DEFINITION AND CLASSIFICATION
Fever of unknown origin (FUO) was defined by Petersdorf and Beeson in 1961 as (1) temperatures of >38.3°C (>101°F) on several occasions; (2) a duration of fever of >3 weeks; and (3) failure to reach a diagnosis despite 1 week of inpatient investigation.Although this classification has stood for more than 30 years, Durack and Street have proposed a new system for classification of FUO: (1) classic FUO; (2) nosocomial FUO; (3) neutropenic 101 FUO; and (4) FUO associated with HIV infection. Classic FUO corresponds closely to the earlier definition of FUO, differing only with regard to the prior requirement for 1 week’s study in the hospital. The newer definition is broader, stipulating three outpatient visits or 3 days in the hospital without elucidation of a cause or 1 week of “intelligent and invasive” ambulatory investigation. In nosocomial FUO, a temperature of ≥38.3°C (≥101°F) develops on several occasions in a hospitalized patient who is receiving acute care and in whom infection was not manifest or incubating on admission.Three days of investigation, including at least 2 days’ incubation of cultures, is the minimum requirement for this diagnosis. Neutropenic FUO is defined as a temperature of ≥38.3°C (≥101°F) on several occasions in a patient whose neutrophil count is <500/ìL or is expected to fall to that level in 1–2 days. The diagnosis of neutropenic FUO is invoked if a specific cause is not identified after 3 days of investigation, including at least 2 days’ incubation of cultures. HIV-associated FUO is defined by a temperature of ≥38.3°C (≥101°F) on several occasions over a period of >4 weeks for outpatients or >3 days for hospitalized patients with HIV infection. This diagnosis is invoked if appropriate investigation over 3 days, including 2 days’ incubation of cultures, reveals no source. Adoption of these categories of FUO in the literature has allowed a more rational compilation of data regarding these disparate groups. In the remainder of this chapter, the discussion will focus on classic FUO in the adult unless otherwise specified.
CAUSES OF CLASSIC FUO
Table 9-1 summarizes the findings of several large studies of FUO carried out since the advent of the antibiotic era, including a prospective study of 167 adult patients with FUO encompassing all eight university hospitals in the Netherlands and using a standardized protocol in which the first author reviewed every patient. Coincident with the widespread use of antibiotics, increasingly useful diagnostic technologies—both noninvasive and invasive—have been developed. Newer studies reflect not only changing patterns of disease, but also the impact of diagnostic techniques that make it possible to eliminate many patients with specific illness from the FUO category. The ubiquitous use of potent broad-spectrum antibiotics may have decreased the number of infections causing FUO. The wide availability of ultrasonography, CT, MRI, radionuclide scanning, and positron emission tomography (PET) scanning has enhanced the detection of localized infections and of occult neoplasms and lymphomas in patients previously thought to have FUO. Likewise, the widespread availability of highly specific and sensitive immunologic testing has reduced the number of undetected cases of systemic lupus erythematosus and other autoimmune diseases. Infections, especially extrapulmonary tuberculosis, remain the leading diagnosable cause of FUO. Prolonged mononucleosis syndromes caused by Epstein-Barr virus, cytomegalovirus (CMV), or HIV are conditions whose consideration as a cause of FUO is sometimes confounded by delayed antibody responses. Intraabdominal abscesses (sometimes poorly localized) and renal, retroperitoneal, and paraspinal abscesses continue to be difficult to diagnose. Renal malacoplakia, with submucosal plaques or nodules involving the urinary tract, may cause FUO and is often fatal if untreated. It is associated with intracellular bacterial infection, is seen most often in patients with defects of intracellular bacterial killing, and is treated with fluoroquinolones or trimethoprim-sulfamethoxazole. Occasionally, other organs may be involved.
Osteomyelitis, especially where prosthetic devices have been implanted, and infective endocarditis must be considered. Although true culture-negative infective endocarditis is rare, one may be misled by slow-growing organisms of the HACEK group (Haemophilus aphrophilus, Actinobacillus actinomycetemcomitans, Cardiobacterium hominis, Eikenella corrodens, and Kingella kingae; Chap. 48), Bartonella spp. (previously Rochalimaea), Legionella spp., Coxiella burnetii, Chlamydophila psittaci, and fungi. Prostatitis, dental abscesses, sinusitis, and cholangitis continue to be sources of occult fever. Fungal disease, most notably histoplasmosis involving the reticuloendothelial system, may cause FUO. FUO with headache should prompt examination of spinal fluid for Cryptococcus neoformans. Malaria (which may result from transfusion, the failure to take a prescribed prophylactic agent, or infection with a drug-resistant strain) continues to be a cause, particularly of asynchronous FUO. A related protozoan infection, babesiosis, may cause FUO and is increasing in geographic distribution and in incidence, especially among the elderly and immunosuppressed. In most earlier series, neoplasms were the next most common cause of FUO after infections (Table 9-1). In more recent series, a decrease in the percentage of FUO cases due to malignancy was attributed to improvement in diagnostic technologies—in particular, high-resolution tomography, MRI, PET scanning, and tumor antigen assays. This observation does not diminish the importance of considering neoplasia in the initial diagnostic evaluation of a patient with fever.
A number of patients in these series had temporal arteritis, adult Still’s disease, drug-related fever, and factitious fever. In recent series, ∼25–30% of cases of FUO have remained undiagnosed. The general term noninfectious inflammatory diseases applies to systemic rheumatologic or vasculitic diseases such as polymyalgia rheumatica, lupus, and adult Still’s disease, as well as to granulomatous diseases such as sarcoidosis and Crohn’s and granulomatous hepatitis. In the elderly, multisystem disease is the most frequent cause of FUO, with giant-cell arteritis being the leading etiologic entity in this category. In patients >50 years of age, this disease accounts for 15–20% of FUO cases.Tuberculosis is the most common infection causing FUO in the elderly, and colon cancer is an important cause of FUO with malignancy in this age group. Many diseases have been grouped in the various studies as “miscellaneous.” On this list are drug fever, pulmonary embolism, factitious fever, the hereditary periodic fever syndromes (familial Mediterranean fever, hyper-IgD syndrome, tumor necrosis factor receptor– associated periodic syndrome, familial cold urticaria, and the Muckle-Wells syndrome), and Fabry’s disease. A drug-related etiology must be considered in any case of prolonged fever.Any febrile pattern may be elicited by a drug. Virtually all classes of drugs cause fever, but antimicrobial agents (especially â-lactam antibiotics), cardiovascular drugs (e.g., quinidine), antineoplastic drugs, and drugs acting on the central nervous system (e.g., phenytoin) are particularly common causes.
It is axiomatic that, as the duration of fever increases, the likelihood of an infectious cause decreases, even for the more indolent infectious etiologies (e.g., brucellosis, paracoccidioidomycosis, and malaria due to Plasmodium malariae). In a series of 347 patients referred to the National Institutes of Health from 1961 to 1977, only 6% had an infection (Table 9-2). A significant proportion (9%) had factitious fevers—i.e., fevers due either to false elevations of temperature or to self-induced disease. A substantial number of these factitious cases were in young women in the health professions. It is worth noting that 8% of the patients with prolonged fevers (some of whom had completely normal liver function studies) had granulomatous hepatitis, and 6% had adult Still’s disease. After prolonged investigation, 19% of cases still had no specific diagnosis. A total of 27% of patients had no actual fever during inpatient observation or had an exaggerated circadian temperature rhythm without chills, elevated pulse, or other abnormalities. More than 200 conditions may be considered in the differential diagnosis of classic FUO in adults; the most common of these are listed in Table 9-3. This list applies strictly to the United States. Geographic considerations are paramount. For example, in Japan, human T-cell lymphotropic virus type I is a consideration; in China, infection plays a greater role and tuberculosis is prominent; and in Spain, visceral leishmaniasis may be a more common cause of FUO. The frequency of global travel underscores the need for a detailed travel history, and the continuing emergence of new infectious diseases makes this listing potentially incomplete. The possibility of international and domestic terrorist activity involving the intentional release of infectious agents, many of which cause illnesses presenting with prolonged fever, underscores the need for obtaining an insightful environmental, occupational, and professional history, with early notification of public health authorities in cases of suspicious etiology (Chap. 6).
SPECIALIZED DIAGNOSTIC STUDIES
Classic FUO A stepwise flow chart depicting the diagnostic workup and therapeutic management of FUO is provided in Fig. 9-1. In this flow chart, reference is made to “potentially diagnostic clues,” as outlined by de Kleijn and colleagues; these clues may be key findings in the history (e.g., travel), localizing signs, or key symptoms. Certain specific diagnostic maneuvers become critical in dealing with prolonged fevers. If factitious fever is suspected, electronic thermometers should be used, temperature- taking should be supervised, and simultaneous urine and body temperatures should be measured.Thick blood smears should be examined for Plasmodium; thin blood smears, prepared with proper technique and quality stains and subjected to expert microscopy, should be used to speciate Plasmodium and to identify Babesia, Trypanosoma, Leishmania, Rickettsia, and Borrelia. Any tissue removed during prior relevant surgery should be reexamined; slides should be requested, and, if need be, paraffin blocks of fixed pathologic material should be reexamined and additional special studies performed. Relevant x-rays should be reexamined; reviewing of prior radiologic reports may be insufficient. Serum should be set aside in the laboratory as soon as possible and retained for future examination for rising antibody titers. Febrile agglutinins is a vague term that in most laboratories refers to serologic studies for salmonellosis, brucellosis, and rickettsial diseases. These studies are seldom useful, having low sensitivity and variable specificity. Multiple blood samples (no fewer than three and rarely more than six, including samples for anaerobic culture) should be cultured in the laboratory for at least 2 weeks to ensure that any HACEK group organisms that may be present have ample time to grow (Chap. 48).
Lysiscentrifugation blood culture techniques should be employed in cases where prior antimicrobial therapy or fungal or atypical mycobacterial infection is suspected. Blood culture media should be supplemented with L-cysteine or pyridoxal to assist in the isolation of nutritionally variant streptococci. It should be noted that sequential cultures positive for multiple organisms may reflect self-injection of contaminated substances. Urine cultures, including cultures for mycobacteria, fungi, and CMV, are indicated. In the setting of recurrent fevers with lymphocytic meningitis (Mollaret’s meningitis), cerebrospinal fluid can be tested for herpesvirus, with use of the polymerase chain reaction (PCR) to amplify and detect viral nucleic acid (Chap. 80).A recent report described a highly multiplexed oligonucleotide microarray using PCR amplification and containing probes for all recognized vertebrate virus species and for 135 bacterial, 73 fungal, and 63 parasitic genera and species. The eventual clinical validation of such microarrays will further diminish rates of undiagnosed FUO of infectious etiology. In any FUO workup, the erythrocyte sedimentation rate (ESR) should be determined. Striking elevation of the ESR and anemia of chronic disease are frequently seen in association with giant-cell arteritis or polymyalgia rheumatica—common causes of FUO in patients >50 years of age. Still’s disease is suggested by elevations of ESR, leukocytosis, and anemia and is often accompanied by arthralgias, polyserositis (pleuritis, pericarditis), lymphadenopathy, splenomegaly, and rash.The C-reactive protein level may be a useful cross-reference for the ESR and is a more sensitive and specific indicator of an “acutephase” inflammatory metabolic response. Antinuclear antibody, antineutrophil cytoplasmic antibody, rheumatoid factor, and serum cryoglobulins should be measured to rule out other collagen vascular diseases and vasculitis.
Elevated levels of angiotensin-converting enzyme in serum may point to sarcoidosis.With rare exceptions, the intermediate- strength purified protein derivative (PPD) skin test should be used to screen for tuberculosis in patients with classic FUO. Concurrent control tests, such as the mumps skin test antigen (Aventis-Pasteur, Swiftwater, PA), should be employed. It should be kept in mind that both the PPD skin test and control tests may yield negative results in miliary tuberculosis, sarcoidosis, Hodgkin’s disease, malnutrition, or AIDS. Noninvasive procedures should include an upper gastrointestinal contrast study with small-bowel followthrough and colonoscopy to examine the terminal ileum and cecum. Colonoscopy is especially strongly indicated in the elderly. Chest x-rays should be repeated if new symptoms arise. Sputum should be induced with an ultrasonic nebulizer for cultures and cytology. If there are pulmonary signs or symptoms, bronchoscopy with bronchoalveolar lavage for cultures and cytology should be considered. High-resolution spiral CT of the chest and abdomen should be performed with both IV and oral contrast. If a spinal or paraspinal lesion is suspected, however, MRI is preferred. MRI may be superior to CT in demonstrating intraabdominal abscesses and aortic dissection, but the relative utility of MRI and CT in the diagnosis of FUO is unknown. At present, abdominal CT with contrast should be used unless MRI is specifically indicated. Arteriography may be useful for patients in whom systemic necrotizing vasculitis is suspected. Saccular aneurysms may be seen, most commonly in renal or hepatic vessels, and may permit diagnosis of arteritis when biopsy is difficult.
Ultrasonography of the abdomen is useful for investigation of the hepatobiliary tract, kidneys, spleen, and pelvis. Echocardiography may be helpful in an evaluation for bacterial endocarditis, pericarditis, nonbacterial thrombotic endocarditis, and atrial myxomas.Transesophageal echocardiography is especially sensitive for these lesions. Radionuclide scanning procedures using technetium (Tc) 99m sulfur colloid, gallium (Ga) 67 citrate, or indium (In) 111–labeled leukocytes may be useful in identifying and/or localizing inflammatory processes. In one study, Ga scintigraphy yielded useful diagnostic information in almost one-third of cases, and it was suggested that this procedure might actually be used before other imaging techniques if no specific organ is suspected of being abnormal. It is likely that PET scanning, which provides quicker results (hours vs days), will prove even more sensitive and specific than 67Ga scanning in FUO. 99mTc bone scan should be undertaken to look for osteomyelitis or bony metastases; 67Ga scan may be used to identify sarcoidosis or Pneumocystis infection (Chap. 111) in the lungs or Crohn’s disease in the abdomen. 111In-labeled white blood cell (WBC) scan may be used to locate abscesses.With these scans, falsepositive and false- negative findings are common. Fluorodeoxyglucose F18 (FDG) PET scanning appears to be superior to other forms of nuclear imaging.
The FDG used in PET scans accumulates in tumors and at sites of inflammation and has even been shown to accumulate reliably at sites of vasculitis. Where available, FDG PET scanning should therefore be chosen over 67Ga scanning in the diagnosis of FUO. Biopsy of the liver and bone marrow should be considered in the workup of FUO if the studies mentioned above are unrevealing and if fever is prolonged. Granulomatous hepatitis has been diagnosed by liver biopsy, even when liver enzymes are normal and no other diagnostic clues point to liver disease. All biopsy specimens should be cultured for bacteria,mycobacteria, and fungi. Likewise, in the absence of clues pointing to the bone marrow, bone marrow biopsy (not simple aspiration) for histology and culture has yielded diagnoses late in the workup.When possible, a section of the tissue block should be retained for further sections or stains. PCR technology makes it possible in some cases to identify and speciate mycobacterial DNA in paraffinembedded, fixed tissues at some research centers. Thus, in some cases, a retrospective diagnosis can be made on the basis of studies of long-fixed pathologic tissues. In a patient over age 50 (or occasionally in a younger patient) with the appropriate symptoms and laboratory findings,“blind biopsy” of one or both temporal arteries may yield a diagnosis of arteritis.Tenderness or decreased pulsation, if noted, should guide the selection of a site for biopsy. Lymph node biopsy may be helpful if nodes are enlarged, but inguinal nodes are often palpable and are seldom diagnostically useful. Exploratory laparotomy has been performed when all other diagnostic procedures fail but has largely been replaced by imaging and guided-biopsy techniques.
Laparoscopic biopsy may provide more adequate guided sampling of lymph nodes or liver, with less invasive morbidity. Nosocomial FUO (See also Chap. 13) The primary considerations in diagnosing nosocomial FUO are the underlying susceptibility of the patient coupled with the potential complications of hospitalization. The original surgical or procedural field is the place to begin a directed physical and laboratory examination for abscesses, hematomas, or infected foreign bodies. More than 50% of patients with nosocomial FUO are infected. Intravascular lines, septic phlebitis, and prostheses are all suspect. In this setting, the best approach is to focus on sites where occult infections may be sequestered, such as the sinuses of intubated patients or a prostatic abscess in a man with a urinary catheter. Clostridium difficile colitis may be associated with fever and leukocytosis before the onset of diarrhea. In ∼25% of patients with nosocomial FUO, the fever has a noninfectious cause. Among these causes are acalculous cholecystitis, deep-vein thrombophlebitis, and pulmonary embolism. Drug fever, transfusion reactions, alcohol/drug withdrawal, adrenal insufficiency, thyroiditis, pancreatitis, gout, and pseudogout are among the many possible causes to consider.As in classic FUO, repeated meticulous physical examinations, coupled with focused diagnostic techniques, are imperative. Multiple blood, wound, and fluid cultures are mandatory. The pace of diagnostic tests is accelerated, and the threshold for procedures—CT scans, ultrasonography, 111In WBC scans, noninvasive venous studies—is low. Even so, 20% of cases of nosocomial FUO may go undiagnosed. Like diagnostic measures, therapeutic maneuvers must be swift and decisive, as many patients are already critically ill.
IV lines must be changed (and cultured), drugs stopped for 72 h, and empirical therapy started if bacteremia is a threat. In many hospital settings, empirical antibiotic coverage for nosocomial FUO now includes vancomycin for coverage of methicillin-resistant Staphylococcus aureus, as well as broad-spectrum gram-negative coverage with piperacillin/tazobactam, ticarcillin/clavulanate, imipenem, or meropenem. Practice guidelines covering many of these issues have been published jointly by the Infectious Diseases Society of America (IDSA) and the Society for Critical Care Medicine and can be accessed on the IDSA website (www.journals.uchicago. edu/IDSA/ guidelines). Neutropenic FUO (See also Chap. 11) Neutropenic patients are susceptible to focal bacterial and fungal infections, to bacteremic infections, to infections involving catheters (including septic thrombophlebitis), and to perianal infections. Candida and Aspergillus infections are common. Infections due to herpes simplex virus or CMV are sometimes causes of FUO in this group. Although the duration of illness may be short in these patients, the consequences of untreated infection may be catastrophic; 50–60% of febrile neutropenic patients are infected, and 20% are bacteremic. The IDSA has published extensive practice guidelines covering these critically ill neutropenic patients; these guidelines appear on the website cited in the previous section. In these patients, severe mucositis, quinolone prophylaxis, colonization with methicillin-resistant S. aureus, obvious catheter-related infection, or hypotension dictates the use of vancomycin plus ceftazidime, cefepime, or a carbapenem with or without an aminoglycoside to provide empirical coverage for bacterial sepsis. HIV-Associated FUO HIV infection alone may be a cause of fever. Infection due to Mycobacterium avium or Mycobacterium intracellulare, tuberculosis, toxoplasmosis, CMV infection, Pneumocystis infection, salmonellosis, cryptococcosis, histoplasmosis, non-Hodgkin’s lymphoma, and (of particular importance) drug fever are all possible causes of FUO.
Mycobacterial infection can be diagnosed by blood cultures and by liver, bone marrow, and lymph node biopsies. Chest CT should be performed to identify enlarged mediastinal nodes. Serologic studies may reveal cryptococcal antigen, and 67Ga scan may help identify Pneumocystis pulmonary infection. FUO has an infectious etiology in >80% of HIV-infected patients, but drug fever and lymphoma remain important considerations. Treatment of HIV-associated FUO depends on many factors and is discussed in Chap. 90. Treatment:
FEVER OF UNKNOWN ORIGIN
The focus here is on classic FUO. Other modifiers of FUO—neutropenia, HIV infection, a nosocomial setting— all vastly affect the risk equation and dictate therapy based on the probability of various causes of fever and on the calculated risks and benefits of a guided empirical approach. The age and physical state of the patient are factors as well: the frail elderly patient may merit a trial of empirical therapy earlier than the robust young adult. The emphasis in patients with classic FUO is on continued observation and examination, with the avoidance of “shotgun” empirical therapy. Antibiotic therapy (even that for tuberculosis) may irrevocably alter the ability to culture fastidious bacteria or mycobacteria and delineate ultimate cause. However, vital-sign instability or neutropenia is an indication for empirical therapy with a fluoroquinolone plus piperacillin or the regimen mentioned above (see “Nosocomial FUO”), for example. Cirrhosis, asplenia, intercurrent immunosuppressive drug use, or recent exotic travel may all tip the balance toward earlier empirical anti-infective therapy. If the PPD skin test is positive or if granulomatous hepatitis or other granulomatous disease is present with anergy (and sarcoid seems unlikely), then a therapeutic trial with isoniazid and rifampin (and possibly a third drug) should be undertaken, with treatment usually continued for up to 6 weeks. A failure of the fever to respond over this period suggests an alternative diagnosis. The response of rheumatic fever and Still’s disease to aspirin and nonsteroidal anti-inflammatory drugs (NSAIDs) may be dramatic.
The effects of glucocorticoids on temporal arteritis, polymyalgia rheumatica, and granulomatous hepatitis are equally dramatic. Colchicine is highly effective in preventing attacks of familial Mediterranean fever but is of little use once an attack is well underway. The ability of glucocorticoids and NSAIDs to mask fever while permitting the spread of infection dictates that their use be avoided unless infection has been largely ruled out and unless inflammatory disease is both probable and debilitating or threatening. When no underlying source of FUO is identified after prolonged observation (>6 months), the prognosis is generally good, however vexing the fever may be to the patient. Under such circumstances, debilitating symptoms are treated with NSAIDs, and glucocorticoids are the last resort. The initiation of empirical therapy does not mark the end of the diagnostic workup; rather, it commits the physician to continued thoughtful reexamination and evaluation. Patience, compassion, equanimity, and intellectual flexibility are indispensable attributes for the clinician in dealing successfully with FUO.