Figure 1

Figure 1. Satellite view of distribution of patients with disseminated leishmaniasis (DL; black circles) and patients with mucosal leishmaniasis (ML; red circles) in Corte de Pedra, Brazil, 1999–2003. Vertical line divides the region...

A satellite view of Corte de Pedra with an arbitrary line dividing it into inner and coastal regions of approximately equal areas is shown in Figure 1. A total of 15 patients with DL were distributed in the coastal region and 15 patients with DL were distributed in the inner region (p>0.05, by χ² test). Patients with ML were rarely observed in the coastal area during the study period; 87% of patients with ML were observed in the inner region (p<0.01, by χ² test). To confirm the differences observed, we compared distribution of patients with ML and those with DL by using the Cuzick and Edwards test, which directly compares 2 sets of geographic events. Results were highly significant (p = 0.00005), which indicated that these 2 types of leishmaniasis spread differently throughout Corte de Pedra.

Because we had detected different subpopulations (clades) of L. braziliensis genotypes defined by random amplified polymorphic DNA analysis in this area (8), we also determined whether distributions of patients with ML and those with DL overlapped distributions of some of those clades. Distributions of patients with ML and those with DL overlapped exclusively clades C and A plus D, respectively (p>0.05, by Cuzick and Edwards test). All other comparisons showed significant differences in distributions (p<0.03). Thus, different types of ATL caused by the same parasite species are distributed differently, even within a specific ATL-endemic region. Overlap between specific L. braziliensis subpopulations and patients with ML or DL also suggests that this phenomenon may be influenced by distribution of parasite genotypes in the region. However, this proposal needs to be tested by using a method that is capable of accurately identifying genotypes for a large panel of isolated parasites.

Figure 2

Figure 2. Satellite view of distribution of patients with disseminated leishmaniasis (DL) limited to the skin, patients with mucosal leishmaniasis (ML), and patients with DL showing mucosal involvement (MDL) in Corte de Pedra,...

Observation of higher frequencies of ML in the inner region of Corte de Pedra suggested that extrinsic local factors not related to subpopulations of parasites might influence disease outcome. Because <40% of patients with DL also have mucosal involvement, we addressed this issue by comparing the distribution of patients with MDL with those with DL or ML. Patients with MDL showed a distribution pattern similar to that of patients with DL in this region (p = 0.8, by Cuzick and Edwards test; Figure 2) but different from that of patients with ML (p = 0.00003).

Figure 3

Figure 3. Satellite view of progressive spread of disseminated leishmaniasis in Corte de Pedra, Brazil, 1993–2002. Cumulative distributions of cases within affected areas are indicated by white circles. A) 1993, B) 1993–1996, C)...

Figure 4

Figure 4. Distribution of American tegumentary leishmaniasis (ATL) (red line) and incidence of disseminated leishmaniasis/total ATL cases (black line) in Corte de Pedra, Brazil, 1993–2003.

Figure 5

Figure 5. Linear regression comparing number of cases of disseminated leishmaniasis (past cases) diagnosed in the 12 months preceding a newly diagnosed case of DL (recent case) and distance to these recent cases,...

The broader distribution of DL compared with ML was surprising because only during the past decade has disseminated disease become more frequently diagnosed in Corte de Pedra. To better understand the dynamics of the spread of DL, we mapped the distribution of this disease in this region during 1993–2002. Our results show that up to 1996, DL, similar to ML, was concentrated mostly in the inner region of Corte de Pedra (Figure 3). However, the next 6 years showed progressive spread of DL to the coastal region until it reached an even distribution over the entire area (Figure 3). Two peaks in the incidence of DL paralleled the spread of DL in both regions of Corte de Pedra in the past decade (Figure 4). An increase was observed in the incidence of DL, with progressive spread into the inner region up until 1996, and another increase resulted in spread of this disease into the coastal region during 1998–2001. These 2 increases and the general increasing trend in incidence (Figure 4) also indicate that DL occurs in a pattern distinct from those of ML and CL. Linear regression (Figure 5) showed a significant inverse correlation between distance from a newly diagnosed case and frequency of diagnosis of DL patients in the preceding 12 months. These findings reinforce our previous suggestion that DL is an emerging disease (9) and that this form of leishmaniasis spreads in part through multiple outbreaks.

A distinct geographic clustering of disease forms secondary to L. braziliensis infection was found in the study region. The overlap of ML and DL with genotypes of parasites associated with such outcomes in this area (clades C and A plus B, respectively) (8) suggests a cause–effect relationship. However, small sample sizes used to represent subpopulations of L. braziliensis genotypes found in Corte de Pedra and the low significance level obtained in comparisons prompt further studies before we can conclude whether the intraspecies parasite polymorphism affects disease distribution over an affected area. Our findings suggest a complex organization of the types of ATL within foci of active disease transmission.

The combination of human population movement in the study region and the usual long latency period of ML suggest cautious interpretation of our data. For ML, months to years may elapse between parasite acquisition and development of mucosal lesions. Thus, residences of patients with ML at the time of diagnosis may not reflect actual distributions of patients at the time of infection with the parasites. However, the likelihood of this possibility was precluded because participants’ mean time of residence at their addresses at the time of ATL diagnosis was 17 years; mean times did not differ between the study groups.

Use of patients’ homes as primary sources of geographic coordinates was another limitation intrinsic to the retrospective study design. A more accurate approach would be to delineate the personal activity spaces of each patient with ML or DL and then perform the geographic analyses. However, estimation of personal activity spaces is based on recollection of data gathered by questionnaires administered to the study participants regarding past information such as places of residence, history of migrations, work-related activities, and sites for these activities. Such information would be more reliable in a prospective study. Nonetheless, taking into account that residents of Corte de Pedra are mostly engaged in agriculture, often conducted within walking distance from their homes, and that population migration within this region is limited, we expect collection of geographic coordinates based on place of residence to be a fair approximation of actual places of infection with Leishmania spp. in most cases. More specifically, in our sample, >90% of patients had lived on farms for more than a decade.

Wide geographic differences in distribution of types of ATL have been reported for Ecuador and Peru (6,7). Our findings extend this observation to a smaller geographic setting that involves foci of endemic parasite transmission to humans. Distinct distributions of ML and DL over an area of only 10,000 km² support the complexity of L. braziliensis reported for this region (8). We hypothesize that some subpopulations of the parasite may be associated with disease manifestations and with factors that affect the transmission dynamics of L. braziliensis strains, such as various sandfly vectors present in the study area (9,10). However, other nonhuman hosts and reservoirs may also play a role.

African trypanosomiasis and schistosomiasis can illustrate the effects of parasite and vector heterogeneities on geographic distribution of these diseases. In eastern and southern Africa, infections with Trypanosoma brucei rhodesiense are characterized by an acute form of sleeping sickness. In western and central Africa, infections with T. brucei gambiense are characterized by a chronic form of this disease (11,12). Regions in which Schistosoma hematobium and S. mansoni are endemic are affected by the presence of snails of the genera Biomphalaria and Bulinus, respectively, in infested bodies of water. Although S. hematobium, which causes urinary schistosomiasis, was the predominant species in Egypt up to the 1930s (13), S. mansoni, which causes hepatointestinal disease, has progressively replaced S. hematobium in the Nile Delta and more recently in Upper Egypt (14). This change was paralleled by the concomitant replacement of Bulinus truncatus snails by Biomphalaria alexandrina snails in the affected areas, largely caused by human intervention and modification of the ecology for irrigation purposes (15–17).

Our findings also show that DL, which is a novel severe and difficult-to-treat form of leishmaniasis (18), is rapidly emerging and spreading within Brazil, 1 of the 5 countries with 90% of human cases of tegumentary leishmaniasis worldwide (1). The unique pattern of DL incidence (Figure 4), which shows 2 peaks, indicates that this form of leishmaniasis may occur as outbreaks. Conversely, the increased frequency of DL in persons living near persons with recent cases of this disease may be caused by other factors, such as uneven human population distribution and vector densities in the area and other potential environmental factors that affect parasite reservoirs. Human-to-vector-to-human transmission of parasites may also play a role in this form of American leishmaniasis. However, anthroponotic cycles are not considered to be predominant in ATL, except for a few reports suggesting that this mode of transmission may occur with L. chagasi within large urban areas in northeastern Brazil (19–21).

The 2 increases in the incidence of DL preceded similar increases in the total number of ATL cases by ≈2 years. One possible explanation would be that factors affecting transmission of DL respond faster to changes in environmental parameters than those of CL and ML. The roles of climate and ecologic changes in leishmaniasis in regions near Corte de Pedra have been reported by Franke et al. (22). These authors reported a significant correlation between the southern oscillation of the El Niño phenomenon and the incidence of visceral leishmaniasis in the state of Bahia, Brazil. The major increase in the incidence of visceral leishmaniasis detected in that study occurred during 1995–1996, a period coincident with the first peak of DL shown in Figure 4. Although mechanisms responsible for the phenomena we describe remain elusive, we believe that information on clustering of disease types, increased frequency of DL among persons living near persons with recently diagnosed cases of the same disease, and predictive behavior of the incidence of DL relative to that of ATL may be used for better management and control of ATL.