This 14-month-old boy presented with fever of 6 days' duration without any symptoms. Specifically, there was no history of vomiting, diarrhea, rhinitis, cough, rash, arthritis, red eyes, red lips, or red palms.
His medical history was unremarkable. His development was normal. Travel history was remarkable in mat he visited Italy and Ireland on separate trips 3 months before the onset of the fever. In Italy, they had stayed in a villa; the child had played on the grounds, where there had been some chickens and other farm animals. There had been no foreign visitors at home.
He was nursed until age 10 months but has been eating a wide variety of table foods and whole milk since that time. There were two adult cats at home. There was no history of tuberculosis exposure.
On physical examination, he was a healthy-appearing boy who was sitting comfortably on his mother's lap. Weight, height, and head circumference were all in the 25th percentile. His temperature was 37.6° rectally in the doctor's office, although it had been 39.4° C earlier that day; he had received acetaminophen. The physical examination was entirely unremarkable.
Robert Listernick, MD, moderator: How long does fever have to continue before you worry about it?
Robert Tanz, MD, general academic pediatrician: That's a difficult question to answer in the abstract. Assuming that the child is well and has no ominous symptoms, most viral infections have fever for 3 days or less. I would guess that the two standard deviations out for an uncomplicated viral infection would be 5 or 6 days. As such, I'd be concerned that "something else" might be going on at the 6th day of fever. At a minimum, a complete physical examination at this point would be indicated.
Dr. Listernick: Is how the fever responds to antipyretics predictive of its etiology?
Dr. Tanz: Absolutely not. Some practitioners have commented that viral infections respond "more completely" to antipyretics, as opposed to the response seen in bacterial infections. This absolutely has been disproved.
Dr. Listernick: Does the setting in which the patient is being seen influence one's approach?
Jennifer Trainor, MD, pediatric emergency medicine specialist: In the emergency department, we're much more likely to perform laboratory testing on a patient than a primary care physician might, primarily because of a concern about follow-up. We don't have the luxury of knowing that the family will call or return if new symptoms arise or if the child's condition deteriorates.
Dr. Listernick: These parents were worried that the temperature was 104° rather than 101°. How high does a fever have to be before the physician worries?
Heather Haukness, MD, general pediatrician: Obviously, it's more important to the physician how a child appears and is behaving rather than the actual height of the temperature. Parents worry about febrile seizures. Febrile seizures are usually the heralding signs of an infection; it would be unusual to seize in the middle of a febrile illness. We have to educate them that fever is just a symptom, rather than an ominous condition üiat will harm their child.
Stanford Shulman, MD, pediatric infectious disease specialist: You didn't ask me how long before I start worrying about fever.
Dr. Listernick: That's because I'm afraid of your response!
Dr. Shulman: Nonetheless, I'll throw in my two cents. Obviously, I see a very biased sample of patients. However, when fever passes the 5day mark in a young child, I worry about an incomplete, unusual presentation of Kawasaki disease. I understand that Fm always carping about Kawasaki disease. However, the unfortunate fact is that we do see young children who have prolonged fever without any other signs or symptoms who ultimately develop coronary artery aneurysms. Kawasaki disease should at least cross the physician's mind. However, I admit that even I wouldn't recommend performing an echocardiogram at this point in this child.
Dr. Listernick: What should we do now?
Dr. Haukness: At a minimum, I would perform a complete blood count (CBC), a urinalysis, and urine culture. If there's an undiagnosed bacterial infection at this point, it would most likely be a urinary tract infection.
Dr. Listernick: No laboratory testing was performed at this office visit. He was seen 3 days later. During the interim, he had continued to have daily fever to as high as 39.6° C. He had no new symptoms. However, on physical examination, his abdomen was mildly protuberant. The liver was palpable 2 cm below the right costal margin and the spleen was palpable 4 cm below the left costal margin.
Dr. Shulman: Fortunately, except in very rare cases in which it has been associated with hemophagocytic syndrome, you can cross Kawasaki disease off the list because of the hepatosplenomegaly. The possibility of an unusual infection or a malignancy rises to the top of this list.
Dr. Listernick: Does he have infectious mononucleosis?
Dr. Shulman: I'd be much happier if he was 1 6 years rather than 16 months, but it certainly happens at this age. The heterophile test for Epstein Barr virus (EB V) associated infectious mononucleosis at this age is likely to be negative. Therefore, I would do IgM and IgG serology for EBV infection as well as for cytomegalovirus (CMV) and toxoplasmosis. Obviously, other testing is indicated at this point as well.
Dr. Listernick: Blood test results showed hemoglobin at 8.3 g/dL with MCV 8 1 ; white blood cell count of 4,100/mm3 with 8% neutrophils, 1 1% immature neutrophils, and 76% lymphocytes; and platelet count at 204,000/mm3. The reticulocyte count was 0.5% and the erythrocyte sedimentation rate was 31 mm/hour. SGPT 44 IU/L, SGOT 69 IU/L, and uric acid were normal. EBV, CMV, and toxoplasma IgM and IgG were all normal. What are your thoughts now?
Reggie Duerst, MD, pediatric oncologist: He's anemic and leukopenic. While malignancy is a possibility, I think that it's not probable. Leukemia would be the most likely malignancy in this age group, but any infiltrative process of the bone marrow generally produces thrombocytopenia as an initial manifestation. Hemolytic anemia is unlikely given the very low reticulocyte count, but I would want to see a Coombs test. Obviously, further testing is warranted, but at a first glance, I would say that infection would be at the top of my list. The mild anemia and leukopenia may be the result of hypersplenism, sequestration by a large spleen.
Morris Kletzel, MD, pediatric oncologist: I agree with everything that has been said, but as a rule of thumb, if two cell lines are depressed, I believe that examination of the bone marrow is warranted. We've certainly seen children with leukemia who have had either perfectly normal CBCs or any combination of cell lines suppressed. I would also culture the bone marrow for bacteria, viruses, tuberculosis, and fungi.
Dr. Listernick: Are there any classic findings on the CBC that would point to a bone marrow infiltrative process?
Dr. Duerst: A leukoerythroblastic picture would be the most telling. The appearance of metamyelocytes, myelocytes, and other immature white blood cell forms in the peripheral smear suggests that they are being "forced out" of the bone marrow. In the normal state, there are control mechanisms which prevent the release of these cells into the circulation.
Dr. Listernick: At the time of the initial evaluation at Children's Memorial, an abdominal ultrasound was performed to assess the quality of the liver and spleen as well as to look for any abdominal masses.
Ellen Benya, MD, pediatric radiologist: The liver appeared normal. The spleen was enlarged, and within its body were numerous hypoechoic lesions measuring 2 to 3 cm. There were no abdominal masses.
Dr. Duerst: I know everyone in the room is thinking about malignancy. However, this presentation would be extremely unusual even for lymphoma. Metastatic disease to the spleen, even in neuroblastoma, is very rare.
Dr. Listernick: Actually, after I reviewed the ultrasound I was fairly convinced that I had a diagnosis - systemic cat scratch disease (CSD). I was so sure that I sent serologic testing and started him on oral rifampin. Tell us about cat scratch disease.
Dr. Shulman: Lymphadenitis accounts for roughly 90% of confirmed cases of CSD. Classically, a papule develops 3 to 10 days after exposure to the cat at the site of inoculation. Regional lymphadenitis occurs 1 to 3 weeks later. Systemic symptoms such as low grade fever, malaise, and fatigue are common complaints. Although involvement of a single regional node is most common, multiple areas of lymphadenopathy may occur. Although spontaneous resolution over weeks to months is often the course, at least 20% of the lymph nodes suppurate and require aspiration. At times, chronic lymphadenopathy is mistaken for a malignancy, only to be identified as CSD on excisional biopsy. The classic etiologic agent is Bartonella henselae. The risk of transmission has been found to be highest with exposure to kittens, the majority of whom are chronically bacteremic with Bartonella.
Dr. Listernick: What about the remaining 10% "atypical" cases?
Dr. Shulman: Other classic, but less common, manifestations of Bartonella infection include Parinauds oculoglandular syndrome (conjunctivitis and ipsilateral preauricular lymphadenopathy), encephalitis, and osteomyelitis. Finally, CSD may present as a fever of unknown origin with granulomatous hepatitis and splenitis. In these cases, biopsy of the liver shows that it's studded with small granulomas.
Dr. Benya: To complete the picture, the ultrasonographic findings are classic for this form of systemic CSD.
Dr. Listernick: How is the diagnosis of CSD confirmed?
Judith Guzman, DO, pediatric infectious disease specialist: Traditionally, the diagnosis of CSD is made serologically through the use of an indirect fluorescent antibody (IFA) test. The sensitivity of the IFA is thought to be between 85% to 95%, while me specificity is above 90%. Unfortunately, the results of serologic testing for CSD from commercial laboratories are less reliable; the best results are obtained when the serology is performed by the Centers for Disease Control and Prevention (CDC).
Dr. Shulman: On occasion, we have been able to culture Bartonella henslae from the pus of aspirated lymph nodes if the pus is inoculated into a blood culture bottle. In general, it's a very difficult organism to isolate.
Dr. Guzman: The lab has to be told to hold the bottle for at least 6 weeks, as it is a very slow-growing organism.
Dr. Listernick: What about the CSD skin test?
Dr. Guzman: The CSD skin test is a thing of the past. It used to be prepared from the pus of aspirated lymph nodes from patients with confirmed CSD. It was injected intradermally, and one looked for a delayed hypersensitivity T-cell response. For obvious reasons, we no longer use it.
Dr. Shulman: I still have a vial in the freezer that was prepared from 20 cc of aspirated pus from a patient with CSD. It was boiled for 5 hours, so the risk of transmission of any infectious agent is nil. Regardless, I wouldn't use it now.
Dr. Listernick: Wise move. Moving on, I treated this child with rifampin. The CSD titers, sent first to a commercial lab and next to the CDC, were negative. He remained febrile for the next 10 days and had continued splenomegaly. His hemoglobin dropped to a nadir of 6.7 g/dL with a low reticulocyte count, white blood count to 2,900 with a similar differential count as before, and platelet count to a low of 90,000/mm3. A bone marrow aspirate and biopsy were normal, as were bacterial, viral, fungal, and Mycobacterium tuberculosis cultures. Serologic testing for histoplasmosis, blastomycosis, tularemia, and brucellosis were negative. Multiple blood cultures were sterile. Tuberculin skin testing was negative. Any thoughts?
Dr. Duerst: This is not a malignancy. However, I would perform a chest x-ray and computerized tomography (CT) scan of the abdomen to better define the splenic lesions and to look for other areas of disease.
Dr. Benya: The chest x-ray was normal. Multiple low density lesions in the spleen, similar to those seen on ultrasound, were identified. They were not hypervascular as would be seen in hemangiomatosis. I have not seen anything like this in either leukemia or lymphoma; there is no regional lymphadenopathy suggestive of these diseases. I would have to say that a granulomatous process is still highest on my list.
Dr. Listernick: At this point I was stymied, although I still thought that he had CSD despite the negative serology. I asked the surgeons if they would consider what I thought of as a cringe-worthy procedure, a splenic biopsy. Is this a particularly common request?
Robert Arensman, MD, pediatric surgeon: This is the first splenic biopsy that I can remember being asked to perform. As the saying goes, "When all else fails, call a surgeon."
Dr. Haukness: Did anybody consider asking interventional radiology to perform a needle biopsy?
Dr. Benya: Technically, it would have been easy. The major issue is whether one could have obtained enough tissue to make a definitive diagnosis.
Dr. Arensman: We discussed our approach and decided to perform a laparoscopic biopsy and get as large a biopsy sample as necessary. It was also felt to be a good way to control any unwanted hemorrhaging. At surgery, we found the spleen covered with 1 to 2 cm cystic lesions filled with serosanguineous fluid. We performed several biopsies trying to get as large a sample as possible.
Pauline Chou, MD, pediatric pathologist: This is a well-formed granuloma, a collection of histiocytes. There is no central necrosis. Gram stain and fungal stains were negative. However, the WarthinStarry stain was positive for bacillary forms; this is very consistent with Bartonella infection. Our final diagnosis was noncaseating granulomas with bacillary forms highly suggestive of CSD.
Dr. Listernick: What is the classic pathology of CSD?
Dr. Chou: Classically, we see rounded or stellate granulomas with central granular debris. The Warthin-Starry stain, although not specific for Bartonella infection, shows these pleomorphic bacilli, aiding us in the diagnosis.
Dr. Listernick: The serology performed twice at the CDC was negative. Cultures of the splenic tissue and the bone marrow have been negative. In addition, the family had the cats tested for Bartonella and the serology and cultures were negative. After reviewing a recent article by Arisoy1 detailing 19 cases of hepatosplenic CSD, I treated with a 4-week regimen of oral azithromycin and rifampin. After 2 days, he became afebrile and remained so. After 2 weeks, his hemoglobin and white blood cell and platelet counts returned to normal. One month later, he had persistent splenomegaly. However, after an additional month, the splenomegaly has disappeared and he has remained well.
Interestingly, none of the 19 patients in the article cited had any evidence of cytopenias or hypersplenism, whereas 80% had elevated white blood cell counts. What did he have?
Dr. Tanz: Could he have had a different Bartonella infection?
Dr. Guzman: There are four strains of Bartonella known to cause human disease. Bartonella henselae is the agent of CSD. Bartonella quintana causes trench fever, which this doesn't resemble clinically; in addition, trench fever is only seen in homeless and immunocompromised patients. I found one case of Bartonella elizabethae causing endocarditis, and Bartonella bacilliformis causes Oroya fever, a highly fatal disease if untreated that occurs exclusively in the Andes.
Dr. Listernick: What else stains positive on a Warthin-Starry stain?
Dr. Guzman: Helicobacter pylori, some Gram-negative bacilli that are not pathogenic in humans, and a few spirochetes, which this organism obviously isn't.
Dr. Listernick: Where does that leave us?
Dr. Guzman: One remote possibility is an organism called Afipia felis, which was discovered in 1988 at the Armed Forces Institute of Pathology, hence its name. It had been isolated from a few patients who had clinical CSD. However, it eventually was discredited as the agent of CSD and has never been implicated as the cause of hepatosplenic CSD.
Dr. Listernick: We're trying to find a lab that assesses Bartonella DNA by polymerase chain reaction. Unfortunately, we only have formalin-fixed tissue. For the moment, I'm saying that he had CSD but I'm open to any better ideas. Thank you, everybody.
1. Arisoy ES, Correa AG, Wagner ML, Kaplan SL. Hepatosplenic cat-scratch disease in children: selected clinical features and treatment. Clin Infect Dis. 1999;28(4):778-84.