From October 2003 to April 2004, DENV-3 circulated on the island, and 108 cases of dengue fever were serologically or virologically confirmed in both the Lepers Laboratory and the IMTSSA laboratory, and 12 persons were hospitalized (10). If one takes into account the 180 suspected cases observed during this period, the incidence of dengue fever during the epidemic was ≈0.62% (180/29,000) on the French part of the island; this value was consistent with values already observed in the Caribbean region (http://www.carec.org/data/dengue/1998/table2.html). No data were available from the Dutch half.

During this period, an additional 26 blood samples from patients with dengue-like syndromes at early stage of the infection were received in IMTSSA Tropical Virology Laboratory in Marseille. Among them, 13 were positive for dengue infection by presence of IgM, elevation (4-fold) of specific IgG, or both. Virus was isolated from 6 other patients, and DENV-3 was identified by IF using serotype-specific monoclonal antibodies. No fluorescence was observed when DENV-1, DENV-2, or DENV-4 monoclonal antibodies were used. Moreover, IF tests were negative when antibody to West Nile virus, Toscana virus, Bunyamwera virus, alphaviruses, phleboviruses and California serogroup viruses was used. These cases were considered confirmed dengue fever; all the patients recovered fully.

DENV-3 was genetically identified in the 6 IF positive isolates (D3StMart1–D3StMart6, referred to as GenBank accession no. AY750713–AY750718) by using BLAST-NCBI software. Comparison of partial sequences showed a high degree of identity between the strains isolated from patients on Saint Martin: paired identity at the nucleotide level ranged from 99.3% to 100%. When compared to 00PuertoR1 strain of the Latin American cluster defined by Messer (5), the values were 99.5%–99.6%. Nucleotide comparison gave the following values: 98.3%–98.4% for East African cluster (5), 98.8%–98.9% for Group B (5), and 96.8%–96.9% for Group A (5). This comparison indicated a close relationship between Saint Martin and Latin American isolates.

The strains from Saint Martin shared an 11-nucleotide insertion between position 10,275 and position 10,276 in the 3´UTR (AGTGAAAAAGA). The same insertion was found in isolates from Martinique 2 years earlier (6) and from DENV-3 Sri Lanka (6), indicating a probable common ancestor. This finding was consistent with a Sri Lankan origin for DENV-3 subtype III circulating in the Latin American and Caribbean regions (5) and with extension of the virus from a single importation event.

A phylogenetic tree was constructed based on partial sequences of the prM/M-E gene region (position 437–1144) of the genomes, including the strains from Saint Martin and several other previously characterized strains from different origins (16–19). DENV isolated from Saint Martin was grouped in subtype III, as defined by Lanciotti (17) (data not shown). Another phylogenetic tree was constructed with the strains analyzed by Messer (5) (Figure 3). Despite the high overall similarity of the subtype III sequences, distinct groups could be distinguished. The 6 isolates from Saint Martin clustered together. The small lineage difference of D3StMart2 is supported by a 95-bootstrap value, but no evidence suggests that such difference implies multiple introductions or a longer transmission period. However, Saint Martin DENV-3 fell in the Latin American cluster defined by Messer (5), which also includes isolates from Guatemala, Nicaragua, and Mexico. Isolates from Saint Martin were particularly close to Martinique isolates (6) and the 00PuertoR1 isolate, which were the geographically and temporally closest isolates of the cluster. The analysis of the deduced amino acid sequences generated a similar phylogram. Altogether, these results indicate that the Saint Martin viral isolate belongs to the genotype Puerto Rico 2000 previously reported (5). These results also confirm a common origin for all DENV-3 circulating in the Caribbean and Latin American region; their ancestor probably originated from Sri Lanka. Additional American DENV-3 strain sequences, encompassing the insertion site, should help to confirm this hypothesis.

The question arose whether the introduction from Sri Lanka was a single event in the Latin American region, first appearing in Panama (5), then spreading through the Caribbean subregion and South America. In fact, viruses belonging to subtype III were first reported in Nicaragua and Panama in 1994 (20). They were then identified in Guatemala from 1996 to 1998; in Puerto Rico, Barbados, and Jamaica in 1998 (18); and in Martinique in 1999 (6). The introduction of DENV-3 in Saint Martin was reported in 2000 (http://www.carec.org/annrep00/index.html), but the subtype was not determined. No virus isolation was reported for 3 years, until the viruses we characterized were isolated. The rapid extension of DENV-3 belonging to subtype III, which have almost completely replaced preexisting DENV of serotype 1 and 2, is indicative that these particular viruses have adapted to Caribbean and Latin American conditions. Together with previous studies (5,6), identification of viruses belonging to subtype III in another area of the Caribbean region confirms the remarkable epidemiologic success of this particular lineage of dengue viruses.

Comparable substitution of local DENV by a new genotype was previously described in the Caribbean region (12). New viruses, often originating from Southeast Asia, arise as successive waves, replacing previous ones. In this regard, sampling of DENV-3 subtype III virus nucleotide sequences from all countries in the Caribbean and the peri-Caribbean area need to be expanded so that we can understand DENV circulation in the region.