Abstract

METHODS

RESULTS

Identification of Homotypic DENV Reinfections

Of 2982 C-case samples tested, 130 DENV-positive cases with quantifiable viremia levels were identified (4.5%). Data from these cases were then added to a database of all symptomatic dengue cases identified in the PDCS during 2004–2013 (599 A/B dengue cases). Thirty-two patients with repeat symptomatic DENV infections were identified, including 1 patient with 3 documented infections, and serotype information was available for both infections in 29 infection pairs (87.8%). Among these 29 infection pairs, homotypic DENV reinfections were identified in 4 (13.8%). Clinical histories, examination findings, and clinical diagnoses for each infection from these 4 patients are shown in Table ​Table1.1. In all cases, patients had a history of fever, although they did not necessarily have documented fever during the visit. Each patient experienced 1 DENV-positive C case and 1 A/B dengue case. In patients 1 and 4, the DENV-positive C case preceded the A/B dengue case. Patient 1 required hospitalization for severe dengue during the second infection (A case); no other patients were hospitalized during these infections.

Table 1.

Clinical Information on 4 Patients Who Developed Homotypic Dengue Virus (DENV) Reinfections

Patient, Infection Date	Sex	Age at First Infection, y	DENV Serotype	Time Between Infections, d	Category	Day of Fever	Clinical History	Physical Examination Finding(s)	Clinical Diagnosis
1	M	13	1	341	…	…	…	…	…
Oct 2011	…	…	…	…	C	1	Sore throat, cough	Temp, 39.2°C	URTI, influenza suspected
Sep 2012	…	…	…	…	A	2	Headache, retro-orbital pain, cervical lymphadenopathy, arthralgia, myalgia, rash, dry mouth	Temp, 37.1°C; conjunctival injection; erythematous rash; cold extremities; cyanosis; delayed capillary refill	Severe dengue, hospitalized
2	F	4	2	621	…	…	…	…	…
Sep 2005	…	…	…	…	A	1	Headache, retro-orbital pain, arthralgia	Temp, 38.9°C; HR, 131 beats/min; RR, 35 breaths/min; BP, 90/60 mm Hg	Classic dengue
May 2007	…	…	…	…	C	1	Rhinorrhea, cough	Temp, 37.4°C	URTI
3	F	4	3	343	…	…	…	…	…
Oct 2009	…	…	…	…	A	2	Headache, cervical lymphadenopathy, myalgia, loss of appetite, rash	Temp, 39.0°C; conjunctival injection; generalized maculopapular rash; positive tourniquet test	Classic dengue
Sep 2010	…	…	…	…	C	2	Headache, cervical lymphadenopathy, cough, nasal congestion, loss of appetite, rhinorrhea, ear ache	Temp, 39.5°C; HR 112 beats/min; RR, 28 breaths/min; pharyngeal erythema	URTI
4	F	3	3	325	…	…	…	…	…
Dec 2009	…	…	…	…	C	1	Diarrhea, sore throat, loss of appetite, hoarseness, cough	Temp, 38.3°C; HR, 98 beats/min; RR, 32 breaths/min; pharyngeal erythema	Diarrhea with dehydration
Nov 2010	…	…	…	…	B	2	Intermittent abdominal pain	Temp, 37.1°C	Acute febrile syndrome

Open in a separate window

Abbreviations: BP, blood pressure; HR, heart rate; RR, respiratory rate; Temp, temperature; URTI, upper respiratory tract infection.

The results of laboratory testing for DENV are shown in Table ​Table2.2. Homotypic reinfections involved DENV-1, DENV-2, and DENV-3. The identified serotypes were consistent with the prevalent serotypes in the PDCS during the years in which these infections occurred [11, 16]. All A/B cases were positive by hemi-nested RT-PCR, virus isolation, and serological testing of paired acute-phase (day 1–2) and convalescent-phase (day 15–18) samples. All A/B cases seroconverted anti-DENV IgM, as determined by MAC-ELISA (from negative to positive). Three cases also demonstrated seroconversion of anti-DENV antibody by inhibition ELISA (patients 1, 3, and 4; acute-phase titers, <10; convalescent-phase titers, 26–69), while patient 2 showed a ≥4-fold rise in antibody titer (from 320 to 20480), indicating a secondary DENV infection. DENV-positive C cases were initially identified by a screening real-time RT-PCR [21] and confirmed using the DENV-specific multiplex real-time RT-PCR and hemi-nested RT-PCR [18, 22]. Serotype calls in the DENV-specific multiplex real-time RT-PCR and hemi-nested RT-PCR, which target different regions of the DENV genome, were concordant. To reconfirm DENV detection in C cases, nucleic acids were extracted from a second serum aliquot (made at the time of the C-case presentation, for patients 1, 2, and 4) or reextracted from the only available aliquot (for patient 3) and tested in the DENV-specific multiplex real-time RT-PCR [18]. DENV viremia levels were 2.75–6.44 log₁₀ copies/mL in C cases and 3.70–8.05 log₁₀ copies/mL in A/B cases. DENV-1 was also isolated in culture from the C case of patient 1.

Table 2.

Results of Dengue Virus (DENV) Laboratory Testing

Patient, Case Category	Hemi-Nested RT-PCR Result	DENV Multiplex RT-PCR Result	Viremia Level, Log10 Copies/mL serum	Virus Isolated	Sequencing Confirmed	IgM Seroconversion	Inhibition ELISA		Neutralizing Antibodies
							Acute- and Convalescent- Phase Seraa	Annual Samplesb	Annual Samplesc
1
C	DENV-1	DENV-1	2.75	Yes	Yes	NDd	NDd	Negative	NDe
A	DENV-1	DENV-1	7.59	Yes	Yes	Yes	Positive	Positive	NDe
2
A	DENV-2	DENV-2	8.00	Yes	Yes	Yes	Positive	Positive	Positive
C	DENV-2	DENV-2	4.98	No	Yes	NDd	NDd	Negative	Negativef
3
A	DENV-3	DENV-3	3.70	Yes	Yes	Yes	Positive	Positive	Positive
C	DENV-3	DENV-3	6.44	No	Yes	NDd	NDd	NDe	NDe
4
C	DENV-3	DENV-3	3.37	No	Yes	NDd	NDd	Negative	Negative
B	DENV-3	DENV-3	8.05	Yes	Yes	Yes	Positive	Positive	Positive

Open in a separate window

Abbreviations: ELISA, enzyme-linked immunosorbent assay; IgM, immunoglobulin M; ND, not determined; NT₅₀, 50% neutralization titer; RT-PCR, reverse transcription polymerase chain reaction.

^a Positive result defined as a ≥4-fold rise in DENV-specific inhibition ELISA titer in paired acute- and convalescent-phase samples.

^b Positive result defined as a ≥4-fold rise in DENV-specific inhibition ELISA titer in paired preinfection and postinfection annual samples [20].

^c Positive result defined as a ≥4-fold rise in NT₅₀ in paired preinfection and postinfection annual samples [9].

^d Convalescent-phase sample was not collected.

^e Sample not available.

^f Three-fold rise in NT₅₀ to DENV-2.

Inhibition ELISA of annual samples collected before and after a DENV-positive C case revealed no evidence of recent DENV infection (3 patients had specimens available for analysis), whereas annual samples indicated a DENV infection in the year in which the A/B cases occurred (Table ​(Table2).2). Three patients developed an A/B dengue case after a DENV-positive C case (secondary infection). In these instances, inhibition ELISA results from acute- and convalescent-phase samples were consistent with primary infection. This analysis includes patient 3, who had 3 separate DENV infections during the study: 2 DENV-3 infections (an A case in 2009 and a C case in 2010), as well as a DENV-1 C case in 2007. Patient 2 was not DENV naive at study entry and experienced a secondary DENV-2 A-case infection in 2005, as well as a DENV-2 C case in 2007.

Annual samples from patients 2–4 were available for neutralizing antibody testing. NT₅₀ and inhibition ELISA results were generally consistent (Table ​(Table2).2). A/B dengue cases resulted in an appropriate, ≥4-fold rise in NT₅₀ between preinfection and postinfection samples. C-case DENV infections were not captured by neutralization antibody assay for patient 2 (DENV-2 C case in 2007), patient 3 (2007 DENV-1 infection; no sample was available to test for C case in 2010), or patient 4 (DENV-3 C case in 2009). However, patient 2, who experienced a secondary DENV-2 infection in 2005, displayed a 3-fold rise in DENV-2 NT₅₀ following their DENV-2 C case in 2007.

Sequencing and Phylogenetic Analysis

Complete E gene sequences were obtained from both infections for patients 1, 3, and 4 and from the A case for patient 2. The complete E gene sequence could not be obtained from the C case for patient 2; however, the 511–base pair amplicon generated during the first step of the hemi-nested RT-PCR [22], covering portions of the capsid and prM genes, was sequenced for this case. NCBI Nucleotide BLAST results for all sequences matched the serotype detected by RT-PCR analysis.

Further E gene sequence analysis revealed 97% nucleotide identity between paired DENV-1 strains from patient 1 and 99.4% and 99.2% nucleotide identity between paired DENV-3 strains from patients 3 and 4, respectively (Table ​(Table3).3). Phylogenetic trees generated using these sequences and sequences obtained from contemporaneous Nicaraguan isolates confirmed that the paired strains were nonidentical but closely related (Figure ​(Figure1).1). In addition, these strains were closely related to the viruses circulating in Nicaragua during the years when the study patients were infected. Translation of the sequences from the strains responsible for homotypic reinfections demonstrated 1–3 amino acid differences in the Envelope protein as compared to the first infecting virus (Table ​(Table33).

Table 3.

Envelope Nucleotide and Amino Acid Sequence Comparison of C-Case and A/B-Case Dengue Virus Strains

Patient	Nucleotide Identity, %	Amino Acid Change(s)a
1	97.0	I54V, F95L, T181A
3	99.4	A328V, R468K
4	99.2	S184P

Open in a separate window

^a Amino acids from the C case are displayed first.

Open in a separate window

Figure 1.

Phylogeny of dengue virus (DENV) strains from patients with homotypic reinfections. A, DENV-1 from patient 1. B, DENV-3 from patients 3 and 4. Phylograms were generated by neighbor joining analysis of all unique envelope (E) gene sequences in the National Center for Biotechnology Information nucleotide database from contemporaneous Nicaraguan isolates of DENV-1 and DENV-3. Patient DENV sequences are labeled with a dot. Reference DENV-1 and DENV-3 E gene sequences representing established genotypes were also included (open diamonds). Sequences are named using the following convention: GenBank accession number/place of origin/year of collection. The scale bars represent the number of nucleotide substitutions per site. Abbreviations: FI, Fiji; MY, Malaysia; NI, Nicaragua; PH, Philippines; SL, Sri Lanka; TH, Thailand; US-FL, Florida; US-HI, Hawaii; US-PR, Puerto Rico.

DISCUSSION

In the current report, we describe the first virologically confirmed cases of homotypic DENV reinfection. Each DENV infection and serotype was detected by ≥2 methods, including viral isolation and distinct RT-PCR assays that target nonoverlapping regions of the DENV genome. Additionally, full-length E gene sequence from both infections for patients 1, 3, and 4 and the A case for patient 2 confirmed the serotypes identified by RT-PCR analysis, and the serotype of the C case for patient 2 was confirmed by sequencing a region of the capsid and prM genes. Phylogenetic analysis of paired E gene sequences further demonstrated that patients were infected with closely related but distinct homotypic DENV strains during their A/B and C infections.

Protective immunity that results from natural DENV infection is presumed to be lifelong and serotype specific, but whether protection results from sterile immunity or the prevention of clinical disease remains unresolved [27]. In the research performed by Albert Sabin, study subjects were protected from clinical disease following rechallenge with DENV-1 at 18 months, but they were not evaluated for viremia [2]. In primate studies, animals typically have not developed viremia upon rechallenge with homologous virus [28–30]. However, much of this research was performed before the routine use of RT-PCR. In a more recent study, Bernardo et al demonstrated low-level viremia only detectable by hemi-nested RT-PCR analysis in 4 of 5 macaques rechallenged with DENV-2 [31]. In the current study, patients developed DENV viremia 1–2 years after a DENV infection of the same serotype. This argues against sterile immunity as the mechanism of protection following natural DENV infection and also demonstrates that, in some cases, serotype-specific immunity to DENV may be short-lived. Furthermore, patients were not protected against clinical disease, as demonstrated by patients 1 and 4. In these patients, an A/B dengue case followed a primary C case of the same serotype, and for patient 1, this resulted in hospitalization with severe dengue.

While findings consistent with homotypic reinfections have been reported in human studies, all lack virologic confirmation and rely primarily on serologic testing [6, 12–14]. Endy et al [12] described one patient with mild dengue fever from DENV-2 who had preexisting monotypic neutralizing antibodies against DENV-2 detected at low levels (50% plaque reduction neutralization titer [PRNT₅₀], 11). Sirivichayakul et al [14] reported 2 cases of patients with monotypic PRNT₅₀ to DENV-2 and -1, respectively, who subsequently developed infections with the same serotype, as determined by mosquito isolation. However, specimens were not available to confirm the first infections in these 3 cases [12, 14]. A possible homotypic DENV-2 reinfection was identified in Puerto Rico. In this case, infections occurred 16 months apart, but the initial infection could not be confirmed, and the patient was removed from further analysis [13]. Finally, Gibbons et al [6] found one possible homotypic DENV-2 reinfection in Thailand, but enzyme immunoassay and RT-PCR analysis yielded discordant serotype results for the first infection [6].

In our patients, homotypic DENV reinfections were identified by testing C cases in the PDCS with real-time RT-PCR assays for DENV that demonstrate improved sensitivity and equivalent specificity as compared to widely-used molecular DENV diagnostic tests [18, 32]. The combination of testing a unique study population with highly sensitive diagnostic tests may have allowed for the detection of homotypic reinfections when they have not been previously identified in large studies of typical, symptomatic dengue cases [6, 9, 10, 13]. When C-case data were added to all symptomatic DENV infections identified in the PDCS, homotypic DENV reinfections accounted for 13.8% of all repeat DENV infections with a confirmed serotype (12.1%, if repeat symptomatic infections without serotype confirmation were included). In the PDCS, symptomatic dengue cases only account for approximately 20% of all DENV infections [11, 33]. Subclinical or asymptomatic DENV infections are identified by inhibition ELISA of annual samples, but the causative serotype in these infections cannot always be accurately determined. Therefore, homotypic DENV reinfections may be more common than indicated by our study. Such repeat infections may have a significant impact on force of infection estimates for DENV, which in turn are used to estimate necessary vaccine coverage and to guide disease control programs [34]. In addition, since the viremia level was detectable and quite high in some of the C cases, these homotypic infections could conceivably contribute to transmission and thus need to be considered in modeling of DENV transmission dynamics.

The viremia level detected at presentation in DENV-positive C cases was as high as 6.44 log₁₀ copies/mL. However, these infections generally did not elicit long-lived humoral immune responses, based on inhibition ELISA of (1) healthy annual samples and (2) samples obtained during A/B dengue cases that followed DENV-positive C cases, as well as analysis of neutralizing antibody titers in healthy annual samples. Inhibition ELISA has demonstrated equivalent sensitivity and specificity to hemagglutination inhibition [23], and testing of annual samples by this method is 80% sensitive for the capture of A/B dengue cases [11]. Taken together, these data are consistent with an altered immune response to C-case DENV infections. Such infections may only elicit a short-lived humoral immune response, or they may be adequately controlled by cellular or innate immune responses without eliciting a detectable humoral response. Interestingly, a 3-fold rise in NT₅₀ was observed following the DENV-2 C case in patient 2, who had experienced a prior secondary A case, indicating that homotypic reinfections may play a role in the maintenance of DENV immunity in endemic regions [25].

Limitations of the current study include the retrospective detection of DENV infections among C cases. Convalescent-phase samples were, therefore, unavailable for serological testing, but annual serum samples were tested for anti-DENV antibodies by inhibition ELISA to evaluate long-lived antibody responses to DENV infection. The dampened immune response to DENV-positive C cases, described earlier, limited our ability to test the possible effect of sequence differences identified in the envelope protein of A/B and C-case DENV strains from patients 1, 3, and 4. Thus, the importance of these amino acid changes in relation to immune escape remains unclear. A final limitation to this study is that DENV was isolated from only 1 C case. However, viral culture is known to be less sensitive than real-time RT-PCR analysis for the detection of DENV. In one prospective method comparison, viral culture was only 27.5% sensitive as compared to RT-PCR analysis [35]. Furthermore, specimens in our study had been stored at −80°C for up to 8 years and had undergone 2 freeze-thaw cycles before attempted viral isolation. These conditions are expected to significantly decrease culture yield, perhaps by as much as 5-fold to >10-fold (E. H., unpublished data) [36, 37].

In conclusion, we present 4 patients with homotypic DENV reinfections, including infections with DENV-1, DENV-2, and DENV-3. To our knowledge, these represent the first virologically confirmed homotypic reinfections in the field of dengue research. Such cases challenge the paradigm of lifelong, serotype-specific DENV immunity following natural infection and have implications for the modeling of DENV transmission.

Notes

References