Case 27-2006 — A 17-Year-Old Boy with Fever and Lesions in the Liver and Spleen
http://www.100md.com
《新英格兰医药杂志》
Presentation of Case
Dr. Melissa Sue Lee (Pediatrics): A 17-year-old boy was admitted to this hospital because of fever and nodules in the liver and spleen. The patient had been well until two and a half months before admission, when night sweats developed. Two weeks earlier, he had moved from the United States to Santo Domingo, Dominican Republic. During the six weeks before admission to our hospital, he had fevers each morning with a maximal temperature of 40.2°C, associated with rigors.
Five and a half weeks before admission, he saw a physician. His weight was 64.9 kg, and his vital signs were reported to be normal. A complete blood count was normal. Tests for rheumatoid factor, malaria, and antistreptolysin O were negative, as was a skin test for tuberculosis. The following week, ultrasonography of the abdomen and pelvis revealed hepatomegaly, a normal spleen and kidneys, and no masses.
Four weeks before admission, the patient began to have a nonproductive, dry cough. Urinalysis revealed turbid urine with 6 to 8 white cells per high-power field. Culture of the urine grew enterobacter species that was sensitive to cefepime and had intermediate sensitivity to fosfomycin. Cultures of the blood were negative. A urinary tract infection was diagnosed, and the patient was admitted to another hospital for the intravenous administration of cefepime and fosfomycin. He was discharged on the fourth day; however, his fevers continued.
Two and a half weeks before admission, computed tomography (CT) of the abdomen (Figure 1) showed hepatomegaly with multiple hypodense nodules in the spleen and liver. One week later, a follow-up urinalysis revealed 2 to 4 white cells per high-power field, and cultures of the urine, blood, and stool were negative. The patient's fever persisted (maximal temperature, 40.2°C) and occasionally peaked in the evening.
Figure 1. CT Scans of the Abdomen.
A contrast-enhanced image of the liver (Panel A) shows multiple low-density lesions with central septations and surrounding halos (arrows). Lesions are also present in the spleen. An image obtained slightly below the image shown in Panel A (Panel B) shows larger lesions, as well as lymphadenopathy around the celiac axis (arrows). An image obtained below the image shown in Panel B (Panel C) shows additional lymphadenopathy in the peripancreatic and periportal regions (arrow).
Five days before admission, the patient was evaluated by an infectious-disease specialist and readmitted to the other hospital for intravenous administration of a three-day course of imipenem–cilastatin and levofloxacin. His electrolyte levels and liver-function tests were normal. Serologic tests for leptospira, Salmonella typhi, toxoplasma, human immunodeficiency virus (HIV) types 1 and 2, hepatitis B surface antigen, and antibodies against hepatitis C virus were negative. Results of additional laboratory tests are shown in Table 1. The patient's parents elected to return to the United States so that he could be further evaluated, and he was admitted to this hospital.
Table 1. Laboratory Data.
He had no history of headache, diarrhea, vomiting, chest pain, shortness of breath, recent sore throat, rashes, or dysuria, and his appetite was relatively preserved. He was allergic to sulfa medications. His immunizations were up to date. Since his early childhood and until his return to the Dominican Republic, he had resided with his family in an urban area in the northeastern United States. After he returned to the Dominican Republic, he had drunk untreated water, had had no exposure to raw meat, and had eaten cooked pork and beef. There were cats and dogs in the family home in the Dominican Republic. He had had no sexual contact with anyone and had no known exposure to tuberculosis. No family members or contacts had been ill. The patient's maternal grandmother had a history of arteriovenous malformations in the liver with bleeding. Other family members were in good health, and there was no family history of cancer or other tumors.
On physical examination, the patient had diaphoresis, was sleepy, and appeared chronically ill, with a thin body habitus and an occasional nonproductive cough. His temperature was 37.5°C, with a blood pressure of 100/62 mm Hg, a pulse of 97 beats per minute, respirations of 16 per minute, and weight of 61.4 kg. The oxygen saturation was 99 percent while he was breathing ambient air. He had pharyngeal erythema, tonsillar hypertrophy without exudate, a right anterior cervical lymph node less than 1 cm in diameter, and one palpable femoral node on the left. A 2/6 systolic ejection murmur was heard best at the lower sternal border. The liver edge was palpable 1 cm below the costovertebral margin, and the results of the remainder of the examination were normal.
Serum levels of electrolytes, amylase, lipase, and carcinoembryonic antigen were normal, as were the results of renal-function testing and a urinalysis. Specimens of blood were sent for culture, testing for antibodies against brucella and bartonella, and other studies (Table 1). Chest radiography revealed no abnormalities. CT of the abdomen and pelvis showed numerous lesions in the liver and spleen. The hepatic lesions had central areas of low density and peripheral enhancement; the largest was 3 cm in diameter. Lymphadenopathy was present in the periportal, peripancreatic, and aortocaval regions, and a wedge-shaped defect was seen in the left kidney. Small amounts of fluid were seen in the pelvis.
The patient's maximal temperature during the first 48 hours of hospitalization was 37.3°C. A tuberculin skin test was administered. Gram's staining of the stool showed abundant mixed flora and no leukocytes; a specimen was sent for culture. On the third hospital day, a diagnostic procedure was performed.
Differential Diagnosis
Dr. Stephen I. Pelton: May we review the radiologic studies?
Dr. John Y. Kim: The initial axial, contrast-enhanced CT image (Figure 1A) revealed multiple low-density lesions throughout the liver ranging from less than 1 cm to 3 cm in diameter. There are also similar low-density lesions throughout the spleen. At the level of the pancreas, there are enlarged lymph nodes around the celiac axis, superior mesenteric artery, and the retroperitoneum and periportal region (Figure 1B and 1C).
Dr. Pelton: This 17-year-old adolescent, who had previously been well, then had an illness characterized by spiking fever, rigors, weight loss, progressive anemia, multiple hypodense lesions with peripheral enhancement on CT of the liver and spleen, and enlarged periportal, prepancreatic, and aortocaval lymph nodes. The differential diagnosis of these findings is broad, and possible causes are listed in Table 2. However, taken together, the findings are most suggestive of liver abscess.
Table 2. Differential Diagnosis of Multiple Lesions in the Liver and Spleen.
The presenting signs and symptoms of liver abscess typically include fever, rigors, and pain in the right upper quadrant. The clinical presentation is usually acute, although sepsis is rare, and some cases have an indolent course. Before the availability of CT and ultrasonography, liver abscess was difficult to diagnose and the diagnosis was often delayed. In patients without underlying diseases, the two most probable causes of liver abscess are pyogenic bacteria and amebae. Distinguishing between them is important, since the prognosis, management, and treatment differ.
Several features can distinguish pyogenic from amebic liver abscesses (Table 3). Lodhi et al.1 reported that there were multiple liver abscesses in 48 percent of 577 cases caused by pyogenic organisms but in only 23 percent of those caused by Entamoeba histolytica. Pyogenic abscess is most common in persons over the age of 50 years, who typically have concomitant gastrointestinal conditions such as cholecystitis, ruptured appendix, or intestinal perforation. Pyogenic abscess has been reported as a complication of Crohn's disease,2 acute leukemia, perforated appendix, colonic diverticulitis, pelvic inflammatory disease,3 and hemochromatosis. The sepsis syndrome has specifically been reported with disease caused by a fusobacterium.4 Amebic abscess usually occurs in young men residing in or visiting areas in which the disease is endemic, including Central America and much of the Caribbean. In general, amebic dysentery precedes the development of amebic abscess. Pulmonary abnormalities such as atelectasis, pulmonary infiltrates, or effusion are frequently found with either pyogenic or amebic abscess. In this adolescent, the absence of abdominal pain or leukocytosis and the presence of extensive hepatic and splenic involvement with no intraabdominal source of infection on CT suggest a hematogenous origin of infection, rather than contiguous spread from an intraabdominal source. Although he had traveled to the Caribbean, the absence of an episode of dysentery and the presence of multiple abscesses argue against amebic abscess.
Table 3. Characteristics of Pyogenic and Amebic Abscesses in the Liver.
Multiple pathogens have been reported in association with abscesses of the liver and spleen, some as a result of hematogenous seeding and others resulting from contiguous spread from an intraabdominal source. Brook and Frazier5 reported that mixed aerobic and anaerobic bacteria were common pathogens recovered from liver and spleen abscesses. The most common aerobes were Escherichia coli, group D streptococci, Klebsiella pneumoniae, and Staphylococcus aureus. S. aureus and -hemolytic streptococci were associated with trauma, group D streptococci and klebsiella species with biliary disease, and anaerobic infection with colonic disease. Additional case reports document actinomyces, Nocardia asteroides, Yersinia pseudotuberculosis and Y. enterocolitica, Listeria monocytogenes, Campylobacter jejuni, Legionella pneumophila, Mycobacterium tuberculosis, S. typhi or S. paratyphi, Candida albicans, and Bartonella henselae as the cause of liver or liver and spleen abscesses.
Some infectious agents are unlikely to be the cause of the abscesses in this young, otherwise healthy patient. Hepatosplenic abscesses from candida species are usually seen in newborns or immunocompromised patients, especially those with leukemia in whom fungemia develops. Multiple hepatosplenic abscesses caused by Y. pseudotuberculosis or Y. enterocolitica have been reported in patients with underlying liver disease and iron overload.6,7,8 Abscess formation caused by C. jejuni was observed as a complication of enteritis.9 Disseminated M. tuberculosis, with multiple liver and spleen abscesses, has also been described in both immunocompetent and immunocompromised hosts.10,11,12 Patients most often have multiorgan disease without isolated hepatic or splenic involvement. S. typhi or S. paratyphi can involve the liver or spleen, or both; it may be secondary to bacteremia but is more often thought to result from an ascending cholangitis. All these infections are unlikely in this healthy adolescent with no history of enteritis or exposure to tuberculosis.
Bacteremic staphylococcal disease occurs spontaneously in adolescents, possibly from trivial skin or soft-tissue foci. It most often results in joint or bone infection, but liver and spleen abscesses have been reported. In most series, staphylococci are among the most common causes of liver abscesses, especially when both the liver and spleen are involved and there is no underlying intraabdominal disease.13 In this patient, staphylococcal infection is a serious consideration.
We were told that there were cats in the patient's household in the Dominican Republic; cats are the major vector of B. henselae. Systemic infection from bartonella occurs most commonly in immunocompromised patients. Bacillary angiomatosis and hepatic peliosis are the common clinical syndromes, especially in those infected with HIV.14 In immunocompetent persons, lymphadenopathy in regional nodes that drain the sites of local inoculation (i.e., the site of a cat scratch) is the most common clinical presentation. However, some patients may have atypical cat scratch disease, with prolonged fever (longer than two weeks), malaise, fatigue, myalgia, and arthralgia; some patients have skin eruptions, weight loss, and splenomegaly.15 Rizkallah et al.16 described four children with cat scratch disease who had lymphadenopathy, high fever for more than four weeks, abdominal pain, and hepatic lesions with or without splenic lesions on ultrasonography and CT. Many aspects of this patient's presentation would be consistent with cat scratch disease, except for the absence of abdominal pain.
Most patients with either localized or disseminated disease caused by B. henselae infection have a history of exposure to cats and either a history or the presence of a papule at the site of inoculation on physical examination. Serologic testing for bartonella (immunofluorescence antibody testing for serum antibodies against bartonella) and DNA testing to detect B. henselae in tissue and other substances are available by special request at the Centers for Disease Control and Prevention (CDC). This patient did not report having been scratched by a cat, and no primary lesion was described by the patient or noted on examination.
Further diagnostic studies that would be valuable in defining the cause of this patient's illness include additional blood cultures, serologic testing and polymerase-chain-reaction (PCR) assay for B. henselae, serologic testing for amebae, and possibly, echocardiography to identify bacterial endocarditis. At this point, my leading diagnosis would be multiple bacterial abscesses of the liver and spleen caused by S. aureus or salmonella species, with secondary concern about the possibility of infection caused by B. henselae.
Dr. Nancy Lee Harris (Pathology): Dr. Pasternack, would you tell us your thinking at the time?
Dr. Mark S. Pasternack (Infectious Disease Unit, MassGeneral Hospital for Children): This child came to us after a prolonged febrile illness and an extensive evaluation for fever of unknown origin performed in the Dominican Republic. The radiologic studies focused our attention on the liver and spleen. What was striking in evaluating him at the bedside was that he did not appear to be acutely ill, despite his obvious weight loss. He did not have leukocytosis, and his liver chemical values were normal. Dr. Pelton pointed out that in the presence of widely dispersed liver lesions one has to think about hematogenous seeding, as opposed to intraabdominal seeding from appendicitis or other intraabdominal diseases, but the fact that he did not appear to be acutely ill and had relatively normal results on laboratory studies pointed me away from pyogenic liver abscess. The history of multiple cats in the household in which he resided in the Dominican Republic made me think that we needed to pursue the diagnosis of bartonellosis. In a case such as this, in which a neoplasm is in the differential diagnosis, obtaining tissue for evaluation is essential. We proceeded to obtain an expedited serologic evaluation for brucellosis and bartonellosis and arranged for a liver biopsy. My leading diagnosis was hepatosplenic bartonellosis.
A Physician: How did you evaluate the lymph node in the groin? Did you think that was an important part of the patient's clinical picture?
Dr. Pasternack: He had several enlarged lymph nodes, particularly in the right anterior and posterior cervical chain, that were of potential pathologic significance. I would say that the inguinal lymphadenopathy was within the realm of what I consider to be normal for a teenage boy. I did not think that any of the lymph-node findings suggested involvement by cat scratch disease.
Clinical Diagnosis
Bartonella henselae infection with abscesses of the liver and spleen.
Dr. Stephen I. Pelton's Diagnosis
Bacterial abscesses of the liver and spleen caused by Staphylococcus aureus or salmonella species, with secondary concern about the possibility of infection caused by Bartonella henselae.
Pathological Discussion
Dr. Richard L. Kradin: The diagnostic procedure was a needle biopsy of the liver. Examination of the biopsy specimen showed a portal region with chronic lymphocytic inflammation extending across the limiting plate of an adjacent hepatic lobule (Figure 2A). There are no granulomas. A Warthin–Starry silver stain shows several extracellular bacilli within the collagenous matrix (Figure 2B). Gram's stain, acid-fast stain, and modified acid-fast stain did not react with the organisms. These findings are consistent with, although not diagnostic of, infection by bartonella species.
Figure 2. Needle-Biopsy Specimen of the Liver.
Hematoxylin-and-eosin staining reveals periportal lymphocytic infiltrates extending into the adjacent hepatic lobule (Panel A). Warthin–Starry silver staining reveals several extracellular bacilli within the collagenous matrix (Panel B).
Bartonella are slender (0.5 to 2.0 μmol), curved, gram-negative rods.17 They can be cultured with difficulty on chocolate agar with carbon dioxide supplementation. They do not react with oxidase, urease, or catalase. Specific identification requires DNA analysis with the use of PCR, immunochemistry, or gas chromatography.
The histopathological characteristics of bartonella infection can range from a complete lack of immune response to necrotizing inflammation with abscess formation. B. henselae proliferates in the walls of capillaries and lymphatics and is ingested by macrophages. Vessels become thickened and occluded by the organisms, and macrophages showing epithelioid transformation accumulate around the infected vasculature. Subcapsular stellate necrosis18 is commonly seen in the lymph nodes. The nonspecific pathological findings in the present case may reflect a sampling error due to the small size of the specimen.
The identification of organisms in situ requires a silver-impregnation technique, and Warthin–Starry staining is the preferred method. Wear19 has proposed three criteria for the identification of bartonella species: the presence of filamentous bacilli in collagen fibers at any stage of the infection, Y-shaped branching bacteria (Chinese figures) that are intracellular or extracellular in either viable areas or areas of necrosis, and single bacilli within macrophages in the same lesion and in consecutive sections.
At least six species of bartonella are recognized as a cause of human disease. In the immunocompetent host, the chief clinical syndromes are Oroya fever and Verruga peruana complex, which are caused by B. bacilliformis. Trench fever, a generally nonlethal, louse-borne disease causing repeated bouts of fever lasting for approximately five days, is caused by B. quintana.20 Cat scratch disease is most often caused by B. henselae, although a small number of cases result from infection by the bacterium Afipia felis.21
Dr. Harris: Dr. Pasternack, will you tell us about the serologic testing and treatment of this patient?
Dr. Pasternack: Expedited serologic testing for bartonella antibodies was performed at the CDC. The patient's IgG titers to B. henselae and B. quintana exceeded 1:1024, which is the greatest dilution typically reported. Interestingly, both IgM antibody titers were less than the assay threshold of 1:20.
Follow-up serologic testing was performed about two months later, at which time IgG titers to both B. henselae and B. quintana remained positive at greater than 1:1024, but the B. henselae IgM antibody titer was 1:80 and the B. quintana IgM titer was still undetectable (less than 1:20). These studies were performed by immunofluorescence assay, and there is cross-reactivity between the quintana and henselae IgG antibodies, although the IgM antibody is thought to be more species specific. The clinical syndrome of B. quintana infection in nonwartime situations has largely been restricted to homeless persons. Body lice are thought to be the vector of transmission, and this patient had no clinical risk factors for that infection. We interpreted the results of the serologic studies as consistent with a diagnosis of B. henselae infection.
Infections attributable to this organism have been treated with a variety of antibiotics. In retrospective reviews ciprofloxacin, rifampin, trimethoprim–sulfamethoxazole, and gentamicin have all been associated with clinical benefit. In HIV-infected persons with bacillary angiomatosis, macrolides have been shown to be effective.22 In a single prospective, randomized trial of azithromycin in the treatment of cat scratch lymphadenitis, a standard five-day course of therapy was reported to be highly efficacious.23 We treated this adolescent with three weeks of clarithromycin at a dose of 500 mg twice daily.
Dr. Harris: I gather that none of the antibiotics that he was given before the diagnosis would be expected to have been effective?
Dr. Pasternack: Fluoroquinolones are thought to have activity against bartonella; however, three days of treatment would probably be inadequate to effect a substantial change.
Dr. Pelton: As I understand it, there is still some debate regarding the effectiveness of antibiotic therapy. The condition of most patients improves without antibiotics, and the studies may show some improvement with antibiotic therapy, but it is not clear how much or how much more quickly the improvement is with therapy.
Dr. Pasternack: Dr. Pelton makes an excellent point, which is that the use of antibiotics has been heavily debated. The trial of azithromycin is the only trial that suggests that therapy is beneficial. We think antibiotics may have a minor role, but many patients with this diagnosis do have an improvement without therapy. This patient had such a dramatic febrile illness that we wanted to treat him.
Dr. Pelton: I am not suggesting that this illness should not be treated. Indeed, the more traditional treatment is with parenteral gentamicin for some period before switching to oral therapy.
Dr. Pasternack: He did not appear acutely ill and had started to regain his appetite by the time of his admission to this hospital. He was so eager to leave the hospital that we embarked on a trial of oral therapy.
Dr. Harris: Dr. Goldstein, can you give us more recent follow-up?
Dr. Mark A. Goldstein (Adolescent Medicine): Within two weeks after antibiotic therapy was instituted, the patient reported subjective improvement in his appetite and diminished fatigue. He had gained almost 2 kg. Six weeks after the initiation of clarithromycin, the patient reported increased energy and complete return of his appetite. At the final follow-up visit, approximately 12 weeks after the initiation of antibiotic therapy and 5 1/2 months after he first become ill, the patient was asymptomatic and the results of physical examination were normal. He weighed 68.2 kg, which was more than he had before his illness. His hematocrit and erythrocyte sedimentation rate were normal.
Dr. Kim: CT two months after the initial CT showed nearly complete resolution of the lesions in the liver and spleen and resolution of the abdominal lymphadenopathy.
Dr. Harris: Dr. Pelton, would you like to make any final comments?
Dr. Pelton: I think the missing features that would have made the diagnosis more obvious would be abdominal pain and the clear presence of a papule or a lesion associated with a cat scratch. Fortunately, the treating physicians recognized the possible importance of the patient's exposure to cats.
Anatomical Diagnosis
Hepatitis caused by Bartonella henselae.
Dr. Pelton reports having served on advisory boards of Wyeth-Vaccines and GlaxoSmithKline, and having received speaking fees and grant support from Sanofi-Aventis.
Source Information
From the Department of Pediatric Infectious Disease, Boston Medical Center (S.I.P.); the Department of Pediatrics, Boston University School of Medicine (S.I.P.); and the Departments of Radiology (J.Y.K.) and Pathology (R.L.K.), Massachusetts General Hospital and Harvard Medical School — all in Boston.
References
Lodhi S, Sarwari R, Muzammil M, Salam A, Smego RA. Features distinguishing amoebic from pyogenic liver abscess: a review of 577 adult cases. Trop Med Int Health 2004;9:718-723.
Manjunatha S, McIntyre PB, Lynn A. Multiple liver abscesses in Crohn's disease. Br J Hosp Med 1992;47:375-376.
Gelfand MS, Hodgkiss T, Simmons BP. Multiple hepatic abscesses caused by Streptococcus milleri in association with an intrauterine device. Rev Infect Dis 1989;11:983-987.
Lechiche C, Corne P, Riviere S, Jonquet O, Le Quellec A. Fusobacterium necrophorum septic shock from colonic origin with multiple abscesses. Rev Med Interne 2001;22:198-199.
Brook I, Frazier EH. Microbiology of liver and spleen abscesses. J Med Microbiol 1998;47:1075-1080.
Farrer W, Kloser P, Ketyer S. Yersinia pseudotuberculosis sepsis presenting as multiple liver abscesses. Am J Med Sci 1988;295:129-132.
Strungs I, Farrell DJ, Matar LD, Dekker L, Franz RJ. Multiple hepatic abscesses due to Yersinia enterocolitica. Pathology 1995;27:374-377.
Hopfner M, Nitsche R, Rohr A, Harms D, Schubert S, Folsch UR. Yersinia enterocolitica infection with multiple liver abscesses uncovering a primary hemochromatosis. Scand J Gastroenterol 2001;36:220-224.
Brmbolic B. Multiple abscesses of the liver caused by Campylobacter jejuni. J Clin Gastroenterol 1995;20:307-309.
Chen CH, Fang YH, Chiang PM, Lin DT, Huang LM. Disseminated tuberculosis presenting as multiple hepatosplenic microabscesses and pancytopenia in a teenage boy. J Formos Med Assoc 2004;103:939-942.
Zheng RQ, Kudo M, Ishikawa E, Zhou P. Multiple tuberculous abscesses of the liver and the brain in a patient with acute leukemia. J Gastroenterol 2004;39:497-499.
Sood A, Midha V, Sood N, Sekhon A. Multiple tubercular liver abscesses in an immunocompetent person. Trop Gastroenterol 2001;22:144-145.
Verbanck J, Ponette J, Verbanck M, Vandewiele I, Segaert M. Sonographic detection of multiple Staphylococcus aureus hepatic microabscesses mimicking Candida abscesses. J Clin Ultrasound 1999;27:478-481.
Bass J, Vincent J, Person DA. The expanding spectrum of Bartonella infections. II. Cat-scratch disease. Pediatr Infect Dis J 1997;16:163-179.
Margileth AM, Wear DJ, English CK. Systemic cat scratch disease: report of 23 patients with prolonged or recurrent severe bacterial infection. J Infect Dis 1987;155:390-402.
Rizkallah MF, Meyer L, Ayoub EM. Hepatic and splenic abscesses in cat-scratch disease. Pediatr Infect Dis J 1988;7:191-195.
Eaton M, Nelson AM. Bartonella infections. In: Horsburgh CR, Nelson AM, eds. The pathology of emerging infections. Washington, D.C.: ASM Press, 1997.
Relman DA, Loutit JS, Schmidt TM. The agent of bacillary angiomatosis: an approach to the identification of uncultured pathogens. N Engl J Med 1990;323:1573-1580.
Wear DJ. Cat scratch disease and bacillary angiomatosis. In: Marty AM, Nelson AM, eds. Advances in diagnostic pathology of infectious disease. Washington, D.C.: Armed Forces Institute of Pathology, 1994:29-36.
Schwartzmann WA. Cat scratch disease and other Bartonella infections. Curr Treat Options Infect Dis 2000;2:155-62.
Jackson LA, Perkins BA, Wenger JD. Cat scratch disease in the United States: an analysis of three national databases. Am J Public Health 1993;83:1707-1711.
Koehler JE, Tappero JW. Bacillary angiomatosis and bacillary peliosis in patients infected with human immunodeficiency virus. Clin Infect Dis 1993;17:612-624.
Bass JW, Freitas BC, Freitas AD, et al. Prospective randomized double blind placebo-controlled evaluation of azithromycin for treatment of cat-scratch disease. Pediatr Infect Dis J 1998;17:447-452.(Stephen I. Pelton, M.D., )
Dr. Melissa Sue Lee (Pediatrics): A 17-year-old boy was admitted to this hospital because of fever and nodules in the liver and spleen. The patient had been well until two and a half months before admission, when night sweats developed. Two weeks earlier, he had moved from the United States to Santo Domingo, Dominican Republic. During the six weeks before admission to our hospital, he had fevers each morning with a maximal temperature of 40.2°C, associated with rigors.
Five and a half weeks before admission, he saw a physician. His weight was 64.9 kg, and his vital signs were reported to be normal. A complete blood count was normal. Tests for rheumatoid factor, malaria, and antistreptolysin O were negative, as was a skin test for tuberculosis. The following week, ultrasonography of the abdomen and pelvis revealed hepatomegaly, a normal spleen and kidneys, and no masses.
Four weeks before admission, the patient began to have a nonproductive, dry cough. Urinalysis revealed turbid urine with 6 to 8 white cells per high-power field. Culture of the urine grew enterobacter species that was sensitive to cefepime and had intermediate sensitivity to fosfomycin. Cultures of the blood were negative. A urinary tract infection was diagnosed, and the patient was admitted to another hospital for the intravenous administration of cefepime and fosfomycin. He was discharged on the fourth day; however, his fevers continued.
Two and a half weeks before admission, computed tomography (CT) of the abdomen (Figure 1) showed hepatomegaly with multiple hypodense nodules in the spleen and liver. One week later, a follow-up urinalysis revealed 2 to 4 white cells per high-power field, and cultures of the urine, blood, and stool were negative. The patient's fever persisted (maximal temperature, 40.2°C) and occasionally peaked in the evening.
Figure 1. CT Scans of the Abdomen.
A contrast-enhanced image of the liver (Panel A) shows multiple low-density lesions with central septations and surrounding halos (arrows). Lesions are also present in the spleen. An image obtained slightly below the image shown in Panel A (Panel B) shows larger lesions, as well as lymphadenopathy around the celiac axis (arrows). An image obtained below the image shown in Panel B (Panel C) shows additional lymphadenopathy in the peripancreatic and periportal regions (arrow).
Five days before admission, the patient was evaluated by an infectious-disease specialist and readmitted to the other hospital for intravenous administration of a three-day course of imipenem–cilastatin and levofloxacin. His electrolyte levels and liver-function tests were normal. Serologic tests for leptospira, Salmonella typhi, toxoplasma, human immunodeficiency virus (HIV) types 1 and 2, hepatitis B surface antigen, and antibodies against hepatitis C virus were negative. Results of additional laboratory tests are shown in Table 1. The patient's parents elected to return to the United States so that he could be further evaluated, and he was admitted to this hospital.
Table 1. Laboratory Data.
He had no history of headache, diarrhea, vomiting, chest pain, shortness of breath, recent sore throat, rashes, or dysuria, and his appetite was relatively preserved. He was allergic to sulfa medications. His immunizations were up to date. Since his early childhood and until his return to the Dominican Republic, he had resided with his family in an urban area in the northeastern United States. After he returned to the Dominican Republic, he had drunk untreated water, had had no exposure to raw meat, and had eaten cooked pork and beef. There were cats and dogs in the family home in the Dominican Republic. He had had no sexual contact with anyone and had no known exposure to tuberculosis. No family members or contacts had been ill. The patient's maternal grandmother had a history of arteriovenous malformations in the liver with bleeding. Other family members were in good health, and there was no family history of cancer or other tumors.
On physical examination, the patient had diaphoresis, was sleepy, and appeared chronically ill, with a thin body habitus and an occasional nonproductive cough. His temperature was 37.5°C, with a blood pressure of 100/62 mm Hg, a pulse of 97 beats per minute, respirations of 16 per minute, and weight of 61.4 kg. The oxygen saturation was 99 percent while he was breathing ambient air. He had pharyngeal erythema, tonsillar hypertrophy without exudate, a right anterior cervical lymph node less than 1 cm in diameter, and one palpable femoral node on the left. A 2/6 systolic ejection murmur was heard best at the lower sternal border. The liver edge was palpable 1 cm below the costovertebral margin, and the results of the remainder of the examination were normal.
Serum levels of electrolytes, amylase, lipase, and carcinoembryonic antigen were normal, as were the results of renal-function testing and a urinalysis. Specimens of blood were sent for culture, testing for antibodies against brucella and bartonella, and other studies (Table 1). Chest radiography revealed no abnormalities. CT of the abdomen and pelvis showed numerous lesions in the liver and spleen. The hepatic lesions had central areas of low density and peripheral enhancement; the largest was 3 cm in diameter. Lymphadenopathy was present in the periportal, peripancreatic, and aortocaval regions, and a wedge-shaped defect was seen in the left kidney. Small amounts of fluid were seen in the pelvis.
The patient's maximal temperature during the first 48 hours of hospitalization was 37.3°C. A tuberculin skin test was administered. Gram's staining of the stool showed abundant mixed flora and no leukocytes; a specimen was sent for culture. On the third hospital day, a diagnostic procedure was performed.
Differential Diagnosis
Dr. Stephen I. Pelton: May we review the radiologic studies?
Dr. John Y. Kim: The initial axial, contrast-enhanced CT image (Figure 1A) revealed multiple low-density lesions throughout the liver ranging from less than 1 cm to 3 cm in diameter. There are also similar low-density lesions throughout the spleen. At the level of the pancreas, there are enlarged lymph nodes around the celiac axis, superior mesenteric artery, and the retroperitoneum and periportal region (Figure 1B and 1C).
Dr. Pelton: This 17-year-old adolescent, who had previously been well, then had an illness characterized by spiking fever, rigors, weight loss, progressive anemia, multiple hypodense lesions with peripheral enhancement on CT of the liver and spleen, and enlarged periportal, prepancreatic, and aortocaval lymph nodes. The differential diagnosis of these findings is broad, and possible causes are listed in Table 2. However, taken together, the findings are most suggestive of liver abscess.
Table 2. Differential Diagnosis of Multiple Lesions in the Liver and Spleen.
The presenting signs and symptoms of liver abscess typically include fever, rigors, and pain in the right upper quadrant. The clinical presentation is usually acute, although sepsis is rare, and some cases have an indolent course. Before the availability of CT and ultrasonography, liver abscess was difficult to diagnose and the diagnosis was often delayed. In patients without underlying diseases, the two most probable causes of liver abscess are pyogenic bacteria and amebae. Distinguishing between them is important, since the prognosis, management, and treatment differ.
Several features can distinguish pyogenic from amebic liver abscesses (Table 3). Lodhi et al.1 reported that there were multiple liver abscesses in 48 percent of 577 cases caused by pyogenic organisms but in only 23 percent of those caused by Entamoeba histolytica. Pyogenic abscess is most common in persons over the age of 50 years, who typically have concomitant gastrointestinal conditions such as cholecystitis, ruptured appendix, or intestinal perforation. Pyogenic abscess has been reported as a complication of Crohn's disease,2 acute leukemia, perforated appendix, colonic diverticulitis, pelvic inflammatory disease,3 and hemochromatosis. The sepsis syndrome has specifically been reported with disease caused by a fusobacterium.4 Amebic abscess usually occurs in young men residing in or visiting areas in which the disease is endemic, including Central America and much of the Caribbean. In general, amebic dysentery precedes the development of amebic abscess. Pulmonary abnormalities such as atelectasis, pulmonary infiltrates, or effusion are frequently found with either pyogenic or amebic abscess. In this adolescent, the absence of abdominal pain or leukocytosis and the presence of extensive hepatic and splenic involvement with no intraabdominal source of infection on CT suggest a hematogenous origin of infection, rather than contiguous spread from an intraabdominal source. Although he had traveled to the Caribbean, the absence of an episode of dysentery and the presence of multiple abscesses argue against amebic abscess.
Table 3. Characteristics of Pyogenic and Amebic Abscesses in the Liver.
Multiple pathogens have been reported in association with abscesses of the liver and spleen, some as a result of hematogenous seeding and others resulting from contiguous spread from an intraabdominal source. Brook and Frazier5 reported that mixed aerobic and anaerobic bacteria were common pathogens recovered from liver and spleen abscesses. The most common aerobes were Escherichia coli, group D streptococci, Klebsiella pneumoniae, and Staphylococcus aureus. S. aureus and -hemolytic streptococci were associated with trauma, group D streptococci and klebsiella species with biliary disease, and anaerobic infection with colonic disease. Additional case reports document actinomyces, Nocardia asteroides, Yersinia pseudotuberculosis and Y. enterocolitica, Listeria monocytogenes, Campylobacter jejuni, Legionella pneumophila, Mycobacterium tuberculosis, S. typhi or S. paratyphi, Candida albicans, and Bartonella henselae as the cause of liver or liver and spleen abscesses.
Some infectious agents are unlikely to be the cause of the abscesses in this young, otherwise healthy patient. Hepatosplenic abscesses from candida species are usually seen in newborns or immunocompromised patients, especially those with leukemia in whom fungemia develops. Multiple hepatosplenic abscesses caused by Y. pseudotuberculosis or Y. enterocolitica have been reported in patients with underlying liver disease and iron overload.6,7,8 Abscess formation caused by C. jejuni was observed as a complication of enteritis.9 Disseminated M. tuberculosis, with multiple liver and spleen abscesses, has also been described in both immunocompetent and immunocompromised hosts.10,11,12 Patients most often have multiorgan disease without isolated hepatic or splenic involvement. S. typhi or S. paratyphi can involve the liver or spleen, or both; it may be secondary to bacteremia but is more often thought to result from an ascending cholangitis. All these infections are unlikely in this healthy adolescent with no history of enteritis or exposure to tuberculosis.
Bacteremic staphylococcal disease occurs spontaneously in adolescents, possibly from trivial skin or soft-tissue foci. It most often results in joint or bone infection, but liver and spleen abscesses have been reported. In most series, staphylococci are among the most common causes of liver abscesses, especially when both the liver and spleen are involved and there is no underlying intraabdominal disease.13 In this patient, staphylococcal infection is a serious consideration.
We were told that there were cats in the patient's household in the Dominican Republic; cats are the major vector of B. henselae. Systemic infection from bartonella occurs most commonly in immunocompromised patients. Bacillary angiomatosis and hepatic peliosis are the common clinical syndromes, especially in those infected with HIV.14 In immunocompetent persons, lymphadenopathy in regional nodes that drain the sites of local inoculation (i.e., the site of a cat scratch) is the most common clinical presentation. However, some patients may have atypical cat scratch disease, with prolonged fever (longer than two weeks), malaise, fatigue, myalgia, and arthralgia; some patients have skin eruptions, weight loss, and splenomegaly.15 Rizkallah et al.16 described four children with cat scratch disease who had lymphadenopathy, high fever for more than four weeks, abdominal pain, and hepatic lesions with or without splenic lesions on ultrasonography and CT. Many aspects of this patient's presentation would be consistent with cat scratch disease, except for the absence of abdominal pain.
Most patients with either localized or disseminated disease caused by B. henselae infection have a history of exposure to cats and either a history or the presence of a papule at the site of inoculation on physical examination. Serologic testing for bartonella (immunofluorescence antibody testing for serum antibodies against bartonella) and DNA testing to detect B. henselae in tissue and other substances are available by special request at the Centers for Disease Control and Prevention (CDC). This patient did not report having been scratched by a cat, and no primary lesion was described by the patient or noted on examination.
Further diagnostic studies that would be valuable in defining the cause of this patient's illness include additional blood cultures, serologic testing and polymerase-chain-reaction (PCR) assay for B. henselae, serologic testing for amebae, and possibly, echocardiography to identify bacterial endocarditis. At this point, my leading diagnosis would be multiple bacterial abscesses of the liver and spleen caused by S. aureus or salmonella species, with secondary concern about the possibility of infection caused by B. henselae.
Dr. Nancy Lee Harris (Pathology): Dr. Pasternack, would you tell us your thinking at the time?
Dr. Mark S. Pasternack (Infectious Disease Unit, MassGeneral Hospital for Children): This child came to us after a prolonged febrile illness and an extensive evaluation for fever of unknown origin performed in the Dominican Republic. The radiologic studies focused our attention on the liver and spleen. What was striking in evaluating him at the bedside was that he did not appear to be acutely ill, despite his obvious weight loss. He did not have leukocytosis, and his liver chemical values were normal. Dr. Pelton pointed out that in the presence of widely dispersed liver lesions one has to think about hematogenous seeding, as opposed to intraabdominal seeding from appendicitis or other intraabdominal diseases, but the fact that he did not appear to be acutely ill and had relatively normal results on laboratory studies pointed me away from pyogenic liver abscess. The history of multiple cats in the household in which he resided in the Dominican Republic made me think that we needed to pursue the diagnosis of bartonellosis. In a case such as this, in which a neoplasm is in the differential diagnosis, obtaining tissue for evaluation is essential. We proceeded to obtain an expedited serologic evaluation for brucellosis and bartonellosis and arranged for a liver biopsy. My leading diagnosis was hepatosplenic bartonellosis.
A Physician: How did you evaluate the lymph node in the groin? Did you think that was an important part of the patient's clinical picture?
Dr. Pasternack: He had several enlarged lymph nodes, particularly in the right anterior and posterior cervical chain, that were of potential pathologic significance. I would say that the inguinal lymphadenopathy was within the realm of what I consider to be normal for a teenage boy. I did not think that any of the lymph-node findings suggested involvement by cat scratch disease.
Clinical Diagnosis
Bartonella henselae infection with abscesses of the liver and spleen.
Dr. Stephen I. Pelton's Diagnosis
Bacterial abscesses of the liver and spleen caused by Staphylococcus aureus or salmonella species, with secondary concern about the possibility of infection caused by Bartonella henselae.
Pathological Discussion
Dr. Richard L. Kradin: The diagnostic procedure was a needle biopsy of the liver. Examination of the biopsy specimen showed a portal region with chronic lymphocytic inflammation extending across the limiting plate of an adjacent hepatic lobule (Figure 2A). There are no granulomas. A Warthin–Starry silver stain shows several extracellular bacilli within the collagenous matrix (Figure 2B). Gram's stain, acid-fast stain, and modified acid-fast stain did not react with the organisms. These findings are consistent with, although not diagnostic of, infection by bartonella species.
Figure 2. Needle-Biopsy Specimen of the Liver.
Hematoxylin-and-eosin staining reveals periportal lymphocytic infiltrates extending into the adjacent hepatic lobule (Panel A). Warthin–Starry silver staining reveals several extracellular bacilli within the collagenous matrix (Panel B).
Bartonella are slender (0.5 to 2.0 μmol), curved, gram-negative rods.17 They can be cultured with difficulty on chocolate agar with carbon dioxide supplementation. They do not react with oxidase, urease, or catalase. Specific identification requires DNA analysis with the use of PCR, immunochemistry, or gas chromatography.
The histopathological characteristics of bartonella infection can range from a complete lack of immune response to necrotizing inflammation with abscess formation. B. henselae proliferates in the walls of capillaries and lymphatics and is ingested by macrophages. Vessels become thickened and occluded by the organisms, and macrophages showing epithelioid transformation accumulate around the infected vasculature. Subcapsular stellate necrosis18 is commonly seen in the lymph nodes. The nonspecific pathological findings in the present case may reflect a sampling error due to the small size of the specimen.
The identification of organisms in situ requires a silver-impregnation technique, and Warthin–Starry staining is the preferred method. Wear19 has proposed three criteria for the identification of bartonella species: the presence of filamentous bacilli in collagen fibers at any stage of the infection, Y-shaped branching bacteria (Chinese figures) that are intracellular or extracellular in either viable areas or areas of necrosis, and single bacilli within macrophages in the same lesion and in consecutive sections.
At least six species of bartonella are recognized as a cause of human disease. In the immunocompetent host, the chief clinical syndromes are Oroya fever and Verruga peruana complex, which are caused by B. bacilliformis. Trench fever, a generally nonlethal, louse-borne disease causing repeated bouts of fever lasting for approximately five days, is caused by B. quintana.20 Cat scratch disease is most often caused by B. henselae, although a small number of cases result from infection by the bacterium Afipia felis.21
Dr. Harris: Dr. Pasternack, will you tell us about the serologic testing and treatment of this patient?
Dr. Pasternack: Expedited serologic testing for bartonella antibodies was performed at the CDC. The patient's IgG titers to B. henselae and B. quintana exceeded 1:1024, which is the greatest dilution typically reported. Interestingly, both IgM antibody titers were less than the assay threshold of 1:20.
Follow-up serologic testing was performed about two months later, at which time IgG titers to both B. henselae and B. quintana remained positive at greater than 1:1024, but the B. henselae IgM antibody titer was 1:80 and the B. quintana IgM titer was still undetectable (less than 1:20). These studies were performed by immunofluorescence assay, and there is cross-reactivity between the quintana and henselae IgG antibodies, although the IgM antibody is thought to be more species specific. The clinical syndrome of B. quintana infection in nonwartime situations has largely been restricted to homeless persons. Body lice are thought to be the vector of transmission, and this patient had no clinical risk factors for that infection. We interpreted the results of the serologic studies as consistent with a diagnosis of B. henselae infection.
Infections attributable to this organism have been treated with a variety of antibiotics. In retrospective reviews ciprofloxacin, rifampin, trimethoprim–sulfamethoxazole, and gentamicin have all been associated with clinical benefit. In HIV-infected persons with bacillary angiomatosis, macrolides have been shown to be effective.22 In a single prospective, randomized trial of azithromycin in the treatment of cat scratch lymphadenitis, a standard five-day course of therapy was reported to be highly efficacious.23 We treated this adolescent with three weeks of clarithromycin at a dose of 500 mg twice daily.
Dr. Harris: I gather that none of the antibiotics that he was given before the diagnosis would be expected to have been effective?
Dr. Pasternack: Fluoroquinolones are thought to have activity against bartonella; however, three days of treatment would probably be inadequate to effect a substantial change.
Dr. Pelton: As I understand it, there is still some debate regarding the effectiveness of antibiotic therapy. The condition of most patients improves without antibiotics, and the studies may show some improvement with antibiotic therapy, but it is not clear how much or how much more quickly the improvement is with therapy.
Dr. Pasternack: Dr. Pelton makes an excellent point, which is that the use of antibiotics has been heavily debated. The trial of azithromycin is the only trial that suggests that therapy is beneficial. We think antibiotics may have a minor role, but many patients with this diagnosis do have an improvement without therapy. This patient had such a dramatic febrile illness that we wanted to treat him.
Dr. Pelton: I am not suggesting that this illness should not be treated. Indeed, the more traditional treatment is with parenteral gentamicin for some period before switching to oral therapy.
Dr. Pasternack: He did not appear acutely ill and had started to regain his appetite by the time of his admission to this hospital. He was so eager to leave the hospital that we embarked on a trial of oral therapy.
Dr. Harris: Dr. Goldstein, can you give us more recent follow-up?
Dr. Mark A. Goldstein (Adolescent Medicine): Within two weeks after antibiotic therapy was instituted, the patient reported subjective improvement in his appetite and diminished fatigue. He had gained almost 2 kg. Six weeks after the initiation of clarithromycin, the patient reported increased energy and complete return of his appetite. At the final follow-up visit, approximately 12 weeks after the initiation of antibiotic therapy and 5 1/2 months after he first become ill, the patient was asymptomatic and the results of physical examination were normal. He weighed 68.2 kg, which was more than he had before his illness. His hematocrit and erythrocyte sedimentation rate were normal.
Dr. Kim: CT two months after the initial CT showed nearly complete resolution of the lesions in the liver and spleen and resolution of the abdominal lymphadenopathy.
Dr. Harris: Dr. Pelton, would you like to make any final comments?
Dr. Pelton: I think the missing features that would have made the diagnosis more obvious would be abdominal pain and the clear presence of a papule or a lesion associated with a cat scratch. Fortunately, the treating physicians recognized the possible importance of the patient's exposure to cats.
Anatomical Diagnosis
Hepatitis caused by Bartonella henselae.
Dr. Pelton reports having served on advisory boards of Wyeth-Vaccines and GlaxoSmithKline, and having received speaking fees and grant support from Sanofi-Aventis.
Source Information
From the Department of Pediatric Infectious Disease, Boston Medical Center (S.I.P.); the Department of Pediatrics, Boston University School of Medicine (S.I.P.); and the Departments of Radiology (J.Y.K.) and Pathology (R.L.K.), Massachusetts General Hospital and Harvard Medical School — all in Boston.
References
Lodhi S, Sarwari R, Muzammil M, Salam A, Smego RA. Features distinguishing amoebic from pyogenic liver abscess: a review of 577 adult cases. Trop Med Int Health 2004;9:718-723.
Manjunatha S, McIntyre PB, Lynn A. Multiple liver abscesses in Crohn's disease. Br J Hosp Med 1992;47:375-376.
Gelfand MS, Hodgkiss T, Simmons BP. Multiple hepatic abscesses caused by Streptococcus milleri in association with an intrauterine device. Rev Infect Dis 1989;11:983-987.
Lechiche C, Corne P, Riviere S, Jonquet O, Le Quellec A. Fusobacterium necrophorum septic shock from colonic origin with multiple abscesses. Rev Med Interne 2001;22:198-199.
Brook I, Frazier EH. Microbiology of liver and spleen abscesses. J Med Microbiol 1998;47:1075-1080.
Farrer W, Kloser P, Ketyer S. Yersinia pseudotuberculosis sepsis presenting as multiple liver abscesses. Am J Med Sci 1988;295:129-132.
Strungs I, Farrell DJ, Matar LD, Dekker L, Franz RJ. Multiple hepatic abscesses due to Yersinia enterocolitica. Pathology 1995;27:374-377.
Hopfner M, Nitsche R, Rohr A, Harms D, Schubert S, Folsch UR. Yersinia enterocolitica infection with multiple liver abscesses uncovering a primary hemochromatosis. Scand J Gastroenterol 2001;36:220-224.
Brmbolic B. Multiple abscesses of the liver caused by Campylobacter jejuni. J Clin Gastroenterol 1995;20:307-309.
Chen CH, Fang YH, Chiang PM, Lin DT, Huang LM. Disseminated tuberculosis presenting as multiple hepatosplenic microabscesses and pancytopenia in a teenage boy. J Formos Med Assoc 2004;103:939-942.
Zheng RQ, Kudo M, Ishikawa E, Zhou P. Multiple tuberculous abscesses of the liver and the brain in a patient with acute leukemia. J Gastroenterol 2004;39:497-499.
Sood A, Midha V, Sood N, Sekhon A. Multiple tubercular liver abscesses in an immunocompetent person. Trop Gastroenterol 2001;22:144-145.
Verbanck J, Ponette J, Verbanck M, Vandewiele I, Segaert M. Sonographic detection of multiple Staphylococcus aureus hepatic microabscesses mimicking Candida abscesses. J Clin Ultrasound 1999;27:478-481.
Bass J, Vincent J, Person DA. The expanding spectrum of Bartonella infections. II. Cat-scratch disease. Pediatr Infect Dis J 1997;16:163-179.
Margileth AM, Wear DJ, English CK. Systemic cat scratch disease: report of 23 patients with prolonged or recurrent severe bacterial infection. J Infect Dis 1987;155:390-402.
Rizkallah MF, Meyer L, Ayoub EM. Hepatic and splenic abscesses in cat-scratch disease. Pediatr Infect Dis J 1988;7:191-195.
Eaton M, Nelson AM. Bartonella infections. In: Horsburgh CR, Nelson AM, eds. The pathology of emerging infections. Washington, D.C.: ASM Press, 1997.
Relman DA, Loutit JS, Schmidt TM. The agent of bacillary angiomatosis: an approach to the identification of uncultured pathogens. N Engl J Med 1990;323:1573-1580.
Wear DJ. Cat scratch disease and bacillary angiomatosis. In: Marty AM, Nelson AM, eds. Advances in diagnostic pathology of infectious disease. Washington, D.C.: Armed Forces Institute of Pathology, 1994:29-36.
Schwartzmann WA. Cat scratch disease and other Bartonella infections. Curr Treat Options Infect Dis 2000;2:155-62.
Jackson LA, Perkins BA, Wenger JD. Cat scratch disease in the United States: an analysis of three national databases. Am J Public Health 1993;83:1707-1711.
Koehler JE, Tappero JW. Bacillary angiomatosis and bacillary peliosis in patients infected with human immunodeficiency virus. Clin Infect Dis 1993;17:612-624.
Bass JW, Freitas BC, Freitas AD, et al. Prospective randomized double blind placebo-controlled evaluation of azithromycin for treatment of cat-scratch disease. Pediatr Infect Dis J 1998;17:447-452.(Stephen I. Pelton, M.D., )