A prospective case-control study to investigate retinal microvascular changes in acute dengue infection

Dengue infection can affect the microcirculation by direct viral infection or activation of inflammation. We aimed to determine whether measured retinal vascular parameters were associated with acute dengue infection. Patients with acute dengue were recruited from Communicable Diseases Center, Singapore and age-gender-ethnicity matched healthy controls were selected from a population-based study. Retinal photographs were taken on recruitment and convalescence. A spectrum of quantitative retinal microvascular parameters (retinal vascular caliber, fractal dimension, tortuosity and branching angle) was measured using a semi-automated computer-based program. (Singapore I Vessel Assessment, version 3.0). We included 62 dengue patients and 127 controls. Dengue cases were more likely to have wider retinal arteriolar and venular calibers (158.3 μm vs 144.3 μm, p < 0.001; 227.7 μm vs 212.8 μm, p < 0.001; respectively), higher arteriolar and venular fractal dimensions (1.271 vs 1.249, p = 0.002; 1.268 vs. 1.230, p < 0.001, respectively), higher arteriolar and venular tortuosity (0.730 vs 0.546 [x104], p < 0.001; 0.849 vs 0.658 [x104], p < 0.001; respectively), compared to controls. Resolution of acute dengue coincided with decrease in retinal vascular calibers and venular fractal dimension. Dengue patients have altered microvascular network in the retina; these changes may reflect pathophysiological processes in the immune system.

Scientific RepoRts | 5:17183 | DOI: 10.1038/srep17183 retinal venular widening, higher retinal vascular fractal dimension and more tortuous retinal vessels) reflecting early changes of the structure and function of systemic small vessels are markers of systemic diseases and its risk factors [5][6][7][8][9][10] . Retinal microvascular changes in infectious diseases have been relatively understudied. However previous studies on cardiovascular risk factors and inflammation found evidence that retinal venular caliber may be influenced by systemic inflammation. Larger retinal venular calibers were associated with systemic markers of inflammation (high-sensitivity C-reactive protein (CRP) and IL6) and endothelial dysfunction (plasminogen activator inhibitor and soluble intercellular adhesion molecule) 11,12 In this study, we aimed to first determine whether quantitatively measured retinal microvascular parameters were associated with dengue infection, compared with age-gender-race-matched healthy controls. We then further compared retinal microvascular parameters in patients with dengue between the acute and convalescent stages, and correlated retinal microvascular parameters with hematological and biochemical markers. We hypothesize that the systemic inflammation and endothelial dysfunction caused by dengue infection can trigger changes in the systemic microcirculation, which in turn are reflected as microvascular alterations in the retina.

Methods
Study population. This is a prospective case-control study conducted at the Communicable Diseases Center, Singapore between September 2011 and June 2012. Patients were recruited upon confirmation of their dengue status. Dengue was confirmed by real-time reverse transcriptase polymerase chain reaction or non-structural 1 (NS1) antigen. Probable dengue cases required fulfilling World Health Organisation (WHO) diagnostic criteria in the presence of positive rapid dengue IgM or IgG serology 13 . 127 age-gender-ethnicity matched normal healthy controls without fever were selected from the Singapore Prospective Study Program and Singapore Cardiovascular Cohort Study 2. The methodology of this population-based study is described in detail elsewhere 12 . Informed consent was obtained from all patients after explanation of the nature and possible consequences of the study. All study protocols were performed in accordance with the Declaration of Helsinki revised in 1989 and approved by the Institutional Review Board of the National Healthcare Group, Singapore (DSRB/A/2011/01796).

Data collection.
Patients' demographic data, concomitant medical co-morbidities and laboratory results (hemoglobin, platelet count, total white cell count, liver function test, creatinine) were collected.
Definitions of dengue severity. The definitions for dengue severity used in this study were based on the WHO 2009 dengue case definitions 14 . This is divided into dengue without warning signs, dengue with warning signs (abdominal pain or tenderness, persistent vomiting, clinical fluid accumulation, mucosal bleeding, lethargy, liver enlargement and hematocrit increase with rapid platelet drop) and severe dengue (severe plasma leakage, severe bleeding or severe organ impairment).
Retinal photography. Study cases underwent 2 visits: at presentation and at convalescence (4 weeks after recruitment) with retinal photographs taken at both visits. The method for obtaining digital fundal photographs was described in previous studies 5 . Digital fundus photography was performed using a 45° retinal camera (Canon CR-DGi; Canon, Tokyo, Japan) with a digital camera back (10D SLR; Canon). Only the optic disc centered image was used for assessment.
Quantitative measurements of retinal microvasculature. We used a semi-automated computer-assisted program (Singapore I Vessel Assessment [SIVA], version 3.0; National University of Singapore) to quantitatively measure a range of retinal vascular parameters (vascular caliber, vascular tortuosity, fractal dimension and branching angles) from digital photographs according to a standardized protocol.
The measured area of retinal vascular branching, retinal vascular fractal dimension, and retinal vascular tortuosity was standardized and defined as the region from 0.5 to 2.0 disc diameters away from the disc margin (Fig. 1). Trained graders followed a standardized protocol and performed visual evaluations of the automated measurements. The graders were masked to the subjects' identity and other measured parameters in the Singapore Advanced Imaging Laboratory Research. Inter-and intra-observer reliability measurements of the trained graders were previously performed 15 .
Retinal vascular caliber. Retinal vascular caliber was measured following a standardized protocol, based on the revised Knudtson-Parr-Hubbard formula, as described in other publications 16 . A pair of indices, the central retinal arteriolar and venular equivalents (CRAE and CRVE), summarizing the average arteriolar and venular calibers for each eye, was then calculated.
Retinal vascular fractal dimension. Fractal dimension was calculated from the skeletonized line tracing using a boxcounting method; a "global" measure summarizing the whole branching pattern of the retinal vascular tree 17,18 . Larger values indicated a more complex branching pattern. Retinal vascular tortuosity. Retinal vascular tortuosity is defined as the integral of the curvature square along the path of the vessel, normalized by the total path length 19,20 . A smaller tortuosity value indicates a straighter vessel. The estimates were summarized as retinal arteriolar tortuosity and retinal venular tortuosity, representing the average tortuosity of arterioles and venules of the eye, respectively.

Retinal vascular branching.
Retinal branching angle is defined as the first angle subtended between two daughter vessels at each vascular bifurcation 17 .

Statistical analysis.
Retinal vascular parameters were compared between patients with dengue and normal controls. Changes in retinal microvasculature were compared in patients with dengue between acute and convalescent visits. These differences were tested using the independent-samples t test and paired t tests for continuous variables, and the Pearon's chi-squared test for categorical variables.
Ordinary least square (OLS) regression is sensitive towards outliers and influential observations. To overcome this problem, robust regression was performed to determine relationships between retinal vascular parameters and hematological or biochemical markers. Robust regression 21 first fits the OLS regression and excludes influential observations. Huber and Biweights are used to assign weights to absolute residuals. The general idea is to downweigh large residuals. The standard error estimate uses the pseudovalues method. Hence, this method is robust toward outliers and non-normality yet does not compromise efficiency 21,22 , Logistic regression was used to ascertain associations of retinal vascular parameters with severe disease and warning signs. Analyses were performed with Stata 13.0 (Stata Corporation, Texas, U.S.A.) at a 5% significance level.

Patient demographic information.
A total of 62 dengue patients were recruited in the study.
Patient demographic data and laboratory results at baseline and convalescence are shown in Table 1. The mean age was 33 years. Majority were males (87%) and of Chinese ethnicity (74.0%). Thirty-six patients (58%) had warning signs and 7 patients (11%) developed severe disease. There were significant changes in laboratory values in the patients from recruitment to convalescence. There were significant increases in platelet count (115 × 10 9 /L to 217 × 10 9 /L), total white cell count (3.43 × 10 9 /L to 5.90 × 10 9 /L), neutrophils (49.7% to 55.6%) and albumin level (40.1 g/L to 41.7 g/L). There were significant decreases in hemoglobin level (14.9 g/dL to 13.9 g/dL) and creatinine level (73.2 to 68.8). Liver transaminases also significantly normalized with aspartate aminotransferases decreasing from 77.9 U/L to 27.0 U/L and alanine aminotransferases decreasing from 74.4 U/L to 31.4 U/L. Significant decreases were found in hemoglobin level (14.9 g/dL to 13.9 g/dL) and creatinine level (73.2 μ mol/L to 68.8 μ mol/L).    Table 3 and Fig. 2. Comparing parameters at enrolment to those at convalescence showed significant reduction of CRAE (p = 0.001), CRVE (p < 0.001) and venular fractal dimension (p = 0.007). The changes in arteriolar fractal dimension, vessel tortuosity and vascular branching angles were not significant.  Table 3. Retinal vascular parameters in dengue patients at acute and convalescence visits (n = 62).

Retinal vascular parameters in
* Paired t-test between acute and convalescent visit. Values are presented as mean (standard deviation). The pathogenesis of systemic dengue infections is currently believed to be multifactorial, complex and not yet fully understood. Factors postulated in its pathogenesis include direct viral invasion or both humoral and cell-mediated immune responses. The severity of the infection is influenced by viral genotypes and host determinants 23 . An immune pathogenesis is likely involved given the natural history of dengue eye disease 24 and resolution after steroid treatment in selected cases 25 . We do not believe that the retinal vasculature changes are secondary to a febrile illness, as we have previously published retinal vascular changes in afebrile patients with human immunodeficiency virus 26 .
In vitro data and autopsy studies suggest that three organ systems play an important role in the pathogenesis of dengue hemorrhagic fever or dengue shock syndrome: the immune system, the liver and endothelial cell linings of blood vessels 3 . Postulations of dengue infected endothelial cells contributing to pathogenesis include: increasing the viral load, secreting inflammatory mediators of cytokines, modulating complement pathways, or transforming the endothelium into an immunologic target of cellular and humoral immune responses 3,27 . The resulting immune-mediated reactions give rise to endothelial dysfunction and capillary permeability.
It has been proposed that the circulation of high levels of secreted NS1 in the presence of pre-existing heterologous non-neutralizing antibody may mediate complement activation and trigger plasma leakage 28 . In addition, it is believed that secreted NS1 from infected cells can activate the complement factors present in the fluid phase directly giving rise to further inflammation 29 . Our study findings of changes in retinal vascular calibers associated with systemic inflammation echoes previous studies on rheumatoid patients. Increased retinal venular calibers have been described in patients with autoimmune rheumatic diseases 30 or rheumatoid arthritis 31 compared to other hospital controls. Okada et al. also found that patients with autoimmune rheumatic diseases and elevated CRP levels had wider retinal venules than those with low or normal CRP 30 .
We postulate that the above mechanisms may account for suboptimal microcirculatory conditions during the acute phase of the dengue infection. The microcirculation thus responds with changes in microvasculature with dilated vessels and increased complexity of the microvasculature. However with recovery from the dengue infection and resolution of systemic inflammation, our results further showed that the microvasculature changes were reversible. Future in vitro studies will be required to correlate the pathophysiological changes with retinal microvasculature abnormalities.
Our study found associations between AST levels and CRVE. The liver is commonly involved in dengue infection in humans and mouse models 32 . Aminotransferase levels were found to increase in conjunction with dengue severity in a local study 33 . Significant elevation of liver aminotransferases were also found in cases of dengue warning signs and severe dengue as compared to cases without warning signs in a study on children from Bangkok 34 . In our study, AST was negatively associated with CRVE with increasing AST levels associated with narrower venular calibers. This is a useful finding and suggests that CRVE values may be associated with disease severity. There is promise that measurements of venular calibers via retinal photography may be a useful non-invasive tool in disease severity monitoring.
Hemoglobin, hematocrit and creatinine levels were associated with venular fractal dimension. We postulate that dengue patients with capillary leakage may suffer from hypovolemia. The resulting hypoxic   Table 5. Association between retinal vascular parameters with dengue outcomes.
and hypovolemic environment may give rise to a suboptimal microcirculation manifested as changes in fractal dimension.
There are a few limitations to our study. The study population is small but we have attempted to control this by comparing with age-gender-ethnicity matched normal controls. Due to the relatively young age of our dengue cohort, only one patient had cardiovascular co-morbidities of hypertension and hyperlipideamia. Although cardiovascular co-morbidities can affect retinal vasculature, we did not control for this. However in view of the small number, it is unlikely to change the results of the analysis. The small sample size might restrict our ability to demonstrate a significant association retinal vascular changes and severe forms of dengue. Despite the standardized protocols used, the retinal vasculature grading includes measurement errors related to subjective grader input and variability in image quality.
In summary, this is the first paper published on retinal microvasculature changes in patients with acute dengue infection. These changes are reversible with resolution of dengue infection and are promising non-invasive avenues of monitoring disease severity. It will be interesting to know if retinal vasculature changes predict severe forms of dengue and whether the changes persist in patients with severe disease. Larger prospective studies should be conducted for these purposes.