Index of Suspicion
http://www.100md.com
《新英格兰医药杂志》
In this Journal feature, information about a real patient is presented in stages (boldface type) to an expert clinician, who responds to the information, sharing his or her reasoning with the reader (regular type). The authors' commentary follows.
A 26-year-old woman with end-stage renal disease from primary membranoproliferative glomerulonephritis presented with a two-week history of intermittent fever, with temperatures as high as 39°C. She had received her second cadaveric renal transplant 11 months previously.
Disease categories to consider in a febrile patient who has received a transplanted organ include infection, rejection, a post-transplantation lymphoproliferative disorder, and the reappearance of an underlying inflammatory disease that resulted in previous organ failure and transplantation. In this patient, both routine community-acquired infections and opportunistic infections associated with immunosuppression deserve consideration; the differential diagnosis of the latter would be shaped by the patient's epidemiologic characteristics (e.g., her travel history) and additional clinical and laboratory data. Fever caused by cellular rejection is often accompanied by allograft dysfunction, so a new elevation in the level of serum creatinine in conjunction with fever would raise this possibility. Post-transplantation lymphoproliferative disorder, which may occur six weeks or more after transplantation, is also possible. If this patient had received treatment for prior episodes of rejection it would place her at increased risk for this complication.
The patient had received her first transplant 10 years before presentation; she had received the second transplant as a result of allograft failure due to chronic rejection. Two weeks before presentation, she had a fever almost daily; the fevers were associated with diffuse myalgias and drenching sweats and chills. She was evaluated by her local physician, who was unable to identify an obvious infectious cause; no new treatment was prescribed. She reported generalized malaise, anorexia, and a weight loss of 1.4 kg (3 lb) since the onset of the fevers. In addition, she noted slight swelling and nontender erythema of her left index finger, which had started one week before presentation. She reported that she had not had any other symptoms or recent contact with sick persons and that she had not traveled out of Michigan, her state of residence.
Whereas myalgias are nonspecific, a new, localized change involving a swollen finger and erythema is highly clinically significant and frames the differential diagnosis. If the abnormalities appear to be confined to the skin, the morphologic features become important; well-demarcated erythema implies cellulitis, whereas a nonblanching lesion increases the likelihood of an embolic infection (e.g., endocarditis or fungal disease) or a vasculitis. The tempo of this presentation and the appearance of erythema after one week of fever argue strongly against one of the usual gram-positive causes of cellulitis. Certain opportunistic infections — such as nocardial or mycobacterial infection — may cause an indolent soft-tissue infection that mimics cellulitis.
The medications she took were cyclosporine, mycophenolate mofetil, prednisone (10 mg per day), propranolol, amlodipine, and valacyclovir (prescribed by an emergency department physician five days before presentation for presumed herpetic whitlow, without confirmatory tests). Approximately six weeks before presentation, the patient received bolus doses of corticosteroids (250 mg of intravenous methylprednisolone daily for three days) as treatment for moderate acute cellular rejection. She was married and did not have children or pets. She worked as a veterinary assistant and had recently had contact with various domestic animals, mostly cats and dogs.
Despite the low current dose of corticosteroids, the patient must be considered to have clinically significant cellular immunodeficiency in the light of her treatment with two other agents that inhibit T-cell function and the recent bolus doses of methylprednisolone for rejection. Patients who have received a solid-organ transplant remain at increased risk for herpesvirus infection; a diagnosis of herpetic whitlow is possible, although the finger is not a common site of herpesvirus reappearance.
The patient's occupational history is intriguing. Several zoonotic infections that can be worrisome in immunocompromised hosts are transmitted from cats. Toxoplasmosis is most often acquired through inadvertent ingestion of feline feces and may be manifested with atypical symptoms in a transplant recipient. Enteric pathogens include salmonella and campylobacter species; the latter especially should be considered, since infection with any of several campylobacter-like organisms may manifest as bacteremia and migratory cellulitis, rather than enteritis, in immunocompromised patients. Another cat-borne illness that is heralded by fever and new cutaneous findings is infection with Bartonella henselae. The lesions of bacillary angiomatosis are typically nonblanching and vascular, which may lead to their confusion with Kaposi's sarcoma, another opportunistic complication of transplantation in persons previously exposed to human herpesvirus 8. The physical examination would help distinguish among these possibilities.
The patient's temperature was 37.2°C, her heart rate was 74 beats per minute, her respiratory rate was 14 breaths per minute, and her blood pressure was 96/50 mm Hg. She was a thin, pale woman in no distress. The oropharynx was free of exudates, and the neck was supple. The chest was clear on auscultation; heart sounds were regular, and a grade 3 systolic ejection murmur was heard throughout the precordium. Abdominal examination revealed no abnormalities; in the region of the bilateral renal allografts, no audible bruits or palpable tenderness was detected. There were multiple enlarged, soft, rubbery, mobile left axillary lymph nodes, with the largest measuring approximately 2 cm in diameter. No epitrochlear or clavicular lymph nodes were palpable. The skin was normal except for a slightly nodular, soft, erythematous lesion, 8 mm in diameter, on the medial aspect of the distal phalanx of the left index finger (Figure 1).
Figure 1. Photograph Showing a Nodular, Erythematous Lesion on the Medial Aspect of the Distal Phalanx of the Left Index Finger.
The patient's white-cell count was 4000 per cubic millimeter, with a differential count of 81 percent neutrophils, 12 percent lymphocytes, 3 percent monocytes, and 4 percent eosinophils. The hematocrit was 22 percent, and the platelet count was 125,000 per cubic millimeter. The results of liver-function tests were normal, and the lactate dehydrogenase level was 156 U per liter. The creatinine level was 2.0 mg per deciliter (176.8 μmol per liter), similar to previously observed levels, and the results of urinalysis were normal.
The patient's low blood pressure raises the possibility of sepsis syndrome due to an underlying infection. The cardiac murmur probably represents a flow murmur due to her anemia, but with a distal digital lesion, endocarditis must be considered. The morphologic features and position of the digital lesion do not support the possibility of herpetic whitlow. Hematogenously spread bacterial, fungal, or mycobacterial infections could have this appearance. The tempo of the illness suggests the presence of indolent organisms such as bartonella and nocardia. If the lesion were caused by direct inoculation, I would think primarily of certain mycobacteria (e.g., Mycobacterium marinum) and environmental fungi, such as aspergillus or cryptococcus. Several of these organisms are relevant to her occupation, assuming the appropriate contact with animals such as fish, birds, cats, or dogs.
Given the hematologic abnormalities, I must also consider bone marrow involvement with a disseminated infection. Granulomatous infections such as fungal diseases (especially histoplasmosis and cryptococcosis) and mycobacterial diseases are the most common marrow-infiltrative infections. The presence of adenopathy may also indicate disseminated infection or cancer (e.g., lymphoma).
Cultures of blood and urine obtained on admission revealed no growth. The results of serologic tests for Epstein–Barr virus (EBV), cytomegalovirus (CMV), parvovirus B19, and bartonella were all negative for acute infection; serum polymerase-chain-reaction (PCR) assays for EBV and CMV were also negative. Transthoracic echocardiography revealed a bicuspid aortic valve but no evidence of endocarditis. Computed tomography of the chest, abdomen, and pelvis revealed left axillary lymph-node enlargement but no other remarkable findings. The patient was presumed to have an unspecified viral illness. The mycophenolate mofetil was discontinued, and she was discharged home, with close outpatient observation anticipated.
Clinicians must be cautious in the interpretation of the results of serologic tests in immunosuppressed patients. These patients may have a component of humoral immunodeficiency that impairs the ability to mount typical antibody responses to acute or recrudescent infection. Thus, the absence of IgM antibodies to EBV or CMV, for example, would have limited negative predictive value. Genomic or antigen-based studies are generally more reliable in this setting. Parvovirus B19 infection was a consideration because of the hematologic cytopenias, but it would not fully explain the skin findings. The absence of detectable parvovirus B19 in a PCR study of the blood provides strong evidence against this infection.
At this point, I am uncomfortable ascribing the fever to an undefined viral illness. The clinical setting, duration of illness, and associated physical findings warrant additional evaluation for specific treatable causes. Since less invasive tests have been unrevealing, histologic evaluation of the skin lesion, enlarged lymph nodes, or bone marrow should be considered next.
Two days after discharge from the hospital, the patient was readmitted because of persistent fever, malaise, and anorexia. The small erythematous lesion on her left index finger had persisted without change. While she was hospitalized, the patient continued to have a fever, with temperatures up to 40.5°C. Blood cultures from the previous admission remained sterile after a one-week incubation period. The results of all the following investigations were negative: repeated blood and urine cultures; PCR assays for human herpesvirus 6, CMV, EBV, hepatitis C virus, ehrlichia, and babesia; serologic tests for human immunodeficiency virus, hepatitis B virus, and toxoplasma; a test for urinary histoplasma; and tests for rheumatoid factor and antinuclear antibody. The erythrocyte sedimentation rate and the C-reactive protein level were normal, and the results of a serum PCR assay for polyomavirus type BK (BK virus) was positive, at 5200 copies per milliliter.
Reactivation of BK virus occurs commonly in transplant recipients who are infected with this agent; the most common associated clinical syndrome is hemorrhagic interstitial cystitis. I doubt that BK virus infection explains the entire picture here. A general principle in the evaluation of a patient with fever of unknown origin is to focus on and pursue any focal abnormalities. Biopsy of the involved tissue may have been a more efficient approach in this case.
At this point, I still favor a diagnosis of disseminated infection with lymphatic and bone marrow involvement; infection with bartonella is an intriguing possibility because of the patient's occupation and the presence of a skin lesion. Mycobacteria and fungi have not yet been ruled out as possible causes; post-transplantation lymphoproliferative disorder would be lower on my differential diagnosis.
The patient's white-cell count decreased to 900 per cubic millimeter (with an absolute neutrophil count of 200 per cubic millimeter), the hematocrit decreased to 20 percent, and the platelet count decreased to 75,000 per cubic millimeter. A bone marrow biopsy revealed normocellular marrow, which was negative for signs of cancer, infection with acid-fast bacilli, fungal infection, or other, atypical infections.
Bone marrow examination has value in establishing the diagnosis of an infection such as histoplasmosis or tuberculosis when there are abnormalities in the peripheral-blood count. The results must be interpreted cautiously in this situation, since immunosuppressive therapy may mask important histopathologic clues, particularly granulomas and lymphoid aggregates suggestive of lymphoproliferative disease.
Examination of a specimen obtained by core biopsy of the left axillary lymph node revealed vascular proliferation that appeared to be morphologically benign. However, the possibility of cancer could not be definitively ruled out because there was insufficient tissue for analysis.
The vascular proliferation seen in the lymph-node aspirate from the core-biopsy specimen must be pursued with an excisional lymph-node biopsy. The diagnoses suggested by the vascular proliferation are Kaposi's sarcoma, angioimmunoblastic lymphadenopathy, and bartonellosis. A larger specimen would be needed to demonstrate the typical findings in Kaposi's sarcoma, which would be slit-like vascular proliferation with extravasation of erythrocytes. When Kaposi's sarcoma involves the lymph nodes or other visceral organs, fever may be a prominent feature. In angioimmunoblastic lymphadenopathy, one of the atypical lymphoproliferative disorders that are often associated with viral oncogenesis, systemic symptoms are common. Finally, lymphatic involvement with bartonella should be easily demonstrable on examination of a lymph-node biopsy specimen if proper staining is done for the causative organism. Bartonellosis and disseminated Kaposi's sarcoma may be very difficult to distinguish without a biopsy, but I favor the former, given the patient's occupation and the relatively low probability of previous infection with human herpesvirus 8.
An excisional biopsy of the patient's left axillary lymph node was performed. Pathological examination showed multiple coalescent nodules of proliferated blood vessels surrounded by amorphous, granular, eosinophilic deposits (Figure 2). A Warthin–Starry stain revealed the amorphous deposits to be aggregates of small, rod-shaped bacteria (Figure 3); the results of further study with special stains for acid-fast bacilli and fungi were negative. These findings were diagnostic of bacillary angiomatosis.
Figure 2. Biopsy Specimen of the Left Axillary Lymph Node, Showing Multiple Amorphous, Granular Deposits (Arrow) between and around Proliferated Blood Vessels (Arrowheads) (Hematoxylin and Eosin, x40).
Figure 3. Biopsy Specimen of the Left Axillary Lymph Node Showing Dark, Granular Deposits (Warthin–Starry, x40).
The deposits are aggregates of rod-shaped bacteria.
The patient remained persistently febrile until treatment with oral levofloxacin was initiated for bartonella infection. Repeated serologic testing for bartonella was performed, three weeks after the previous (negative) titers had been obtained. Whereas the IgM titers for both B. henselae and B. quintana remained normal (range, less than 1:20), the IgG titers for both species were now positive, at greater than 1:1024. Two weeks after antibiotic therapy was initiated, the patient reported that her symptoms had dissipated, and the lesion on her finger and the lymphadenopathy had almost resolved.
This case highlights the potential limitations of some diagnostic tests for opportunistic infections, particularly in immunosuppressed patients. Not only were the genomic and serologic evaluations extremely expensive; the PCR assay yielded a positive result (for polyomavirus) that was unrelated to the final diagnosis, and the initial serologic tests yielded a false negative result (for bartonella). Histologic and microbiologic examinations of clinically involved tissues may be warranted even when the results of serologic testing are negative.
Commentary
Fever of unknown origin is defined as a temperature of 38.3°C or greater on several occasions over a period of more than three weeks, accompanied by an uncertain diagnosis after three days of inpatient investigation or three outpatient visits.1 The majority of cases are caused by infections, tumors, and various noninfectious inflammatory diseases.1,2 However, the spectrum of disease in immunocompromised patients who present with fever of unknown origin tends to include infections more often than other illnesses.1 The increased risk of rapid clinical deterioration in immunocompromised patients underscores the importance of prompt diagnosis.3 Errors in the processing of available information during the diagnostic evaluation can result in delays and missed diagnoses,2,4 as occurred in the case under discussion.
The primary impediment to a more timely diagnosis in this case was the excessive reliance on noninvasive diagnostic tests rather than tissue biopsy for direct examination early in the presentation. As the discussant notes, the findings of an erythematous lesion on the left index finger and of ipsilateral axillary lymphadenopathy warranted directed evaluation. The negative results on the serologic tests markedly reduced the treating physicians' suspicion of bartonella infection and thus delayed biopsy and appropriate therapy.
Several aspects of the diagnostic challenge merit review. Undirected serologic and immunologic tests have poor predictive value for evaluating fever of unknown origin because of the low prevalence of the disorders for which they screen. Even with tests with high specificity, the rate of false positive results remains unacceptably high.1 The positive PCR assay for BK virus provides an example of this general principle. Conversely, an example of the opposite phenomenon occurred in the search for bartonella infection (since there was a high pretest probability of the infection and a false negative test result).
Although they are commonly used to evaluate the possibility of bartonella infection,5 serologic tests have variable sensitivities and specificities.6,7 These discrepancies are probably related to differences in the available tests and to spectrum bias, which can occur when a diagnostic test is used in a patient who has characteristics different from those of the reference population in which the test characteristics were defined.8 The discussant correctly emphasizes the low predictive value of serologic testing given that the patient's antibody response was probably blunted because of impaired T-cell function.3,9 Although the results of blood cultures were also negative, the yield of this test is generally low.10
A PCR test is available for bartonella,11 but it requires examination of tissue samples. In this case, PCR examination of the specimen from the core biopsy might have provided an earlier diagnosis. Repeated serologic testing after a reduction in the level of immunosuppression revealed a delayed antibody response, with elevated IgG titers. The diagnosis of bacillary angiomatosis often requires a biopsy of lesions and the subsequent detection of characteristic histopathological features or identification of the bacillus by Warthin–Starry staining or electron microscopy.5
Both B. henselae and B. quintana have been associated with cat scratch disease, bacillary angiomatosis, bacillary peliosis, splenitis, osteomyelitis, bacteremia, and endocarditis.10,12,13 Whereas cat scratch disease most often affects healthy people, bacillary angiomatosis occurs predominantly in patients with the acquired immunodeficiency syndrome. Cases also have been described in organ-transplant recipients as well as in immunocompetent patients.5,14 Exposure usually occurs through the scratch or bite of a cat.13,15 This patient was probably exposed while working as a veterinary assistant.
Patients with bacillary angiomatosis often present with cutaneous or subcutaneous vascular proliferative lesions at the site of inoculation, as did this patient. These lesions can be difficult to distinguish clinically from those associated with Kaposi's sarcoma.16 Extracutaneous lesions may involve bone, mucosal surfaces, the central nervous system, and even bone marrow,5 providing a potential explanation for this patient's pancytopenia.17 Although the blood cultures were sterile, a bacteremic syndrome probably accounts for this patient's insidious, prolonged constellation of symptoms.5 In general, prompt treatment of bacillary angiomatosis may prevent the illness and death associated with progressive disease.18
Despite their early recognition of bartonella infection as a potential cause of this patient's fever of unknown origin, the physicians pursued a noninvasive investigation. Appropriate interpretation of negative serologic tests in an immunosuppressed patient requires that a high index of suspicion for probable causes be maintained.
Supported by a Career Development Award from the Health Services Research and Development Program of the Department of Veterans Affairs and a Patient Safety Developmental Center Grant from the Agency for Healthcare Research and Quality (P20-HS11540, to Dr. Saint).
We are indebted to Camilla Payne, R.N., for her careful review of an earlier draft of this manuscript; to Kamal Amin, M.D., for his generous provision of the photograph used in Figure 1; and to Riccardo Valdez, M.D., for his assistance with photomicrographs and pathologic interpretation.
Source Information
From the Department of Pediatrics, University of Michigan Medical School, Ann Arbor (U.D.P.); the Department of Medicine, University of California, San Francisco, School of Medicine, San Francisco (H.H.); and the Department of Veterans Affairs Health Services Research and Development Center of Excellence, and the Department of Internal Medicine, University of Michigan — both in Ann Arbor (U.D.P., S.S.).
Address reprint requests to Dr. Patel at the University of Michigan, Robert Wood Johnson Clinical Scholars Program, 6312 Medical Sciences Bldg. I, 1150 W. Medical Center Dr., Ann Arbor, MI 48109-0604, or at patelu@umich.edu.
References
Knockaert DC, Vanderschueren S, Blockmans D. Fever of unknown origin in adults: 40 years on. J Intern Med 2003;253:263-275.
Arnow PM, Flaherty JP. Fever of unknown origin. Lancet 1997;350:575-580.
Pizzo PA. Fever in immunocompromised patients. N Engl J Med 1999;341:893-900.
Kassirer JP, Kopelman RI. Cognitive errors in diagnosis: instantiation, classification, and consequences. Am J Med 1989;86:433-441.
Adal KA, Cockerell CJ, Petri WA. Cat scratch disease, bacillary angiomatosis, and other infections due to Rochalimaea. N Engl J Med 1994;330:1509-1515.
Sander A, Posselt M, Oberle K, Bredt W. Seroprevalence of antibodies to Bartonella henselae in patients with cat scratch disease and in healthy controls: evaluation and comparison of two commercial serological tests. Clin Diagn Lab Immunol 1998;5:486-490.
Bergmans AM, Peeters MF, Schellekens JF, et al. Pitfalls and fallacies of cat scratch disease serology: evaluation of Bartonella henselae-based indirect fluorescence assay and enzyme-linked immunoassay. J Clin Microbiol 1997;35:1931-1937.
Ransohoff DF, Feinstein AR. Problems of spectrum and bias in evaluating the efficacy of diagnostic tests. N Engl J Med 1978;299:926-930.
Denton MD, Magee CC, Sayegh MH. Immunosuppressive strategies in transplantation. Lancet 1999;353:1083-1091.
Cotell SL, Noskin GA. Bacillary angiomatosis: clinical and histologic features, diagnosis, and treatment. Arch Intern Med 1994;154:524-528.
Relman DA, Loutit JS, Schmidt TM, Falkow S, Tompkins LS. The agent of bacillary angiomatosis: an approach to the identification of uncultured pathogens. N Engl J Med 1990;323:1573-1580.
Koehler JE, Quinn FD, Berger TG, LeBoit PE, Tappero JW. Isolation of Rochalimaea species from cutaneous and osseous lesions of bacillary angiomatosis. N Engl J Med 1992;327:1625-1631.
Zangwill KM, Hamilton DH, Perkins BA, et al. Cat scratch disease in Connecticut: epidemiology, risk factors, and evaluation of a new diagnostic test. N Engl J Med 1993;329:8-13.
Cline MS, Cummings OW, Goldman M, Filo RS, Pescovitz MD. Bacillary angiomatosis in a renal transplant recipient. Transplantation 1999;67:296-298.
Tappero JW, Mohle-Boetani J, Koehler JE, et al. The epidemiology of bacillary angiomatosis and bacillary peliosis. JAMA 1993;269:770-775.
Tappero JW, Koehler JE. Bacillary angiomatosis or Kaposi's sarcoma? N Engl J Med 1997;337:1888-1888.
Kiriakos J, Ranchin B, Gillet Y, et al. Systemic bacillary angiomatosis after kidney transplantation. Pediatr Nephrol 2003;18:C46-C46. abstract.
van der Wouw PA, Hadderingh RJ, Reiss P, Hulsebosch HJ, Walford N, Lange JM. Disseminated cat-scratch disease in a patient with AIDS. AIDS 1989;3:751-753.(Uptal D. Patel, M.D., Har)
A 26-year-old woman with end-stage renal disease from primary membranoproliferative glomerulonephritis presented with a two-week history of intermittent fever, with temperatures as high as 39°C. She had received her second cadaveric renal transplant 11 months previously.
Disease categories to consider in a febrile patient who has received a transplanted organ include infection, rejection, a post-transplantation lymphoproliferative disorder, and the reappearance of an underlying inflammatory disease that resulted in previous organ failure and transplantation. In this patient, both routine community-acquired infections and opportunistic infections associated with immunosuppression deserve consideration; the differential diagnosis of the latter would be shaped by the patient's epidemiologic characteristics (e.g., her travel history) and additional clinical and laboratory data. Fever caused by cellular rejection is often accompanied by allograft dysfunction, so a new elevation in the level of serum creatinine in conjunction with fever would raise this possibility. Post-transplantation lymphoproliferative disorder, which may occur six weeks or more after transplantation, is also possible. If this patient had received treatment for prior episodes of rejection it would place her at increased risk for this complication.
The patient had received her first transplant 10 years before presentation; she had received the second transplant as a result of allograft failure due to chronic rejection. Two weeks before presentation, she had a fever almost daily; the fevers were associated with diffuse myalgias and drenching sweats and chills. She was evaluated by her local physician, who was unable to identify an obvious infectious cause; no new treatment was prescribed. She reported generalized malaise, anorexia, and a weight loss of 1.4 kg (3 lb) since the onset of the fevers. In addition, she noted slight swelling and nontender erythema of her left index finger, which had started one week before presentation. She reported that she had not had any other symptoms or recent contact with sick persons and that she had not traveled out of Michigan, her state of residence.
Whereas myalgias are nonspecific, a new, localized change involving a swollen finger and erythema is highly clinically significant and frames the differential diagnosis. If the abnormalities appear to be confined to the skin, the morphologic features become important; well-demarcated erythema implies cellulitis, whereas a nonblanching lesion increases the likelihood of an embolic infection (e.g., endocarditis or fungal disease) or a vasculitis. The tempo of this presentation and the appearance of erythema after one week of fever argue strongly against one of the usual gram-positive causes of cellulitis. Certain opportunistic infections — such as nocardial or mycobacterial infection — may cause an indolent soft-tissue infection that mimics cellulitis.
The medications she took were cyclosporine, mycophenolate mofetil, prednisone (10 mg per day), propranolol, amlodipine, and valacyclovir (prescribed by an emergency department physician five days before presentation for presumed herpetic whitlow, without confirmatory tests). Approximately six weeks before presentation, the patient received bolus doses of corticosteroids (250 mg of intravenous methylprednisolone daily for three days) as treatment for moderate acute cellular rejection. She was married and did not have children or pets. She worked as a veterinary assistant and had recently had contact with various domestic animals, mostly cats and dogs.
Despite the low current dose of corticosteroids, the patient must be considered to have clinically significant cellular immunodeficiency in the light of her treatment with two other agents that inhibit T-cell function and the recent bolus doses of methylprednisolone for rejection. Patients who have received a solid-organ transplant remain at increased risk for herpesvirus infection; a diagnosis of herpetic whitlow is possible, although the finger is not a common site of herpesvirus reappearance.
The patient's occupational history is intriguing. Several zoonotic infections that can be worrisome in immunocompromised hosts are transmitted from cats. Toxoplasmosis is most often acquired through inadvertent ingestion of feline feces and may be manifested with atypical symptoms in a transplant recipient. Enteric pathogens include salmonella and campylobacter species; the latter especially should be considered, since infection with any of several campylobacter-like organisms may manifest as bacteremia and migratory cellulitis, rather than enteritis, in immunocompromised patients. Another cat-borne illness that is heralded by fever and new cutaneous findings is infection with Bartonella henselae. The lesions of bacillary angiomatosis are typically nonblanching and vascular, which may lead to their confusion with Kaposi's sarcoma, another opportunistic complication of transplantation in persons previously exposed to human herpesvirus 8. The physical examination would help distinguish among these possibilities.
The patient's temperature was 37.2°C, her heart rate was 74 beats per minute, her respiratory rate was 14 breaths per minute, and her blood pressure was 96/50 mm Hg. She was a thin, pale woman in no distress. The oropharynx was free of exudates, and the neck was supple. The chest was clear on auscultation; heart sounds were regular, and a grade 3 systolic ejection murmur was heard throughout the precordium. Abdominal examination revealed no abnormalities; in the region of the bilateral renal allografts, no audible bruits or palpable tenderness was detected. There were multiple enlarged, soft, rubbery, mobile left axillary lymph nodes, with the largest measuring approximately 2 cm in diameter. No epitrochlear or clavicular lymph nodes were palpable. The skin was normal except for a slightly nodular, soft, erythematous lesion, 8 mm in diameter, on the medial aspect of the distal phalanx of the left index finger (Figure 1).
Figure 1. Photograph Showing a Nodular, Erythematous Lesion on the Medial Aspect of the Distal Phalanx of the Left Index Finger.
The patient's white-cell count was 4000 per cubic millimeter, with a differential count of 81 percent neutrophils, 12 percent lymphocytes, 3 percent monocytes, and 4 percent eosinophils. The hematocrit was 22 percent, and the platelet count was 125,000 per cubic millimeter. The results of liver-function tests were normal, and the lactate dehydrogenase level was 156 U per liter. The creatinine level was 2.0 mg per deciliter (176.8 μmol per liter), similar to previously observed levels, and the results of urinalysis were normal.
The patient's low blood pressure raises the possibility of sepsis syndrome due to an underlying infection. The cardiac murmur probably represents a flow murmur due to her anemia, but with a distal digital lesion, endocarditis must be considered. The morphologic features and position of the digital lesion do not support the possibility of herpetic whitlow. Hematogenously spread bacterial, fungal, or mycobacterial infections could have this appearance. The tempo of the illness suggests the presence of indolent organisms such as bartonella and nocardia. If the lesion were caused by direct inoculation, I would think primarily of certain mycobacteria (e.g., Mycobacterium marinum) and environmental fungi, such as aspergillus or cryptococcus. Several of these organisms are relevant to her occupation, assuming the appropriate contact with animals such as fish, birds, cats, or dogs.
Given the hematologic abnormalities, I must also consider bone marrow involvement with a disseminated infection. Granulomatous infections such as fungal diseases (especially histoplasmosis and cryptococcosis) and mycobacterial diseases are the most common marrow-infiltrative infections. The presence of adenopathy may also indicate disseminated infection or cancer (e.g., lymphoma).
Cultures of blood and urine obtained on admission revealed no growth. The results of serologic tests for Epstein–Barr virus (EBV), cytomegalovirus (CMV), parvovirus B19, and bartonella were all negative for acute infection; serum polymerase-chain-reaction (PCR) assays for EBV and CMV were also negative. Transthoracic echocardiography revealed a bicuspid aortic valve but no evidence of endocarditis. Computed tomography of the chest, abdomen, and pelvis revealed left axillary lymph-node enlargement but no other remarkable findings. The patient was presumed to have an unspecified viral illness. The mycophenolate mofetil was discontinued, and she was discharged home, with close outpatient observation anticipated.
Clinicians must be cautious in the interpretation of the results of serologic tests in immunosuppressed patients. These patients may have a component of humoral immunodeficiency that impairs the ability to mount typical antibody responses to acute or recrudescent infection. Thus, the absence of IgM antibodies to EBV or CMV, for example, would have limited negative predictive value. Genomic or antigen-based studies are generally more reliable in this setting. Parvovirus B19 infection was a consideration because of the hematologic cytopenias, but it would not fully explain the skin findings. The absence of detectable parvovirus B19 in a PCR study of the blood provides strong evidence against this infection.
At this point, I am uncomfortable ascribing the fever to an undefined viral illness. The clinical setting, duration of illness, and associated physical findings warrant additional evaluation for specific treatable causes. Since less invasive tests have been unrevealing, histologic evaluation of the skin lesion, enlarged lymph nodes, or bone marrow should be considered next.
Two days after discharge from the hospital, the patient was readmitted because of persistent fever, malaise, and anorexia. The small erythematous lesion on her left index finger had persisted without change. While she was hospitalized, the patient continued to have a fever, with temperatures up to 40.5°C. Blood cultures from the previous admission remained sterile after a one-week incubation period. The results of all the following investigations were negative: repeated blood and urine cultures; PCR assays for human herpesvirus 6, CMV, EBV, hepatitis C virus, ehrlichia, and babesia; serologic tests for human immunodeficiency virus, hepatitis B virus, and toxoplasma; a test for urinary histoplasma; and tests for rheumatoid factor and antinuclear antibody. The erythrocyte sedimentation rate and the C-reactive protein level were normal, and the results of a serum PCR assay for polyomavirus type BK (BK virus) was positive, at 5200 copies per milliliter.
Reactivation of BK virus occurs commonly in transplant recipients who are infected with this agent; the most common associated clinical syndrome is hemorrhagic interstitial cystitis. I doubt that BK virus infection explains the entire picture here. A general principle in the evaluation of a patient with fever of unknown origin is to focus on and pursue any focal abnormalities. Biopsy of the involved tissue may have been a more efficient approach in this case.
At this point, I still favor a diagnosis of disseminated infection with lymphatic and bone marrow involvement; infection with bartonella is an intriguing possibility because of the patient's occupation and the presence of a skin lesion. Mycobacteria and fungi have not yet been ruled out as possible causes; post-transplantation lymphoproliferative disorder would be lower on my differential diagnosis.
The patient's white-cell count decreased to 900 per cubic millimeter (with an absolute neutrophil count of 200 per cubic millimeter), the hematocrit decreased to 20 percent, and the platelet count decreased to 75,000 per cubic millimeter. A bone marrow biopsy revealed normocellular marrow, which was negative for signs of cancer, infection with acid-fast bacilli, fungal infection, or other, atypical infections.
Bone marrow examination has value in establishing the diagnosis of an infection such as histoplasmosis or tuberculosis when there are abnormalities in the peripheral-blood count. The results must be interpreted cautiously in this situation, since immunosuppressive therapy may mask important histopathologic clues, particularly granulomas and lymphoid aggregates suggestive of lymphoproliferative disease.
Examination of a specimen obtained by core biopsy of the left axillary lymph node revealed vascular proliferation that appeared to be morphologically benign. However, the possibility of cancer could not be definitively ruled out because there was insufficient tissue for analysis.
The vascular proliferation seen in the lymph-node aspirate from the core-biopsy specimen must be pursued with an excisional lymph-node biopsy. The diagnoses suggested by the vascular proliferation are Kaposi's sarcoma, angioimmunoblastic lymphadenopathy, and bartonellosis. A larger specimen would be needed to demonstrate the typical findings in Kaposi's sarcoma, which would be slit-like vascular proliferation with extravasation of erythrocytes. When Kaposi's sarcoma involves the lymph nodes or other visceral organs, fever may be a prominent feature. In angioimmunoblastic lymphadenopathy, one of the atypical lymphoproliferative disorders that are often associated with viral oncogenesis, systemic symptoms are common. Finally, lymphatic involvement with bartonella should be easily demonstrable on examination of a lymph-node biopsy specimen if proper staining is done for the causative organism. Bartonellosis and disseminated Kaposi's sarcoma may be very difficult to distinguish without a biopsy, but I favor the former, given the patient's occupation and the relatively low probability of previous infection with human herpesvirus 8.
An excisional biopsy of the patient's left axillary lymph node was performed. Pathological examination showed multiple coalescent nodules of proliferated blood vessels surrounded by amorphous, granular, eosinophilic deposits (Figure 2). A Warthin–Starry stain revealed the amorphous deposits to be aggregates of small, rod-shaped bacteria (Figure 3); the results of further study with special stains for acid-fast bacilli and fungi were negative. These findings were diagnostic of bacillary angiomatosis.
Figure 2. Biopsy Specimen of the Left Axillary Lymph Node, Showing Multiple Amorphous, Granular Deposits (Arrow) between and around Proliferated Blood Vessels (Arrowheads) (Hematoxylin and Eosin, x40).
Figure 3. Biopsy Specimen of the Left Axillary Lymph Node Showing Dark, Granular Deposits (Warthin–Starry, x40).
The deposits are aggregates of rod-shaped bacteria.
The patient remained persistently febrile until treatment with oral levofloxacin was initiated for bartonella infection. Repeated serologic testing for bartonella was performed, three weeks after the previous (negative) titers had been obtained. Whereas the IgM titers for both B. henselae and B. quintana remained normal (range, less than 1:20), the IgG titers for both species were now positive, at greater than 1:1024. Two weeks after antibiotic therapy was initiated, the patient reported that her symptoms had dissipated, and the lesion on her finger and the lymphadenopathy had almost resolved.
This case highlights the potential limitations of some diagnostic tests for opportunistic infections, particularly in immunosuppressed patients. Not only were the genomic and serologic evaluations extremely expensive; the PCR assay yielded a positive result (for polyomavirus) that was unrelated to the final diagnosis, and the initial serologic tests yielded a false negative result (for bartonella). Histologic and microbiologic examinations of clinically involved tissues may be warranted even when the results of serologic testing are negative.
Commentary
Fever of unknown origin is defined as a temperature of 38.3°C or greater on several occasions over a period of more than three weeks, accompanied by an uncertain diagnosis after three days of inpatient investigation or three outpatient visits.1 The majority of cases are caused by infections, tumors, and various noninfectious inflammatory diseases.1,2 However, the spectrum of disease in immunocompromised patients who present with fever of unknown origin tends to include infections more often than other illnesses.1 The increased risk of rapid clinical deterioration in immunocompromised patients underscores the importance of prompt diagnosis.3 Errors in the processing of available information during the diagnostic evaluation can result in delays and missed diagnoses,2,4 as occurred in the case under discussion.
The primary impediment to a more timely diagnosis in this case was the excessive reliance on noninvasive diagnostic tests rather than tissue biopsy for direct examination early in the presentation. As the discussant notes, the findings of an erythematous lesion on the left index finger and of ipsilateral axillary lymphadenopathy warranted directed evaluation. The negative results on the serologic tests markedly reduced the treating physicians' suspicion of bartonella infection and thus delayed biopsy and appropriate therapy.
Several aspects of the diagnostic challenge merit review. Undirected serologic and immunologic tests have poor predictive value for evaluating fever of unknown origin because of the low prevalence of the disorders for which they screen. Even with tests with high specificity, the rate of false positive results remains unacceptably high.1 The positive PCR assay for BK virus provides an example of this general principle. Conversely, an example of the opposite phenomenon occurred in the search for bartonella infection (since there was a high pretest probability of the infection and a false negative test result).
Although they are commonly used to evaluate the possibility of bartonella infection,5 serologic tests have variable sensitivities and specificities.6,7 These discrepancies are probably related to differences in the available tests and to spectrum bias, which can occur when a diagnostic test is used in a patient who has characteristics different from those of the reference population in which the test characteristics were defined.8 The discussant correctly emphasizes the low predictive value of serologic testing given that the patient's antibody response was probably blunted because of impaired T-cell function.3,9 Although the results of blood cultures were also negative, the yield of this test is generally low.10
A PCR test is available for bartonella,11 but it requires examination of tissue samples. In this case, PCR examination of the specimen from the core biopsy might have provided an earlier diagnosis. Repeated serologic testing after a reduction in the level of immunosuppression revealed a delayed antibody response, with elevated IgG titers. The diagnosis of bacillary angiomatosis often requires a biopsy of lesions and the subsequent detection of characteristic histopathological features or identification of the bacillus by Warthin–Starry staining or electron microscopy.5
Both B. henselae and B. quintana have been associated with cat scratch disease, bacillary angiomatosis, bacillary peliosis, splenitis, osteomyelitis, bacteremia, and endocarditis.10,12,13 Whereas cat scratch disease most often affects healthy people, bacillary angiomatosis occurs predominantly in patients with the acquired immunodeficiency syndrome. Cases also have been described in organ-transplant recipients as well as in immunocompetent patients.5,14 Exposure usually occurs through the scratch or bite of a cat.13,15 This patient was probably exposed while working as a veterinary assistant.
Patients with bacillary angiomatosis often present with cutaneous or subcutaneous vascular proliferative lesions at the site of inoculation, as did this patient. These lesions can be difficult to distinguish clinically from those associated with Kaposi's sarcoma.16 Extracutaneous lesions may involve bone, mucosal surfaces, the central nervous system, and even bone marrow,5 providing a potential explanation for this patient's pancytopenia.17 Although the blood cultures were sterile, a bacteremic syndrome probably accounts for this patient's insidious, prolonged constellation of symptoms.5 In general, prompt treatment of bacillary angiomatosis may prevent the illness and death associated with progressive disease.18
Despite their early recognition of bartonella infection as a potential cause of this patient's fever of unknown origin, the physicians pursued a noninvasive investigation. Appropriate interpretation of negative serologic tests in an immunosuppressed patient requires that a high index of suspicion for probable causes be maintained.
Supported by a Career Development Award from the Health Services Research and Development Program of the Department of Veterans Affairs and a Patient Safety Developmental Center Grant from the Agency for Healthcare Research and Quality (P20-HS11540, to Dr. Saint).
We are indebted to Camilla Payne, R.N., for her careful review of an earlier draft of this manuscript; to Kamal Amin, M.D., for his generous provision of the photograph used in Figure 1; and to Riccardo Valdez, M.D., for his assistance with photomicrographs and pathologic interpretation.
Source Information
From the Department of Pediatrics, University of Michigan Medical School, Ann Arbor (U.D.P.); the Department of Medicine, University of California, San Francisco, School of Medicine, San Francisco (H.H.); and the Department of Veterans Affairs Health Services Research and Development Center of Excellence, and the Department of Internal Medicine, University of Michigan — both in Ann Arbor (U.D.P., S.S.).
Address reprint requests to Dr. Patel at the University of Michigan, Robert Wood Johnson Clinical Scholars Program, 6312 Medical Sciences Bldg. I, 1150 W. Medical Center Dr., Ann Arbor, MI 48109-0604, or at patelu@umich.edu.
References
Knockaert DC, Vanderschueren S, Blockmans D. Fever of unknown origin in adults: 40 years on. J Intern Med 2003;253:263-275.
Arnow PM, Flaherty JP. Fever of unknown origin. Lancet 1997;350:575-580.
Pizzo PA. Fever in immunocompromised patients. N Engl J Med 1999;341:893-900.
Kassirer JP, Kopelman RI. Cognitive errors in diagnosis: instantiation, classification, and consequences. Am J Med 1989;86:433-441.
Adal KA, Cockerell CJ, Petri WA. Cat scratch disease, bacillary angiomatosis, and other infections due to Rochalimaea. N Engl J Med 1994;330:1509-1515.
Sander A, Posselt M, Oberle K, Bredt W. Seroprevalence of antibodies to Bartonella henselae in patients with cat scratch disease and in healthy controls: evaluation and comparison of two commercial serological tests. Clin Diagn Lab Immunol 1998;5:486-490.
Bergmans AM, Peeters MF, Schellekens JF, et al. Pitfalls and fallacies of cat scratch disease serology: evaluation of Bartonella henselae-based indirect fluorescence assay and enzyme-linked immunoassay. J Clin Microbiol 1997;35:1931-1937.
Ransohoff DF, Feinstein AR. Problems of spectrum and bias in evaluating the efficacy of diagnostic tests. N Engl J Med 1978;299:926-930.
Denton MD, Magee CC, Sayegh MH. Immunosuppressive strategies in transplantation. Lancet 1999;353:1083-1091.
Cotell SL, Noskin GA. Bacillary angiomatosis: clinical and histologic features, diagnosis, and treatment. Arch Intern Med 1994;154:524-528.
Relman DA, Loutit JS, Schmidt TM, Falkow S, Tompkins LS. The agent of bacillary angiomatosis: an approach to the identification of uncultured pathogens. N Engl J Med 1990;323:1573-1580.
Koehler JE, Quinn FD, Berger TG, LeBoit PE, Tappero JW. Isolation of Rochalimaea species from cutaneous and osseous lesions of bacillary angiomatosis. N Engl J Med 1992;327:1625-1631.
Zangwill KM, Hamilton DH, Perkins BA, et al. Cat scratch disease in Connecticut: epidemiology, risk factors, and evaluation of a new diagnostic test. N Engl J Med 1993;329:8-13.
Cline MS, Cummings OW, Goldman M, Filo RS, Pescovitz MD. Bacillary angiomatosis in a renal transplant recipient. Transplantation 1999;67:296-298.
Tappero JW, Mohle-Boetani J, Koehler JE, et al. The epidemiology of bacillary angiomatosis and bacillary peliosis. JAMA 1993;269:770-775.
Tappero JW, Koehler JE. Bacillary angiomatosis or Kaposi's sarcoma? N Engl J Med 1997;337:1888-1888.
Kiriakos J, Ranchin B, Gillet Y, et al. Systemic bacillary angiomatosis after kidney transplantation. Pediatr Nephrol 2003;18:C46-C46. abstract.
van der Wouw PA, Hadderingh RJ, Reiss P, Hulsebosch HJ, Walford N, Lange JM. Disseminated cat-scratch disease in a patient with AIDS. AIDS 1989;3:751-753.(Uptal D. Patel, M.D., Har)