The outbreak investigation protocol was approved by the Ethical Committee, Ministry of Health of Madagascar. The outbreak area contained 7 villages along a field path in the communes of Ambarakaraka and Anaborano, Ambilobe District, in northern Madagascar at an altitude <500 m. In January 2011, two brothers (boys) were working in a copper mine in Beramanja, Madagascar. On January 6, the boys returned home to Ankatakata (distance 50 km) (Figure 1). During the journey, fever, headache, and chills developed in 1 of the boys (age 13 years). Subsequently, he experienced severe chest pain, a deep cough, and hemoptysis, which are the clinical signs of pneumonic plague. He died at home on January 14 (Figure 2). It can be assumed retrospectively that this boy (index case-patient 1) had been infected with Y. pestis in Beramanja or during the journey home (Figures 1,2). On January 21, his mother, who had provided him with treatment until he died, also died (case-patient 2 in household A). Her husband (case-patient 3), daughter (case-patient 4), and granddaughter (case-patient 5) died on January 22. Symptoms of pneumonic plague had developed in these 3 persons (Figure 2; Table 1). These 3 persons were provided treatment by neighbors in household B (Figure 2).

On January 26, a second wave of human pneumonic plague cases was observed in 12 other persons (Figure 2; Table 1). During this second wave, household A (case-patients 6, 7, 12, 13, and 16) and household B (case-patients 8, 10, 11, 14, and 17) in Ankatakata were affected. Case-patient 9 was probably infected while visiting her sister-in-law (case-patient 2) in Ankatakata and transmitted the pathogen to her home (household C) in Antsakoabe (Figures 1,2). Case-patients 15 and 18 were also present in Ankatakata and transmitted the infection to Tsaratanana (household D) and Ankiganyo (household E). Case-patient 18 was responsible for a third small wave of pneumonic plague that started on January 31 and spread to Ambarakaraka (Figure 2; Table 2). This case-patient infected her brother (case-patient 19), who had carried his sister to a traditional healer (case-patient 20). Symptoms of pneumonic plague developed in all 3 persons; all 3 died (Figure 2; Table 1).

Three plague pandemics and numerous plague epidemics have been caused by Y. pestis in the past 1,500 years (2,4,7,12,24). During the second medieval pandemic, which also included the Black Death period, 50% (≈50 million persons) of the human population in Europe reportedly died of plague (2,4). At the beginning of the 20th century, 10,000 persons died during 2 plague epidemics in Manchuria, China (12). Although plague is still endemic to other countries, the reported numbers of plague patients has decreased to an average of 4,000/year since 1954. Also, there has been a slight decrease regarding the highest reported case number from countries in Asia and Africa (8,9).

Since the introduction of plague to Madagascar in 1898, the Institute Pasteur in Antananarivo was assigned to control this disease. The pneumonic plague outbreak in 2011 shows that despite introduction of education programs, a plan to investigate plague outbreaks promptly, and a trained task force, plague outbreaks cannot be prevented. However, in contrast to control of historical plague, when treatment with antimicrobial drugs was not available and the disease could spread unhindered, the present outbreak was stopped quickly after 27 days because of successful treatment with antimicrobial drugs (Figure 2). The good response of the patients to these drugs suggests that the Y. pestis strain that caused this outbreak was susceptible to streptomycin. However, streptomycin-resistant Y. pestis strains were isolated in Madagascar in 1995 and also during the outbreak in 2011 (25; M. Rajerison, pers. comm.). Therefore, drug-resistant Y. pestis strains may pose a new challenge to health authorities.

During the outbreak in 2011, persons shared single-room houses, lived in extended families, and closely cared for each other. Patients were not isolated. Because of social conventions, some persons trusted a traditional healer rather than physicians. This finding led to the third wave of the outbreak, which included 3 case-patients with pneumonic plague who died (Figure 2; Table 1).

WHO lists plague as a disease for which patients should be quarantined and requires that pneumonic plague patients are isolated from healthy persons (9). However, during the latency period before hemoptysis, sputum contains hardly any infectious organisms (11,26). Simple countermeasures, such as protective facial masks, are efficient in preventing transmission by droplets. Also, turning one’s head away from or turning one’s back toward a healthy person has a major prophylactic effect (26). This finding might explain why c26, who shared the same bed with case-patient 11 until his death, was not infected (Tables 1, 2). Thus, knowledge of the pathogenesis of Y. pestis in humans is essential for persons who live in plague-endemic countries.

It has been suggested that patients with bubonic plague and patients who have died of plague are not directly infectious to other humans (9,26). This suggestion is consistent with findings in the present study because contacts (c32–c41) who only attended the funerals did not show symptoms or seroconversion (Table 2; Figure 2).

Plague is endemic to Madagascar, especially in the central highlands (7). At an altitude >800 m, large numbers of rodent species and insectivores live in the rain forest. Those animals represent the classical natural focus for Y. pestis (7). The low-elevation seaport villages of Mahajanga and Antananarivo are exceptions to this altitude factor. Because of trade and stockpiling of grain and other food products, homophilous species, such as the black rat (R. rattus) and the Norwegian rat (R. norvegicus), play a major role in the urban lifecycle of plague (6,7,27). The present outbreak occurred at an altitude <500 m in a region that does not have much commercial or economic activity. Because of this finding, there was a low prevalence of small mammals in this area, and only a low number of rodents and insectivores were trapped during the epidemiologic investigation (Table 3). This factor resulted in the plague outbreak not being immediately recognized.

Although the outbreak Y. pestis strain could not be isolated, information was obtained by molecular analyses of human and animal samples. All samples contained the Madagascar-specific 1.ORI3-k genotype of Y. pestis, as previously reported (1,6). Results of CRISPR typing identified >1 genotype, which indicated that the outbreak area was a natural plague focus before the outbreak in 2011 (Table 4). This result is supported by an unusual high prevalence (12%, 8/69) of Y. pestis–positive animal samples (Table 3) compared with prevalences in previous studies (28–30). We suggest that Y. pestis strains containing the major CRISPR profile, which was found in 6 of 10 samples, was responsible for the present outbreak (Table 4). This CRISPR genotype has also been found in samples from the central highlands. The isolate from case-patient 13 lost the b5 spacer (Table 4). This phenomenon has been reported for other CRISPR profiles, and a different genotype has been assessed (21–23). A third Y. pestis CRISPR genotype was found in the R. rattus rodent R16 sample and in the human Bealanana 013 sample from 2010; this genotype includes the new element c3′–c12 of phage origin (Table 4) (21). The loss of 1 nt at the end of the spacer, as observed in in c3>c3′, has been previously reported (22).

Despite performing several PCRs, we could not amplify the complete Ypc locus in the R. rattus rodent R05 sample from Antanabe (Table 4), a finding that has been previously reported (21). We suggest that 4 CRISPR genotypes of Y. pestis were present in the outbreak area, which indicates that Ambilobe was a natural plague focus even before the outbreak. Unnoticed presence of pathogens near human populations requires higher surveillance activity, as recently reported (7). In contrast, it has been reported that hereditary resistance against Y. pestis might develop in rats (7,31).

One question that also needs to be addressed is why plague caused millions of deaths during devastating pandemics in the past while today plague is restricted to some geographic locations. One possible explanation would be the presence of additional virulence factors in historical plague strains, which have been lost from current Y. pestis strains.

To answer this question, 2 ancient Y. pestis genomes were sequenced and compared with sequences of current Y. pestis isolates. The ancient genomes were from a pandemic European Y. pestis pestis biovar Antiqua isolate (genotype 1.ANT) that originated during the Black Death period (3), and from a pandemic Y. pestis pestis biovar Antiqua isolate (genotype 0.ANT), which was isolated from a patient who died during the plague of Justinian in Germany (5). Sequences of current Y. pestis genomes used for comparison were from the nonhuman pathogenic Chinese Y. pestis microtus strain (91001: 0.PE4, biovar Xilingolensis) and from the Y. pestis pestis strain (CO92: 1.ORI1, biovar Orientalis) (3,5).

Analyses showed that genomes of ancient Y. pestis strains did not contain additional virulence genes that might explain higher virulence. (3,5). Furthermore, because plague epidemics were caused by different biovars or genotypes in Asia, Europe, and Africa (3,5,6,10,12–15), we assume that various Y. pestis subtypes are similar in virulence. Thus, factors other than differences in virulence might better explain the decrease in the reported plague cases. It is more likely that the general perception and understanding of infectious agents; improved hygiene; trade management; knowledge of plague pathogenesis in humans; vector control; specific outbreak management, including selective isolation of infectious patients; and chemoprophylaxis and treatment with antimicrobial drugs, have accounted for the decrease in plague cases (9,11,15,26).

In conclusion, the 1.ORI3-k genotype of Y. pestis identified in the present study has virulence comparable with that of ancestral genotypes that caused other epidemics. The course and outcome of a human plague epidemic depend on this virulence, as well as on education, public awareness, life style, infrastructure, isolation of patients, and medical care. However, there is an unpredictable pathogenic potential in drug-resistant strains that has not been estimated and needs to be studied.