Combination of inflammatory and vascular markers in the febrile phase of dengue is associated with more severe outcomes

Article activity feed

1. 

   Version published to 10.7554/elife.67460 on eLife

   Jun 22, 2021
2. 

   eLife

   May 12, 2021

   Author Response:

   Reviewer #1:

   Weaknesses: The main aim of the study is to identify biomarkers that predict S/MD dengue early in the course of dengue. This requires biomarkers of which the levels change early after symptom onset. However, levels of several of the biomarkers did not change markedly between the two time points (early vs late), suggesting that the levels of these biomarkers had not yet changed on day 1-3, thereby questioning their use as 'early biomarkers'.

   Thank you, we acknowledge that the levels of some of the biomarkers are not markedly different between early and late time points. However this does not affect the aims of the study; firstly the late time-point may not represent the patient’s baseline as this time-point was within 2-3 weeks of the acute illness and secondly, our focus was on the first 3 days of …

   Author Response:

   Reviewer #1:

   Weaknesses: The main aim of the study is to identify biomarkers that predict S/MD dengue early in the course of dengue. This requires biomarkers of which the levels change early after symptom onset. However, levels of several of the biomarkers did not change markedly between the two time points (early vs late), suggesting that the levels of these biomarkers had not yet changed on day 1-3, thereby questioning their use as 'early biomarkers'.

   Thank you, we acknowledge that the levels of some of the biomarkers are not markedly different between early and late time points. However this does not affect the aims of the study; firstly the late time-point may not represent the patient’s baseline as this time-point was within 2-3 weeks of the acute illness and secondly, our focus was on the first 3 days of illness, in order to identify early predictors, noting that this may not represent the peak for many of the biomarkers, which would be in the critical phase. However, we still were able to achieve our main aim which was to compare biomarkers on days 1-3 between patients who progressed to more severe outcomes and those who did not.

   The authors selected the biomarkers based on earlier pathophysiology studies. An alternative approach might have been to first measure a larger set of candidate biomarkers in a selection of patients and select only those biomarkers showing a clear change in the early phase.

   Thank you for your suggestion. For this study, due to the limited number of outcomes (moderate-severe events - 281 cases) and limited volume blood samples, we selected 10 biomarkers as the events-per-variable should be greater than 10 and we also would like to investigate the non- linear effect and interaction of the biomarkers [Heinze et al., Biom J 2018]. We therefore selected the most promising biomarkers systematically based on pilot data and published literature.

   Reference: Heinze G, Wallisch C, Dunkler D. Variable selection - A review and recommendations for the practicing statistician. Biom J 2018; 60(3): 431-49.

   The predictive values of many of the biomarkers was only modest or absent. In addition, some of the findings appear a bit counterintuitive. Examples include the trend of the association of IP-10 with S/MD dengue that changed from positive to negative in the global model, and the opposite trends of some of the biomarkers (e.g. IL-8, ferritin) in adults and children. The authors acknowledge the existence of differences in dengue pathology between children and adults, but could discuss the possible biological reasons in more detail. For example, why would specifically IL-8 or ferritin have an oppositie effect in children and adults.

   The trend of the association of IP-10 with S/MD changed from the single to global model does not diminish the possibility of that biomarker being selected in the best combinations. In this study we do not try to elucidate causal pathways. Another biomarker in our model may be a mediator or confounder of IP-10 in the pathway to the outcome. This could be IL-1RA, as its association with S/MD was similar between the single and global model, and the correlation between IP-10 and IL-1RA was strong (Spearman’s rank correlation coefficient was 0.75). A change in direction after correction for another variable is often referred to as Simpson’s paradox. We have added this point to the discussion of the revised manuscript (page 14, lines 10-16).

   The opposing effect in children and adults is likely to be due to the composite endpoint of severe and moderate dengue. As shown in the analysis of severe dengue alone (figure S5, table S6), the effects of IL-8 and ferritin were similar in children and adults, which suggests these biomarkers are still associated with severe disease in all age groups and that the difference is driven by the moderate dengue group. In addition, uncomplicated dengue in adults have higher ferritin levels compared to in children, with increasing age and chronic conditions in adults likely contributing to this. We have added this point to the discussion in the revised manuscript (page 14, lines 21-26 and page 15, lines 1-2).

   The study does not include a validation cohort. The authors conclude that their findings 'assist the development of biomarker panels for clinical use.' Can the authors put into perspective the performance of their current combined biomarker panel to rule out S/MD dengue.

   Thank you for your comment, this is a case-control and preliminary study to investigate the potential combination of biomarkers associated with dengue clinical outcomes. We quantify importance by means of AIC and p-value. Another dataset without selection by outcome is needed to validate the findings in relation to predictive value. We have added to the limitations that this was not a prediction study, therefore, the performance of the combined biomarker panel with respect to predictive value was not performed (page 16, lines 14-17).

   Overall, the authors show convincingly in a unique cohort that biomarkers can be helpful to triage dengue patients already in the first days from symptom onset. Identification of the best biomarkers for this goal, validation in other cohorts, and a better understanding of differences between children and adults are required before such panels can be introduced in daily clinical practice.

   Thank you for your comment.

   Reviewer #2:

   The main weakness is the exclusion of virological markers, such as plasma/serum viral RNA levels or NS1 antigenaemia. Indeed, previous observations have found severe dengue patients to have higher viraemia in the acute phase of illness compared to those with uncomplicated dengue. More recently, several mechanistic studies have suggested that dengue virus NS1 protein could bind endothelial cells to disrupt its integrity, leading to vascular leakage. Indeed, the authors have pointed out these findings in lines 20-25 on page to lines 1-2 on page 6. Despite these reports, it is curious that the authors have not included either viraemia or NS1 antigenaemia as possible biomarkers for severe dengue.

   Thank you, we acknowledge that plasma viremia and NS1 antigenaemia levels are important factors in dengue disease outcomes. In this study, only enrolment viremia levels were available, but NS1 antigenaemia levels were not. We have previously investigated the association between viremia levels and clinical outcomes using a pooled dataset of the IDAMS international study and other three studies in Vietnam. We found that higher plasma viremia was associated with increased dengue severity [Vuong et al., Clin Infect Dis 2020]. For this study, the main aim was to investigate host biomarkers which could be combined in a multiplex test panel.

   However, as suggested, we have added the information of viremia levels to table S3 (which was previously table 2) of the revised manuscript. Also, we have performed a sensitivity analysis to include viremia levels as a potential biomarker and we have found that: (1) higher plasma viremia was associated with increased the risk of severe/moderate dengue in both single and global models, and (2) viremia was not selected in children but was selected fourth in adults when performing the best subset procedure. We have added this information in the Statistical analysis (page 10, lines 20- 24) and Results sections (page 13, lines 17-20), and the Supplementary file (appendix 8, figure S8, tables S13-S15, pages 30-34).

   Reference: Vuong NL, Quyen NTH, Tien NTH, et al. Higher plasma viremia in the febrile phase is associated with adverse dengue outcomes irrespective of infecting serotype or host immune status: an analysis of 5642 Vietnamese cases. Clin Infect Dis 2020.

   The manuscript in its present form may favour those with a strong statistical background to fully appreciate the nuances. Clearer explanations on the statistical findings would, I think, be helpful to those without such statistical background but who would nonetheless be in positions to translate these findings into clinical practice.

   We have added more explanation in the Statistical analysis, Results and Discussion sections to clarify statistical methods used in this study and the interpretation of the results.

   Most of the cases included in this study had DENV-1 infection. The biomarkers identified in this study may thus be DENV-1 specific and may not be readily applied to triage dengue cases caused by other DENV infection.

   In our study, DENV-1 accounted for 42% of all cases. We have performed a sensitivity analysis taking into account differences between serotypes. The results showed that there was no significant difference between serotypes with respect to the association between the biomarkers and primary endpoint in both the single and global models. This suggests that the study’s results are applicable for all serotypes. This information has been added in the Statistical analysis (page 10, lines 18-20) and Results sections (page 12, lines 18-20), and the Supplementary file (appendix 5, figures S3-S4, tables S4-S5, pages 13-17).

   Reviewer #3:

   1. For general ease of readership, it would greatly help if the authors can explain the choice of the statistical method used in the data analysis and perhaps briefly explain the model and how AIC should be interpreted in the main rather than the supplementary text).

   We have clarified in more details in the Statistical analysis section of the revised manuscript.

   1. While this reviewer understands that the authors want to focus on host immune and inflammatory biomarkers but it would be helpful if NS1 and viremia data are also shown ( at least in supplementary data) if these have been found not to correlate with disease severity.

   Thank you please see response to comment #1 of reviewer #2. Quantitative NS1 results were not available in this study. We have added viremia in a sensitivity analysis and the results showed that higher viremia was associated with increased risk of severe/moderate dengue, similar to our previous study [Vuong et al., Clin Infect Dis 2020]. In the best subset procedure, viremia was not selected in children and was selected fourth in adults.

   Reference: Vuong NL, Quyen NTH, Tien NTH, et al. Higher plasma viremia in the febrile phase is associated with adverse dengue outcomes irrespective of infecting serotype or host immune status: an analysis of 5642 Vietnamese cases. Clin Infect Dis 2020.

   1. It is Interesting to note that some biomarkers ( particularly the vascular markers) in severe group do not return to the same baseline as mild cases at convalescence even after >20 days. Whether such individuals already are at higher inflammatory state at baseline (pre-infection) as a result of underlying co-morbidities such as obesity or diabetes? Table 1 did not provide such information but would be interesting to show if there is any difference in health state in the 2 groups especially for obesity.

   We have added the information of obesity and diabetes in table 1, Results section (page 11, lines 13-14). There were 5 patients with diabetes; obesity was balanced between groups (14% in control group and 10% in S/MD group).

   1. It is rather confusing that the 2nd paragraph of discussion stated "Balancing model fit, robustness, and parsimony, we suggest the combination of five biomarkers IL-1RA, Ang-2, IL-8, ferritin, and IP-10 for children, and the combination of three biomarkers SDC-1, IL-8, and ferritin for adults to be used in practice."

   But the concluding paragraph went on to state "The best biomarker combination for children includes IL-1RA, Ang-2, IL-8, ferritin, IP-10, and SDC-1; for adults, SDC-1, IL-8, ferritin, sTREM-1, IL-1RA, IP-10, and sCD163 were selected." This should be clarified further.

   Thank you for pointing this out. The conclusion was based on the best combinations (taking into account AIC only), which consisted of 6 biomarkers for children and 7 biomarkers for adults. In the discussion, we reduced the number of biomarkers, taking into consideration not only the AIC, but also parsimony for clinical translation purposes, while keeping the model fit as good as possible (taking a difference of AIC of less than 5 compared to the best combination). We therefore suggested a combination of 5 biomarkers for children and 3 biomarkers for adults, considering these 3 factors - model fit, robustness and parsimony. We have clarified this point in the Discussion section of the revised manuscript (page 15, lines 20-25).

   Read the original source
3. 

   eLife

   May 12, 2021

   Reviewer #3 (Public Review):

   Nguyen Lam Vuong et al performed a nested case control study of a multisite, multicountry prospective dengue study (IDAMS) to identify early biomarkers at day 1-3 of illness onset that predicts for severe dengue of ten biomarkers. Ten biomarkers from the inflammatory, immune or vascular pathways (VCAM-1,SDC-1, Ang-2, IL-8, IP-10, IL-1RA, sCD163, sTREM-1, ferritin, CRP) were chosen based on prior literature and understanding of dengue pathogenesis in severe disease. The biomarkers were measured at two time points: at enrollment (illness day 1-3) and after recovery(day 10-31 ). They find moderate-to strong positive correlations for some markers, particular IP-10 and IL-1RA, and IP-10 and VCAM-1, ( Spearman's rank correlation coefficients above 0.6). Interestingly, in their single modal analysis, they also find …

   Reviewer #3 (Public Review):

   Nguyen Lam Vuong et al performed a nested case control study of a multisite, multicountry prospective dengue study (IDAMS) to identify early biomarkers at day 1-3 of illness onset that predicts for severe dengue of ten biomarkers. Ten biomarkers from the inflammatory, immune or vascular pathways (VCAM-1,SDC-1, Ang-2, IL-8, IP-10, IL-1RA, sCD163, sTREM-1, ferritin, CRP) were chosen based on prior literature and understanding of dengue pathogenesis in severe disease. The biomarkers were measured at two time points: at enrollment (illness day 1-3) and after recovery(day 10-31 ). They find moderate-to strong positive correlations for some markers, particular IP-10 and IL-1RA, and IP-10 and VCAM-1, ( Spearman's rank correlation coefficients above 0.6). Interestingly, in their single modal analysis, they also find differences in biomarkers levels in children compared to adults, Associations between SDC-1 and IL-8 and the S/MD endpoint were stronger in adults than children, while the effects of IL-1RA and ferritin were stronger in children than adults. When global analysis was performed, only SDC-1 and IL-1RA were the most stable relative to the single models for both children and adults. And the the differences of the associations between children and adults were more marked, particularly for Ang-2, IL-8 and ferritin. When the biomarkers were combined, for children, the best subset that showed the clearest association with S/MD was the combination of the six markers IL-1RA, Ang-2, IL-8, ferritin, IP-10, and SDC-1 with an AIC of 465.9. For adults, the best subset included the seven markers SDC-1, IL-8, ferritin, sTREM-1, IL-1RA, IP-10, and sCD163 This manuscript certainly provides useful insight into the biomarkers that are involved in the early phase of dengue before onset of vascular leakage or severe dengue which is valuable as most previous publications mainly focused on measurement of these markers after onset of severe disease which was often too late for meaningful interpretation of the disease biology or of limited clinical utility. The conclusions of this paper are mostly well supported by data, but some aspects of study and data analysis need to be clarified in order to improve understanding of the statistical methodology and readability.

   Major Strengths:

   • More than 7000 participants ( children and adults) in eight countries across Asia and Latin America were enrolled in the IDAMS study
   • Prospective and systematic blood sampling starting from day 1 of illness onset
   • Cases were laboratory confirmed via PCR or NS1 testing
   • Cases and control were fairly well- matched
   • Strong rationale for selection of host biomarkers

   Weaknesses:

   • Three quarter of cases from one country
   • Serotype-1 biased

   Specific comments to address:

   1. For general ease of readership, it would greatly help if the authors can explain the choice of the statistical method used in the data analysis and perhaps briefly explain the model and how AIC should be interpreted in the main rather than the supplementary text).

   2. While this reviewer understands that the authors want to focus on host immune and inflammatory biomarkers but it would be helpful if NS1 and viremia data are also shown ( at least in supplementary data) if these have been found not to correlate with disease severity.

   3. It is Interesting to note that some biomarkers ( particularly the vascular markers) in severe group do not return to the same baseline as mild cases at convalescence even after >20 days. Whether such individuals already are at higher inflammatory state at baseline (pre-infection) as a result of underlying co-morbidities such as obesity or diabetes? Table 1 did not provide such information but would be interesting to show if there is any difference in health state in the 2 groups especially for obesity.

   4. It is rather confusing that the 2nd paragraph of discussion stated "Balancing model fit, robustness, and parsimony, we suggest the combination of five biomarkers IL-1RA, Ang-2, IL-8, ferritin, and IP-10 for children, and the combination of three biomarkers SDC-1, IL-8, and ferritin for adults to be used in practice."

   But the concluding paragraph went on to state "The best biomarker combination for children includes IL-1RA, Ang-2, IL-8, ferritin, IP-10, and SDC-1; for adults, SDC-1, IL-8, ferritin, sTREM-1, IL-1RA, IP-10, and sCD163 were selected." This should be clarified further.

   Read the original source
4. 

   eLife

   May 12, 2021

   Reviewer #2 (Public Review):

   Summary:

   In this study, Vuong and colleagues conducted a case control study nested within a larger longitudinal and multi-country observational study to evaluate 10 biomarkers. These biomarkers were selected based on the strength of evidence in the literature and the current understanding of dengue pathogenesis. Using a 1:2 ratio of severe/moderately severe dengue cases to uncomplicated dengue controls, the authors examined the trends of the expression of these biomarkers during acute illness (days 1-3 from illness onset) as well as early (10-20 days from illness onset) and late (>20 days from illness onset) convalescence. The identified several biomarkers that were expressed at higher levels during acute illness in cases than controls and showed that these could be used in combination to predict those at …

   Reviewer #2 (Public Review):

   Summary:

   In this study, Vuong and colleagues conducted a case control study nested within a larger longitudinal and multi-country observational study to evaluate 10 biomarkers. These biomarkers were selected based on the strength of evidence in the literature and the current understanding of dengue pathogenesis. Using a 1:2 ratio of severe/moderately severe dengue cases to uncomplicated dengue controls, the authors examined the trends of the expression of these biomarkers during acute illness (days 1-3 from illness onset) as well as early (10-20 days from illness onset) and late (>20 days from illness onset) convalescence. The identified several biomarkers that were expressed at higher levels during acute illness in cases than controls and showed that these could be used in combination to predict those at increased risk of severe/moderately severe dengue. Notably, the authors identified different sets of biomarkers for paediatric and adult dengue cases, suggesting that the underlying pathophysiology of severe disease may differ in these groups of dengue cases. The authors concluded that the biomarkers they identified would be a major public health benefit to allocate healthcare resources during dengue outbreaks, and suggested that these biomarkers could also be applied as biological endpoints in dengue clinical trials.

   Strengths:

   This is a fairly sizeable study involving 281 severe/moderately severe dengue cases and 556 uncomplicated dengue controls. The authors combined data clinical observation data with those derived from serum protein measurements and analysed them using sophisticated statistical approaches.

   The search for biomarkers predictive of the risk of severe dengue has spanned decades. This study distinguishes itself from others in its study design and the systematic selection of biomarkers for evaluation. Avoidance of untargeted screening reduces the likelihood of chance discovery and makes the findings more statistically robust.

   The findings have useful practical applications. Severe dengue manifests typically around the period of fever defervescence at around days 4-7 from fever onset. Application of biomarkers in the acute febrile phase of illness could thus provide a 2- to 3-day lead time to triage those at risk of severe dengue for closer monitoring and management.

   Weaknesses:

   The main weakness is the exclusion of virological markers, such as plasma/serum viral RNA levels or NS1 antigenaemia. Indeed, previous observations have found severe dengue patients to have higher viraemia in the acute phase of illness compared to those with uncomplicated dengue. More recently, several mechanistic studies have suggested that dengue virus NS1 protein could bind endothelial cells to disrupt its integrity, leading to vascular leakage. Indeed, the authors have pointed out these findings in lines 20-25 on page to lines 1-2 on page 6. Despite these reports, it is curious that the authors have not included either viraemia or NS1 antigenaemia as possible biomarkers for severe dengue.

   The manuscript in its present form may favour those with a strong statistical background to fully appreciate the nuances. Clearer explanations on the statistical findings would, I think, be helpful to those without such statistical background but who would nonetheless be in positions to translate these findings into clinical practice.

   Most of the cases included in this study had DENV-1 infection. The biomarkers identified in this study may thus be DENV-1 specific and may not be readily applied to triage dengue cases caused by other DENV infection.

   Overall impression:

   This study provides two interesting findings. Firstly, that there are biomarkers that can be further developed into clinical tests to triage dengue patients for management. Although this possibility will require further assay development - the Luminex platform used for multiplex measurements of these biomarkers is unlikely to be available in most clinical laboratories - this study does show proof-of-concept to justify the development of simpler and perhaps even point-of-care assays. Secondly, the finding that adult and paediatric dengue require different biomarkers to indicate risk of severe disease should also trigger more detailed clinical and basic science investigation into how age influence host response to infection.

   Read the original source
5. 

   eLife

   May 12, 2021

   Reviewer #1 (Public Review):

   Dengue is the most common arboviral infections in humans. Better tools to effectively triage patients at risk for severe dengue are urgently needed to optimize use of healthcare resources. This well written manuscript by Nguyen Lam Vuong and colleagues assessed the associations of a panel of blood biomarkers on day 1-3 from symptom onset with the development of severe or moderate (S/MD) dengue in a large cohort of children and adults. Ten candidate biomarkers were selected, each representing important pathogenic processes in dengue. Overall, higher concentrations of the biomarkers increased the risk for S/MD dengue. Important differences between adults and children were found for the performance of several biomarkers. The performance of the individual biomarkers, as well as the best combination was assessed …

   Reviewer #1 (Public Review):

   Dengue is the most common arboviral infections in humans. Better tools to effectively triage patients at risk for severe dengue are urgently needed to optimize use of healthcare resources. This well written manuscript by Nguyen Lam Vuong and colleagues assessed the associations of a panel of blood biomarkers on day 1-3 from symptom onset with the development of severe or moderate (S/MD) dengue in a large cohort of children and adults. Ten candidate biomarkers were selected, each representing important pathogenic processes in dengue. Overall, higher concentrations of the biomarkers increased the risk for S/MD dengue. Important differences between adults and children were found for the performance of several biomarkers. The performance of the individual biomarkers, as well as the best combination was assessed for children and adults.

   Strengths: Particular strengths of this study are the uniqueness of the prospective cohort with a large number of participants from different countries and the availability of blood samples early in the course of infection.

   Other strengths include the enrolment of both children and adults, which is important given the observed differences in dengue pathology, the consistent data collection and the use of standardized outcome definitions.

   The authors selected the candidate biomarkers based on earlier pathogenesis studies, reflecting different pathogenic pathways in dengue (e.g. activation mononuclear cells, vascular pathology).

   State-of-the-art statistical modelling was used to assess the performance of the biomarkers.

   Weaknesses: The main aim of the study is to identify biomarkers that predict S/MD dengue early in the course of dengue. This requires biomarkers of which the levels change early after symptom onset. However, levels of several of the biomarkers did not change markedly between the two time points (early vs late), suggesting that the levels of these biomarkers had not yet changed on day 1-3, thereby questioning their use as 'early biomarkers'. The authors selected the biomarkers based on earlier pathophysiology studies. An alternative approach might have been to first measure a larger set of candidate biomarkers in a selection of patients and select only those biomarkers showing a clear change in the early phase.

   The predictive values of many of the biomarkers was only modest or absent. In addition, some of the findings appear a bit counterintuitive. Examples include the trend of the association of IP-10 with S/MD dengue that changed from positive to negative in the global model, and the opposite trends of some of the biomarkers (e.g. IL-8, ferritin) in adults and children. The authors acknowledge the existence of differences in dengue pathology between children and adults, but could discuss the possible biological reasons in more detail. For example, why would specifically IL-8 or ferritin have an oppositie effect in children and adults.

   The study does not include a validation cohort. The authors conclude that their findings 'assist the development of biomarker panels for clinical use.' Can the authors put into perspective the performance of their current combined biomarker panel to rule out S/MD dengue.

   Overall, the authors show convincingly in a unique cohort that biomarkers can be helpful to triage dengue patients already in the first days from symptom onset. Identification of the best biomarkers for this goal, validation in other cohorts, and a better understanding of differences between children and adults are required before such panels can be introduced in daily clinical practice.

   Read the original source
6. 

   Version published to 10.1101/2021.03.13.21253501v1 on medRxiv

   Mar 15, 2021