TRANSACTIONS OFTHE ROYAL SOCIETY OFTROPICAL MEDICINE AND HYGIENE (1997) 91,544-546

An outbreak of bartonellosis in Zamora Chinchipe Province in Ecuador

Philip Cooper1y2y4, Ronald Guderian I, Paulina Orellanal, Carlos Sandovall, Hector Olalla3, Macias Valdez3, Manuel Calvop%al, Angel Guevaral and George Griffin4 ‘Department of Clinical Investigations, Hos- pital Vozandes, Quito, Ecuador; 2Laboratory of Parasitic Diseases, National Institutes of Health, Bethesda, Maryland, USA; 3Hospital Cantonal de Zumba, Zumba, Ecuador;4Divi>ion of Infectibm Diseases, St George’s Hospital Medical School, Lon- don, UK

Abstract We report an outbreak of human bartonellosis in Zamora Chinchipe Province in Ecuador, which occurred in 1995-l 996. Nineteen cases were seen, of which 18 presented with classical oroya fever (fever and pro- found anaemia) and one with verruga peruana; 11 of the cases (58%) had positive blood films containing Bartonella baciUijknis.The houses of cases and neighbouring controls were visited; blood samples for thin films and cultures were collected born members of each house and a questionnaire was administered to investigate possible risk factors for disease transmission. In none of those sampled wasB. bacilliformis bac- teriologically demonstrable. All case houses were located in isolated areas at the margin of forest and the presence of dead rodents was reported only in case houses (BO.05). We suggest that human bartonellosis is a zoonosis with a natural rodent reservoir and that migrant humans infected in this way may become a temporary reservoir host in populated areas.

Keywords: bartonellosis, BartoneZZu bacilliformis, animal reservoir, Ecuador

Introduction Human bartonellosis is caused by infection with Bar-

tonella bacillijknis, a Gram-negative pleomorphic coc- cobacillus which is transmitted to humans through the bite of sandflies of the genus Lutzomyia. The disease is endemic at altitudes of 1000 to 3000 m in inter-Andean valleys in Peru and Ecuador (HE-R, 1990).

Infection with B. bacilltjixmis may cause an acute fe- brile illness, oroya fever, which may be fatal and which is characterized by severe haemolysis and secondary in- fections, particularly with Salmonella spp. (CUADRA, 1956). Immigrants to endemic areas suffer more severe disease (FERRER, 1990), while the indigenous popula- tion tends to suffer a milder illness or remain asympto- matic. Untreated disease mav be followed bv the development of haemangiomat&s verrugas, verruga pe- ruana, after a period of weeks to months.

Bartonellosis in Ecuador is a sporadic infection and cases have been reported from several provinces (COOP- ER et al., 1996). Outbreaks have been reported in Zamo- ra Chinchipe Province in communities living in the Provincial District of Zumba since 1938 (ALVAR4DO COBO, 1942; BONILLA, 1944; &ON & LEON, 1986). The largest recent outbreak was seen over the period 1979 to 1980 when an estimated 200 cases were seen (V. Reyes, unpublished report). The last reported out- break was in 1984 when an estimated 13 cases and 3 deaths were reported (E. Beltran, personal communica- tion1 . From 1984 to 1995. a total of 17 cases was renort- ed (COOPER et al., 1996) and the disease rem&ed sporadic, with isolated cases being reported from several communities.

We report here a recent cluster of cases of bartonello- sis from several communities in the Provincial District of Zumba and describe the results of an investigation into the possible factors which may have contributed to this small outbreak.

Materials and Methods Study population

All cases were identified (by H.O. and M.V.) at the Hospital Cantonal de Zumba (HCZ) in Ecuador. The cases came fiorn 12 communities in the Provincial Dis- trict of Zumba (El Chito, Romerillos, Bellavista, Sala- paca, San Antonio, Palanuma, Las Sidras, Irachi Palanda, Palanda, Muyuche, Jesus de1 Gran Poder, and

Address for correspondence: Philip Cooper, Laboratory of Parasitic Diseases, Building 4, Room 126, National Institutes of Health, Bethesda, Maryland 20892, USA.

Tolisos). These communities, many consisting of only 3-4 houses, are situated in a sparsely populated area of low mountainous tropical forest at altitudes of 800 to 2000 m. All members of each community are engaged in agriculture (principally of coffee, maize, and yucca) and most keep a large number of domestic animals, including dogs, cats, chickens, guinea-pigs, and pigs, many of which had direct access into the houses.

The house of each case was visited and a question- naire was administered to the head of each household. Venous blood samples were taken from all members of each household and examined by thin blood film prep- aration and culture. For each case, the 2 nearest houses were selected as controls, which were questioned and sampled in the same way.

Sample processing Standard methods were used for staining with Giem-

sa’s stain and examination of thin blood films. Culture medium (Colombia agar, Difco) was prepared accord- ing to the manufacturer’s instructions and autoclaved; 10% heparinized human whole blood was added and the medium left to set in sterile plastic Petri dishes. Ap- proximately 50 & of patient’s whole blood was plated on individual dishes and the cultures were incubated at 28°C and examined weekly for the presence of bacterial colonies.

Questionnaire A questionnaire to assess risk factors for disease trans-

mission and document the clinical symptoms experi- enced by each member of each household was administered to the head of case and control house- holds. The questionnaire was modified from one used previously in studies of human bartonellosis (COOPER et al., 1996). Briefly, the head of household was ques- tioned regarding the following: place of birth, deaths in household over the previous 12 months, type of con- struction of house, water source, use of bed nets, wheth- er the household had been sprayed recently with insecticides, presence of domestic animals inside and outside the house, presence of rodents in the house, presence of dead rodents over the past year, whether any domestic animal had been sick or had died over the past year, clinical symptoms over the past 6 months of each household member (e.g. fever, pallor, jaundice, cough, diarrhoea, and verrugas), any antibiotic received, and the outcome of treatment. The head of household was also presented with a set of numbered photographs which included verruga peruana and the main differen- tial diagnoses (leishmanial ulcers, molluscum contagio-

BARTONELLOSISINECUADOR 545

sum, common warts, yaws, chicken-pox, and tropical ulcer) and asked if any member of the family had suf- fered from any of these lesions over the past 6 months.

Statistical analysis Proportions were compared using the x2 test or Fish-

er’s exact test as appropriate.

Results A total of 19 cases of bartonellosis was identified at

HCZ over the period September 1995 to July 1996. The median age of cases was 8 years (range 2-57 years) and most (74%) were males. A diagnosis of bartonellosis was confirmed microscopically in 11, a further 8 were reported as microscopically negative on a single blood film but a clinical diagnosis of acute bartonellosis was made based on a highly suspicious clinical presentation, and one was diagnosed with verruga peruana (micro- scopically negative for B. bacilliformis). Blood cultures were not-prepared from the cases. The median haemat- ocrit was 4.7 e/dL (range 2.3-l 1.3 g/dL). Clinical fea- tures in orde:of frequ&cy were: t&er’(95%), pallor (84%), jaundice (47%), oedema (26%), hepato- splenomegaly (2 1%)) generalized lymphadenopathy (1 l%), petechiae (5%), gastrointestinal bleeding (5%), and verruga peruana (5%). Two patients with acute bar- tonellosis died despite the start of treatment with intra- venous chloramphenicol and 16 patients responded successfully to adequate doses of oral or intravenous chloramphenicol. One case (who had had a splenecto- my following trauma in 1991) failed to clear B. bacilli- fomzis from the blood during high-dose chloramphenicol treatment but responded successfully to a 7 d course of doxycycline (100 mg oral daily).

The study was performed in October 1996, approxi- mately 3 months after the last reported case. A iota1 of 48 households was visited 116 case households and 32 controls) and 253 subjects were bled for blood films and cultures (60 members of case households and 193 con- trols), all of which were negative. All case houses were located at least 05 km from other dwellings in isolated areas of cleared forest. No 2 cases occurred in the same household.

From the questionnaire, the frequencies of sick or dead animals with access to the houses were as follows: dogs (69% in case houses vs. 75% in control houses). cats (51% vs. 41%), guinea-pigs (69% vs. 59%), chick: en (88% vs. 78%), and pigs (56% vs. 44%). The fre- quencies of sick or dead animals were: dogs (25% in case houses vs. 31% in control houses), cats (none vs. 19%), guinea-pigs (13% vs. 38%), chickens (63% vs. 69%), and pigs (6% vs. 16%). The only risk factor which was increased in case households compared to controls was the presence of peridomiciliary dead ro- dents (RO.05). A large proportion of the inhabitants of both case and control houses reported illness (fever, pal- lor, diarrhoea, etc.) within the past 6 months (data not shown). The use of antibiotics over the past 6 months was reported by 23% of members of case households and 21% of controls. One case of verruga peruana was identified in a control household; the onset was within 2 months of the diagnosis of the case of acute bartonello- sis in the corresponding case house.

Discussion We have previously reported that human bartonello-

sis is sporadic in this region of Ecuador (COOPER et al., 1996) and attributed this change in endemicity to the widespread use of residual insecticides and the easy availabilitv of antibiotics (aenerallv chloramohenicol) in even the smallest commumties. There are several possi- ble reasons for the appearance of this outbreak or cluster of sporadic cases.

Firstly, in the adjacent Peruvian focus of human bar- tonellosis across the Ecuador-Peru frontier in San Igna- cio Province, the number of cases of bartonellosis

reaches a peak every 10 to 15 years (G. Huatuco, per- sonal communication): this pattern of disease may also be true in Zamora Chinchipe Province. The number of cases renorted between 1983 and 1996 reached neaks in 1984-1985and1995-1996.

Secondly, there was an increased diagnostic aware- ness in HCZ, as 2 laboratory technicians working at the hosnital were trained at the Denartment of Clinical In- ve&gations at Hospital Vozandes in Quito in the recog- nition of B. baczllifrmis in blood films in June 1995, 2 months before the identification of the first case, and the apparent outbreak may merely reflect improved recog- nition. We think this is unlikely to account for the in- crease in the number of cases, although it is likely to have increased clinical suspicion of bartonellosis among both clinicians and laboratory technicians.

Further, it has been suggested that bartonellosis in Zamora Chinchipe Province is not autochthonous, but rather is imported by infected migrant workers from the adjacent Peruvian Province of San Ignacio, and there- fore increased disease activity in Canton Chinchipe may reflect disease activity in Peru. However, none of the cases had visited Peru in the past 2 years and the frontier had been closed since the border conflict of 1994.

Finally, the increase in number of cases of acute bar- tonellosis may reflect an increase in the reservoir of non- immune persons: the head of most case households (63%) was not born in the Provincial District of Zumba in Zamora Chinchipe Province and had moved into the area within the previous 10 years (data not shown). This might explain why almost all cases (95%) presented with severe acute bartonellosis, rather than verruga pe- ruana which is more typical of the indigenous popula- tion in endemic areas (HERRER, 1990; COOPER et al., 1996).

The source of this outbreak is unclear. Most workers believe that humans are the reservoir of B. bacillifrmis infection (HIXRER, 1990) and asymptomatic infection may be present in up to 10% of individuals living in en- demic areas (HERRER,1990).However,someobserva- tions in humans and animals suggest that there may be a natural animal reservoir.

Humans may become infected in uninhabited areas (T0~~~~~~,1913;H~~~~~,1990)andthisis support- ed by our findings in this study: none of the cases was infected in populated communities but rather in isolated houses at the margin of forest. The presence of asymp- tomatic human infection is seasonal and disappears dur- ing the ‘dry’ season (FERRER, 1953). In a previous study (COOPER et al., 1996), and also in this investiga- tion, we were unable to detect a single individual with asymptomatic infection either among schoolchildren, who are most affected by acute bartonellosis and verru- ga peruana, or in family members of case or control households. Both these observations argue strongly against the maintenance of a reservoir of infection in hu- mans.

The evidence for a natural animal reservoir is more compelling. An increase in the rodent population was reported before a previous epidemic in Peru (GRAY et al., 1990) and, in this study, the presence of dead rats was significantly more frequent in case houses. Bar- tonellosis can be induced inexperimental animals by in- oculation of B. bactiiformis and the ranee of suscentible animals includes monkeys, dogs, donkiys, rabbits, and guinea-pigs (HERRBR, 1990). Further, the inhabitants of endemic areas have reported that domestic animals are susceptible to bartonellosis and may develop verru- ga-like nodules (f%IANNON, 1929; GAMARRA, 1964). In a previous investigation of an outbreak in Peru, the pres- ence of animals inside the house was associated with an increased risk of acquiring infection (GRAY et al., 1990). In our study, animals were present in the houses of a high proportion of both cases and controls (15/16 of case houses and 32/32 of controls) (data not shown). We have previously reported an association between the

development of verruga peruana and the presence of sick or dead chickens in the same household (COOPER et al., 1996). In this study, at the time of the appearance of the first few cases (i.e., August-September 1995), an epidemic of a rapidly fatal chicken illness occurred which was characterized by fever, diarrhoea, and dysp- noea. The few avian survivors developed blood-contain- ing ‘verrugas’ which were most noticeable on the head. The epidemic spread slowly between communities, and most case households (63%) reported its appearance before the development of disease in the human case.

We propose that human bartonellosis is a zoonosis with wild rodents as the natural reservoir. Recently, an- other member of the same genus, Z3. henselue, the caus- ative agent of cat-scratch disease, has been identified as a natural infection of domestic cats (KOEHLER et al., 1994) and other Bartonella spp. usually found in small rodents have been implicated in human disease (PUNTARK et al., 1994). Domestic animals, particularly chickens, may be susceptible to sandfly transmission of B. bacilliforks infection and may act as a secondary res- ervoir. This outbreak may have followed increased ro- dent invasion of houses due to food shortages because of the drought. Epidemic disease may occur when in- fected inhabitants move to more populated areas where vector species of sandflies are also present and establish a human-to-human transmission cycle: in endemic are- as in both Ecuador and Peru, many town-dwellers also have farms in the nearby hills where they work and oc- casionally stay. Outbreaks in populated areas in Peru have become increasingly infrequent and this is proba- bly largely due to the widespread use of residual insecti- cides in the control of malaria (HBRRBR, 1990).

Acknowledgements P.C. was supported by a Wellcome Trust Research Travel

Grant.

References Alvarado Cobo, A. (1942). Bartonellosis en el Ecuador. Retitu

1 Announcements (

PHILIP COOPER ET/X..

de la Asociacion Medica de Cuenca, 3, 15-25. Bonilla, A. (1944). Verruga peruana o enfermedad de carrion.

Revista de1 Centro de Estudiantes de Meduicina ‘Eugenio Espe- jo’, Quito, 2,45-V.

Cooper, P., Guderian, R., Paredes, W., Daniels, R., Perera, D., Espinel, M., Valdez, M. & Griffin, G. (1996). Bartonellosis in Zamora Chinchipe province in Ecuador. Transactions ofthe Royal Society of Tropical Medicine and Hygiene, 90,241-243.

Cuadra, M. (1956). Salmonellosis complication of human bar- tonellosis. lkas Reports in Biology and Medicine, 14,97-l 13.

Gamarra, A. (1964). Geografia medica de la verruga peruana. Verruga peruana en la zona fronteriza de Canchis-Chin- chipe. Revtita Sanidad de Policia, 24, 50-53.

Gray, G. C., Johnson, A. A., Thornton, S. A., Smith, W. A., Knobloch, J., Kelley, P W., Escudero, L. O., Huayda, M. A. & Wignall, F. S. (1990). An epidemic of Oroya fever in the Peruvian Andes. American Journal of Tropical Medicine and Hygiene, 42,2 15-22 1.

Herrer, A. (1953). Carrion’s disease. II. Presence of Bartonella baciUijormti in the peripheral blood of patients with the be- nign form. American Journal of Tropical Medicine and Hygiene, 2,645-649.

Herrer, A. (1990). Eptdemiologia de la Verrugu Peruana. Lima: Pribaceb.

Koehler, J. E., Glaser, C. A. &Tappero, J. W. (1994). Rochali- maea henselue infection. A new zoonosis with the domestic cat as reservoir. Journal of the American Medical Association, 271,531-535.

Leon, L. A. & Leon, R. (1986). Bartonelkxis o Enfermedad de Carrion en la Republica de1 Ecuador. Enfermedades Nativas No. 1. Quito: Institnto de Investigaciones, Facultad de Ciencias Medicas, Universidad Central de1 Ecuador.

Punatric,V., Borcic, D. & Bejuk, D. (1994). Haemottopic bac- teria in man. Lancet, 343,356-360.

Shannon, R. C. (1929). Entomological investigations in con- nection with Carrion’s disease. American Journal of Hygiene, 10,78.

Townsend, C. H.T. (1913).The possible and probable etiology and transmission of verruga fever. Journal of Economic Ento- mology, 6,21 l-225.

Received ZbJanuary 1997; revised 11 March 1997; accept- edforpublication 11 March 1997

Second International Conference on Mycobacterium ulceram University of Melbourne, Australia: l-3 April 1998

The conference will mark the 50th anniversary of the publication by MacCallumet al. describing the disease and the bacterium; the paper originated from the University and the Alfred Hospital, Melbourne.

The conference will be followed by a two days’ bus tour of the two most important endemic areas invictoria, incidentally passing through some of the most beautiful scenery in southern Australia.

Further information can be obtained from A/Professor John Hayman, Department of Pathology, Box Hill Hospital, Box Hill,VIC 3128,Australia; phone +61 (0) 3 9895 3146, fax +61 (0) 3 9899 6132.

Hepatology, Gastroenterology and Infectious Diseases Cairo, Egypt: 16-l 9 December 1997

Joint Meeting with the Royal Society of TropicaZ Medicine and Hygiene

An International Congress organized by the Egyptian Society of Hepatology, Gastroenterology and Infectious Diseases.

Further information can be obtained from Professor Kabil S. M., Cairo GIT and Liver Centre, 41 Talaat Harb Street, Central Cairo, Egypt; phone +20 2 3915 115, fax +20 2 3938723.