prediction, and prevention.

1. Introduction movements are leading to a wider distribution, emphasizing the global

health emergence status of the disease [1–3]. In children, the most
Sickle cell anemia (SCA), the homozygous form of the c:20A > T common sub-phenotypes are cerebral vasculopathy (CV), acute chest
mutation in the beta-globin gene, is the most common and severe syndrome, hyposplenism, renal disease and painful crises. CV is a major
presentation of sickle cell disease (SCD). High birth rates and child complication and comprises overt stroke, silent cerebral infarcts (SCIs),
mortality are most frequent in sub-Saharan Africa, however, population transient ischemic attacks and frequently neurocognitive complications

Abbreviations: SCA, sickle cell anemia; SCD, sickle cell disease; CV, cerebral vasculopathy; SCI, silent cerebral infarction(s); LDH, lactate dehydrogenase; TAMMV,
time-averaged mean velocity in the middle cerebral artery; TCD, transcranial Doppler ultrasound; MRI, magnetic resonance imaging; MRA, magnetic resonance
angiography; HC, hydroxycarbamide; HbS, hemoglobin S; HbF, fetal hemoglobin; Hb, total hemoglobin; RBC, red blood cells; WBC, white blood cells; MCV, mean
corpuscular volume; MCH, mean corpuscular hemoglobin; RDW, red cell distribution width; NGS, next-generation sequencing; VCAM-1, vascular cell adhesion
molecule 1; ITGA4, very-late antigen 4; eNOS, endothelial nitric oxide synthase; NO, nitric oxide; TFBS, transcription factor binding site(s)
⁎
Corresponding author at: Departamento de Genética Humana, Instituto Nacional de Saúde Dr. Ricardo Jorge, Avenida Padre Cruz, 1649-016 Lisboa, Portugal.
E-mail address: [email protected] (P. Faustino).
1
Deceased co-author.

https://doi.org/10.1016/j.bcmd.2020.102436
Received 12 February 2020; Received in revised form 5 April 2020; Accepted 5 April 2020
Available online 13 April 2020
1079-9796/ © 2020 Elsevier Inc. All rights reserved.
M. Silva, et al. Blood Cells, Molecules and Diseases 83 (2020) 102436

at a later stage. Children with SCA have a much higher risk of stroke treatment with HC and more than > 120 days after receiving a blood
than the general pediatric population. The prevalence of overt stroke transfusion. The patients were analyzed according to CV outcome, or
approaches 11% by age 20 years. On the other hand, SCIs have been stroke risk, in the following groups (i) overt ischemic stroke (≥1 epi-
found in up to 37% of children with SCA [4]. sode), henceforth designated as “stroke” (n = 62); (ii) “silent cerebral
The current standard of care for stroke prevention in SCA children is infarct”(≥1 event), SCI (n = 53); (iii) stroke and/or SCI, as confirmed
Transcranial Doppler (TCD) screening – through measurement of time- by MRI or MRA, (n = 62); or “stroke risk” (n = 60), with risk strati-
averaged mean velocity (TAMMV) in the middle cerebral artery – fol- fication provided by the TAMMV values, obtained during TCD, as fol-
lowed by regular blood transfusions therapy and hydroxycarbamide lows: (a) high risk – TAMMV ≥200 cm/s; (b) conditional (or moderate)
(HC) treatment. Despite its high sensitivity, TCD still does not allow risk – 200 cm/s > TAMMV ≥170 cm/s; and (c) low risk –
identification of all SCA children that will experience a stroke and, TAMMV < 170 cm/s.
conversely, children with high TAMMV (> 200 cm/s) may not develop
stroke [5]. Moreover, blood transfusion/HC therapies are not without 2.3. Genotyping
limitations or adverse side effects [6]. On the other hand, although
diagnosis with magnetic resonance imaging, MRI, (or angiography, Genomic DNA was isolated from peripheral blood samples of each
MRA) is recommended for early diagnosis of SCIs and recognition of patient using the MagNA Pure LC Instrument (Roche Diagnostics
large vessel stenosis, MRI/MRA are not useful to identify patients at risk GmbH, Mannheim, Germany). All samples were anonymized and spe-
of developing SCIs or large vessel stenosis [7]. A more specific and cific genotypes could be linked to phenotypes only through the main
sensitive panel of biomarkers for CV prediction and prognosis, that study database.
includes genetic variants with disease modifying effects, is therefore of The homozygous status for the SCA mutation in the HBB gene
the utmost importance. (c.20A > T) was confirmed by polymerase chain reaction followed by
In previous studies, we identified variants in VCAM1, NOS3 and restriction fragment length analysis (PCR-RFLP) with the endonuclease
HBA with a positive association with chronic hemolysis, a known pa- Bsu36 I. Beta-globin cluster haplotypes were assigned after examining
thophysiological SCA mechanism [8]. Building on those results we six restriction endonucleases sites within the cluster: Xmn I (5′ to
aimed, in this work, to assess if those variants were also associated with HBG2), Hind III (within the HBG2 and HBG1) Hinc II (within and 3′ to
pediatric CV in a sub-Saharan SCA population. Our candidate gene ψHBB) and Hinf I (3′ to HBB). Aliquots of the amplified products were
approach also focused on the VCAM-1 ligand gene, ITGA4, and for digested with the appropriate enzymes under the conditions re-
comparison purposes, we included the three genetic variants (PON1 commended by the manufacturers. Haplotypes were determined by the
rs662, ENPP1 rs1044498 and GOLGB1 rs3732410) previously reported presence or absence of cleavage at each site and by analyzing the
in association with pediatric stroke in SCA patients [7,9,10]. compiled pattern through comparison to known patterns [11]. The
-α3.7kb–thal deletion was assessed by gap-PCR [12].
2. Materials and methods Putative modifier genes were identified through previous studies by
our group [8] and from other published reports based on the influence
2.1. Ethical statement on the phenotypes of interest. These candidate genes were used to
identify and genotype SNPs and other variants in patient samples. For
Ethical approval for the study was granted by the institutional re- VCAM1, NOS3, PON1, ENPP1 and GOLGB1 genes genotyping was per-
view boards of Instituto Nacional de Saúde Dr. Ricardo Jorge (INSA) formed using TaqMan-based PCR with commercially available or cus-
and of participant hospitals, in line with the principles of the tomized primers.
Declaration of Helsinki. The aim and study procedures were explained
to the children's parents (or legal guardians) and they provided in- 2.3.1. Screening for ITGA4 variants by next-generation sequencing (NGS)
formed written consent prior to their enrolment in this study. In order to search for variants in the regulatory region of ITGA4
gene, NGS analysis was used on a long PCR fragment (4.1 kb), including
2.2. Study population its flanking regions. PCR was performed using the primers FW5’-CAG
AGGCTCATTAGGACCC-3′ and Rv5′-CCTTGCGGTACTATCCAGGC-3′
This case-control study was performed at INSA in cooperation with and the FailSafe enzyme with the PreMix G (Epicentre, Illumina, San
four hospitals in the Greater Lisbon area – Hospital D. Estefânia, Diego).
Hospital de Santa Maria, Hospital Fernando Fonseca and Hospital The NGS workflow consisted in five steps: PCR product purification
Garcia de Orta, the four largest centers of that metropolitan area. These using paramagnetic beads (Agentcourt, Ampure XP); double-stranded
centers combined receive the highest numbers of SCA pediatric patients DNA quantification in a Qubit fluorometer; DNA library preparation
in our country. Seventy unrelated pediatric patients (≥3 years old) of using the Nextera XT kit (Illumina, San Diego) following the manu-
direct sub-Saharan African ancestry diagnosed with SCA were selected. facturer's instructions; high throughput sequencing in a MiSeq benchtop
Exclusion criteria included age < 3 years old, non-African ancestry, sequencer (Illumina, San Diego); data analyses were performed using
previous HC treatment or having received a blood transfusion in the the following tools: Sequencing Analysis Viewer (v1.8.46, Illumina) and
120 days prior to enrolment. FastQC (v0.11.5, Babraham Bioinformatics,https://www.
Data obtained from patients and parents (or legal guardians) in- bioinformatics.babraham.ac.uk/projects/fastqc/) were used for
terviews, which included demographic characteristics (age, gender, quality analysis. The MiSeq® Reporter software package (v2.6.2,
parents' geographic origin), were collected. Hemoglobin profiles (HbS, Illumina) was used for read mapping (with Burrows-Wheeler Aligner)
HbF), hematological parameters [RBC, leukocyte, neutrophil and pla- and variant calling and filtering (with Genome Analysis Toolkit). FastQ
telet counts, mean corpuscular volume (MCV), mean corpuscular he- screen (v.0.9.3, Babraham Bioinformatics) was used to screen for con-
moglobin (MCH), red cell distribution width (RDW)] as well as hemo- tamination between samples. Variant Effect Predictor (www.ensembl.
lysis biochemical and hematological biomarkers [serum LDH and total org) was used to annotate variants and Integrative Genomics Viewer
bilirubin and reticulocyte count] were retrospectively collected from (v.2.3.86) (Broad Institute; [13]) was used for visualization of reads and
patients' hospital records. All these parameters were obtained by stan- variants. Validation of the variants was performed using automated
dard procedures and HbF levels, in particular, were measured by high- Sanger sequencing, after amplification with customized primers in the
performance liquid chromatography (HPLC). The data collected for 3130XI Genetic Analyser, (Applied Biosystems). The results were ana-
each parameter were the result of, at least, three different time-point lyzed using FinchTV v.1.4.0 software (Geospiza, Inc.). The genotyping
measurements, performed in steady-state periods, and prior to any results were added to the previously created database.

2
M. Silva, et al. Blood Cells, Molecules and Diseases 83 (2020) 102436

2.3.2. Haplotype reconstruction Table 1

Haplotype reconstruction was performed using PHASE software Demographic, neurological status and laboratory parameters of the population
v2.1 developed by Mathew Stephens at Washington University, ac- in this study.
cording to the developer's instructions (https://els.comotion.uw.edu/ Age (years) 3–16
express_license_technologies/phase). Haplotypes were reconstructed for Gender n %
genetic variants in NOS3 (rs2070744, intron 4_27 bp VNTR, Male 40 57.1
Female 30 42.9
rs1799983), in the promoter of VCAM1 (rs1409419, rs3917024,
Parental geographic origin n %
rs3917025, rs3783598, rs1041163, rs3783599) and for genetic variants Angola 42 60.0
of ITGA4 (rs1375493, rs35723031, rs10562650; rs1839269 and São Tomé and Príncipe Islands 8 11.4
rs1839268). Cape Verde 5 7.1
Guinea-Bissau 7 10.0
Guinea-Conakry 1 1.4
2.3.3. In silico analysis
Nigeria 1 1.4
Population allele and genotype frequencies were recorded for each Double origin 6 8.6
observed variant using the Ensembl browser (www.ensembl.org). SNPs' Neurological status (n = 70) n %
sequences were retrieved using the NCBI SNP search engine (http:// Stroke Yes 15 24.2
(n = 62) No 47 75.8
ncbi.nlm.gov/snp).
SCI Yes 9 16.9
Transcription factor binding sites (TFBS) analysis was performed for (n = 53) No 44 83.0
the variants identified in the regulatory regions, using the MatInspector CV Yes 24 38.7
tool (Genomatix, Munich, Germany), to evaluate potential effects on (n = 62) No 38 61.3
the regulation of gene expression by altering putative TFBS. Only re- Stroke risk High (TAMMV≥200 cm/s) 25 41.7
(n = 60) Moderate (170 cm/s ≤ TAMMV ≤ 200 cm/s) 6 10.0
sults above the 0.85 threshold were considered, which corresponds to a
Low risk (TAMMV < 170 cm/s) 29 48.3
maximum of 15% dissimilarity between the identified sequence and the Hematological parameters Median IQR
consensus sequence. A comparison with previously reported consensus Hb S (%) 79.9 14.5
sequences of TFBS for the VCAM1 [14] and the ITGA4 [15] promoters Hb F (%) 10.7 11.7
Hb (g/dL) 8.0 1.3
was performed. The sequences of the identified variants were not found
RBC (×1012/L) 3.0 0.7
to overlap with any of the previously reported TFBS consensus se- MCV (fL) 81.3 14.3
quences. Hence, no strong effects for these genes' expression are to be MCH (pg) 26.9 6.0
expected as a result of the presence of those variants. Reticulocytes (%) 9.9 6.3
RDW (%) 21.2 4.5
Platelets (×109/L) 404.1 167.7
2.3.4. Statistical analysis
WBC (×109/L) 12.6 4.8
The analyses were performed using the SPSS software (version 25.0, Neutrophils (×109/L) 5.6 2.65
IBM Inc., Chicago, USA). For descriptive analysis, continuous variables Biochemical parameters Median IQR
were represented as medians and interquartile ranges (IQR). To eval- LDH (U/L) 636.3 473.4
Total bilirubin (mg/dL) 2.6 1.9
uate the Gaussian distribution of variables, Shapiro-Wilk normality
HBB cluster haplotype n %
tests where applied. We used the Mann-Whitney U test to compare the Bantu/Bantu 38 54.3
medians of variables. Categorical variables are represented as number, Senegal/Senegal 11 15.7
frequencies and percentages. The chi-square test or the Fisher's exact Benin/Benin 3 4.3
test were used to compare categorical variables on bivariate analysis. Compound heterozygous 17 24.2
Atypical 1 1.4
Statistical significance was defined as p < 0.05.
The minor allele for each variant was evaluated for potential asso-
ciation with stroke, SCI, cerebral vasculopathy (stroke and SCI com- A total of seventy-one genetic variants were identified of which
bined) or risk (as measured by TCD-TAMMV values), via 2 × 2 phe- twenty-eight (MAF > 0.05) were used in the association studies –
notype × genotype contingency tables. Only polymorphisms with a seven in VCAM1, five in NOS3, three in GOLGB1, one in PON1, one in
minor allele frequency (MAF) > 5% were considered for association ENPP1, one in HBA (-α3.7kb–thal) and ten in ITGA4 (Appendix Table
analysis. Phenotypes were classified as follows: (i) “stroke” (at least one A.1). We were able to reconstruct 16 haplotypes, and used ten of them
previous overt ischemic stroke event, as confirmed by MRI/MRA) or (frequency > 0.05) for statistical analysis. Concerning the HBB gene
“no stroke” (no clinically/imaging identified stroke event); (ii) “SCI” (at cluster haplotypes, only the more frequent genotypes, Bantu/Bantu and
least one event as identified by MRI/MRA) or “no SCI” (no SCI events, Senegal/Senegal, were used for statistical analysis (frequency > 0.05).
as confirmed by MRI); (iii) “cerebral vasculopathy” (at least one overt
ischemic stroke and/or SCI event) or “no cerebral vasculopathy”; and
(iv) “risk of stroke” (high/moderate: TAMMV ≥ 200 cm/s or 3.2. In silico analysis
199 ≥ TAMMV ≥ 170 cm/s; low: TAMMV < 170 cm/s).
Each variant was also evaluated for potential association with bio- In silico analysis of the VCAM1 gene promoter variants was focused
chemical and hematological parameters, including hemolysis bio- on those with MAF > 0.05, except rs3917025 due to its occurrence in
markers (LDH, total bilirubin, reticulocyte count). only one haplotype (Haplotype 3). The rs1041163_C, rs1409419_T and
rs3917025_delCT VCAM1 alleles were classified (according to ClinVar
3. Results and Ensembl's VEP and Mat Inspector) as intergenic variants with po-
tential modifying impact, although with no major pathologic effects.
3.1. Population description and genotyping Potential changes resulting from the presence of the rs1041163_C allele
include (i) TFBS alteration for RXRF transcription factor, substituting it
This study was performed on seventy unrelated SCA patients, (age for a PRDF site, and (ii) loss of a FHXB TFBS. The presence of
range: 3–16 years, 40 males, 30 females), living in Portugal but of direct rs1409419_T would lead to a potential gain of binding sites, in parti-
sub-Saharan ancestry, with parental geographic origin mainly from cular, for EVI1, Oct1 and Barx2. Regarding rs3917025_delCT, a po-
Angola, São Tomé and Príncipe and Cape Verde (Table 1). tential gain of a FAST1 (FoxH1) TFBS was indicated.

3
M. Silva, et al. Blood Cells, Molecules and Diseases 83 (2020) 102436

Table 2 Table 3
Association of biochemical and hematological parameters of SCA patients with Genetic variants association with the hematological and biochemical para-
stroke and stroke risk. meters.
Parameter (units) n Stroke p Gene Variant Allele change or Parameter p
haplotype (unit; nr. patients)
Yes No
HbS (%; n = 60)
Medians (IQR) Medians (IQR) Var No var
VCAM1 rs1041163 T>C 74.5 83.5 0.019
HbF (%) 64 3.2 (9.3) 11.9 (10.3) 0.018 Haplotype 3 C/C/CT/T/C/C 73.7 81.4 0.034
MCH (pg) 70 21.2 (20.8) 27.4 (6.0) 0.005 rs33988592 G>A 11.3 13.4 0.030
ITGA4 rs113276800 C>A 69.0 80.4 0.012
rs3770138 C>T 66.1 80.7 0.003
Parameter (units) n Stroke risk p HbF (%; n = 64)
Var No var
High + moderate Low VCAM1 rs1041163 T>C 13.0 7.0 0.014
Haplotype 3 C/C/CT/T/C/C 14.5 9.3 0.005
Medians (IQR) Medians (IQR) HBB Haplotype Sen/Sen 13.6 9.2 0.038

HbF (%) 64 8.5 (10.2) 12.1 (10.7) 0.043 Hb (g/dL; n = 65)

Platelets (×109/L) 61 442.0 (156.4) 363.2 (124.2) 0.017 Var No var
Neutrophils (×109/L) 61 6.3 (2.5) 4.9 (2.6) 0.009 HBB Haplotype Sen/Sen 8.1 7.9 0.022
LDH (U/L) 65 761.5 (535.7) 510.0 (325.3) < 0.001 RBC (×1012/L;
n = 64)
Values indicated as medians (interquartile range in brackets); n - number of HBA -α3.7kb del αα > -α3.7kb 3.1 2.7 0.008
patients; HbF - hemoglobin F; MCH - mean corpuscular hemoglobin; LDH -
MCV (fL; n = 69)
lactate dehydrogenase.
NOS3 rs2070744 T>C 87.3 79.9 0.024
Haplotype IV T/4a/G 78.7 83.2 0.032
3.3. Association of biochemical and hematological parameters with cerebral HBA -α3.7kb del αα > -α3.7kb 75.7 85.7 < 0.001
vasculopathy HBB Haplotype Sen/Sen 90.1 80.6 0.039

MCH (pg; n = 70)

We observed significant associations of both stroke and stroke risk Var No var
with several biochemical and hematological parameters (Table 2). HBA -α3.7kb del αα > -α3.7kb 25.1 29.0 0.001

Lower HbF percentages and MCH values were positively associated RDW (n = 70)
with stroke, while stroke risk was associated not only to lower HbF Var No var
NOS3 rs2070744 T>C 19.6 21.6 0.031
percentage but also to higher levels of coagulation, inflammation and
Haplotype I C/4a/G 19.6 21.5 0.044
hemolysis markers (Table 2). HBA -α3.7kb del αα > -α3.7kb del 20.8 21.2 0.021

Platelets (×109/L;
n = 61)
3.4. Association of genetic variants with biochemical and hematological Var No var
parameters PON1 rs662 G>A 442.0 378.1 0.028

WBC (×109/L;
Genetic variants analyzed in our study, using the dominant genetic n = 67)
test model, showed association with both hematological and biochem- GOLGB1 rs33988592 G>A 11.3 13.4 0.030
HBB Haplotype Sen/Sen 9.9 13.0 0.030
ical parameters, whether individually or as part of specific haplotypes
(Table 3). VCAM1 rs1041163 (CC + TC), VCAM1_haplotype 3, ITGA4 Neutrophils (×109/L;
n = 61)
rs113276800 (CA) and ITGA4 rs3770138 (TT + CT) showed an asso-
Var No var
ciation with lower levels of HbS. Conversely, higher HbF percentages GOLGB1 rs33988592 G>A 5.0 6.2 0.013
were observed in association with rs1041163 (CC + TC), VCAM1_ha- HBA -α3.7kb del αα > -α3.7kb 5.7 5.5 0.010
plotype 3 and with Senegal/Senegal haplotype. HBB Haplotype Bantu/Bantu 6.2 5.0 0.036
Namely, VCAM1 as well as ITGA4 variants, individually or within a Haplotype Sen/Sen 4.7 6.2 0.014

haplotype context, were significantly associated with traditional bio- LDH (U/L; n = 65)
markers of disease severity, such as lower HbS percentage and higher Var No var
VCAM1 rs1409419 C>T 748.0 517.0 < 0.001
LDH and total bilirubin values.
Haplotype 7 T/C/CT/T/T/C 748.0 517.0 < 0.001
As for PON1 rs662, the AA and GA genotypes showed a positive ITGA4 Haplotype A G/GA/TT 611.5 1269.0 0.003
association with high platelets counts characteristic of a pro-coagulant
Bilirubin (mg/dL;
state. Regarding GOLGB1, no significant association was found between n = 69)
the presence of rs3742410_C and hematological or biochemical para- Var No var
meters. However, we found two other GOLGB1 SNPs while analyzing VCAM1 rs3783613 G>C 3.2 2.4 0.026
rs3732410_C – rs61746571_G and rs33988592_A, a synonymous and a
missense variant, respectively. The rs61746571_G seems to be in
Var – presence of variant allele or haplotype; No var – absence of variant allele
linkage disequilibrium with variant rs3732410_C. On the other hand,
or haplotype; HbS – hemoglobin S; HbF – fetal hemoglobin; Hb – total he-
rs33988592 AA and GA genotypes showed an association with lower moglobin; RBC – red blood cells; MCV – mean corpuscular volume; MCH –
values of inflammation markers (Table 3). mean corpuscular hemoglobin; RDW – red cell distribution width; WBC – white
blood cells; LDH – lactate dehydrogenase; Sen - Senegal.

4
M. Silva, et al. Blood Cells, Molecules and Diseases 83 (2020) 102436

Table 4
Genetic variants association with cerebral vasculopathy.
Gene Variant Stroke (n = 62)

Yes No p OR 95% CI

VCAM1 rs1409419_TT+CT 12 23 0.041 4.17 1.04–16.73

rs1409419_CC 3 24
Haplotype 7 12 23 0.041 4.17 1.04–16.73
Haplotype X 3 24
ITGA4 rs113276800_CA 4 3 0.025 7.62 1.39–41.65
rs113276800_CC 7 40
rs3770138_TT+CT 4 4 0.045 5.57 1.12–27.67
rs3770138_CC 7 39

Gene Variant SCI (n = 53)

Yes No p OR 95% CI

NOS3 VNTR_4b 6 42 0.030 0.10 0.01–0.69

VNTR_4a+4c 3 2
Haplotype V 4 36 0.031 0.18 0.04–0.81
Haplotype Y 5 8

Gene Variant CV (n = 62)

Yes No p OR 95% CI

NOS3 Haplotype VII 1 23 0.006 0.08 0.01–0.69

Haplotype Z 13 25

Gene Variant Stroke risk (n = 60)

High + moderate Low p OR 95% CI

VCAM1 rs1409419_TT+CT 23 11 0.009 4.71 1.57–14.13

rs1409419_CC 8 18
Haplotype 7 23 11 0.009 4.71 1.57–14.13
Haplotype X 8 18
NOS3 Haplotype V 18 24 0.050 0.29 0.09–0.96
Haplotype Y 13 5
ENPP1 rs1044498_AA+CA 14 5 0.026 4.03 1.21–13.42
rs1044498_CC 16 23

Haplotype X – presence of any of the other VCAM1 promoter haplotypes studied; Haplotype Y – presence of any of the other NOS3 haplotypes studied; SCI – silent
cerebral infarction; CV – cerebral vasculopathy; 95% CI – 95% confidence interval.

3.5. Association of genetic variants with cerebral vasculopathy and cerebral In silico analysis of the VCAM1 rs1409419_T allele indicated parti-
vasculopathy risk cularly interesting potential TFBS gain, namely for EVI1, Oct1 and
Barx2 transcription factors. EVI1 is of special note due to its complexity,
We found a significant association between the presence of several multiple targets and modulation of several numerous processes, in-
of the variants identified and CV/CV risk (Table 4). Namely, VCAM1 cluding cell migration, adhesion and proliferation [16]. It may co-
rs1409419 (TT + CT), haplotype 7, ITGA4 promoter rs113276800 (CA) operate with FOS transcription factor to limit cell adhesion while en-
and rs3770138 (TT + CT), showed a positive association with stroke. hancing cell proliferation [16]. Conversely, Oct1 is a TF known to
ITGA4 variant rs3770138 (TT + CT) was also positively associated with promote a transcriptional repression/silencing effect, which would
CV as a whole. Positive associations were also found between high potentially result in VCAM1 down-regulation. On the other hand, a
TAMMV values and VCAM1 rs1409419 (TT + CT), VCAM1 promoter Barx2 binding site gain, predicted as a result of the rs1409419_T pre-
haplotype 7, and also with the ENPP1 rs1044498_A allele. On the other sence, has been shown to promote murine muscle cell differentiation,
hand, NOS3 intron 4 VNTR_4b allele and haplotype V were negatively by interacting with muscle regulatory factors, whereas its loss would
associated with SCI, while haplotype VII showed a negative association lead to decreased adhesion properties [17]. Hence, it is reasonable to
with CV overall. assume that a gain could result in increased adhesion properties. All the
TFs which expression may be affected by the significantly associated
4. Discussion variants are mainly involved in development and in different tissues
following the proposed role of VCAM-1 [18]. It is important to em-
Our study aimed to assess demographic, clinical, biochemical, he- phasize that while we can address the potential effects of the individual
matological, genotyping and imaging data to design a potential bio- VCAM1 variants' genotypes, the most important modifying role on
marker panel for CV prognosis in children with SCA. With this ap- disease manifestation would probably arise in the context of the hap-
proach, we were able to identify statistically significant associations of lotypes that encompass them. The one exception seems to be
biochemical, as well as hematological parameters, with genetic variants rs1409419_T given the overlapping findings observed for this variant
and CV. and haplotype 7, which includes it.

5
M. Silva, et al. Blood Cells, Molecules and Diseases 83 (2020) 102436

We found that variants in the gene that encodes the VCAM-1 ligand, allele was not found in our patients as in Flanagan's studies but contrary
ITGA4, namely rs113276800_A and rs3770138_T were also positively to 1.67% in Martella's report. Homozygotes for the PON1 rs662_C allele
associated with stroke. Furthermore, the latter was associated with CV occurred in a frequency of 10,3% in our study, whereas Martella et al.
globally. Given its role in WBC, reticulocyte and sickle erythrocyte [7] and Flanagan et al. [9] reported 45% and 13.7%, respectively. Of
adhesion to the activated SCA endothelium, this finding further un- the three variants, only ENPP1 rs1044498 AA and AC genotypes
derlines the role of cell-endothelium adhesion in the severe CV sub- showed a positive association with stroke risk. Notably, rs1044498_A is
phenotype. ITGA4 rs113276800 has been previously described in as- the minor allele, in our study group, while the variant allele (C) is the
sociation with multiple sclerosis and, as in our case, no AA individuals most frequent, which is in line with the reference population (African
were observed [19,20]. This variant is located in the ITGA4 promoter Yoruba) and contrary to what is described for the other reference po-
region near the AP-2 binding sites and the AA genotype may, therefore, pulations. This may reflect a negative selection for the less advanta-
cause a negative expression of the integrin α4 subunit [21]. We also geous allele - rs1044498_A in these populations. As for the PON1
found that the ITGA4 rs1375493, rs35723031 and rs10562650 variants rs662_C variant, albeit no association with stroke or global CV was
behaved similarly in our group of patients, most of whom were het- apparent, we observed a positive association with high platelet levels,
erozygous for the three of them simultaneously. Furthermore, the co- indicating a potential impact on hemostasis and inflammation.
occurrence of minor alleles in haplotype A – which, to our knowledge, The only consensual modifiers of SCD disease severity, so far, have
has not been previously described – was associated with lower LDH been the persistence of HbF beyond infancy and the presence of
values and potentially to a protective effect against hemolysis. -α3.7kb–thal deletion. The co-inheritance of -α3.7kb–thal and homo-
NOS3 (or eNOS) is the major NO-producer enzyme in the cardio- zygous HbS mutation has been associated with an overall ameliorating
vascular system, playing a crucial role in vascular tone control, vascular effect on anemia, particularly a protective effect against stroke in
remodeling and proliferation. Furthermore, in SCA, NO bioavailability children [28]. We did not find any direct relationship between the
plays a very important modulating role, primarily through scavenging presence of the -α3.7kb–thal deletion and stroke, although we did find
by cell-free hemoglobin [22]. The rs2070744 variant, located at posi- that patients with α–thal showed a higher RBC count, lower MCV and
tion −786 of the NOS3 gene 5′ flanking region, has been correlated MCH, consistent with previous reports [29,30] of ameliorating anemia
with cardiovascular disease, namely its C allele, although there is still factors. Other authors have also reported an absence of association
debate about how it affects mRNA and protein levels. In our study, we between -α3.7kb–thal presence and stroke protection [31,32]. Despite
did not find any association of this variant with CV or stroke risk, which the small sample size in our study, we cannot exclude that population
may be in accordance with previous reports of no significant differences heterogeneity or other specific population characteristics may con-
between CC and TT genotypes effect on NOS3 promoter activity [23]. tribute for the lack of association observed. Additionally, the un-
The fact that the rs2070744_C has different distributions in different expected association with increased neutrophil count might lower the
ethnic backgrounds [24] may also be responsible, to some extent, for above mentioned potentially favorable effect by indicating a proin-
these differences, since that allele is more frequent in Caucasians and flammatory role. The latter was also found in subjects with the Bantu
our study population is of sub-Saharan origin. However, we observed a haplotype while the Senegal haplotype seems to have the opposite ef-
significant association between CC and TC genotypes of this variant and fect, ameliorating inflammation and the hematological indices. Never-
lower RDW, which has been discussed as a possible biomarker of lower theless, in our study, no HBB cluster haplotypes were found to be as-
cerebrovascular disease risk [25]. Lower RDW values (or reduced ani- sociated with CV.
socytosis) would potentially act as a protective factor in consonance
with what we observed for NOS3_haplotype V, which includes allele C, 5. Conclusion
and CV protection. Although rs2070744 has been described in asso-
ciation with cardiovascular disease [26], its role in ischemic stroke has Although the sample size in our study limits extrapolation to the
not been consensual. NOS3_haplotype V also includes intron 4 VNTR_4b general SCA pediatric population, our results seem to reinforce the
allele, a variant that showed a protective effect against SCI. However, importance of genetic modulators in the pathophysiology of cerebral
no relationship with any of the CV presentations studied here was ob- vasculopathy and provide clues for the discovery of novel targets for
served. On the other hand, the NOS3 rs1799983_T allele, which leads to SCA therapy. Our findings lead us to suggest that a comprehensive
aspartate for glutamate substitution at eNOS position 298, has been biomarker panel that includes biochemical, hematological, imaging as
previously reported to be related to deficient eNOS caveolar localiza- well as genetic data may be very important for CV prediction, and
tion and deficient shear stress response leading to reduced enzymatic prevention. Even though the patients we studied are subjected to en-
activity. This SNP has been found in some populations to be more vironmental, both physical and social, factors different from those to
prevalent in patients with coronary artery disease, ischemic stroke and which the populations their parents originated from have been exposed,
arterial hypertension [27]. However, in our study population, we did the genetic modifiers described in our study, namely VCAM1 haplotype
not observe any relationship of rs1799983 TT or GT genotypes with CV, 7 and rs1409419, may provide further tools for CV prevention in SCA.
biochemical or hematological parameters. Functional studies are of the utmost importance, not only for con-
Several studies have identified other candidate gene polymorphisms firmation purposes, but also to assess the mechanisms by which the
as potentially affecting the risk of CV. However, the results published so phenotype modulation may occur, and the potential use of these var-
far have been conflicting. A GWAS study by Flanagan et al. [9], per- iants as novel genetic biomarkers of disease prognosis.
formed in a large cohort of mainly African American SCA pediatric
patients, showed a decreased risk of clinically overt stroke in associa- Author statement
tion with GOLGB1 rs3732410_G (Y1212C) and ENPP1 rs1044498_C
(K173Q) mutations, whereas PON1 rs662_C (Q192R) was associated Due to the sensitive nature of the question asked in this study,
with increased risk of stroke [9]). In the same study, GOLGB1 Y1212C survey respondents were assured raw data would remain confidential
was associated with reduced TCD velocities and lower frequencies of and would not be shared.
SCIs. Conversely, reports from Martella et al. [7] and Belisário et al.
[10] indicated a link between ENPP1 rs1044498_A and increased stroke CRediT authorship contribution statement
risk as well as high TCD velocities. In our study, the ENPP1
rs1044498_A allele was found in 18% of patients compared to 68.33% Marisa Silva:Methodology, Software, Formal analysis,
of Martella et al. [7], 26.08% of Belisário et al. [10] and 5.08% of Investigation, Data curation, Writing - original draft.Sofia
Flanagan et al. [9], while homozygosity for the GOLGB1 rs3732410_G Vargas:Methodology, Software, Formal analysis, Investigation,

6
M. Silva, et al. Blood Cells, Molecules and Diseases 83 (2020) 102436

Data curation.Andreia Coelho:Methodology, Software, Formal [8] A. Coelho, A. Dias, A. Morais, B. Nunes, E. Ferreira, I. Picanço, P. Faustino,
analysis, Investigation, Data curation.Emanuel Ferreira:Method- J. Lavinha, Genetic variation in CD36, HBA, NOS3 and VCAM1 is associated with
chronic haemolysis level in sickle cell anaemia: a longitudinal study, Eur. J.
ology, Software, Investigation.Joana Mendonça:Methodology, Haematol. 92 (2014) 237–243, https://doi.org/10.1111/ejh.12226.
Software.Luís Vieira:Methodology, Software, Funding [9] J.M. Flanagan, V. Sheehan, H. Linder, T.A. Howard, Y.D. Wang, C.C. Hoppe,
acquisition.Raquel Maia:Resources, Writing - review & editing. B. Aygun, R.J. Adams, G.A. Neale, R.E. Ware, Genetic mapping and exome se-
quencing identify 2 mutations associated with stroke protection in pediatric pa-
Alexandra Dias:Resources, Writing - review & editing.Teresa tients with sickle cell anemia, Blood 121 (2013) 3237–3245, https://doi.org/10.
Ferreira:Resources, Writing - review & editing.Anabela 1182/blood-2012-10-464156.
Morais:Resources.Isabel Mota Soares:Resources, Writing - review [10] A.R. Belisário, R.R. Sales, N.E. Toledo, C. Velloso-Rodrigues, C.M. Silva, M.B. Viana,
Association between ENPP1 K173Q and stroke in a newborn cohort of 395 Brazilian
& editing.João Lavinha:Funding acquisition, Writing - review & children with sickle cell anemia, Blood 126 (2015) 1259–1260, https://doi.org/10.
editing.Rita Silva:Resources, Writing - review & editing.Paula 1182/blood-2015-05-645176.
Kjöllerström:Resources, Writing - review & editing.Paula [11] S.H. Orkin, H.H. Kazazian, S.E. Antonarakis, S.C. Goff, C.D. Boehm, J.P. Sexton,
et al., Linkage of β-thalassaemia mutations and β-globin gene polymorphisms with
Faustino:Project administration, Funding acquisition,
DNA polymorphism in human β-globin gene cluster, Nature 296 (1982) 627–631.
Methodology, Supervision, Writing - review & editing. [12] C. Dodé, R. Krishnamoorthy, J. Lamb, Rapid analysis of α3.7 thalassaemia and
αααanti3.7 triplication by enzymatic amplification analysis, Br. J. Haematol. 83
Declaration of competing interest (1993) 105–111.
[13] J.T. Robinson, H. Thorvaldsdóttir, W. Winckler, M. Guttman, E.S. Lander, G. Getz,
J.P. Mesirov, Integrative genome viewer, Nat. Biotechnol. 29 (2011) 24–26,
The authors have no competing interests. https://doi.org/10.1038/nbt.1754.Integrative.
[14] M.F. Iademarco, J.J. McQuillan, G.D. Rosen, D.C. Dean, Characterization of the
promoter for vascular cell adhesion molecule-1 (VCAM-1), J. Biol. Chem. 267
Acknowledgments (1992) 16323–16329 http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
Retrieve&db=PubMed&dopt=Citation&list_uids=1379595%5Cnhttp://www.jbc.
org/content/267/23/16323.abstract.
The authors wish to thank the SCA patients and their parents for [15] G.D. Rosen, T.M. Birkenmeier, D.C. Dean, Characterization of the α4 integrin gene
their participation in this study. We would also wish to thank Prof. promoter, Proc. Natl. Acad. Sci. U. S. A. 88 (1991) 4094–4098, https://doi.org/10.
Constança Coelho for her support in statistical analyses and insightful 1073/pnas.88.10.4094.
[16] E.A. Bard-Chapeau, J. Jeyakani, C.H. Kok, J. Muller, B.Q. Chua, J. Gunaratne,
review of the manuscript.
a. Batagov, P. Jenjaroenpun, V.A. Kuznetsov, C.-L. Wei, R.J. D’Andrea, G. Bourque,
N.A. Jenkins, N.G. Copeland, Ecotopic viral integration site 1 (EVI1) regulates
Financial support multiple cellular processes important for cancer and is a synergistic partner for FOS
protein in invasive tumors, Proc. Natl. Acad. Sci. 109 (2012) 2168–2173, https://
doi.org/10.1073/pnas.1119229109.
This work was partially funded by Fundação para a Ciência e a [17] R. Meech, K.N. Gonzalez, M. Barro, A. Gromova, L. Zhuang, J.A. Hulin,
Tecnologia (FCT) grant PIC/IC/83084/2007, ISAMB, and INSA project H.P. Makarenkova, Barx2 is expressed in satellite cells and is required for normal
2012DGH720. Additionally, it is a result of the GenomePT project muscle growth and regeneration, Stem Cells 30 (2012) 253–265, https://doi.org/
10.1002/stem.777.
(POCI-01-0145-FEDER-022184), supported by COMPETE 2020 - [18] M.F. Iademarco, J.J. McQuillan, D.C. Dean, Vascular cell adhesion molecule 1:
Operational Programme for Competitiveness and Internationalisation contrasting transcriptional control mechanisms in muscle and endothelium, Proc.
(POCI), Lisboa Portugal Regional Operational Programme Natl. Acad. Sci. U. S. A. 90 (1993) 3943–3947 http://www.pubmedcentral.nih.gov/
articlerender.fcgi?artid=46422&tool=pmcentrez&rendertype=abstract.
(Lisboa2020), Algarve Portugal Regional Operational Programme [19] G.A. Heymann, H. Kiesewetter, A. Salama, Frequencies of α4 A3061G variants and
(CRESC Algarve2020), under the PORTUGAL 2020 Partnership identification of three new variants of the human integrin α4-subunit, Mol.
Agreement, through the European Regional Development Fund (ERDF), Immunol. 39 (2003) 855–860, https://doi.org/10.1016/S0161-5890(03)00009-9.
[20] V. Andreoli, R. Cittadella, P. Valentino, F. Condino, A. La Russa, M. Liguori,
FCT. I. Manna, P. Spadafora, R. Nisticò, D. Pirritano, A. Clodomiro, A. Quattrone, The
role of VLA4 polymorphisms in multiple sclerosis: an association study, J.
Appendix A. Supplementary data Neuroimmunol. 189 (2007) 125–128, https://doi.org/10.1016/j.jneuroim.2007.
06.015.
[21] K. Hilger-Eversheim, M. Moser, H. Schorle, R. Buettner, Regulatory roles of AP-2
Supplementary data to this article can be found online at https:// transcription factors in vertebrate development, apoptosis and cell-cycle control,
doi.org/10.1016/j.bcmd.2020.102436. Gene 260 (2000) 1–12, https://doi.org/10.1016/S0378-1119(00)00454-6.
[22] C.H.D.R. Reiter, X.U.W. Ang, J.O.S.E.E.T.A. Antos, N.E.I.L.H. Ogg, R.I.O.C.A. Iii, A.
L.A.N.N.S. Chechter, M.A.R.K.T.G. Ladwin, FreeHb limits NO availability in SCD,
References Nat. Med. 8 (2002) 1383–1389. doi:https://doi.org/10.1038/nm799.
[23] A.S. Sim, J. Wang, D. Wilcken, Xing Li Wang, Mspl polymorphism in the promoter
[1] D.J. Weatherall, The inherited diseases of hemoglobin are an emerging global of the human endothelial constitutive NO synthase gene in Australian caucasian
health burden, Blood 115 (2010) 4331–4336, https://doi.org/10.1182/blood- population, Mol. Genet. Metab. 65 (1998) 62, https://doi.org/10.1006/mgme.
2010-01-251348. 1998.2741.
[2] S.D. Grosse, I. Odame, H.K. Atrash, D.D. Amendah, F.B. Piel, T.N. Williams, Sickle [24] J.E. Tanus-Santos, M. Desai, D.A. Flockhart, Effects of ethnicity on the distribution
cell disease in Africa: a neglected cause of early childhood mortality, Am. J. Prev. of clinically relevant endothelial nitric oxide variants, Pharmacogenetics 11 (2001)
Med. 41 (2011) S398–S405, https://doi.org/10.1016/j.amepre.2011.09.013. 719–725, https://doi.org/10.1097/00008571-200111000-00011.
[3] F.B. Piel, S.I. Hay, S. Gupta, D.J. Weatherall, T.N. Williams, Global burden of sickle [25] N. Li, H. Zhou, Q. Tang, Red blood cell distribution width: a novel predictive in-
cell anaemia in children under five, 2010–2050: modelling based on demographics, dicator for cardiovascular and cerebrovascular diseases, Dis. Markers 2017 (2017),
excess mortality, and interventions, PLoS Med. 10 (2013), https://doi.org/10. https://doi.org/10.1155/2017/7089493.
1371/journal.pmed.1001484. [26] D. Colomba, G. Duro, S. Corrao, C. Argano, T. Di Chiara, D. Nuzzo, F. Pizzo,
[4] K. Ohene-Frempong, S.J. Weiner, L.A. Sleeper, S.T. Miller, S. Embury, J.W. Moohr, G. Parrinello, R. Scaglione, G. Licata, Endothelial nitric oxide synthase gene poly-
D.L. Wethers, C.H. Pegelow, F.M. Gill, Cerebrovascular accidents in sickle cell morphisms and cardiovascular damage in hypertensive subjects: an Italian case-
disease: rates and risk factors, Blood. 91 (1998) 288–294. control study, Immun. Ageing 5 (2008) 4, https://doi.org/10.1186/1742-4933-5-4.
[5] I.D. Buchanan, A. James-Herry, I. Osunkwo, The other side of abnormal: a case [27] I. Armenis, V. Kalotychou, R. Tzanetea, P. Kollia, Z. Kontogeorgiou,
series of low transcranial doppler velocities associated with stroke in children with D. Anastasopoulou, M. Mantzourani, M. Samarkos, K. Pantos, K. Konstantopoulos,
sickle cell disease, J. Pediatr. Hematol. Oncol. 35 (2013) 543–546, https://doi.org/ I. Rombos, Prognostic value of T786C and G894T eNOS polymorphisms in sickle
10.1097/MPH.0b013e318279caae. cell disease, Nitric Oxide Biol. Chem. 62 (2017) 17–23, https://doi.org/10.1016/j.
[6] G. Buchanan, E. Vichinsky, L. Krishnamurti, S. Shenoy, Severe sickle cell disease– niox.2016.11.002.
pathophysiology and therapy, Biol. Blood Marrow Transplant. 16 (2010) S64–S67, [28] G.J. Kato, M.T. Gladwin, M.H. Steinberg, Deconstructing sickle cell disease: re-
https://doi.org/10.1016/j.bbmt.2009.10.001. appraisal of the role of hemolysis in the development of clinical subphenotypes,
[7] M. Martella, N. Quaglia, A.C. Frigo, G. Basso, R. Colombatti, L. Sainati, Association Blood Rev. 21 (2007) 37–47, https://doi.org/10.1016/j.blre.2006.07.001.
between a combination of single nucleotide polymorphisms and large vessel cere- [29] M.B. Rumaney, V.J. Ngo Bitoungui, A.A. Vorster, R. Ramesar, A.P. Kengne,
bral vasculopathy in African children with sickle cell disease, Blood Cells Mol. Dis. J. Ngogang, A. Wonkam, The co-inheritance of alpha-thalassemia and sickle cell
61 (2016) 1–3, https://doi.org/10.1016/j.bcmd.2016.07.005. anemia is associated with better hematological indices and lower consultations rate

7
M. Silva, et al. Blood Cells, Molecules and Diseases 83 (2020) 102436

in Cameroonian patients and could improve their survival, PLoS One 9 (2014) A.S. Araújo, A.R. Lucena-Araujo, M.A. Bezerra, Influence of the βs haplotype and α-
e100516, , https://doi.org/10.1371/journal.pone.0100516. thalassemia on stroke development in a Brazilian population with sickle cell
[30] M.M. Aleluia, C.C. Da Guarda, R.P. Santiago, T.C.C. Fonseca, F.I. Neves, R.Q. De anaemia, Ann. Hematol. 93 (2014), https://doi.org/10.1007/s00277-014-2016-1.
Souza, L.A. Farias, F.A. Pimenta, L.M. Fiuza, T.N. Pitanga, J.R.D. Ferreira, [32] S.L. Saraf, R.E. Molokie, M. Nouraie, C.A. Sable, L. Luchtman-Jones, G.J. Ensing,
E.V. Adorno, B.A.V. Cerqueira, M. de S. Gonçalves, Association of classical markers A.D. Campbell, S.R. Rana, X.M. Niu, R.F. Machado, M.T. Gladwin, V.R. Gordeuk,
and establishment of the dyslipidemic sub-phenotype of sickle cell anemia, Lipids Differences in the clinical and genotypic presentation of sickle cell disease around
Health Dis. 16 (2017) 1–9, https://doi.org/10.1186/s12944-017-0454-1. the world, Paediatr. Respir. Rev. 15 (2014) 4–12, https://doi.org/10.1016/j.prrv.
[31] I.F. Domingos, D.A. Falcão, B.L. Hatzlhofer, A.F. Cunha, M.N. Santos, 2013.11.003.
D.M. Albuquerque, K.Y. Fertrin, F.F. Costa, R.C. Azevedo, C.G. Machado,