This cross-sectional study assessed the influence of different serotypes of dengue virus on the clinical outcomes of 485 laboratory cases of dengue infection detected in Vitória, Brazil, between 2009 and 2013. The results of the present study demonstrated that infections caused by DENV-2 were more associated with SD. In this study, gender and age did not influence the associations between the serotypes and the presentation of SD.
Previous studies confirm the present findings, showing an increased proportion of severe outcomes , such as DHF [2, 7, 15, 16], and dengue shock syndrome (DSS) [15, 17] in infections caused by DENV-2. Introduction of DENV-2 was a determinant factor for the emergence of SD in different global regions , and epidemics with a high number of severe hemorrhagic cases had DENV-2 as the predominant serotype [18, 19].
DENV-2 apparently played a crucial role in the emergence of severe cases in Vitória, considering the epidemiological overview seen in the period analyzed. In the municipality, the years with higher proportion of SD were 2009 (n = 5; 22.7 %), 2010 (n = 125; 23.4 %), and 2011 (n = 268; 10.9 %), when considering cases with laboratory confirmation. In the time series analyzed, 2009, 2010 and 2011 were the only years when DENV-2 was detected. In 2009, DENV-2 was the only serotype detected, but unfortunately, the number of laboratory tests performed was low in this period. Thereby, the restricted number of SD with laboratory confirmation impairs the link of this year with an increasing severity, especially that related to DENV-2 circulation. In 2010, DENV-2 was isolated from 52.9 % of samples with serotyping and was responsible for 88.9 % (n = 8) of severe cases with serotyping, providing evidence for its relevance in the emergence of SD during the epidemic.
The mechanism involved in DENV-2 virulence is not clear. One possible factor is the stimulatory effect of DENV-2 on nitric oxide production, causing toxic and inflammatory effects, inducing apoptosis in host cells . Another factor responsible for the enhanced pathogenicity of DENV-2 is its efficient replication . As a consequence, DENV-2 infections presented high viral load .
In the DENV-2 cases of the sample analyzed, milder hemorrhagic manifestations, such as petechiae and positive tourniquet test, were more common than serious ones, such as hematemesis, melena, hematuria, menorrhagia and epistaxis, as was found in a previous study . It was not possible to generate evidence for severe manifestations being linked to DENV-2, such as plasma leakage [15, 21], cavity effusion , hypovolemic shock, internal hemorrhage , liver dysfunction , thrombocytopenia [21, 23] and hemoconcentration . However, ten cases of SD caused by DENV-2 were present and larger sample sizes would be necessary to establish a statistically sound relation between DENV-2 and clinical manifestations.
DENV-3 was anteriorly related to severe forms of dengue , including DHF  and DSS [22, 25] and to other severe manifestations [25, 26], such as liver involvement [15, 22]. However, no conclusion can be drawn from the present study, as only one case was detected.
The present results suggest that those infected by DENV-1 evolved less frequently to severe outcomes. Less severe cases were linked to DENV-1 in previous investigations as well [22, 24], and plasma leakage was rarely observed . In 2011, DENV-1 was isolated from 96.7 % (n = 355/367) of cases with serotyping in Vitória and was responsible for 88.2 % (n = 15/17) of the severe cases where serotyping was performed. Additionally, 2011 met other conditions related to increasing severity. In that period, there was a co-circulation of the serotypes DENV-1 and DENV-2, and probably a considerable number of secondary dengue infections occurred after years with wide circulation of DENV-2. Even with these two factors, DENV-1 presented better outcomes than other serotypes. Similarly to the present findings, Balmaseda et al. (2006) found DENV-1 to be associated with milder hemorrhagic manifestations, such as positive tourniquet test and petechiae . The present study did not capture the presence of gastrointestinal symptoms in DENV-1, as reported by Thomas et al. .
There are discordant results in other studies. Fox et al. (2011) showed that DENV-1 and DENV-2 cases had similar chances to progress to DHF . However, their study included only hospitalized cases, which may have introduced a bias. Yung et al. (2015) demonstrated that patients with DENV-1 had a higher chance of developing DHF than those with DENV-2 in Singapore. The DENV-2 Cosmopolitan genotype circulating in that period in Singapore  was different from the DENV-2 American/Asian genotype, circulating in Vitória , affecting the disease presentation. There is an association between the introduction of the Asian genotype in the Americas and the emergence of hemorrhagic cases. Up until 2003, all DENV-2 isolated from DHF cases on the American continent belonged to Asian genotype .
Cases with DENV-4 infection did not present increased or lower association with SD in the sample evaluated. Previous studies demonstrated that infections caused by DENV-4 presented less severe clinical manifestations [16, 21] and lower viral titers than other serotypes . In Vitória, DENV-4 was detected for the first time in 2012. In 2013, DENV-4 was circulating in a highly susceptible population and caused the largest epidemic ever registered in Vitória, with more than 19,000 cases reported. Despite the impressive incidence, considering laboratory confirmed cases, SD affected only 7.6 % of cases in 2012 and 10.6 % of cases in 2013. Both years presented a lower occurrence of SD than 2009, 2010 and 2011, indicating the limited capacity of DENV-4 to cause SD. Halsey et al. (2012) showed the relation of DENV-4 and hemorrhagic cutaneous manifestations . In the present study, no case with petechiae and positive tourniquet test was found in DENV-4 cases. However, the presence of only five severe cases caused by this serotype impaired the analysis on clinical manifestations.
The difference observed in serotypes association with severity is a concern in face of the recent approval of the first dengue vaccine in Brazil, since DENV-2 presented an efficacy of 42.3 % (varying from 14 % to 61.1 %) the lowest compared to other serotypes. . A phase III trial for the vaccine has been conducted in Vitória since June 2011. However, no case of DENV-2 has been detected in the population since the beginning of the trial. Therefore, so far it is not possible to perceive the effect of the vaccination in this site in a scenario with DENV-2 circulation, regarding protection against dengue infection, emergence of severe cases and demand for hospitalizations. This issue will most likely be clarified in future, with the vaccine implementation and surveillance information.
The present study had some limitations. Potential influencing factors, such as secondary dengue infection, the sequence of serotypes responsible for the secondary infection , or co-morbidity, which could contribute to dengue severity, have not been assessed. As data collection was performed for surveillance purpose, it was dependent on professional precision by the involved health care workers when recording clinical information. Consequently, data on age and clinical manifestations were missing and could have been inaccurately documented. However, standardization of the report forms and the control of documented information by the Epidemiological Surveillance Service minimize information bias caused by misclassification. Determination of serotype was conducted in 485 systematically selected cases, who attended sentinel sites aiming at surveillance of circulating serotypes, which did not relate to patient care. Between 2009 and 2013, Vitória reported 30,027 suspected cases of dengue fever. In 1.6 % of them, serotyping was performed. Furthermore, the collection of blood occurred in the viremic phase, before the emergence of warning signs. As this was a test requested by the surveillance service during this period, it is unlikely that these cases were selected based on their clinical manifestations. Despite the fact that the study includes only 1.6 % of all dengue cases in the catchment area reported in the period, its sample was large enough to have sufficient power to detect the association between serotype and severe dengue. Dengue serotypes present genetic variations and some genotypes have a higher association with SD. Molecular studies of dengue in Vitória are necessary to define the genotypes circulating and their relation with severe epidemics.
Future investigations with prospective approaches in hyperendemic sites or multicenter settings could contribute to elucidate the role of different factors that influence the progression to severe dengue. Some of the factors that should be included are related to the virus, such as serotypes and genotypes, and others are related to the human hosts, such as demographic characteristics, co-morbidities and immunological status. Thereby, an integrated analysis of these factors could contribute to understand the complexity influencing severe dengue outcome.