Serum samples from patients with suspected DENV infection were referred to the Public Health Virology Laboratory, Queensland Health Forensic and Scientific Services, following the directive of Queensland Public Health medical officers, or were obtained through the public or private laboratory network. Acute-phase specimens underwent RT-PCR and serologic testing, and those that successfully yielded an RT-PCR product (after specific DENV serotype amplication) were sequenced and genotyped by phylogenetic analyses to assist public health investigations. This work was approved by the Ethics Committee of Queensland Health Forensic and Scientific Services.

RNA was extracted from 200 µL of serum, either manually (QIAamp viral RNA extraction kit; QIAGEN, Hildren, Germany) or by using the EZ1 Virus Mini Kit and (QIAGEN) according to the manufacturer’s instructions. Amplification was performed for each DENV serotype by using the Superscript III/Platinum Taq High Fidelity One-Step RT-PCR System (Invitrogen, Carlsbad, CA, USA) with specific RT-PCR primers (Table 1). Nucleotide sequencing of the complete envelope gene region (DENV-1, DENV-2, and DENV-4: 1,485 bp; DENV-3: 1,479 bp) was performed by using the Big Dye Terminator v3.1 cycle sequencing kit (Applied Biosystems, Foster City, CA, USA). Sequence data obtained were deposited in GenBank (Table 2).

Full-envelope protein sequences for each serotype were aligned by using the multiple alignment tool of MEGA5 (www.megasoftware.net). Unrooted trees were then constructed by using a maximum likelihood estimation with a Jukes-Cantor model and γ-distributed rates, and by constructing 1,000 replicates to generate bootstrap support values. Divergence time from a common ancestor was estimated by using the molecular clock calculator.

Previous reports (3,11) and anecdotal evidence indicated that there has been an increase in the number of dengue outbreaks occurring in Queensland. To investigate this apparent trend, we combined recent and historical outbreak data (3) over a 20-year period. A 5-year moving average does indeed show trends of increasing dengue outbreak incidence and increasing diversity of DENV serotypes that cause such outbreaks (Figure 1, panels A, B). A line of best fit revealed a significant increase with time (r² = 0.48; p<0.05 by Student t test, 2-tailed). All 4 DENV serotypes caused outbreaks; DENV-2 was the most common cause (50.0%), followed by DENV1 and DENV-3 (19.4% each) and DENV-4 (11.1%) (Figure 1, panel C). The increase in outbreak incidence reflects changes in international travel over this period, which has increased 3.5-fold since the early 1990s (12). This increase is consistent with increased importations of virus carried by viremic travelers and the recognized increase in DENV infections throughout the world (13). Also of note is the dramatic increase in infections caused by imported viruses in 2010 (Figure 1, panel D).

We ascertained the number and diversity of imported DENV serotypes from infected travelers during 2002–2010 (Table 3). This period was chosen because the most comprehensive patient sequence data were available. Information was analyzed from viremic travelers, for whom an RT-PCR amplification product and serotype designation could be obtained. The possible strain origins, which were determined after phylogenetic analyses, were compared with available travel histories to ascertain likely geographic origins of the infecting virus. The data were categorized into 4 separate regions: Asia, Papua New Guinea (PNG), the Pacific Islands, and countries outside of the Asia-Pacific region (non–Asia-Pacific).

Most infected travelers (77.9%) reported spending time abroad in Asia. In particular, 26.4% of all virus importations could be traced to Indonesia alone. All 4 DENV serotypes were detected in the specimens sequenced from persons with a travel history to that country. Notably, patients reported traveling to other Asian countries, including Thailand and the Philippines, where all 4 DENV serotypes have been found. Most other countries to which travel was reported had at least 3 DENV serotypes (Timor-Leste, PNG, and Vietnam), 2 serotypes (Cambodia, India, Fiji, Malaysia, and Laos), or 1 serotype (Brazil, Guyana, Samoa, Singapore, Solomon Islands, Sri Lanka, Tonga, and Vanuatu). A greater degree of diversity cannot be excluded in many of these countries because sampling numbers for individual countries were often low.

We calculated the proportion of the 4 DENV serotypes for infected travelers (Figure 2, panel A), which was slightly different from the proportion associated with outbreaks within Queensland (Figure 1, panel C). The serotype most commonly imported by travelers was DENV-1 (39.3%), followed by DENV-2 (25.7%), DENV-3 (21.4%), and DENV-4 (13.6%). Strains of all 4 DENV serotypes originated mainly in Asia (Figure 2, panel B). DENV-1 had the most diverse origins, with patients reporting travel mainly to Asia, but also to PNG, the Pacific Islands, and non–Asia-Pacific regions. In addition to Asia, DENV-2 was found to originate in PNG; DENV-3 originated in PNG and a non–Asia-Pacific area (Brazil); and DENV-4 originated in the Pacific Islands.

Infected travelers were either returning Queensland residents or international visitors. Returning residents were the largest proportion of patients (96.4%) who sought treatment from the health system with dengue viremia. Further analysis was conducted on the subset of returning residents (135 of 140 patients with imported cases). Similarly to the analysis of all travelers above, infected returning residents (Table 4) reported that they had most frequently returned from Asia (77.0%), followed by PNG (13.3%), then the Pacific Islands (8.1%), and least often from non–Asia-Pacific areas (1.5%). The overall ratio was significantly different from that expected on the bases of the proportion of all Queensland residents who reported returning from dengue-endemic countries of those 4 regions, as calculated by using data from the Australian Bureau of Statistics for 2002–2010 (χ² analysis, p = 0.0004). The proportion of patients reporting travel to Asia, and to PNG in particular, was higher than expected. The overrepresentation of cases from PNG is best explained by a recent increase in DENV activity in that country, which is consistent with a large number of importations from PNG (47.4%) during 2010 (Figure 1, panel D) and, in addition, other recent reports (14). In comparison, the proportion of patients who reported travel to the Pacific Islands was lower than expected. Because only 2 cases originated in the non–Asia-Pacific region, it was subsequently difficult to draw conclusions about this region for statistical purposes.

The genotypic mix for the various regions from which dengue was imported is shown in Table 3 and Figure 3, Figure 4, Figure 5, Figure 6. Across all regions, viruses could be classified into 1 of 2 genotypic groups within each serotype; the exception was DENV-1, which had 3 groups. In countries which were a source of imported viruses, generally 1 genotypic group for each serotype predominated.

DENV genotypic groups generally circulate in particular regions (15). The viruses imported into Queensland were consistent with DENV genotypes which had previously been reported to circulate in those countries to which patients had reported travel (16–19). For example, DENV-4 genotypic group II has been reported in Indonesia, Tahiti, the Caribbean Islands, and Central and South America (17). In 2007–9, DENV-4 was introduced into the Pacific Islands, displacing DENV-1 in the process (20,21). Our genotypic analysis confirms classification of the Pacific Island DENV-4 in genoptypic group II, as recently reported (21). This was the first time this genotypic group had been reported in the Pacific region, and suggested that the origin of this strain of DENV-4 may have been Southeast Asia.

In support of this suggestion, a closely related DENV-4 strain from the Torres Strait (Figure 6, JN575595) with 99.1% envelope nucleotide identity to a DENV-4 strain from Samoa (Figure 6, JN575592), was detected before the Pacific outbreak in 2005. A maximum likelihood test of the phylogenetic tree determined that a molecular clock was applicable (H_o not rejected; p = 0.06). Using a previously published substitution rate for dengue 4 of 1 × 10⁻³ substitutions/site/year (22), we calculated that divergence from a common ancestor occurred in ≈2002 with an error of ± 2 years. Thus, the Pacific Island outbreak strain (Figure 6, Pacific Island clade) is geographically and temporally closely related to the Torres Strait 2005 virus. Both of these virus strains are mostly closely related to DENV-4 strains which originated in Indonesia (Figure 6, JN575583). These data support suggestions the Pacific Island outbreak strain originated in Indonesia and made its way to the Torres Strait (Australia) in 2005, probably through PNG, and into the Pacific in 2007 where it is currently circulating. A virus most closely related to the Pacific Island strain was then imported into Innisfail in northern Queensland in 2009, where it caused an outbreak (Figure 6). This incident highlights the epidemic potential of DENV strains that are imported into Queensland (3,23,24).