34 Brucella, bartonella and streptobacillus
Brucellosis; Oroya fever; trench fever; cat scratch disease; bacillary angiomatosis; rat bite fever
Key points
• Brucellae are highly infectious coccobacilli that cause a septicaemic illness, undulant fever. Most human disease is caused by Brucella melitensis, B. abortus or B. suis.
• The disease is a typical zoonosis most commonly acquired from infected animals, or from infected meat or dairy products.
• Brucellosis is diagnosed by isolation of the organism from blood; alternatively serology or polymerase chain reaction tests can be used.
• Brucellosis is treated with a tetracycline, usually in combination with an aminoglycoside or rifampicin.
• Bartonella bacilliformis is a highly infectious agent causing the sandfly-disseminated diseases, Oroya fever (Carrion’s disease) and verruga peruana in parts of South America.
• The organism infects blood cells and can be diagnosed in stained blood or tissue aspirates. Alternatively, PCR methods are used.
• Other bartonellae cause trench fever and cat scratch disease. Endocarditis can complicate these infections.
• Chloramphenicol, macrolides, aminoglycosides, fluoroquinolones and tetracyclines are used in the treatment of bartonella infections.
• Streptobacillus moniliformis is one of the causes of a septicaemic illness, rat bite fever. Treatment with penicillin or a tetracycline is usually effective.
The genus Brucella comprises a group of Gram-negative coccobacilli that can infect a wide range of mammals ranging from rodents to killer whales. They are of particular zoonotic and economic importance as a cause of highly transmissible disease in cattle, sheep, goats and pigs. Infection in pregnant animals often leads to abortion, and involvement of the mammary glands may cause the organisms to be excreted in milk for months or even years. Human infections arise through direct contact with infected animals, including handling of infected carcasses; indirectly from a contaminated environment; or through consumption of infected dairy produce or meat.
Brucellosis is a typical zoonosis, and person-to-person infection does not play a significant role in transmission. Infection may remain latent or subclinical, or it may give rise to symptoms of varying intensity and duration. Brucellosis can present as an acute or subacute pyrexial illness that may persist for months or develop into a focal infection that can involve almost any organ system. The characteristic intermittent waves of increased temperature that gave the name undulant fever to the human disease are now usually seen only in long-standing untreated cases.
The Brucella genus comprises a group of closely related bacteria that probably represent variants of a single species. For convenience these have been classified into nomen species that differ from one another in their preferred animal host, genetic arrangement, phage sensitivity pattern, and oxidation of certain amino acids and carbohydrates. The main human pathogens are Brucella abortus, B. melitensis, B. suis and B. canis. The first three may be further subdivided into biovars associated with various animal hosts. B. abortus has a preference for cattle and other Bovidae, B. melitensis for sheep and goats. The first three biovars of B. suis preferentially infect pigs, whereas the fourth and fifth biovars have reindeer or caribou and rodents, respectively, as natural hosts. The biovars differ in their sensitivity to dyes, in production of hydrogen sulphide and in agglutination by sera monospecific for A and M epitopes. Various molecular typing methods are also used to differentiate subtypes down to the level of individual strains.
B. cetaceae and B. pinnipediae, isolated from dolphins, porpoises, killer whales and seals, appear to be pathogenic for man. Newly designated species include B. microti from voles and B. inopinata isolated from a breast implant recipient with a brucellosis-like syndrome.
Brucellae are Gram-negative coccobacilli or short bacilli, occurring singly, in groups or short chains. They are non-motile, non-capsulate and non-sporing.
Brucella spp. are aerobic. However, B. ovis and many strains of B. abortus, when first cultured, are unable to grow without the addition of 5–10% carbon dioxide. All strains grow best at 37°C in a medium enriched with animal serum and glucose.
On clear solid medium, smooth, transparent and glistening (‘honey droplet’) colonies appear after several days. However, the organisms can mutate, especially in liquid media, forming ‘rough’ colonies on subculture. There is a corresponding loss in virulence and an antigenic change, so that they are no longer readily agglutinated by homologous antisera prepared against normal smooth strains. Identification can be made by the polymerase chain reaction (PCR) with appropriate primers or by a combination of biochemical, cultural, phage typing and serological tests. Rapid gallery tests may misidentify Brucella spp. and this has resulted in laboratory-acquired infections; they are not recommended.
Brucellae may be killed at a temperature of 60°C for 10 min, but dense suspensions, such as laboratory cultures, can require more drastic heat treatment to ensure their inactivation. Infected milk is rendered safe by efficient pasteurization. Brucellae are very sensitive to direct sunlight and moderately sensitive to acid, so that they tend to die out in sour milk and in cheese that has undergone lactic acid fermentation. The organisms can survive in soil, manure and dust for weeks or months, and remain viable in dead fetal material for even longer. They have been isolated from butter, cheese and ice cream prepared from infected milk. They may survive in carcass meat, pork and ham for several weeks under refrigeration. Pickling and smoking reduce survival. They are susceptible to common disinfectants if used at appropriate concentration and temperature. They are sensitive in vitro to a wide range of antibiotics, only a few of which are effective therapeutically.
In all smooth strains the dominant surface antigen is a lipopolysaccharide (LPS) O chain, which, depending on the three-dimensional structure of the polysaccharide portion, forms A, M or C epitopes. These are common to all smooth species, but the distribution of A and M depends on biovar. Rough strains do not produce the O chain but have a common R epitope. The LPS has endotoxin activity of relatively low pyrogenicity and elicits limited antibody-mediated protection. The organisms behave as ‘stealth pathogens’ and evade innate immune responses; effective immunity is dependent on specific cell-mediated and cytotoxic responses elicited by a variety of protein antigens.
The incubation period is usually about 10–30 days, although infection may persist for several months before causing any symptoms. B. melitensis and B. suis tend to cause more severe disease than B. abortus or B. canis. Infection by any species may give rise to a variety of non-specific symptoms and, without the fluctuating temperature to act as a guide, diagnosis may be difficult. Pointers to the diagnosis are a history of occupational exposure or recent travel to endemic areas with consumption of milk products.
Brucellae can enter the body through skin abrasions, through mucosal surfaces of the alimentary or respiratory tracts, and sometimes through the conjunctivae, to reach the bloodstream by way of regional lymphatics. The organisms are facultative intracellular parasites and subsequently localize in various parts of the reticulo-endothelial system with the formation of abscesses or granulomatous lesions, resulting in complications that may involve any part of the body. Brucellae surviving within cells may cause relapses of acute disease, or a chronic syndrome may develop that is associated with continued illness and vague symptoms of malaise, low-grade fever, lassitude, insomnia, irritability and joint pain. Such ‘chronic brucellosis’ may follow an acute attack or develop insidiously over several years without previous acute manifestations. It rarely responds to antibiotic therapy and is probably a post-infectious response similar to the ‘myalgic encephalomyelitis syndrome’.
Brucellosis is confirmed in man by isolating the organisms from blood or other tissue samples and by serological and other tests. In animals, culture may be attempted from abortion material, placenta, milk, semen or samples of lymphoid tissue, mammary gland, uterus or testis collected post mortem.
Brucellae are easily transmitted by aerosols, ingestion and percutaneous inoculation. Samples suspected to contain brucellae must be treated as high risk. Cultures must be handled under containment conditions appropriate to hazard group 3 pathogens.
When brucellosis is suspected, blood culture should be attempted repeatedly, not only during the febrile phase. Because the organisms may be scanty, at least 10 mL of blood should be withdrawn on each occasion, 5 mL being added to each of two blood culture bottles containing glucose–serum broth. One of these bottles should be incubated in an atmosphere containing 10% carbon dioxide. Preliminary lysis and centrifugation of the blood improves the isolation rate. Other materials such as bone marrow, solid tissue samples or exudates are also suitable for culture.
Subculture should be made on to serum–dextrose agar every few days; alternatively, a two-phase Castaneda culture system, in which the broth is periodically allowed to flow over agar contained within the blood culture bottle, may be used. Blood cultures should be retained for 6–8 weeks before being discarded as negative. Automated blood culture systems may also be used.
B. melitensis and B. suis are more frequently isolated from blood than are B. abortus or B. canis.
In the absence of positive cultures, the diagnosis of brucellosis usually depends on serological tests, the results of which tend to vary with the stage of the infection (Table 34.1).
In some rural communities the sera from a proportion of the normal population are reactive in low dilutions in serological tests because of previous subclinical infection.
This test usually becomes positive 7–10 days after onset of symptoms. During the acute stage of the disease, levels of agglutinins associated with both immunoglobulin (Ig) M and IgG continue to rise. As high-titre sera may not cause agglutination in low dilution (the prozone effect), a range of serum dilutions from 1 in 10 to greater than 1 in 1000 should be made.
As the disease progresses from the acute to the chronic phase and the organisms become localized intracellularly in various parts of the body, the IgM antibodies decrease; the agglutination titre falls and may become undetectable even while the patient is still ill. The absence of agglutination therefore does not rule out the possibility of infection. Persisting antibodies that are no longer capable of agglutinating may be detected by complement fixation, antiglobulin or enzyme-linked immunosorbent assay (ELISA) tests. In latent or chronic infection, the complement fixation test is likely to be positive, whereas in cases of past infection it is negative.
Low-titre agglutinins due to residual IgM may persist for months or even years after the infection has cleared. Testing in the presence of mercaptoethanol, which destroys the agglutinating ability of IgM, was formerly used to indicate the continuing presence of IgG and the likelihood of persisting infection, but is no longer considered reliable.
The ELISA for IgG and IgA antibodies shows a good correlation with active disease, especially in long-standing infection. It has largely replaced the anti-human globulin (Coombs’) test, formerly used for detecting non-agglutinating (IgG) antibodies.
The O chain of smooth Brucella LPS is structurally related to the LPS antigens of various other Gram-negative bacteria. False-positive cross-reactions in agglutination, complement fixation and ELISA tests are produced by antibodies to Yersinia enterocolitica O9, Salmonella O30, Escherichia coli O157, Francisella tularensis, Stenotrophomonas maltophilia and Vibrio cholerae. Tests using protein antigens are more specific but may cross react with Ochrobactrum spp. – environmental bacteria closely related to Brucella that are occasionally implicated in opportunistic infections.
PCR methods can detect Brucella specifically and also give an indication of species and biovar. Promising results have been obtained in clinical studies, but no standard procedure is yet available.
The Rose Bengal plate test, a rapid slide agglutination test with a buffered stained antigen, is widely used as a screening test in farm animals, but also gives good results in human brucellosis. It is not affected by prozones or immunoglobulin switching. Positive results should be confirmed by a quantitative method.
The brucellin skin test, similar to the tuberculin test (see p. 216), does not differentiate active from past or subclinical infection and is no longer recommended.
Brucella infections respond to a combination of streptomycin or gentamicin with tetracycline, or to rifampicin and doxycycline. Tetracycline alone is often adequate in mild cases. Fluoroquinolones may be used in combination with rifampicin or tetracyclines but are not recommended for monotherapy. Treatment should be continued for at least 6 weeks. Co-trimoxazole and rifampicin can be used in children. In patients with endocarditis and neurobrucellosis, a combination of a tetracycline, aminoglycoside and rifampicin is recommended.
Serum antibody titres usually decline sharply after effective treatment. The chronic post-infectious form without localizing lesions responds poorly to treatment.
B. abortus has been eradicated from cattle in most developed countries, although there has been a resurgence in some parts of Europe. It was formerly common in dairy farmers, veterinarians and abattoir workers, but is now rare. Nearly all human cases in the UK are now acquired abroad; most are caused by B. melitensis, which is still prevalent in Mediterranean countries, the Middle East, central and southern Asia, and parts of Africa and South America.
Human brucellosis due to B. suis is largely an occupational disease arising from contact with infected pigs or pig meat. It was once common in the USA, chiefly among those who handled raw meat shortly after slaughter. It occurs in feral pigs in Australia and the USA, and is a hazard to hunters. It is widespread in domesticated pigs in various African, Asian and South American countries; biovar 4 is found only in the Arctic regions of North America and Russia.
Brucellae have potential as agents of biological warfare or bioterrorism and this possibility should be borne in mind in the event of unexplained outbreaks.
The live-attenuated B. abortus strain S19 vaccine has been used to protect cattle from abortion and so reduce spread of the disease. It can interfere with subsequent diagnostic serology and has been widely replaced by the rough strain B. abortus RB51, which may give comparable protection, but does not induce interfering antibodies and is less hazardous to man, though not innocuous.
The live-attenuated smooth strain B. melitensis Rev I is used to protect sheep and goats from B. melitensis infection. Vaccination of pigs is not widely practised, although the attenuated B. suis strain 2 has been used in China.
Effective and non-reactogenic vaccines are not currently available for human vaccination. Pasteurization eliminates the risk of brucellosis from the consumption of infected milk or milk products. However, there remains the possibility of infection due to contact with infected animals or their tissues. Veterinary surgeons, farmers and laboratory workers are particularly at risk.
Eradication depends on elimination of the infection from domestic animals by a policy of compulsory testing of the animals and slaughtering of positive reactors.
The genus Bartonella, which is distantly related to Brucella, comprises at least 20 species of very small Gram-negative bacilli, most of which have been implicated in various febrile and localizing diseases in man.
• Bartonella bacilliformis is the cause of Oroya fever or Carrion’s disease and verruga peruana. It is spread by sandflies.
• Bart. quintana is the cause of louse-borne trench fever.
• Bart. henselae and Bart. clarridgeiae are the most common causes of cat scratch disease and can be transmitted by fleas and possibly ticks.
Other Bartonella species have been identified as pathogens of dogs and other mammals. Some, including Bart. vinsoni (and its various subspecies), Bart. elisabethae, Bart. alsatica, Bart. koehlerae and Bart. mayotimonensis occasionally cause a range of syndromes involving many organ systems in man including bacteraemia and endocarditis. Bart. grahami has been implicated in ocular disease.
Bart. bacilliformis is responsible for outbreaks of a severe and often fatal disease of man in the mountainous regions of Peru, Colombia and Ecuador. The name Oroya fever was given after an epidemic of the disease in 1870 during the building of a railway between Lima and Oroya, when 7000 labourers died within a few weeks. The infection is spread by sandflies, usually Lutzomyia verrucarum and L. peruensis.
After recovery from Oroya fever the patient may develop a skin eruption known as verruga peruana. Individuals may remain bacteraemic and act as reservoirs of infection long after recovery from the illness, or after asymptomatic infection, which probably occurs in more than 50% of those exposed. Bart. bacilliformis is pathogenic only to man.
Bart. bacilliformis is a small, strictly aerobic, Gram-negative coccobacillus. The organisms occur singly, in pairs, chains or clumps. In older cultures they tend to be extremely pleomorphic. They are motile through a cluster of about ten flagella situated at one end of the cell. The organism grows best at 25–28°C and at pH 7.8.
After an incubation period of about 20 days, Oroya fever presents as a high fever followed by progressively severe anaemia due to blood cell destruction. There may be enlargement of the spleen and liver, and haemorrhages into the lymph nodes.
The case fatality rate in untreated cases of Oroya fever may be over 40%, although the overall fatality rate for all forms of the infection is probably only 0.1%. Verruga peruana is a form of bacillary angiomatosis; it may occur without the initial attack of Oroya fever or may develop several weeks after recovery. A pleomorphic skin eruption of reddish, round, elevated, hard nodules may become secondarily infected, producing ulcers and haemorrhagic lesions. The rash usually appears mainly on the legs, arms and face, although all parts of the body may be affected. The condition may persist for as long as a year, but is rarely fatal.
In both Oroya fever and verruga peruana, bartonellosis is confirmed by demonstrating the organisms in smears of blood or tissue aspirates stained by Giemsa or immunofluorescent stain. They are seen packing the cytoplasm of the cells and adhering to the cell surfaces.
Bartonella spp. are dangerous pathogens and should be handled under class 3 containment conditions. Bart. bacilliformis is readily cultured in semi-solid nutrient agar supplemented with rabbit serum and haemoglobin, similar to that used for the culture of leptospires (p. 376). Visible growth may take up to 10 days. PCR or serology is used for identification.
Blood culture should be carried out at all stages of infection. It may be difficult to isolate the organisms from the blood when the verruga stage has developed, and culture from the skin lesions is rarely productive.
PCR tests offer a rapid and reliable diagnosis, but are usually available only from reference laboratories. Bart. bacilliformis antibodies can be detected by various serological tests, but they are common in inhabitants of endemic areas and not necessarily diagnostic of active disease.
Chloramphenicol can drastically reduce the mortality rate in Oroya fever and the frequently associated salmonella infections. Penicillin, streptomycin, tetracyclines, rifampicin, fluoroquinolones and clarithromycin may also be effective in uncomplicated cases. A combination of two antimicrobials is recommended in severe cases. Blood transfusion may be necessary in severe cases of anaemia.
This organism was formerly classified among the rickettsiae as Rochalimaea quintana. However, unlike the rickettsiae, these organisms can grow in cell-free media and they tend to be epicellular rather than strictly intracellular parasites of man. Unlike Bart. bacilliformis the organism does not possess flagella, although it may exhibit a twitching movement owing to fimbriae.
Bart. quintana was first identified as the cause of the febrile illness known as trench fever among the troops in the First World War. It is transmitted by the body louse, Pediculus humanus, under unhygienic living conditions and is not uncommon among homeless people in some countries. Trench fever is a bacteraemic condition typically associated with periodic febrile episodes lasting for about 5 days. Bart. quintana has also been implicated in cases of angiomatosis and endocarditis. The organism may be isolated from the blood of patients by culture on blood agar. Bart. vinsoni and its subspecies are similar and can cause an identical syndrome.
Bart. henselae has been isolated from the blood and lymph nodes of patients suffering from cat scratch disease, a severe condition of regional lymphadenopathy and fever resulting from the scratch or bite of an infected cat. Cat fleas and ticks may be responsible for transmission. Bart. clarridgeiae, which can be differentiated from Bart. henselae by its flagella, can cause an identical syndrome.
An organism known as Afipia felis has also been implicated in a small proportion of cases of cat scratch disease. It is morphologically similar to Bart. henselae, but differs biochemically, genetically and in being culturally less fastidious.
Bart. henselae and, less frequently, Bart. quintana and other species have been identified in the blood and tissues of individuals suffering from two severe clinical syndromes associated with human immunodeficiency virus (HIV) or other immunosuppressant conditions:
1. Bacillary angiomatosis, which produces proliferative vascular lesions in the skin, regional lymph nodes and various internal organs.
Bart. henselae may be cultured from the pus or lymph node samples of patients with cat scratch disease, and from blood, lymphoid tissue, liver and spleen of patients with bacillary angiomatosis or peliosis. In tissue sections the organisms are best demonstrated by silver stain or an immunospecific stain. ELISA, with various protein antigens, is the most useful serological test.
Conventional blood culture often fails to detect endocarditis. Most Bartonella species can be detected and differentiated by PCR methods.
Streptobacillus moniliformis is one of the causes of rat bite fever in man, the other being Spirillum minus (see p. 322). It is a common commensal of the nasopharynx of rodents, and sometimes causes epizootic disease in mice and rats, resulting in otitis media, multiple arthritis and swelling of the feet and legs. Laboratory workers who handle rodents are most at risk. Rarely, outbreaks of infection occur as a result of the ingestion of milk or other food contaminated by rats.
S. moniliformis is Gram-negative, non-motile, non-capsulate and highly pleomorphic. The organisms appear as short bacilli, forming chains interspersed with long filaments that may show oval or spherical lateral swellings.
It is a facultative anaerobe that benefits from added carbon dioxide and a moist atmosphere. It grows best at 37°C and pH 7.6. Culture media must contain blood, serum or ascitic fluid. Loeffler’s serum medium is satisfactory. Media may be made selective by addition of colistin and nalidixic acid. After incubation for 2 days, discrete, granular, greyish yellow colonies 1–5 mm in diameter are visible on the surface, and minute ‘fried egg’ colonies appear in the depth of the medium. The latter are L-phase variants (see pp. 20–1) that have little or no virulence for laboratory animals. They develop spontaneously and are thought to have a defective mechanism for cell wall formation. Cultural and biochemical variants occur and have been used to define biotypes.
S. moniliformis is killed in 30 min by a temperature of 55°C. In culture it survives for only a few days, although it may remain viable for up to a 1 week in serum broth at 37°C. With the exception of the L-forms, S. moniliformis is susceptible to penicillin, and both forms are sensitive to streptomycin and tetracycline.
In man the organism usually enters the body through wounds caused by rodent bites. It multiplies and invades the lymphatics and bloodstream, causing a feverish illness with severe toxic symptoms and sometimes complications such as arthritis, endocarditis and pneumonia.
Infection acquired by ingestion of contaminated water, milk or food is known as Haverhill fever, a condition characterized by fever, sore throat, rash, polyarthritis and erythema. The duration of the illness varies from a few days to several weeks. Endocarditis, hepatitis and amnionitis may develop as complications. In the pre-antibiotic era a case fatality rate of about 10% was reported; the rate is much lower nowadays with effective treatment.
An acute febrile illness associated with asymmetric arthropathy, a maculopapular rash involving the extremities and a history of contact with rodents may point to the diagnosis.
S. moniliformis can be isolated in culture from the patient’s blood during the acute phase of the illness and from the synovial fluid of those who develop arthritis. Growth occurs in serum broth as a characteristic granular sediment, appearing like ‘cotton wool balls’ that do not disintegrate on shaking.
Mice are highly susceptible to intraperitoneal inoculation of infected blood or joint fluid, as a result of which they develop either a rapidly fatal generalized condition or a more chronic disease with swelling of the feet and legs.
ELISA is the method of choice for detecting antibodies. A PCR test may also be used for detection of human and rat infections. Specific agglutinins may be detected in the patient’s serum as early as 10 days or as late as several weeks after the rat bite, but is no longer recommended. A false-positive reaction in the VDRL test (Venereal Disease Research Laboratory slide test; see p. 370) is seen in about 25% of patients.
Andre JM, Freydiere AM, Benito Y, et al. Rat bite fever caused by Streptobacillus moniliformis in a child: human infection and rat carriage diagnosed by PCR. Journal of Clinical Pathology. 2005;58:1215–1216.
Boot R, Oosterhuis A, Thuis HC. PCR for the detection of Streptobacillus moniliformis. Laboratory Animals. 2002;36:200–208.
Corbel MJ, Beeching NJ. Brucellosis. In: Kasper DL, Braunwald E, Fauci A. Harrison’s Principles of Internal Medicine. ed 17. New York: McGraw Hill; 2008:973–976.
Graves MH, Janda JM. Rat bite fever (Streptobacillus moniliformis): a potential emerging disease. International Journal of Infectious Diseases. 2001;5:151–155.
Jensen WA, Fall MZ, Rooney J, et al. Rapid identification and differentiation of Bartonella species using a single step PCR assay. Journal of Clinical Microbiology. 2000;38:1717–1722.
Lopez-Goni I, Moriyon I. Brucella. Molecular and Cellular Biology. Wymondham: Horizon Bioscience, 2004.
Madkour MM. Madkour’s Brucellosis. Berlin: Springer; 2001.
Murray PR, Corbel MJ. Brucella, ed 10. Borriello SP, Murray PR, Funke G, eds. Topley and Wilson’s Microbiology and Microbial Infections, Vol. 2. London: Hodder Arnold. 2005:1719–1751.
Rolain JM, Brouqui P, Koehler JE, et al. Recommendations for the treatment of human infections caused by Bartonella species. Antimicrobial Agents and Chemotherapy. 2004;48:1921–1933.
Wullenweber M. Streptobacillus moniliformis – a zoonotic pathogen. Taxonomic considerations, host species, diagnosis, therapy, geographical distribution. Laboratory Animals. 1995;29:1–15.