Article Text
PDF
Abstract
Background The Epstein–Barr virus (EBV) and Kaposi's sarcoma associated herpesvirus (KSHV) are consistently associated with lymphoproliferative diseases and cancers in humans, notably in patients with HIV.
Aims Our aim was to evaluate whether EBV and/or KSHV viral loads regularly assessed in peripheral blood mononuclear cells (PBMC) correlate with clinical or laboratorial parameters retrieved for patients living with HIV.
Methods This was a longitudinal study with a cohort of 157 HIV positive patients attending an academic HIV outpatient clinic in São Paulo State, Brazil. For each patient, up to four blood samples were collected over a 1 year clinical follow-up: on enrolment into the study, and after 4, 8 and 12 months. Total DNA was extracted from PBMC, and EBV and KSHV viral loads were assessed by real time quantitative PCR.
Results Higher viral loads for EBV were significantly associated with high HIV viraemia, a greater number of circulating T CD8+ cells and lack of virological response to the antiretroviral treatment. KSHV viral load was undetectable in virtually all samples.
Conclusions EBV viral load in PBMC correlated with the number of circulating T CD8+ lymphocytes and the response to the antiretroviral therapy in HIV infected patients. In contrast, KSHV was undetectable in PBMC, presumably an effect of the antiretroviral treatment. Therefore, either KSHV infection in the population studied was absent or viral load in PBMC was beyond the analytical limit of the assay.
- EBV
- HIV
- IMMUNOCOMPRISED HOST
- INFECTIONS
- KAPOSI'S SARCOMA
Statistics from Altmetric.com
Introduction
The Epstein–Barr virus (EBV) and Kaposi's sarcoma associated herpesvirus (KSHV, also known as human herpesvirus type 8) are two gammaherpesviruses consistently associated with the development of human malignancies. EBV is most frequently associated with subsets of Hodgkin, non-Hodgkin's lymphomas (NHL) and a few epithelial malignancies. Conversely, KSHV is the causal agent of Kaposi's sarcoma (KS) and a few rare lymphoproliferative disorders. EBV and KSHV are more often associated with cancers in immunocompromised individuals.1–4 Remarkably, the introduction of the highly active antiretroviral therapy (HAART) regimens for HIV positive patients has dramatically decreased the incidence of KS,5 as well as the risk of these patients developing EBV associated NHL.6–8
EBV and KSHV viral loads have been evaluated as potential biomarkers to monitor the development and progression of virus induced cancers, notably in the HIV setting.9–13 Previous studies indicate that HIV infected patients have similar levels of EBV viral load in their peripheral blood cells, regardless of whether they are being treated with HAART.14 ,15 Some studies have reported that high EBV viral load impacts the risk of developing NHL.16 ,17 It has been suggested that the viral load for EBV reaches a new set point following HIV seroconversion, which could explain the lack of a predictive value for the development of EBV associated NHL in HIV positive patients.18
In contrast, patients treated with HAART present with lower KSHV viral loads and a reduced risk of KS compared with patients not administered antiretroviral treatment.19 Not only does the risk of KS correlate with KSHV detection in peripheral blood mononuclear cells (PBMC), but in addition, viral load has been associated with the severity and stage of KSHV induced cancers in HIV infected patients.20–23 It has been hypothesised that HAART has a direct effect on the reactivation and replication of EBV and KSHV, and that this accounts for improved control of the primary infection, particularly for KSHV.24
The clinical relevance of EBV and KSHV viral loads in peripheral blood during HIV infection has been a disputed topic in recent years. In this study, viral loads for these gammaherpesviruses were prospectively evaluated in PBMC retrieved from HIV positive patients, in order to access possible associations with clinical and laboratory parameters commonly used to monitor the progression of HIV infection.
We concluded that EBV viral load may be informative about changes in the natural history of HIV disease, and this deserves further investigation. Conversely, because KSHV detection by a sensitive molecular technique in the same group of patients was consistently negative, the relevance of KSHV viral load during HIV infection may be relevant only in specific clinical situations, notably in cases of KSHV associated diseases, such as Kaposi sarcoma, and very rare lymphoproliferative disorders (Castleman disease and primary effusion lymphomas).
Patients and methods
Patients
The study was conducted in accordance to the Declaration of Helsinki, and was approved by the Ethical Committee on Research at the Botucatu Medical School (FMB), São Paulo State University (UNESP), SP, Brazil.
Patients infected by HIV who attended the FMB-UNESP outpatient clinic were prospectively evaluated. The diagnosis of HIV infection was conducted in accordance with national Brazilian policies in effect during the period of patient recruitment.25 Briefly, the diagnosis was based on one blood sample testing positive for HIV-1 using two different immunoassay evaluations (ELISA) followed by a confirmatory test (immunofluorescence or immunoblot). If all three tests were reagent, this first sample was considered positive for HIV-1, and another sample was used to repeat the immunoassay to further confirm the result. The diagnosis of HIV infection was established when investigation of both blood samples resulted in a positive response.
A total of 157 HIV positive patients were enrolled in the study. Cohort size was defined by convenience sampling, taking into account the maximum number of eligible patients according to the following criteria: participants were at least 18 years old, presented no evidence of proliferative disease associated with EBV and/or KSHV, and were treated exclusively with antiretroviral drugs and regimens recommended by the Brazilian Federal Government. Furthermore, they were not simultaneously enrolled in any therapeutic clinical trial. Patients were followed up for 12 months after enrolment, and four peripheral blood samples were obtained from each participant at an average interval of 4 months between samples. Sampling time points were denominated as 0, 1, 2 and 3: time of enrolment, and 4, 8 and 12 months afterwards, respectively. The number of blood samples obtained during each of these time points were 157 (100%), 153 (97.4%), 137 (87.3%) and 115 (73.2%), respectively.
Information on age, sex, declared sexual orientation, presumed route of HIV infection and estimated time since HIV diagnosis were retrieved during medical interviews with the patients. Data on HIV viraemia, number of circulating T CD4+ and T CD8+ lymphocytes, and HIV infection response to antiretroviral therapy were retrieved from the medical records. Required data on HIV clinical course were also retrieved from the medical records in order to define the stage of HIV infection, according to the criteria of the Center for Disease Control (CDC, USA).26
Assessment of EBV and KSHV viral loads
Blood samples obtained by regular peripheral venepuncture were stored in 4 mL BD Vacutainer tubes with EDTA (Beckson-Dickson, Rutherford, New Jersey, USA). Shortly after blood collection, PBMC were separated by centrifugation with Histopaque (Sigma-Aldrich, St Louis, Missouri, USA), followed by DNA extraction with the QIAamp DNA Blood Midi Kit (Qiagen, Valencia, California, USA), as instructed by the manufacturer. Extracted DNA was resuspended, quantified by UV spectrophotometry using NanoVue equipment (GE Healthcare, Boston, Massachusetts, USA) and stored at −20°C when not in use.
EBV and KSHV viral loads were assessed by real time quantitative PCR (qPCR) using Rotor-Gene 6000 (Corbett Research, Mortlake, Australia) and Taqman probes. The oligonucleotides used were retrieved from the literature (table 1).27–29 The primers used to assess EBV viral load flank conserved regions in EBV types 1 and 2, as confirmed using the sequences for the viral prototypes B95.8 (NCBI GenBank accession No VO1555.2; nt121 571–121 622) and AG876 ( accession No DQ279927; nt110 132–110 188), respectively. Thermocycling conditions are detailed in table 1. The qPCR reactions were set up as follows: 12.5 µL of TaqMan Universal Master Mix reagent (Applied Biosystems, Foster City, California, USA), 0.30μM of each primer and 0.15 mM or 0.10 mM of the appropriate TaqMan probe (for EBV and KSHV, respectively), 4 μL of sample DNA (average concentration 65 ng/µL) and DNase/RNase free water up to 20 μL of reaction volume.
Sequences of oligonucleotides (primers and TaqMan probes) and conditions for the quantitative PCR assays for qualitative and quantitative analysis of Epstein–Barr virus and Kaposi's sarcoma associated herpesvirus, and for estimating the number of cells corresponding to the quantity of DNA in samples from patients with HIV (quantitative PCR for amplification of human endogenous retrovirus 3)
EBV and KSHV viral loads were estimated using standard curves prepared with serial dilutions (1:10) of commercial samples of genomic viral DNA (Advanced Biotechnologies, Columbia, Maryland, USA). For EBV and KSHV, genomic samples of DNA were derived from B95-8 and KS-1 cells, respectively. Both standard curves were set to range from 106 to 101 viral copies per qPCR reaction (20 μL).
In order to express the qPCR results in terms of number of viral copies per PBMC, the number of cells was estimated by amplification of human endogenous retrovirus 3 (ERV-3), as previously proposed.29 A plasmid containing the complete ERV-3, kindly provided by Dr Denise Whitby (National Cancer Institute, Frederick, Maryland, USA), was used to prepare the ERV-3 standard curve (1:10 serial dilutions), which is equivalent to 105 to 101 cells per reaction. Reactions for ERV-3 were set up as described for EBV; oligonucleotides and the thermocycling programme used for this assay are described in table 1 and in the online supplementary material 1. Independent of the target, samples were always assayed in duplicate.
Data analysis
Possible associations involving data for age, sex, declared sexual orientation, presumed route for HIV infection, estimated time since HIV diagnosis, stage of retroviral infection,26 HIV viraemia, number of circulating T CD4+ and T CD8+ lymphocytes, and HIV infection response to antiretroviral therapy were evaluated. For response to antiretroviral therapy, patients were divided into the following groups: (i) no data available; (ii) no antiretroviral treatment for at least 6 months; (iii) receiving antiretroviral treatment and no therapeutic failure (ie, HIV viraemia below 50 copies/mL for more than 6 months using the same drugs continuously); (iv) receiving antiretroviral treatment with treatment failure (ie, detectable HIV viraemia or >50 copies/mL after 6 months on HAART); and (v) unclassifiable (ie, the available data do not match any of the previous situations).
The Mann–Whitney and Fisher exact tests were used to evaluate possible associations between sex and age, or sex and declared sexual orientation, respectively. Spearman's test was used to assess possible associations regarding EBV viral load and age, estimated time of HIV diagnosis, HIV viraemia and number of T CD4+ or T CD8+ lymphocytes. Associations with response to antiretroviral treatment were analysed using the χ2 test.
In order to evaluate how often gammaherpesviral DNA was detected in samples during the clinical follow-up, the term ‘viral persistence’ was arbitrarily defined at 8 or 12 months whenever measurable viral load was determined at two or three consecutive time points. This analysis was only possible for EBV. Persistence of EBV infection at 8 and 12 months was evaluated for the following covariates: age, sex, declared sexual orientation, estimated time of diagnosis of retroviral infection, HIV viraemia and number of circulating T CD4+ and CD8+ lymphocytes, and CDC classification data. An adjusted logistic model and analysis of deviance were used to test the effects of these covariates.30 The association between EBV viral load throughout the follow-up and response to antiretroviral treatment, number of circulating T CD4+ and T CD8+ lymphocytes or HIV viraemia was performed using a semiparametric generalised additive model.31
In all analysis, correlations were considered significant when p≤0.05; those values of 0.05≤p≤0.1 were considered trends. Statistical analyses were performed using the SAS/STAT software V.9.1.3 (SAS Institute Inc, Cary, North Carolina, USA). The statistical analyses performed and their p values are shown in the online supplementary material 1.
Results
Study group
Of the 157 HIV infected patients, 92 (58.6%) were men and 65 (41.4%) were women; median ages were 41.5 and 35.0 years, respectively (p=0.003; Mann–Whitney test) (table 2). CDC classification was established for 152/157 (96.8%) patients. Ten (6.6%) patients were asymptomatic and did not show a significant reduction in the number of circulating T CD4+ cells (A1) whereas 65 (42.8%) were classified as presenting with the most advanced stage of HIV disease (C3). At least half of the patients had suffered AIDS associated illnesses at some point (CDC category C, 50.7%) or nadir number of T CD4+ lymphocytes below 200 cells/µL (CDC category 3, 59.9%).
Sex and age of patients infected by HIV included in the study
According to the response to antiretroviral treatment, patients were divided into four groups: no data available (17.2%), no antiretroviral therapy (12.7%), receiving HAART and without therapeutic failure (40.8%), receiving HAART and with therapeutic failure (12.7%) and unclassified, when not matching any of the previous categories (16.6%). Additional information for these groups is available in the online supplementary material 2. Table 3 shows a more detailed characterisation of the patients included in the study.
General features of 150 HIV positive patients evaluated for Epstein–Barr virus and Kaposi's sarcoma associated herpesvirus infection in peripheral blood mononuclear cells
EBV and KSHV viral loads in PBMC
Representative qPCR amplification plots for quantification of ERV-3, EBV and KSHV are shown in the online supplementary material 3.
EBV viral load in PBMC was detectable in 125/157 (79.6%) samples at time point 0, 118/152 (77.6%) at time point 1 (4 months), 101/137 (73.7%) at time point 2 (8 months) and 81/117 (69.2%) at time point 3 (12 months) (table 3). A positive linear correlation was verified between EBV viral load and the estimated time of diagnosis of HIV infection for samples at time points 0, 2 and 3 (p=0.002, p=0.013 and p=0.0096, respectively) (table 4). Furthermore, a positive linear correlation between the number of T CD8+ lymphocytes and EBV viral load in samples from time points 0 and 1 was observed. No significant correlations were observed regarding the number of T CD4+ lymphocytes.
Evaluation of Epstein–Barr virus viral load for the four time points of the study
Patients presenting with therapeutic failure had a higher EBV load in PBMC (p<0.0001) (figure 1). A significant correlation was found between EBV viral load and HIV viraemia at time point 0, and patients with a high HIV viraemia presented with higher EBV viral loads throughout follow-up (p<0.0001, semiparametric generalised additive model). Patients presenting with therapeutic failure frequently showed EBV viral persistence at 8 and 12 months (p=0.0356 and p=0.0307, respectively). A longer estimated HIV diagnosis period was significantly associated with EBV viral persistence at 12 months (p=0.0025), but not at 8 months. A trend was observed in patients with persistent EBV detection at 12 months who showed a higher number of circulating T CD8+ lymphocytes on enrolment in the study (p=0.0632). No other correlations with EBV viral persistence were observed (see online supplementary material 1).
Epstein–Barr virus (EBV) viral load in samples from patients infected with HIV at time points 0, 1, 2 and 3, evaluated according to the response to antiretroviral treatment. ▪p=0.0017; ▪▪p<0.0001; †p=0.0016; ††p<0.0001; *p<0.0001; **p=0.0525 (χ2 test). HAART, highly active antiretroviral therapy; PBMC, peripheral blood mononuclear cells.
At time points 0, 2 and 3, KSHV viral load in PBMC was consistently undetectable or below the calculated analytical sensitivity of the qPCR method performed (1.1 viral copies/µL). At time point 1 (4 months), only one sample was KSHV positive, with a viral load of 1300 viral copies/10−6 PBMC. As KSHV viral load was undetectable in all samples except for one, these data could not be evaluated further.
Discussion
In this study, EBV and KSHV DNA were assessed in peripheral blood samples from HIV infected patients over a 12 month follow-up period. PBMCs was chosen for evaluation of EBV and KSHV viral loads because plasma or whole blood samples would not be appropriate to assess the population of latently infected cells, which is biologically relevant both in terms of viral reactivation and B cell transformation by these oncogenic herpesviruses. In this regard, evaluation of viral load exclusively by whole blood or plasma was recently regarded as inappropriate to infer virus–host interactions in relevant clinical scenarios.32
EBV viral load in PBMCs was detectable in most of the samples evaluated, as reported in other studies.14 ,33 ,34 The lack of an association between EBV detection and number of circulating T CD4+ lymphocytes is in agreement with previous reports.14 ,18 ,35 ,36 Although Ling et al17 verified that the estimated EBV viral load did not correlate with HIV viraemia, Panagiotaki et al37 reported that EBV was more frequently detected when HIV viraemia exceeded 50 copies/mL. In this study, a significant linear correlation between EBV viral load and HIV viraemia was consistently observed (table 5). Furthermore, patients with high HIV viraemia on enrolment to the study (time point 0) presented with higher EBV viral loads throughout follow-up.
Results for correlations involving the Epstein–Barr virus viral load and age of the patients, estimated time of retroviral infection, HIV viraemia and circulating T CD4+ and T CD8+ cell counts
A significant correlation was also found between EBV viral load and number of T CD8+ lymphocytes in samples taken during the first 4 months of follow-up (time points 0 and 1). An increase in the absolute numbers of CD8+ T cells specific to EBV lytic antigens following HIV seroconversion has been previously reported.18 It should be noted that the significantly higher numbers of circulating T CD8+ cells in HIV patients with a high EBV viral load in this study was not characterised in terms of specific responses to EBV antigens. Therefore, it is not possible to assume that the increase in T CD8+ cells is due to expansion of cells responsive to EBV antigens, but it remains a possible explanation.
Even if we consider that the number of circulating T CD4+ lymphocytes in HIV infected patients does not have a significant impact on EBV viral load, immunological disturbances occurring during HIV infection seemingly contribute to increased rates of EBV replication and viral spread within the body. One plausible mechanism for the increase in EBV viral load following HIV seroconversion is B cell activation.18 Broad and abnormal activation of B cells is a common finding during HIV infection, often expressed clinically as hypergammaglobulinaemia.38 Activation of B cells can lead to EBV reactivation, increased expression of EBV antigens and expansion of EBV reactive T CD8+ cells. Patients infected by both HIV and EBV present functional impairment of T CD8+ lymphocytes that is not necessarily associated with a decrease in the number of circulating cells. Functional impairment of T CD8+ lymphocytes follows a deficiency in T CD4+ lymphocytes,9 ,39 which in turns contributes to abnormal B cell activation, increased EBV reactivation and viral spread. Although Jacobson et al did not find an association between EBV or KSHV replication and activation of T CD8+ lymphocytes in HIV infected cells following administration of HAART,40 in that study, viral DNA was investigated in saliva samples, which cannot be regarded as a biological compartment equivalent to peripheral blood.
Impairment of retroviral replication by HAART causes an increase in the number of circulating T CD4+ lymphocytes in HIV infected patients. Although HAART does not seem to cause a significant change in EBV viral load in HIV infected individuals,41 ,42 a reduction in herpesviral reactivation is reported, as well as partial recovery of specific cytotoxic T lymphocyte responses against EBV.39 In this study, higher levels of EBV viral loads were more frequently verified in patients who experienced failure of antiretroviral treatment, defined as an increase in HIV viraemia and a decrease in the number of circulating T CD4+ lymphocytes. As this condition favours EBV lytic replication, the associations involving the response to antiretroviral treatment, changes in proportions of circulating T CD4+ and T CD8+ lymphocytes, HIV viraemia and EBV viral load could be anticipated.
Data concerning the clinical consequences of persistent EBV detection or high herpesviral load in HIV infected hosts remain scarce. Petrara et al43 reported that the expansion of EBV infected cells and appearance of EBV related malignancies could be associated with the persistence of HIV-1 viraemia and immune activation, regardless of peripheral T CD4+ lymphocyte depletion. Conversely, persistently high EBV viral load in PBMC might be a common event in the HIV infection setting, even in patients on HAART, and it does not necessarily change the risk of development of EBV associated diseases.14
Because the EBV load is frequently increased in HIV infected patients, viral reservoirs other than memory B cells might participate in EBV persistence. Richard et al44 suggested that transitional-like B cells must be an alternative EBV reservoir in patients with chronic HIV infection, because they persist concomitantly with high EBV load following the use of HAART, and their relative increase among PBMCs could be associated with a higher risk of developing B cell lymphomas. In this study, no evidence of EBV associated lymphomas was observed in any patient enrolled.
KSHV DNA was undetectable in virtually all samples evaluated in this study. Detection of KSHV in peripheral blood of HIV patients is quite uncommon, notably when antiretroviral drugs are used, and even when KS is present. Using a highly sensitive qPCR assay, Pellet et al45 did not detect KSHV in peripheral blood in any of the eight HIV infected patients with KS evaluated, all of whom were treated with antiretroviral drugs. However, Albrecht et al detected KSHV in 21/260 (7.8%) HIV infected patients without KS, all of whom had low viral loads in PBMC,46 and Minami et al23 reported KSHV detection by qPCR in 37/125 (29.6%) peripheral blood samples of HIV infected patients. In this latter study, detection was lower in the group of patients on antiretroviral treatment (12/58; 20.7%) compared with the group not receiving treatment (25/67; 37.3%); the authors concluded that KSHV detection in peripheral blood by molecular methods is more feasible in a setting of high herpesviral replication, usually associated with uncontrolled HIV viraemia and low numbers of circulating CD4+ T cells.23
None of 803 KSHV seropositive patients from three distinct geographic regions in Brazil (the states of São Paulo, Amazon and Bahia) had KSHV DNA detectable in peripheral blood samples by qPCR. Therefore, even when there is immunological evidence of previous exposure to KSHV, peripheral viraemia in individuals without KSHV associated diseases is below the limit of detection of sensitive molecular techniques.47 It is plausible to reason that the lack of detection of KSHV DNA in PBMC reported here is related to the systematic and universal antiretroviral treatment provided to HIV infected patients enrolled. One of the main limitations of this study is that KSHV seroprevalence could not be assessed; it would be valuable to interpret the lack of KSHV detection in PBMCs in the light of data on exposure of the studied cohort to the viral infection. In Brazil, high KSHV seroprevalences were reported for HIV infected patients with KS (97.4%),48 and in men who had sex with men (28.7%);49 moreover, seroprevalence among HIV negative blood donors was reported to be below 7.4%.50–53
In summary, in this study, EBV viral load in PBMCs correlated with the number of circulating T CD8+ lymphocytes and the response to antiretroviral therapy in HIV infected patients. These results indicate that poorly controlled HIV infection may impact on the EBV burden within the organism. Even if the pathological consequences of the increased EBV viral load could not be inferred in the present study (notably due to the short follow-up of patients), the results indicate that this is a clinically relevant topic that deserves further investigation. This is particularly important because HIV infection now has the status of a manageable chronic disease, although currently incurable; therefore, patients may develop EBV associated illnesses in the long term. In contrast, KSHV was undetectable, presumably an effect of antiretroviral treatment. Furthermore, low exposure to KSHV infection should also be considered, as there are no data on KSHV seroprevalence for the specific population to which patients enrolled in this study belong.
Take home messages
Epstein–Barr virus (EBV) viral load in peripheral blood mononuclear cells (PBMCs) correlated with the number of circulating T CD8+ lymphocytes in HIV infected patients, but not the number of T CD4+ lymphocytes.
Higher levels of EBV viral load in PBMCs were found in HIV positive patients under therapeutic failure.
No correlation was found between EBV viral load and age, sex, declared sexual orientation, presumed route for HIV infection, estimated time since HIV diagnosis, stage of retroviral infection or HIV viraemia.
Kaposi's sarcoma associated herpesvirus (KSHV) was consistently undetectable in the present cohort of HIV positive patients over 1 year of follow-up.
Acknowledgments
The authors would like to thank Liciana Vaz de Arruda Silveira, PhD, and Helio Rubens de Carvalho Nunes, PhD, for their support with the statistical analysis; Denise Whitby, MD, PhD, for providing the ERV-3 plasmid; Lenice do Rosário de Souza, MD, PhD, for assistance with the medical interviews; Carolyne Chaves, MD, for retrieving patient data from the medical records; Carina Alves de Oliveira and Luciene Ferreira Daltin for collecting blood samples; and Domingo Alves Meira, MD, PhD (in memoriam), for setting up the collaboration that allowed this study to be conducted.
References
Supplementary materials
Abstract in Portuguese
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Abstract in Portuguese - Online abstract
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement 1
- Data supplement 2 - Online supplement 2
- Data supplement 3 - Online supplement 3
Footnotes
Handling editor Slade Jensen
Twitter Follow Deilson Elgui de Oliveira at @DeilsonElgui
Collaborators Lenice do Rosário de Souza and Carolyne Chaves.
Contributors APFdS performed the experiments, evaluated the results and wrote the manuscript. LBG and SRdS contributed to the experimental data. ANB and RAMdBA selected the patients and retrieved clinical–epidemiological data. DEdO designed the study, obtained funding, evaluated the results and wrote the manuscript.
Funding This project was funded by the São Paulo Research Foundation (FAPESP), with scholarships awarded to APFdS (Proc MS 2007/06033-7) and LBG (IC 2008/54927-0), and a research grant to DEdO (Proc AP 2006/03643-6).
Competing interests None.
Ethics approval The study was approved by the Ethical Committee on Research (#374/2008) at the Botucatu Medical School (FMB), São Paulo State University (UNESP), SP, Brazil.
Provenance and peer review Not commissioned; externally peer reviewed.