Serum Ferritin As a Prognostic Indicator in Adult Dengue Patients

ABSTRACT

Dengue virus (DENV) infection is increasing with rapid urbanization in India. Treatment of DENV infection is mainly supportive with no specific antiviral therapy. Although most patients show mild illness, some have a severe disease course such as dengue hemorrhagic syndrome, dengue shock, multi-organ failure, and death. The cause for severity is not fully understood. Currently, there are no methods available to predict the course of the illness. Hence, it is crucial to develop an early biomarker to predict the course of dengue illness which can aid in vigorous monitoring and early intervention. Here, we tried to establish a correlation between serum ferritin and severity of dengue illness. We measured ferritin levels in 100 dengue-positive cases on day 1 (D1) (febrile phase) and day 4 (D4) (defervescence or convalescent) of admissions to compare the levels with the severity of the disease. On D1, the serum ferritin level was a “good” predictor of severe dengue, with an area under the curve (AUC) of 0.863 with standard error (SE) = 0.043 and a 95% CI from 0.778 to 0.947 (P < 0.05). On D4, serum ferritin was an “excellent” predictor of severe dengue, with an AUC of 0.947 with SE = 0.021 and a 95% CI from 0.907 to 0.988 (P < 0.05). Serum ferritin is an inexpensive and easily accessible biomarker that can assist in monitoring and prognosticating the dengue-positive patients. This biomarker also directs us to explore the underlying pathogenetic mechanism in severe dengue, which can lay a foundation for future targeted therapeutic options to combat severe illness.

INTRODUCTION

Dengue fever is caused by an RNA virus that belongs to the family Flaviviridae. Its transmission is arthropod-borne through the bite of an infected Aedes aegypti mosquito.¹ Currently, 2.5 billion people reside in at-risk dengue-endemic areas.² The clinical course of dengue fever varies from a mild undifferentiated flu-like illness to a severe spectrum, including dengue hemorrhagic fever (DHF) and dengue shock syndrome (DSS). Dengue fever is diagnosed by NS1 antigen detection by ELISA or by IgM antibody detection by MAC-ELISA.¹ Other rapid diagnostic tests (RDTs) such as biosensor based RDT (viro track, BluSense Diagnostics, København Ø, Denmark) and SD Bioline (Chennai, India) Dengue duo NS1/IgM/IgG, which is a commercially available RDT, have shown comparable results with ELISA and increased sensitivity when combined with serology, making them the potential alternative to ELISA.³ However, there is no specific antiviral treatment for dengue. Currently, the management of dengue includes supportive care and monitoring for warning signs.

In 2009, the WHO classified dengue into dengue fever, dengue fever with warning signs, and severe dengue. Severe dengue includes severe plasma leakage leading to shock (DSS) or respiratory distress and life-threatening bleeding manifestations (DHF), with organ involvement (cardiac involvement, liver enzymes > 1,000, or central nervous system [CNS] impairment).¹ The cause of the severe spectrum of illnesses in a minority of patients is yet to be understood. Risk factors described in DHF are host age, race, nutrition status, host immune response, viral serotype, hyperendemicity, etc. However, the pathogenesis of severe dengue is attributed to a non-neutralizing antibody-dependent enhancement (ADE) in secondary dengue infection and cytokine dysregulation.^4,5 Because there is no specific antiviral therapy, the focus of the treatment is vigorous monitoring and early prediction of a severe course of illness to prevent catastrophic events, such as DHF and DSS.

Macrophage activation syndrome (MAS) exaggerated activation of T cells and macrophages, which results in a cytokine storm that leads to a hyperinflammatory state.⁶ This entity is associated with a few viral infections. Serum ferritin, sCD163 (monocyte marker), and sCD25 (regulatory T-cell marker) are crucial biomarkers of MAS.⁷ Several cases of MAS have been reported in dengue infection.^6,8,9

Thus, we aimed to establish a correlation between serum ferritin and severe dengue considering the aforementioned pathogenesis. An early prediction of the disease course is the key to tackling the disease until specific dengue antiviral therapies are available.

METHODS

Patients who fulfill the inclusion and exclusion criteria who are coming to the Department of General Medicine at Kempegowda Institute of Medical Sciences Hospital, Bangalore, were included. Inclusion criteria: Any patient older than 18 years with a history of fever with a dengue-positive profile (NS1 antigen and IgM antibody positive) by SD BIOLINE RDT. Exclusion criteria: All cases that involved anemia, chronic inflammatory disease, and a recent blood transfusion.

All the dengue-positive cases were examined, and their clinical history was assessed. The dengue patients were classified according to severity as per the WHO 2009 guidelines, and their serum ferritin levels were estimated on day 1 (D1). The patients were monitored daily to assess the progression of the disease. A second sample was drawn on day 4 (D4) to estimate the serum ferritin levels. The ferritin levels on D1 and D4 of admission were compared with the severity of dengue. Day 1 corresponds to 3–7 days of illness, whereas D4 corresponds to 7–11 days of illness. For purpose of our study, we categorized dengue without warning signs or probable dengue (according to the WHO definitions) as grade A, dengue with warning signs as grade B, and severe dengue as grade C (Supplemental Table 1).

Statistical methods.

The sociodemographic data, gender, age distribution, clinical presentation, and intensive care unit (ICU) admission were described by descriptive statistics. A Kruskal-Wallis H test was used to study and compare the severity categories. Analysis of variance was used to compare the means of two or more groups. The results are expressed as median, and a P-value of less than < 0.05 was considered statistically significant. The receiver operating characteristic (ROC) curve was obtained by plotting the different cutoffs values of the serum ferritin levels. The area under the curve (AUC) was obtained by plotting the ROC and helped predict the severity. The larger the AUC, the better was predictive capacity at identifying the cutoff scores to predict the severity.

RESULTS

The current study included 100 subjects, and 70 (70.0%) were males and 30 (30.0%) were females. The majority, that is, 43 (43.0%), of the study subjects were in the 18- to 25-year-old age-group, and 25 (25.0%) were in the 26- to 35-year-old group, 18 (18.0%) were in the 36- to 45-year-old group, seven (7.0%) were in the 46- to 55-year-old group, four (4.0%) were in the 56- to 65-year-old group, and three (3.0%) were in the > 65-year-old group. The mean age was 32.49 + 13.64 years, with a range from 18 to 80 years. The mean age of the male and female patients were 31.53 + 13.99 years and 34.73 + 12.71 years, respectively (Table 1).

Table 1.

Sociodemographic characteristics of study subjects

Age (years)	Males	Females	Total
18–25	32 (37.9)	11 (36.7)	43 (43.0)
26–35	19 (27.1)	06 (20.0)	25 (25.0)
36–45	11 (15.7)	07 (23.3)	18 (18.0)
46–55	04 (5.7)	03 (10.0)	07 (7.0)
56–65	01 (1.4)	03 (10.0)	04 (4.0)
> 65	03 (4.3)	00 (0.0)	03 (3.0)
Total	70 (70.0)	30 (30.0)	100 (100.0)

Many of the study subjects had multiple complaints. The most common complaint was fever (100%), followed by gastrointestinal manifestations, such as vomiting (33.0%), abdominal pain (8.0%), loose stools (5%), abdominal distension (2%), and jaundice (1.0%). Bleeding manifestations were observed in 24% of patients, which included melena (6%), epistaxis (3%), hematemesis (1%), bleeding tongue (1.0%), and per-vaginal bleeding (1.0%). Other symptoms included cough (3%), breathlessness (3.0%), lower limb weakness (1.0%), generalized weakness (1.0%), irrelevant speech (1.0%), headache (6.0%), and arthralgia (1.0%) (Supplemental Table 2). Among the study subjects, 46% were positive for NS1 antigen, 14% were positive for IgM antibodies, and NS1Ag and IgM antibodies were positive in 40% (Supplemental Figure 1).

On D1, the serum ferritin level, platelet count, and hematocrit were assessed. The serum ferritin level increased with the increasing severity of dengue, the platelet count decreased with the increasing severity, and the hematocrit was the same for the grade A and B severity patients and increased slightly within the grade C severity patients (Table 2). On D4, the serum ferritin levels and hematocrit levels increased with the severity of dengue, whereas the platelet counts decreased with the increasing severity (Table 3).

Table 2.

Distribution of medians and range of hematological parameters on day 1 according to the severity of the disease

Hematological parameter	Median (minimum–maximum)
	Severity A	Severity B	Severity C
Platelet count D1	71,000 (13,000–20,3000)	28,500 (14,000–97,000)	30,000 (5,000–108,000)
Hematocrit D1	45.60 (30.4–54.0)	44.90 (34.2–54.4)	48.90 (32.2–59.4)
Serum ferritin 1	900 (90.0–7,374.0)	1,775 (800–3,809)	2,800 (900–4,889.4)

Table 3.

Distribution of medians and range of hematological parameters on day 4 according to the severity of the disease

Hematological parameter	Median (minimum–maximum)
	Severity A	Severity B	Severity C
Platelet count D4	64,000 (17,000–235,000)	45,000 (35,000)	30,000 (10,000–81,000)
Hematocrit D4	43.40 (31.3–51.0)	47.40 (35.8–49.4)	48.00 (36.8–59.0)
Serum ferritin D4	580.7 (120.6–2,000.0)	1,725.00 (900–2,000)	2,000.00 (1,500–3,600)

A majority of the study subjects, that is, 63.0%, 59.0%, 64.0%, and 70.0%, were classified as having grade A severity on days 1, 2, 3, and 4, respectively. In addition, the proportion of the subjects who were classified as having type B severity on days 1, 2, 3, and 4 were 20.0%, 21.0%, 15.0%, and 12.0%, respectively. Finally, the proportion of the subjects who were classified as having severe dengue (grade C) on days 1, 2, 3, and 4 were 17.0%, 20.0%, 21.0%, and 18.0%, respectively (Supplemental Table 3). Among the study subjects, 23.0% had an abnormal electrocardiogram that depicted tachycardia, sinus bradycardia, atrial fibrillation, or ST wave changes. In addition, among the study subjects, 15.0% had an ICU admission. Our study showed 2% mortality (Supplemental Figure 2).

On day 1.

The association of the severity of dengue with respect to the serum ferritin levels, platelet count, and hematocrit varied. For example, the serum ferritin levels increased (H [2] = 28.89, P < 0.001) significantly, and the platelet counts decreased significantly with the increasing severity of dengue (H [2] = 20.02, P = 0.435). Moreover, the hematocrit remained the same with the dengue fever without warning signs and dengue fever with warning signs patients (A and B severities, respectively) and increased slightly with severe dengue (severity C). However, this was not statistically significant (H [2] = 1.66, P < 0.001) (Table 4).

Table 4.

Comparison of medians of different indicators among the different severity A, B, and C on D1

Hematological parameters on D1	Median (IQR)	N	Mean rank	H	Degree of freedom	P-value
Platelet count
Severity A	71,000 (64,000)	63	60.44	20.02	2	< 0.001*
Severity B	28,500 (12,500)	20	33.40
Severity C	30,000 (43,500)	17	33.76
Hematocrit
Severity A	45.60 (6.90)	63	47.63	1.66	2	0.435
Severity B	44.90 (7.45)	20	55.15
Severity C	48.90 (12.15)	17	55.65
Serum ferritin
Severity A	900 (1,328)	63	39.71	28.89	2	< 0.001*
Severity B	1,775 (964)	20	58.92
Severity C	2,800 (1,100)	17	80.59

D1 = day 1.

*Significant statistical difference between the groups with P < 0.05.

On day 4.

The association of the severity of dengue with the serum ferritin levels, platelet count, and hematocrit varied. For example, the serum ferritin levels (H [2] = 53.12, P < 0.001) and hematocrit levels (H [2] = 11.02, P = 0.004) increased significantly, and the platelet counts decreased significantly with the increasing severity of dengue (H [2] = 18.86, P < 0.001) (Table 5).

Table 5.

Comparison of medians of different indicators among the different severity A, B, and C on D4

Hematological parameters on D4	Median (IQR)	N	Mean rank	H	Df	P-value
Platelet count
Severity A	63,000 (45,250)	70	58.39	18.86	2	< 0.001*
Severity B	45,000 (35,000)	12	40.33
Severity C	30,000 (35,500)	18	26.61
Hematocrit
Severity A	43.40 (5.40)	70	44.74	11.02	2	0.004*
Severity B	47.40 (9.17)	12	55.17
Severity C	48.00 (4.90)	18	69.78
Serum ferritin
Severity A	569.0 (488)	70	36.83	53.12	2	< 0.001*
Severity B	1,725.00 (955)	12	75.25
Severity C	2,000.00 (1,150)	18	87.17

D4 = day 4.

*Significant statistical difference between the groups with P < 0.05.

On the day of admission, among the platelet, hematocrit, and serum ferritin levels, the platelet level was a significantly “fair” predictor of warning signs and severe dengue, with an AUC for dengue with warning signs of 0.714 with standard error (SE) = 0.052 and a 95% CI from 0.611 to 0.816 (P < 0.05) and an AUC for severe dengue of 0.702 with SE = 0.072 and a 95% CI from 0.560 to 0.843 (P < 0.05). Moreover, the serum ferritin level was a “good” predictor of severe dengue, with an AUC of 0.863 with SE = 0.043 and a 95% CI from 0.778 to 0.947 (P < 0.05) (Figure 1A).

Figure 1.

Receiver operating characteristic (ROC) curve for serum ferritin levels in predicting the severity of dengue. (A) Serum ferritin levels with severity C on day 1. (B) Serum ferritin levels with severity B on day 4 (D4). (C) Serum ferritin levels with severity C on D4. This figure appears in color at www.ajtmh.org.

On D4 of admission, among the platelet, hematocrit, and serum ferritin levels, the hematocrit level was a “fair” predictor of severe dengue, with an AUC of 0.735 with SE = 0.072 and a 95% CI from 0.594 to 0.876 (P < 0.05). Also, the platelet level was a “fair” predictor of severe dengue, with an AUC of 0.791 with SE = 0.059 and a 95% CI from 0.676 to 0.907 (P < 0.05). Finally, the serum ferritin level was a “fair” predictor of dengue severity B, with an AUC of 0.781 with SE = 0.047 and a 95% CI from 0.690 to 0.873 (P < 0.05) (Figure 1B), and it was an “excellent” predictor of severe dengue, with an AUC of 0.947 with SE = 0.021 and a 95% CI from 0.907 to 0.988 (P < 0.05) (Figure 1C).

Among the study subjects, on D1, the serum ferritin level was a good predictor of severe dengue. Using a cutoff of 1,990 ng/mL, severe dengue was detected with a maximum sensitivity of 94.1% and a specificity of 71.1%. On D4, serum ferritin was a good and excellent predictor of warning signs and severe dengue. Using a cutoff of 895 ng/mL, dengue with warning signs was detected with a maximum sensitivity of 100.0% and a specificity 61.4%, and a cutoff of 1,380 ng/mL detected severe dengue with a maximum sensitivity of 100.0% and a specificity of 85.4%.

Hence, a decrease in serum ferritin on D4 for severe dengue with a cutoff of 1,380 ng/mL and for dengue with warning signs with a cutoff of 895 ng/mL may be used as a prognostic marker to indicate the reduction in severity. However, the various admission criteria for dengue followed by different hospital settings can be a limitation.

DISCUSSION

Recently, dengue fever has shown increasing trends because of rapid urbanization, lifestyle changes, and improper water storage practices in urban and rural areas, leading to the proliferation of mosquito-breeding sites. Until now, there are five known dengue serotypes (Dengue virus [DENV] 1–5) that were isolated from dengue-infected patients. Dengue infection ranges from a mild fever with myalgias to DSS, DHF, and multi-organ failure. The reason for the severe disease symptoms in some patients is yet to be fully understood. The major pathogenic mechanisms described in dengue patients are ADE and erratic cytokine release, leading to hyperinflammation, vascular leakage, and multi-organ damage. Antibody-dependent enhancement occurs during secondary DENV infection with different serotypes, where heterotypic antibodies from primary DENV infection neutralize DENV and facilitating Fc receptor-mediated viral uptake and viral replication within the macrophages.^10–12 This high viral load can be attributed to severe disease spectrum.^13,14 Fc receptor-mediated viral uptake further facilitates viral replication by releasing anti-inflammatory cytokines, such as interleukin (IL)-10.^11,15 However, DENV RNA levels were at undetectable levels in DHF/DSS,^11,16 indicating the presence of alternate mechanisms that lead to the severe illness observed in dengue infection.

Another pathogenic mechanism explored in severe dengue is cytokine dysregulation. Several case reports and studies have established a correlation between a hyperinflammatory and a dysregulated immune response in dengue.^6,8,9 Dengue virus infection of macrophages leads to the erratic activation of macrophage proliferation, which leads to the massive release of pro-inflammatory cytokines, dysregulated T cells, and macrophage proliferation.^17,18 The pro-inflammatory cytokines, such as tumor necrosis factor (TNF), secreted by mononuclear macrophages cause endothelial activation and plasma leakage, which is a hallmark of severe dengue. This pathogenesis is similar to MAS, a hyperinflammatory systemic condition with a fatal outcome without an early diagnosis and intervention.

A Sri Lankan study also emphasizes evidence of mast cell activation and degranulation leading to the release of intracellular inflammatory mediators in DHF. This study observed significantly elevated serum chymase level (sensitivity of 96% with cutoff 1.5 ng/mL) in a subset of patients with DHF compared with other subsets such as dengue fever and dengue fever with bleeding (not fulfilling the DHF criteria).¹⁹ Our study demonstrates serum ferritin could be an excellent prognostic biomarker with higher sensitivity (100%) during the defervescence/recovery phase and a decent biomarker (sensitivity of 94.1%) during the febrile phase to predict severe dengue. A key biomarker used in suspicion of MAS is serum ferritin, which is easily accessible and cost-effective. Hence, our study looked at serum ferritin levels in the spectrum of dengue illness (mild to severe) and their rising trend in severe dengue. Serum ferritin is a readily available and inexpensive test compared with various other biomarkers linked with MAS.

The sociodemographic data from the current study showed that the majority of the patients (70%) were males, and the age-group of 18–25 years was most prominent. The mean age was 32.49 ± 13.64 years. Shah et al.²⁰ found that the most common age-group for dengue manifestation was between 18 and 30 years (71.0%), and majority of them were males (65.6%), which was comparable with the current study’s findings. Similarly, Roy Chaudhuri et al.²¹ noted that the majority of their study subjects were male (56.7%), and the mean age of the study participants was 39.86 + 12.95 years.

Fever was present in all the patients admitted with dengue. Other common presentations included gastrointestinal manifestations, including vomiting, abdominal pain, loose stools, abdominal distension, and jaundice, followed by bleeding manifestations, such as melena, epistaxis, hematemesis, bleeding tongue, and per-vaginal bleeding, cough, breathlessness, lower limb weakness, generalized weakness, irrelevant speech, headache, and arthralgia. According to Chhotala et al.,²² the most common manifestation is fever among all patients, which is followed by headache, myalgia, vomiting, arthralgia, retro-orbital pain, rash, bleeding manifestations, abdominal pain, and diarrhea. Several other Indian studies have also demonstrated that fever is the most common symptom in dengue patients.^20,22–24

A majority of the study subjects, that is, 59–70%, were classified as mild dengue (dengue without warning signs), followed by dengue with warning signs (12–20%), and severe dengue was noted in 17–21% of the patients in the current study. Eregowda et al.²⁵ reported that of 57 cases, 81.0% had dengue with no warning signs, 14.0% had dengue fever with warning signs, and 5.3% had severe dengue in their study. In a Brazilian study, 4.8% presented with dengue without warning signs, 33.3% had warning signs, and 61.9% had severe dengue.²⁶

The electrocardiogram changes noted in our study subjects included sinus tachycardia, sinus bradycardia, atrial fibrillation, and ST segment and T wave changes. Intensive care unit admissions were required in 15% of our patients. The major causes for the ICU admissions were life-threatening bleeding manifestations, severe thrombocytopenia with or without bleeding manifestation, systemic inflammatory response syndrome, DSS, acute respiratory distress syndrome, CNS impairment, and shock with prerenal failure. Platelet transfusion was required in 19% of the study patients. The case fatality rate in our study was 2%. A systemic review and meta-analysis of dengue infection in India by Ganesh Kumar et al.²⁷ showed that the case fatality of dengue was 2.6%.

In the current study, D1 serum ferritin levels increased, and the platelet counts decreased with the increasing severity of dengue. Among the three parameters measured, serum ferritin levels and hematocrit levels increased, and the platelet counts decreased with the increasing severity of dengue. Hyperferritinemia is associated with increasing severity of dengue.

The association of severity of dengue with the serum ferritin levels, platelet count, and hematocrit varied. Among the three parameters measured, serum ferritin levels significantly increased, and platelet counts significantly decreased with the increasing severity of dengue. By contrast, the hematocrit was nearly the same on the day of admission for the differing severities but increased significantly on D4 of admission with the increasing severity of dengue. Hematocrit ≥ 40 and platelet ≤ 50,000/μL were significant independent predictors of dengue severity. Moreover, Chaiyaratana et al.²⁸ found that a high serum ferritin level ≥ 1,200 ng/mL was a predictor of DHF in children.

Petchiappan et al.²⁹ noted higher ferritin levels were reported on D4 onward during the febrile phase of illness (within 7 days of onset of illness) in severe dengue patients than nonsevere dengue cases. Our study demonstrated an increase in ferritin levels as the severity grade increases and analyzed serum ferritin levels both during febrile, defervescence, or convalescent phase, which is critical as the disease progression to severe dengue occurs after the febrile phase. Our study also had a higher number of severe dengue cases than this study.²⁹

Furthermore, in another study in Aruba, which compared serum ferritin levels of 44 dengue patients with those of other febrile illnesses (OFIs), and found significantly elevated ferritin levels on days 6–8 of illness in dengue patients. The higher ferritin levels in dengue patients were associated with higher viral copy numbers on days 2–3 of illness. However, this study did not find a significant difference in ferritin levels between dengue patients with warning signs and dengue patients without warning signs,³⁰ whereas our study demonstrated the contrary. Aruba study findings were validated by the Brazilian dengue cohort, which included 132 dengue patients, including 33 severe dengue patients. The authors observed severe dengue patients showed elevated ferritin and a pro-inflammatory cytokine profile (IL-6, IL-8, MCP, and sIL-2R).³⁰

The ROC curve obtained by plotting the different cutoff values for the serum ferritin levels showed that the serum ferritin level was a significantly “good” predictor of severe dengue with an AUC of 0.863 with SE = 0.043 and a 95% CI from 0.778 to 0.947 (P < 0.05) on the day of admission, whereas on D4 of admission, the serum ferritin level was a “fair” predictor of the warning signs, with an AUC of 0.781 with SE = 0.047 and a 95% CI from 0.690 to 0.873 (P < 0.05) and was a significant “excellent” predictor of severe dengue, with an AUC of 0.947 with SE = 0.021 and a 95% CI from 0.907 to 0.988 (P < 0.05). Similarly, in a study conducted by Roy Chaudhuri et al.,²¹ an ROC analysis for assessing ferritin as a diagnostic marker for dengue versus OFI subset revealed an AUC of 0.942 with a SE of 0.035 and P-value of < 0.0001 (95% CI: 0.874–1.00), which implied that ferritin was a good to excellent differentiator between the dengue group and OFI group.

On the day of admission, the serum ferritin level was a good predictor using a cutoff of 1,990 ng/mL, to detect severe dengue with a maximum sensitivity of 94.1% and a specificity of 71.1%. Serum ferritin on D4 was a good and excellent predictor of warning signs in dengue and severe dengue, respectively. A cutoff of 895 ng/mL detects dengue with warning signs with a maximum sensitivity of 100.0% and a specificity of 61.4%, and a cutoff of 1,380 ng/mL detected severe dengue with a maximum sensitivity of 100.0% and a specificity of 85.4%, which indicated that a decrease in serum ferritin on D4 might be used as a prognostic marker for a reduction in severity. According to Roy Chaudhuri et al.,²¹ the best cutoff for ferritin levels to differentiate dengue from OFIs was 1,291 ng/dL. At this cutoff, the sensitivity is 82.6% and the specificity is 100%.

Similarly, a case–control study conducted in Jawaharlal Institute of Postgraduate Medical Education and Research, India, compared serum ferritin, ceruloplasmin, and alpha1-antitrypsin levels of 48 dengue cases with those of 48 OFIs. They observed significantly elevated ceruloplasmin and ferritin levels in severe dengue patients compared with those in non-severe dengue and OFIs. They reported that only ferritin could predict the disease severity with a highest sensitivity and specificity of 76.9% and 83.3%, respectively, on the day of admission, and these numbers were 90% (sensitivity) and 91.6% (specificity) around defervescence, suggesting that ferritin may serve as a potential biomarker for the early prediction of disease severity,³¹ whereas we divided our study population (100 dengue cases) into three subsets based on severity (according to the WHO 2009 classification) and compared ferritin among different dengue classes. In comparison, our study also revealed better sensitivity on the day of admission and around defervescence and better specificity around defervescence but poorer specificity on admission day.

Other potential biomarkers evaluated are cytokines and chemokines. Major cytokines evaluated are IL-10, interferon (IFN)-℘, TNF-α, IL-6, etc.³² Flores-Mendoza et al.³³ observed elevated IL-10 and IL-6 in dengue patients compared with those in healthy controls and suggested that elevation of these interleukins and elevated socs1 and socs3 expressions can recognize patients with at risk for severe dengue. Braiser et al.³⁴ also suggested that elevated IL-10 could “assist in triage of potential DHF patients” based on their study findings on 51 dengue patients. But Malavige et al.³⁵ concluded that although IL-10 was elevated in severe dengue, it carries a low discriminatory value to differentiate the severe dengue patients from non-severe dengue patients. IFN-℘ levels show varied results in several studies; these discrepancies between the studies are attributed to its rapid kinetics.³² Another study comparing TNF-α, IFN-℘, and IL-6 between severe dengue (n = 30) and non-severe (n = 50) dengue patients found a significant elevation in TNF-α and IFN-℘ but not IL-6 in severe dengue.³⁶ Although IL-10 may appear as a promising biomarker to predict dengue severity, cytokine assays are not as readily available and are expensive tests for low- and middle-income countries.

Limitations.

The sampling procedure was purposive, and thus, the generalizability and the selection of representative samples is a limitation. Hence, the study needs to be conducted using a larger sample size to generalize the obtained results.

CONCLUSION

Serum ferritin levels significantly increased with the increasing severity of dengue both on the day of admission and on D4 of admission. On the day of admission, the serum ferritin level was a significant “good” predictor of the severe dengue, with a cutoff of 1,990 ng/mL, showing a maximum sensitivity of 94.1% and a specificity of 71.1%. On D4 of admission, serum ferritin was a “fair” predictor of the dengue fever with warning signs, with a cutoff of 895 ng/mL, revealing a maximum sensitivity of 100.0% and a specificity of 61.4%. Also, it was a significant “excellent” predictor of severe dengue, with a cutoff of 1,380 ng/mL, displaying a maximum sensitivity of 100.0% and a specificity of 85.3%.

Note: Supplemental tables and figures appear at www.ajtmh.org.