Development and clinical evaluation of multiple investigational monovalent DENV vaccines to identify components for inclusion in a live attenuated tetravalent DENV vaccine

Abstract

The Laboratory of Infectious Diseases at the National Institute of Allergy and Infectious Diseases, National Institutes of Health has been engaged in an effort to develop a safe, efficacious, and affordable live attenuated tetravalent dengue vaccine (LATV) for more than ten years. Numerous recombinant monovalent DENV vaccine candidates have been evaluated in the SCID-HuH-7 mouse and in rhesus macaques to identify those candidates with a suitable attenuation phenotype. In addition, the ability of these candidates to infect and disseminate in Aedes mosquitoes had also been determined. Those candidates that were suitably attenuated in SCID-HuH-7 mice, rhesus macaques, and mosquitoes were selected for further evaluation in humans. This review will describe the generation of multiple candidate vaccines directed against each DENV serotype, the preclinical and clinical evaluation of these candidates, and the process of selecting suitable candidates for inclusion in a LATV dengue vaccine.

INTRODUCTION

Dengue viruses belong to the Flaviviridae family and present as four distinct serotypes; DENV-1, DENV-2, DENV-3, and DENV-4, with each serotype capable of causing the full spectrum of dengue illness [1]. Epidemiological studies have determined that the risk for more severe dengue illness is higher following a second, heterotypic DENV infection than for a primary DENV infection [2, 3]. A potential etiology of the increased risk may be immune enhancement from prior infection resulting in increased virus replication, which has been shown to correlate with disease severity [4]. Although severe dengue illness can occur with a third or fourth DENV infection, this risk appears to be very low [3]. For these reasons, there is consensus that a successful DENV vaccine must ideally protect against all four DENV serotypes. The DENV vaccines furthest along in clinical development are live attenuated (LA) vaccines. Several investigational LA tetravalent DENV vaccine candidates are currently being evaluated in clinical trials [5-7]. However, developing a LA tetravalent DENV vaccine that is sufficiently attenuated for each of the monovalent components yet immunogenic for all four dengue serotypes has been a challenge [6, 8-10].

The Laboratory of Infectious Diseases (LID) at the National Institutes of Allergy and Infectious Diseases, NIH, has been engaged in efforts to develop a LA tetravalent dengue vaccine for more than a decade. A LA vaccine approach was chosen for several reasons. First, natural DENV infection is believed to induce life-long protective homotypic immunity [11]. Because live vaccines stimulate both cellular and humoral immune responses and are therefore able to induce a strong memory response, a durable immune response can be expected. Second, LA vaccines for other related flaviviruses such as yellow fever and Japanese encephalitis virus have been successfully developed. The LA yellow fever 17D vaccine has been in use for more than 70 years and has an excellent safety record. A single subcutaneous injection induces a protective antibody response in more than 95% of vaccinees and the immunity induced by the 17D vaccine is very long lasting, if not lifelong [12]. The live attenuated Japanese encephalitis vaccine SA 14-14-2 is produced in China and has recently been licensed in Nepal, South Korea, Sri Lanka and India [13]. The protective efficacy of this vaccine in children 1 – 10 years of age was greater than 98% [13]. And third, LA vaccines can be very economical to produce, helping to ensure that the countries most in need of a dengue vaccine will have access.

Toward this end, scientists at the LID have used recombinant DNA technology and a directed design approach to develop numerous monovalent LA DENV candidate vaccines. These candidates underwent pre-clinical evaluation in a novel hepatoma mouse model and a nonhuman primate model to identify those candidates with the most favorable attenuation and immunogenicity profile for further evaluation in clinical trials. Eight LA monovalent DENV vaccine candidates have been evaluated in 13 separate NIAID-sponsored clinical trials conducted at the Center for Immunization Research, Johns Hopkins Bloomberg School of Public Health and the University of Vermont School of Medicine (Table 1). The purpose of these trials was to fully characterize the safety profile, the viral replication kinetics, the immunogenicity profile, and the human infectious dose 50% (HID₅₀) of these vaccine viruses. Because difficulties in achieving a balanced immune response from a tetravalent vaccine may be due to viral interaction between the individual components or differences in virus infectivity, one objective in evaluating each component as a monovalent vaccine was to better understand differences in viral kinetics, immunogenicity, and infectivity, as these parameters may modify the performance of a tetravalent formulation [14]. Below we describe the many monovalent DENV vaccines that have been generated by the LID and their evaluation process. Two primary attenuation strategies were employed: creation of nucleotide deletions into the 3′ untranslated region (UTR) and chimerization (Figure 1). Both of these strategies result in mutations that are highly unlikely to revert. Several mutants were found to be either under- or over-attenuated, either in preclinical testing or in clinical trials, precluding further evaluation. Using a systematic design for attenuation strategies and for evaluation of candidate monovalent DENV vaccines, we have identified 6 viruses for inclusion in different tetravalent admixtures.

Table 1.

Monovalent DENV vaccines developed by the Laboratory of Infectious Diseases and evaluated in human subjects

Serotype	Vaccine candidate	Dose (PFU)	No. of Subjects	Study site
1	DEN1Δ30	103	71	CIRa
1	DEN1Δ30	101	15	UVMb
2	DEN2/4Δ30(ME)	103	40	CIR
2	DEN2/4Δ30(ME)	101	15	CIR
3	DEN3/4Δ30(ME)	103	20	CIR
3	DEN3/4Δ30(ME)	105	20	CIR
3	DEN3-3′D4Δ30	103	20	CIR
3	DEN3Δ30/31	103	20	CIR
3	DEN3Δ30/31	101	20	CIR
4	DEN4Δ30	105	20	CIR
4	DEN4Δ30	103	70	CIR, UVM
4	DEN4Δ30	102	20	CIR
4	DEN4Δ30	101	20	CIR
4	DEN4Δ30-4995	103	20	Vanderbilt University
4	DEN4Δ30-200,201	105	20	CIR
4	DEN4Δ30-200,201	103	20	CIR
			Total: 431

^aCenter for Immunization Research, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD.

^bUniversity of Vermont College of Medicine, Burlington, VT.

Figure 1.

Recombinant attenuated DENV vaccine candidates were constructed by deletion of nucleotides from the 3′ UTR (Δ30 and Δ30/31) or by chimerization of genomic regions from different serotypes. Representations for each vaccine candidate are shown with DENV-1 sequence elements shown in white, DENV-2 elements shown in hatched lines, DENV-3 elements shown as speckled, and DENV-4 elements shown in gray. Genome regions are not drawn to scale.

Dengue 4 vaccines

Using recombinant DNA technology, scientists at the LID developed novel, full-length cDNA clones of the DENV-4 Dominica/81 strain 814669 from which infectious virus could be recovered [15, 16]. The 3′ UTR of the flavivirus genome is thought to play an important role in viral RNA replication and, as such, was chosen as a target for mutagenesis. The level of replication of several of these deletion mutants was compared with cDNA-derived wild type (wt) DEN4 and the recombinant mutant viruses were found to be attenuated in a non-human primate model [15]. Following a single subcutaneous dose of 10⁵ plaque forming units (PFU), a mutant with a 30-nucleotide deletion in the 3′ UTR at position 172 – 143 caused viremia in fewer monkeys, demonstrated a reduced mean peak titer of viremia, and a shorter duration of viremia compared to its wt parent virus. [15, 17]. Despite its restricted replication in non-human primates, this virus, designated as rDEN4Δ30, induced neutralizing antibody in all of the immunized monkeys. Consistent with its attenuation phenotype, the antibody titer induced by rDEN4Δ30 was lower than that induced by wt virus. When monkeys were challenged with wt DEN4 virus 42 days later, no animal had detectable viremia, indicating that the neutralizing antibody induced by the attenuated virus was protective. For these reasons, rDEN4Δ30 was selected as a candidate DENV-4 vaccine for evaluation in humans.

rDEN4Δ30 was first evaluated as a single subcutaneous dose of 10⁵ PFU in 20 flavivirus-naïve subjects [17]. Subjects were admitted to an inpatient unit following vaccination for evaluation of the transmissibility of the vaccine virus to mosquitoes [18]. Fourteen vaccinees had detectable vaccine virus in the blood following vaccination with a mean peak titer of 1.6 log₁₀ PFU/mL (0.5 log₁₀ PFU/ml being the lower limit of detection) (Table 2). The vaccine was well tolerated in all subjects, with an asymptomatic maculopapular rash and a temporary elevation of serum alanine amino transferase (ALT) levels being the most commonly reported adverse events (Table 3). Although ALT elevations were mild in four vaccinees, one vaccinee had a moderate increase of serum ALT (peak 238 IU/mL at day 16) which was not accompanied by nausea, vomiting, abdominal pain, or liver enlargement. In this and the other studies described below, serum neutralizing antibody titer was determined as the 60% plaque reduction neutralization titer (PRNT₆₀) by laboratory convention. All vaccinees seroconverted to wt DEN4 as defined by a ≥ 4-fold rise in serum neutralizing antibody by study day 42 [17]..

Table 2.

Level of viremia remains low in recipients of monovalent DENV vaccines

Vaccine candidate	Dose (log10PFU)	N	% with viremia	Mean peak virus titer ± SE (log10 PFU/mL)	Mean day of onset of viremia ± SE	Mean # of days of viremia ± SE
rDEN1Δ30	3	71	60	1.0 ± 0.08 (0.5-2.9)	10.0 ± 0.3 (6-16)	3.3 ± 0.3 (1-7)
rDEN2/4Δ30(ME)	3	40	60	0.5 ± 0.03 (0.5-1.2)	9.2 ± 0.6 (2-16)	3.3 ± 0.6 (1-7)
rDEN3/4Δ30(ME)	3	20	15	1.0 ± 0.3 (0.5-1.4)	12 ± 0	4.3 ± 0.7 (3-5)
rDEN3/4Δ30(ME)	5	20	0	n/a	n/a	n/a
rDEN3-3′D4Δ30	3	20	20	0.6 ± 0.1 (0.5-1)	10 ± 1.1 (8-13)	2.0 ± 0.8 (1-4)
rDEN3Δ30/31	3	20	20	0.5 ± 0.0 (0.5)	8.5 ± 0.6 (7-10)	2.0 ± 0.7 (1-4)
rDEN4Δ30	5	20	70	1.6 ± 0.2 (0.5 – 2.3)	5.6 ± 0.7 (2 – 10)	4.4 ± 0.8 (1 – 10)
rDEN4Δ30	3	70	28	0.6 ± 0.1 (0.5 – 1.7)	10.2 ± 0.6 (6 – 16)	1.8 ± 0.3 (1 – 5)
rDEN4Δ30-4995	5	20	0	n/a	n/a	n/a
rDEN4Δ30-200,201	5	20	0	n/a	n/a	n/a
rDEN4Δ30-200,201	3	20	0	n/a	n/a	n/a

Table 3.

Summary of clinical findings following administration of monovalent DENV vaccines

Vaccine candidate	Dose (log10PFU)	No. of subjects	% viremic	% of subjects with indicated clinical response
				Fever	Rash	Headache	Neutropeniaa	↑ALT
rDEN1Δ30 b	3	71	60	1 c	31	41	45	1
rDEN2/4Δ30(ME)	3	40	60	0	32	30	28	8 d
rDEN3/4Δ30(ME)	3	20	20	0	14	33	10	0
rDEN3/4Δ30(ME)	5	20	0	5	5	50	30	5
rDEN3-3′D4Δ30	3	20	20	5 c	35	30	10	5 c
rDEN3Δ30/31-7164	3	20	20	0	20	25	5	5 c
rDEN4Δ30 (Lot 4-9)	5	20	70	5	50	35	15	25
rDEN4Δ30 (Lot 4-9)	3	20	35	0	55	35	25	5
rDEN4Δ30 (Lot 109A)	3	50	26	0	40	52	6	2
rDEN4Δ30-4995	5	20	0	0	10	50	5	10
rDEN4Δ30-200,201	5	20	0	0	20	30	10	0
rDEN4Δ30-200,201	3	20	0	0	20	25	20	0
Placebo	n/ae	124	0	1	2	32	7	2

^aNeutropenia is defined as an absolute neutrophil count ≤ 1,500/mm³

^bBold typeface indicates vaccine candidates selected for inclusion in a tetravalent admixture currently under evaluation in humans.

^cNot related to vaccination.

^dPeak ALT levels in these subjects ranged from 1.3 to 1.7 × ULN. 2/3 subjects with elevated ALT had an ALT above the ULN on day 0 prior to vaccination despite being normal at screening

^eNot applicable.

Prior to conducting studies of rDEN4Δ30 in the outpatient setting, the ability of rDEN4Δ30 to disseminate in mosquitoes, as well as its capacity for transmission from vaccinees to mosquitoes was evaluated [18]. rDEN4Δ30 was found to be restricted in its capacity to infect the midgut of Aedes aegypti mosquitoes and to disseminate, compared with wt virus. Because the only mutation present in the rDEN4Δ30 virus that distinguishes it from parent virus is the Δ30 mutation, the observed growth restriction in mosquitoes was attributed to the Δ30 mutation. Direct mosquito transmissibility was evaluated in 10 of the 20 vaccinees. Aedes albopictus mosquitoes were fed on vaccinated subjects on study days 7, 8, and 9 post-vaccination; days on which the subjects were expected to be viremic. Five of the ten vaccinees had detectable viremia on at least one of those days, with titers ranging from 1.0 log₁₀ PFU – 2.3 log₁₀ PFU/mL. Vaccine virus was not detected in any of the 352 mosquitoes that fed on the subjects [18].

Although rDEN4Δ30 demonstrated a favorable safety, immunogenicity, and mosquito transmissibility profile when administered at a dose of 10⁵ PFU, a dose de-escalation study was initiated in the outpatient setting for two purposes: 1) to determine the human infectious dose 50% (HID₅₀) of the vaccine and 2) to determine if the incidence of rash and elevated serum ALT levels would be modified with a reduction in dose [19]. Three dose levels were evaluated; 10³ PFU, 10² PFU, and 10¹ PFU. Twenty subjects in each dose cohort received vaccine and 4 subjects received placebo (vaccine diluent) as a single subcutaneous injection. The mean peak virus titers of vaccinees who received the lower doses of vaccine (0.5 log₁₀ PFU/mL – 0.7 log₁₀ PFU/mL) were significantly lower than those who received a dose of 10⁵ PFU (1.6 log₁₀ PFU/mL). Although the incidence of rash and neutropenia was not affected by dose reduction, elevation in serum ALT levels was less frequent with lower doses. Only 1/60 vaccinees in the dose reduction study developed an elevation in serum ALT level, which was mild, compared with 5/20 in the previous study. Although 60% plaque reduction neutralization test (PRNT₆₀) antibody titers were slightly lower than those of vaccinees who received 10⁵ PFU, 58/60 vaccinees (97%) seroconverted to DENV-4, as defined by a ≥ 4-fold rise in PRNT₆₀ antibody titer by study day 42. This study was pivotal in that it established a very low HID₅₀ for this vaccine candidate (< 10 PFU) so that at least 100 times the HID₅₀ was administered with 10³ PFU of rDEN4Δ30, and that other candidate vaccines may be evaluated at an initial dose lower than 10⁵ PFU. For this reason, we opted to begin evaluation of candidate vaccines for the other DENV serotypes at a dose of 10³ PFU. rDEN4Δ30 has been further evaluated as a single subcutaneous dose of 10³ PFU in 50 additional subjects. The clinical, virology, and immunogenicity data for all subjects who received 10³ PFU of rDEN4Δ30 are presented in Tables 2, 3, and 4.

Table 4.

Summary of vaccine immunogenicity following single administration of monovalent DENV candidate vaccines

Vaccine candidate	Dose (log10)	No. of subjects	% Infectedb	Mean neutralizing antibody titer, (range)a			% seroconvertedc
				Day 0	Day 28	Day 42
rDEN1Δ30 d	3	70	94	<5	170 (<5 - 6309)	140(<5 – 2753)	93 e
rDEN2/4Δ30(ME)	3	40	100	<5	88 (11– 1043)	104 (11 - 1377)	100
rDEN3/4Δ30(ME)	3	20	30	<10	37 (<10-271)	210 (65-483)	30
rDEN3/4Δ30(ME)	5	20	25	<10	25 (<10-40)	210 (<10-323)	25
rDEN3-3′D4Δ30	3	20	80	<5	52 (<5 – 368)	72 (14 – 271)	80
rDEN3Δ30/31-7164	3	20	95	<5	136 (15-1715)	116 (28-1341)	95
rDEN4Δ30 (Lot 4-9)	5	20	100	<10	567 (72 – 2455)	399 (45 – 1230)	100
rDEN4Δ30 (Lot 4-9)	3	20	95	<10	139(<10 – 2365)	129 (15 – 1222)	95
rDEN4Δ30 (Lot 109A)	3	50	94	<5	112 (12 – 892)	135 (22 – 748)	93
rDEN4Δ30-4995	5	20	95	<5	150 (74-227)	126 (46-206)	95
rDEN4Δ30-200,201	5	20	100	<10	100 (40 – 530)	79 (19 – 926)	100
rDEN4Δ30-200,201	3	20	95	<5	62 (5 – 900)	71 (12 – 768)	95

^aGeometric mean plaque-reduction neutralization titer (reciprocal PRNT₆₀).

^bInfection is defined as recovery of vaccine virus from the blood or by seroconversion.

^cSeroconversion is defined as a ≥ 4-fold rise in PRNT₆₀ compared to Day 0.

^dBold typeface indicates vaccine candidates selected for inclusion in a tetravalent admixture currently under evaluation in humans.

^eOne subject had vaccine virus recovered from the blood but did not meet the definition of seroconversion.

In parallel with the dose de-escalation studies of rDEN4Δ30, additional recombinant DENV-4 candidate vaccines were developed with the goal of reducing the incidence of serum ALT elevation in vaccinees. Two derivatives of rDEN4Δ30 were identified, rDEN4Δ30-200,201 and rDEN4Δ30-4995. Both viruses were attenuated in severe combined immunodeficiency (SCID) mice engrafted with human hepatoma (HuH) cells (the SCID-HuH-7 mouse model) and in rhesus macaques [20-23]. The mutagenesis strategy differed for the two viruses. For rDEN4Δ30-4995, rDEN4 was subjected to chemical mutagenesis with 5-fluorouracil (5-FU) and more than one thousand resultant viruses were cloned and screened in SCID-HuH-7 mice. Viruses with a temperature sensitive (ts) phenotype and a 10-fold or greater reduction in peak titer compared with the parent virus in SCID-HuH-7 mice were chosen for further evaluation [24]. Three viruses that met both of these criteria contained a mutation at nucleotide position 4995 resulting in a serine to leucine change at amino acid residue 158 of the NS3 protein. The E protein was not affected and remained wild type in sequence. In addition to attenuating the virus, this mutation also promotes efficient replication of the virus in Vero cells. The recombinant virus rDEN4Δ30-4995 was then generated to contain both the 4995 mutation in NS3 and the Δ30 mutation [21]. This virus was more attenuated in both SCID-HuH-7 mice and in rhesus macaques compared with rDEN4. Paired charge-to-alanine mutagenesis of the DENV-4 NS5 protein revealed that mutation of amino acid residues 200 and 201 resulted in a virus that was ts and attenuated [20]. rDEN4Δ30-200,201 was generated by introducing these same changes in the rDEN4Δ30 virus [21]. rDEN4Δ30-200, 201 was more attenuated than rDEN4 and rDEN4Δ30 in SCID-HuH-7 mice and rhesus macaques. Although viremia was not detected in any macaque that received 10⁵ PFU of rDEN4Δ30-200,201, all animals seroconverted to DENV-4 and were protected against wt DEN-4 challenge. The PRNT₆₀ of these animals against DENV-4 was lower than that of rDEN4 and rDEN4Δ30, further reflecting its attenuation phenotype [21].

Both rDEN4Δ30-4995 and rDEN4Δ30-200,201 were initially evaluated as a single subcutaneous dose of 10⁵ PFU in healthy flavivirus-naïve adult subjects [25, 26]. A dose of 10⁵ PFU was chosen for both candidates to evaluate the number of subjects who developed elevated serum ALT levels at this dose compared with those who received the same dose of DEN4Δ30. In each study, 20 subjects received vaccine and 8 received a placebo. Blood for clinical laboratory studies and viremia was sampled every other day for the first 16 days following vaccination. Subjects who received either rDEN4Δ30-4995 or rDEN4Δ30-200,201 did not develop detectable viremia during the 16 days following vaccination whereas 14/20 who received 10⁵ PFU of rDEN4Δ30 in the comparator study did (Table 2) [17, 25, 26]. The incidence of rash was also significantly lower in recipients of rDEN4Δ30-4995 and rDEN4Δ30-200,201 compared with rDEN4Δ30. The incidence of ALT elevation was significantly lower in recipients of rDEN4Δ30-200,201 compared with rDEN4Δ30; no subject who received rDEN4Δ30-200,201 developed an elevation in serum ALT levels (Table 3). Two subjects who received rDEN4Δ30-4995 developed elevated serum ALT levels; one was mild and the other moderate. The subject who had a moderate elevation of serum ALT level had concurrent clinical influenza illness and the ALT elevation was attributed to acute influenza infection. Despite the low incidence of maculopapular rash in recipients of rDEN4Δ30-4995, 17/20 vaccinees developed injection site erythema that was not seen with either rDEN4Δ30 or rDEN4Δ30-200,201. The rash had a median onset of four days post-vaccination and was visible for up to one week [26]. Seroconversion induced by both vaccine candidates was high; 100% of those vaccinated with rDEN4Δ30-200,201 seroconverted to DEN4, as did 95% of those vaccinated with rDEN4Δ30-4995 (Table 4). Although the overall safety profile of rDEN4Δ30-4995 was acceptable, the higher incidence of local reactogenicity elicited by this vaccine made it a less desirable candidate for inclusion in a tetravalent formulation than either rDEN4Δ30 or rDEN4Δ30-200,201. Because of the high infectivity and favorable safety profile of rDEN4Δ30-200,201 when given at a dose of 10⁵ PFU, this candidate was further evaluated at a dose of 10³ PFU in healthy adult subjects. The virologic, safety, and immunogenicity profiles of this vaccine at dose of 10³ PFU was very similar to that of the 10⁵ PFU dose (Tables 3 and 4). rDEN4Δ30 and rDEN4Δ30-200,201 are currently being evaluated at a dose of 10³ PFU as the DENV-4 vaccine components of different live attenuated tetravalent dengue vaccine admixtures.

Dengue serotype 1 vaccines

The structure of the DENV 3′ UTR is well conserved, providing a rationale for the introduction of 30 nucleotide deletion homologous to that of rDEN4Δ30 into the 3′ UTRs of DENV-1, DENV-2 and DENV-3 as a strategy for attenuating these three serotypes [27]. Although the individual nucleotides are not well conserved, the predicted stem loop structures formed by the nucleotides are conserved, making them a target for mutagenesis. The homologous Δ30 mutation was introduced into a cDNA clone of wt DENV-1 Western Pacific strain (WP) and infectious virus was recovered. The virus, designated rDEN1Δ30, was evaluated for attenuation in the SCID-HuH-7 mouse and in rhesus macaques [28]. When compared to recombinant wt DENV-1 WP in SCID-HuH-7 mice, the mean peak titer of rDEN1Δ30 was more than 100-fold lower than that of the parent virus. In rhesus macaques, rDEN1Δ30 replicated with a lower mean peak virus titer, and had a mean duration of viremia less than wt rDEN1. All rDEN1Δ30 inoculated monkeys were completely protected against wt rDEN1 challenge.

Based on the attenuation profile of rDEN1Δ30 in animal models, the candidate DENV-1 vaccine was evaluated in healthy flavivirus-naïve adult subjects. Three trials have been conducted to evaluate the safety and immunogenicity of this DENV-1 candidate vaccine. Two trials evaluated the vaccine given at a dose of 10³ PFU [29], and the third trial, designed to determine the HID₅₀ of the vaccine, evaluated a dose of 10¹ PFU. For the purpose of this review, the safety and immunogenicity results from the two trials evaluating a dose of 10³ PFU are presented together. Seventy-one subjects have received rDEN1Δ30 as a subcutaneous dose of 10³ PFU [30]. Forty-three subjects (60%) had detectable vaccine virus in the blood within the 16 day period following vaccination (Table 2). The pattern of viremia was similar to that observed with rDEN4Δ30, however, the mean peak titer and duration of viremia were greater for rDEN1Δ30 (Table 2). The adverse event profile was also similar to rDEN4Δ30 (Table 3). No subjects developed a fever related to vaccinations and the incidence of rash, headache, and ALT elevation were similar to rDEN4Δ30. A higher incidence of neutropenia was observed in recipients of rDEN1Δ30 compared to rDEN4Δ30 or rDEN4Δ30-200,201 (Table 3). Vaccinated subjects who became neutropenic had a statistically significant lower mean baseline ANC value (2,544/mm³) than did those subjects who did not experience neutropenia (3,858/mm³), α = 0.01. Twenty-two of the 32 vaccinees (69%) who became neutropenic had a baseline ANC of ≤ 3,000/mm³ compared with 7/39 (18%) vaccinees who did not become neutropenic. Although neutropenia was classified as mild (ANC 1,000 – 1,500/mm³) in 72% of neutropenic subjects, moderate (750 – 999/mm3) in 9% of neutropenic subjects, and severe (500 - 750/mm³) in 19%, all subjects remained asymptomatic. Neutropenia was transient (mean duration 2.3 days ± 2.0 SD), and in 5/6 individuals, severe neutropenia was observed only in a single assay over the 16-day post-vaccination period. rDEN1Δ30 was highly immunogenic with an overall infection rate of 94% and an overall seroconversion rate of 93%. One vaccinee had detectable vaccine virus in the blood but did not meet the definition of seroconversion (≥ 4-fold rise in serum neutralizing antibody). All subjects who seroconverted after the first vaccination and had a serum sample taken at study day 180 had detectable neutralizing antibody at that time-point, demonstrating durability of the antibody response.

Forty-six subjects received a second dose of 10³ PFU of the rDEN1Δ30 vaccine subcutaneously at 4 (23 subjects) or 6 months (23 subjects) following the first dose (unpublished data). One of the subjects withdrew prior to day 42 post-second vaccination and the antibody response following the second dose was evaluated in 45/50 vaccinees. Ten placebo recipients were included for comparison of safety and immunogenicity data. No vaccinee developed detectable viremia, fever, or rash following the second vaccination. A total of 4 vaccinees (9%) developed transient neutropenia, as did one placebo-recipient (10%). None of the 45 vaccinees developed a ≥ 4-fold rise in serum neutralizing antibody titer following the second dose of vaccine. Three subjects who did not seroconvert to DENV-1 following the first dose received a second dose of vaccine 4 months after the first dose. None of these subjects developed a ≥ 4-fold rise in serum neutralizing antibody or in ELISA titer following the second dose. However, five subjects who had seroconverted after the first dose of vaccine developed a ≥ 4-fold rise in ELISA titer following the second dose. One of these subjects received the second dose four months after primary vaccination and three subjects received the second dose six months after primary vaccination. These findings suggest that a single dose of rDEN1Δ30 was able to provide sterilizing immunity to infection with the vaccine virus for at least 6 months in 83% of vaccinees. Based on the overall safety and immunogenicity findings from these studies, rDEN1Δ30 was chosen for inclusion as the DENV-1 component in tetravalent admixtures currently undergoing evaluation clinical trials.

Dengue serotype 2 vaccines

Based on the Δ30 strategy employed for DENV-1 and DENV-4 candidate vaccines, a 30-nucletotide deletion homologous to that of rDEN4Δ30 was introduced into cDNA of the wt DENV-2 Tonga/74 strain. The recovered candidate vaccine virus rDEN2Δ30 was evaluated in mice, mosquitoes, and monkeys [31]. Wild type DENV-2 Tonga/74, recombinant cDNA-derived wt DENV-2 (rDEN2), and rDEN2Δ30 virus were all unable to infect the midgut of Aedes aegypti mosquitoes, a phenotype that has been described for other American genotype viruses [32, 33]. Inoculation of rDEN2Δ30 into SCID-HuH-7 mice resulted in a 10-fold reduction in mean peak titer compared with wt DENV-2 Tonga/74 and rDEN2. However, when rhesus macaques were inoculated with rDEN2Δ30, the mean peak titer and duration of viremia in these monkeys was only slightly less than that of monkeys inoculated with wt rDEN2 (1.7 vs 1.9 log₁₀ PFU/mL and 2.8 vs 4.0 days, respectively) [31]. Because it was not possible to demonstrate a significant attenuation phenotype for rDEN2Δ30 compared to what may be a “naturally” attenuated parent virus, the candidate vaccine was not evaluated in clinical trials.

A second candidate DENV-2 vaccine was developed by utilizing two attenuation techniques; chimerization between two different dengue viruses and introduction of the Δ30 mutation into the 3′ UTR of DEN4. This was accomplished by substituting the prM and E coding regions of rDEN4Δ30 with those derived from the DENV-2 NGC prototype strain, resulting in the chimeric virus rDEN2/4Δ30(ME) [34, 35]. Chimerization alone resulted in an attenuation phenotype as infection of SCID-HuH-7 mice with rDEN2/4(ME), lacking the Δ30 mutation, resulted in a ≥ 20-fold reduction in mean peak titer compared to infection with wt parent viruses DENV-2 or DENV-4. When the Δ30 mutation was combined with chimerization in rDEN2/4Δ30(ME), a further reduction in mean peak titer of 0.5 log₁₀ PFU/ml resulted, similar to what was observed for the rDEN4Δ30 virus [24, 35]. rDEN2/4Δ30(ME) was found to be less infectious for Aedes aegypti mosquitoes than either parent virus. Interestingly, chimerization resulted in reduced infectivity of Toxorhynchites mosquitoes as well [35]. rDEN2/4Δ30(ME) was markedly attenuated in rhesus macaques compared with either wt parent. Only 1/4 macaques had detectable viremia following inoculation with rDEN2/4Δ30(ME) compared with 4/4 macaques inoculated with rDEN4 and 6/6 macaques infected with wt DENV-2 NGC. The mean peak titer of rDEN2/4Δ30(ME) was 0.8 log₁₀ PFU/mL and 1.4 log₁₀ PFU/mL lower than that of the parent viruses rDEN4 and DENV-2 NGC, respectively. Although all animals developed a ≥ 4-fold rise in serum neutralizing antibody against DENV-2, the geometric mean antibody titer induced by rDEN2/4Δ30(ME) was lower than that induced by DENV-2 NGC (1:62 compared with 1:312). Despite the attenuation phenotype exhibited by rDEN2/4Δ30(ME), no animal had detectable viremia following challenge with wt DENV-2 NGC [36].

rDEN2/4Δ30(ME) has been evaluated for safety and immunogenicity in three separate clinical trials. Two trials evaluated the vaccine given at a dose of 10³ PFU and the third trial, designed to determine the HID₅₀ of the vaccine, evaluated a dose of 10¹ PFU [37] and unpublished data. A total of 40 subjects have received rDEN2/4Δ30(ME) at a dose of 10³ PFU. The first trial evaluated the vaccine as a single subcutaneous dose given to 20 subjects with 8 subjects receiving a placebo (vaccine diluent). The second trial evaluated two doses of 10³ PFU given to 20 subjects 6 months apart with 10 subjects receiving placebo. For the purpose of this review, the safety and immunogenicity results from these two trials are presented together. Following the first dose of vaccine, rDEN2/4Δ30(ME) was recovered from the blood of 60% of vaccinees over the 16 day post-vaccination period (Table 2). The mean peak titer was 0.5 log₁₀ PFU/mL, the limit of detection of the assay. The reactogenicity profile of rDEN2/4Δ30(ME) was similar to that observed for rDEN1Δ30 and rDEN4Δ30 (Table 3). Rash, headache and neutropenia were the most commonly observed adverse events although the incidence of headache was no greater than that observed in placebo recipients. All vaccinees seroconverted to wt DENV-2 by study day 42 following the first dose of vaccine. Seventeen of 20 subjects who had received the first dose of vaccine received a second 10³ subcutaneous dose of vaccine 6 months following the first dose. Fifteen subjects had detectable antibody at the time of revaccination with titers ranging from 1:11 to 1:190. The geometric mean titer (PRNT₆₀) of all vaccinees at study day 180 was 1:25. Viremia was not detected in any subject following the second dose. One subject developed a mild neutropenia following the second dose of vaccine, as did one placebo recipient. No subject developed a ≥ 4-fold rise in serum neutralizing antibody following the second dose. These data suggest that a single subcutaneous dose of rDEN2/4Δ30(ME) provided sterilizing immunity to reinfection with vaccine virus for at least 6 months in 17/17 subjects. The favorable safety and immunogenicity profile of rDEN2/4Δ30(ME) demonstrated in these studies prompted its inclusion as the DENV-2 component of the LA tetravalent admixtures currently under evaluation.

Dengue serotype 3 vaccines

Several live attenuated DENV-3 candidate vaccines were developed by the LID. The first of these, rDEN3Δ30, contained the homologous Δ30 mutation in the 3′ UTR of the wt DENV-3 Slemen/78. In contrast to what was observed for DENV-1 and DENV-4, the Δ30 mutation in rDEN3Δ30 did not appear to restrict replication of this virus in SCID-HuH-7 mice or in rhesus macaques [38], and therefore this candidate was not further evaluated. The second attenuation strategy employed was to replace the prM and E coding regions of rDEN4 and rDEN4Δ30 with those of DENV-3 Slemen/78 forming the chimeric viruses rDEN3/4(ME) and rDEN3/4Δ30(ME) [38]. Chimerization of the DENV-3 ME genes with rDEN4 conferred a 100-fold restriction of replication in SCID-HuH-7 mice but introduction of the Δ30 did not appear to confer any additional attenuation effect in this model. Both rDEN3/4(ME) and rDEN3/4Δ30(ME) were attenuated in rhesus macaques. Neither virus induced detectable viremia yet all of the animals developed a ≥ 4-fold rising in serum neutralizing antibody, although geometric mean titers of antibody were lower than those induced by the parent viruses, consistent with an attenuation phenotype. All animals were protected against challenge with wt DENV-3 Slemen/78 [38]. Both viruses were restricted in their ability to disseminate from the midgut to the head of orally infected Aedes aegypti mosquitoes and were restricted in their ability to replicate outside of the midgut in Toxorynchites mosquitoes. Because the presence of the Δ30 mutation may enhance the phenotypic stability of the chimeric virus, rDEN3/4Δ30(ME) was chosen for further evaluation in clinical trials.

rDEN3/4Δ30(ME) was initially evaluated as a single subcutaneous dose of 10³ PFU in 20 flavivirus-naïve adults with 8 subjects receiving a placebo (unpublished data). Five subjects had vaccine virus detectable in the blood following vaccination (Table 2). The virus was well tolerated by vaccinees with headache being the most commonly reported adverse event, although it occurred with similar frequency in placebo recipients (Table 3). Only 3 subjects developed a maculopapular rash. Unfortunately, only 30% of vaccinees were infected by rDEN3/4Δ30(ME) at this dose. For this reason, an additional 20 subjects were vaccinated with 10⁵ PFU and 8 subjects received placebo. Viremia was not detected following vaccination and only 25% of vaccinees seroconverted to wt DENV-3 by study day 42 following vaccination (Tables 2 and 3).

Because of the disappointing infectivity and immunogenicity of rDEN3/4Δ30(ME), two additional DENV-3 vaccine candidates were developed using recombinant DNA technology [39]. Both candidates are derived from wt DENV-3 Slemen/78 and contain all DENV-3 structural and non-structural proteins. The first virus, rDEN3Δ30/31, includes the original Δ30 mutation described above and an additional 31 nucleotide deletion in the 3′ UTR at a position located 55 nucleotides upstream of the Δ30 mutation. The second virus, rDEN3-3′D4Δ30, is a chimeric virus in which the entire 3′ UTR of rDEN3 has been replaced the 3′ UTR of rDEN4Δ30; a strategy referred to as the 3′ UTR swap. These viruses were attenuated in SCID-HuH-7 mice and in rhesus macaques. Although the viruses did not cause detectable viremia in rhesus macaques, both induced a ≥ 4-fold rise in serum neutralizing antibody in all inoculated animals and protected them from wt DENV-3 challenge [39]. The ability of rDEN3Δ30/31 to replicate in Toxorhynchites mosquitoes was also evaluated and the virus was found to be restricted in replication compared with wt DENV-3. Because of the attenuated phenotype demonstrated by both of these candidate vaccine viruses, they were evaluated in clinical trials.

rDEN3-3′D4Δ30 and rDEN3Δ30/31 were evaluated in healthy adult flavivirus-naïve adults in two separate clinical trials (unpublished data). Twenty subjects each received 10³ PFU of rDEN3-3′D4Δ30 or rDEN3Δ30/31 and 8 each received placebo. Both vaccines were well tolerated and both resulted in detectable viremia in 20% of vaccinated subjects (Table 2). The mean peak titer of both vaccines was low (0.6 log₁₀ PFU/ml and 0.5 log₁₀ PFU/mL). The reactogenicity of both vaccines was similar, with rash and headache being the most commonly observed adverse events (Table 3). Few vaccinees developed neutropenia in either study. One subject who received rDEN3-3′D4Δ30 developed an elevated temperature on one measurement. This episode of fever was judged to be unrelated to vaccination. Both vaccines were immunogenic, with 80% and 95% seroconversion following rDEN3-3′D4Δ30 and rDEN3Δ30/31 administration, respectively (Table 4). Both vaccines exhibited a favorable safety profile and are currently being further evaluated as components of LA tetravalent admixtures in humans.

Conclusions

Recombinant DNA technology has been systematically used to develop multiple monovalent DENV vaccine candidates with the distinct goal of generating both primary and back-up vaccine candidates for each serotype. Eight of these were evaluated in clinical trials and six vaccine candidates (rDEN1Δ30, rDEN2/4Δ30(ME), rDEN3-3′D4Δ30, rDEN3Δ30/31, rDEN4Δ30, and rDEN4Δ30-200,201) are currently being evaluated in four different LA tetravalent admixtures in humans. Based on data obtained from the monovalent vaccine studies, we were able to characterize the infectivity, replication kinetics, and immunogenicity of each of these candidates. The candidates chosen for inclusion in a tetravalent admixture infect 80 – 100% of vaccinees when given as a monovalent vaccine. The viral replication kinetics of each of the candidates is similar as well. Mean peak titers ranged from 0.5 log₁₀ PFU/mL – 1.0 log₁₀ PFU/mL and mean day of onset of viremia ranged from 8.5 days post-vaccination to 10 days post-vaccination. One monovalent candidate, rDEN3/4Δ30(ME) was found to be over attenuated in humans, even at a dose of 10⁵ PFU. This candidate was less infectious than the other monovalent vaccines, infecting only 25 – 30% of vaccines. It was therefore excluded from further evaluation. If this candidate had been included as part of a LA tetravalent vaccine without knowledge of its low infectivity as a monovalent vaccine, true homotypic immunogenicity to DENV-3 likely would have been poor. Additionally, had the monovalent study not been done, it would not have been possible to determine whether low infectivity or viral interference was the cause of its poor immunogenicity.

Overall, the vaccines were very well tolerated. Only 1 of 431 subjects had a fever that was determined to be possibly, probably, or definitely related to vaccine. This subject had an elevated temperature on only 1 measurement throughout the 16-day follow-up period although subjects record their temperature 3 times a day during this period. No subject met the protocol-defined criteria for dengue-like syndrome which defined as: infection, as determined either by recovery of vaccine virus from the blood and/or seroconversion, accompanied by fever and 2 or more of the following symptoms of Grade 2 or higher; headache lasting ≥12 hours, photophobia lasting ≥12 hours, or generalized myalgia lasting ≥12 hours. The monovalent vaccines had similar safety profiles with rash and transient neutropenia being the most commonly observed adverse events. Although a low level of viremia was detected for the majority of vaccine candidates at some point post-vaccination, there was no association of detectable viremia with rash, neutropenia, or other adverse event. As has been described previously, the rash was asymptomatic and was unnoticed by the majority of affected subjects [19, 29, 37]. The neutropenia observed was transient and was more likely to occur in those subjects who had a baseline absolute neutrophil count of ≤ 3,000/mm³ (p < .0001) and in subjects who were of African decent (p = .0007). Persons of African descent are known to maintain lower baseline white cell counts than other populations [40-42]. More than 60% of subjects enrolled in these studies were persons of African descent, accounting for the lower baseline ANC found in our vaccinees. Studies to evaluate whether or not neutropenia will be observed in subjects who receive a tetravalent admixture or in subjects from endemic areas are currently undergoing or are being planned. An important safety feature of these candidate vaccines is the stability of the attenuating mutations. Virus isolates from the last day of detectable viremia were collected from each subject with detectable viremia and were sequenced in the region of the Δ30 mutation to determine if any nucleotide changes had been introduced during viral replication. Fifty-six isolates representing 581 days of viral replication were tested. No changes in regions flanking the Δ30 mutation were found in any of these isolates, demonstrating the genetic stability of the 3′ UTR deletion mutations.

One concern regarding deployment of a live DENV vaccine is transmission of vaccine virus to non-vaccinated populations, including to those for whom a live vaccine may be contraindicated such as immunosuppressed individuals and pregnant women. For this reason, the ability of such a candidate vaccine to be transmitted by mosquitoes must be evaluated. All of the vaccines we have identified for inclusion in a live attenuated tetravalent admixture have multiple barriers to transmission by mosquitoes. Most importantly, each of the monovalent vaccine viruses replicates to a very low titer in vaccinated subjects, with mean peak titers ranging from undetectable (rDEN4Δ30-200,201) to 1.0 log₁₀ PFU/mL (rDEN1Δ30). Because a mosquito takes only 1 – 2 μL per blood meal, transmission of vaccine virus to a mosquito is highly unlikely. Additionally, as described above, the Δ30 mutation in many cases restricts the ability of virus to infect the midgut and to disseminate in Ae. Aegypti mosquitoes, further reducing the risk of transmission. Based on these data and our experiment demonstrating that vaccine virus was not transmitted to mosquitoes from infected subjects, we are confident that the risk of vaccine transmission in dengue endemic areas is extremely low.

A second concern with respect to dengue vaccines is the effect of waning antibody titers over time on the risk of developing severe disease should a vaccinee subsequently become infected with dengue. Antibody-enhanced dengue virus uptake by monocytes and macrophages is hypothesized to be a mechanism leading to more severe disease in secondary, heterotypic dengue infections [43-45]. Additionally, dengue outbreaks in Cuba demonstrated an increased risk for more severe disease with a longer interval between primary and secondary dengue infection [46]. For these reasons, the ideal dengue vaccine should induce a long-lived, balanced immune response to the four DENV serotypes. If the vaccine induced a poor antibody response to one or more DENV serotypes, or if it induced only a short-lived antibody response, vaccine recipients may theoretically be at risk for developing more severe dengue should they be subsequently infected with a serotype to which they have a sub-neutralizing antibody level. Although we have not evaluated the durability of the antibody response to our monovalent candidate vaccines beyond six months, we determined that the rDEN1Δ30 and rDEN2/4Δ30 candidate vaccines were able to induce sterilizing immunity to a second dose of vaccine out to six months in 83% of rDEN1Δ30 and 100% of rDEN2/4Δ30 vaccinated subjects (unpublished data). Interestingly, even subjects with low to undetectable levels of neutralizing antibody were not infected by a second dose of vaccine. Although these data are encouraging, subjects in endemic areas who received a dengue vaccine must be followed for several years to evaluate the long-term protective efficacy of the vaccine and the risk of the vaccine in increasing the susceptibility of vaccinated individuals to more severe disease over time. Through long-term follow-up, the need for booster doses of the vaccine can be determined.

The goal of evaluating each DENV serotype-specific candidate vaccine individually was to carefully characterize each candidate's safety, virology, and immunogenicity profile prior to inclusion in a tetravalent admixture. Those candidates with the most favorable safety and immunogenicity profiles were chosen for inclusion in a tetravalent admixture. We can now compare the safety, virology, and immunogenicity profile of each component in the tetravalent admixture with its safety, virology, and immunogenicity profile as a monovalent vaccine. It is not yet known whether or not we will observe differences in the immunogenicity of the monovalent components when administered in a tetravalent formulation, however using this approach, we hope to better understand how these viruses interact with one another when given together as a tetravalent formulation. We remain hopeful that our strategies for DENV vaccine development will result in a safe, immunogenic and protective dengue vaccine that will be available, affordable, and accessible in those regions where the need for such a vaccine is greatest.