9 Pyrexia of unknown origin (PUO)


*E-mail: simon.edwards2@nhs.net

9.1 Background

General review definition divides PUO as classical, nosocomial, HIV-related and immunosuppression-related [1]. For HIV infection, pyrexia of unknown origin (PUO) identifies a pattern of fever with temperature higher than 38.3 °C on several occasions over more than 4 weeks for outpatients, or more than 3 days duration in hospital, in which the diagnosis remains uncertain after an initial diagnostic work-up, including at least 2 days of incubation of microbiological cultures [2]. It is a common clinical manifestation in HIV-seropositive patients with severe immunosuppression and probability of an infection-related aetiology for PUO in HIV infection increases with CD4 decline, i.e. greater risk if CD4 count <50 cells/μL than <100 cells/μL than >200 cells/μL [3].

  • Fever is rarely the result of the effects of HIV itself and investigation of a specific cause should be actively pursued [4] (level of evidence IV).

Causes of PUO in cohorts of HIV-seropositive patients vary between published studies and are influenced by diagnostic criteria used, degree of immunosuppression, risk category for HIV infection and geographical location of cohort. Comparison of causes of PUO in HIV-seropositive to seronegative patients shows that infection is the most frequent cause of PUO in patients with HIV infection whilst collagen diseases are more common in patients without HIV infection [5].

Many studies were performed before the widespread availability of antiretroviral therapy where the majority of patients had a very low CD4 cell count. The main causes of PUO in patients with severe immunodeficiency are infections and lymphoma [4,6]. Furthermore, these patients often have multiple diagnoses [6,7].

  • Multiple diagnoses are common, and should be considered in all persons with severe immunosuppression (level of evidence III).

A careful travel history is paramount. The commonest cause of PUO in a study from USA was disseminated Mycobacterium avium infection (DMAC) [6] whereas reports from southern Europe and Brazil have described disproportionately more cases of leishmania species or Mycobacterium tuberculosis [8,9]. Febrile illnesses are well described presentations in both disseminated histoplasmosis [10] and Penicillium marneffei [7,11] in persons who have travelled to or originated from an endemic area.

  • Take a careful history, including a lifetime travel history, as new and reactivation of tropical infections are not uncommon (level of evidence IV).

In the era of HAART, tuberculosis and lymphoma continue to be significant causes of PUO. However, as the HIV-seropositive population ages due to the success of HAART, multisystem diseases (encompassing rheumatic diseases, connective tissue disorders, vasculitis including temporal arteritis, polymyalgia rheumatica, and sarcoidosis) should be considered in the differential diagnosis [12].

PUO may present as a manifestation of antiretroviral therapy with the development of an immune reconstitution syndrome to an underlying pathogen such as DMAC, Mycobacterium tuberculosis or cryptococcus.

Fever persisting for a prolonged time may be the first presenting symptom of patients with systemic infections such as PCP [13], cryptococcal disease [14], HSV [15], syphilis and infective endocarditis. Fever and personality change have been reported for cryptococcal meningitis, HSV and VZV encephalitis.

Another cause of chronic fever in HIV-seropositive individuals, not addressed elsewhere in these guidelines, is Bartonellosis, an infection caused by Bartonella henselae or Bartonella quintana [16]. It is associated with profound immunosuppression, usually with a CD4 count <50 cells/μL [17], so is less common in the post-HAART era. Individuals can present with non-specific features such as fever, lymphadenopathy, hepatosplenomegaly, abdominal pain, anaemia or elevated alkaline phosphatase [17]. More specific features include bacillary angiomatosis in which cutaneous lesions that can be friable, red, vascular and exophytic proliferative lesions, non-descript papules or subcutaneous nodules may be observed [18]. These can be difficult to distinguish from the lesions of Kaposi's sarcoma. Other presentations include osteolytic bone lesions and bacillary peliosis (usually caused by B. henselae) where patients can present with fever, abdominal pain, raised alkaline phosphatase and hypodense lesions on computed tomography of the liver and occasionally the spleen [18]. Rarer presentations include nodular or ulcerated lesions of the gastrointestinal tract, which can present with haemorrhage, respiratory tract lesions or neurological manifestations including aseptic meningitis. Neuropsychiatric presentations have been described [19]. Focal necrotising lymphadenopathy is more commonly associated with higher CD4 T-cell counts. Diagnosis involves culture and PCR of blood or biopsy specimens and serology [20]. Treatment is with erythromycin 500 mg qid orally or doxycycline 100 mg bd for at least 3 months, though other macrolides may also be effective [18].

Other, less common causes of prolonged fever include drug-induced fever and thromboembolic disease.

9.2 Clinical evaluation

9.2.1 A detailed history should include:

  • Symptoms from all major systems;
  • All general complaints: e.g. fever, weight loss, night sweats, headaches and rashes;
  • Travel and residence history including contact with TB, illness whilst away including possible exposure to ticks and other vectors, malaria prophylaxis and pre travel vaccinations received;
  • Previous illnesses including TB, surgery and psychiatric problems;
  • Drug history, including over-the-counter medications, prescription medications and any illicit substances including the use of paraphernalia;
  • Immunization status;
  • Family history;
  • Occupational history, which should include exposure to chemicals, animals;
  • Sexual history, including details of any genital ulceration;
  • Personality change.

9.2.2 Examination of the patient should include:

  • Documentation of fever (the fever should be measured more than once and with another person present if factitious fever is suspected);
  • Awareness that diseases such as brucellosis, borreliosis and Hodgkin disease tend to cause recurrent episodes of fever;
  • Careful examination, particularly looking for rashes, lymph node enlargement, signs of arthritis, new/changing cardiac murmurs, hepatosplenomegaly, abdominal tenderness or rigidity, fundoscopic changes and neurological deficits;
  • Mucosal examination (intraoral and genital);
  • Physical examination repeated daily if the patient is in hospital (careful observation is required if patient is monitored as an outpatient as new signs and symptoms may develop).

9.2.3 Initial investigations

  • CD4 cell count;
  • Full blood count (FBC), urea and electrolytes (U&E), C-reactive protein (CRP), liver function test (LFT), lactate dehydrogenase (LDH);
  • Cryptococcal antigen if CD4 count <200 cells/μL;
  • Syphilis serology;
  • Blood cultures to include mycobacterial culture;
  • Culture urine, sputum (include mycobacterial culture pending CD4 cell count);
  • Chest X-ray;
  • Molecular virology studies (hepatitis screen, CMV);
  • Arterial blood gases and at rest and ambulatory oxygen saturation;
  • Echocardiography if murmur or endocarditis suspected;
  • ANA, rheumatoid factor if connective tissue disease suspected.

9.3 The choice and utility of invasive diagnostic tests

Whilst the majority of diagnoses in PUO may be achieved through the use of simple microbiological tests, such as blood cultures and respiratory specimens, invasive tests may be required when such measures fail to elucidate the cause or when a diagnosis is urgently sought. (See Table 9.1 for a list of common diagnoses).

Table 9.1.   Common diagnoses following the investigation of HIV associated PUO
Mycobacterium tuberculosis
Bacterial infections
Pneumocystic Jirovecii pneumonia
CMV end organ disease
Viral (hepatitis B, C, herpes virus, adenovirus) Neoplastic/Lymphoproliferative
Kaposi's sarcoma
Castleman's disease

Several published studies report on the use of histopathological examination of samples acquired from bone marrow, lymph nodes, liver and lung. Fewer data exist on histopathological examination of tissue from other sites such as intestine, skin, oesophageal, brain, mediastinal nodes and lumbar puncture. Choice of further investigation is likely to be dictated by positive findings from clinical evaluation and baseline investigations (see flow diagram in Fig. 9.1).

Figure 91.

 PUO in an HIV-seropositive patient.

  • When tissue specimens are collected, there should always be one specimen sent to microbiology and one specimen sent to the histopathology laboratory. It is important to give complete clinical information to laboratory staff (including HIV status) to ensure appropriate tests are carried out in a timely fashion by an appropriately qualified person (level of evidence IV).

It is good practice to discuss with the laboratory prior to collecting the sample which diagnoses you are considering as samples may need to be sent to another hospital for analysis.

  • Investigations should be undertaken promptly as immunosuppressed patients are prone to rapid clinical deterioration. Advice from a physician experienced in HIV and opportunistic infections should be sought on choice of investigations and use of HAART (level of evidence IV).

9.3.1 Bone marrow examination (BME)

The diagnostic utility of bone marrow examinations in HIV-seropositive persons with PUO has been reviewed in a number of studies [21–23]. A diagnosis is identified in around one-third to one-half of all patients. Whilst the ultimate diagnosis was frequently obtained at other sites, BME was the only site in a minority of cases in most studies [21–24].

Diagnosis by BME may be achieved through bone marrow culture, visualization of granulomas and/or histologically apparent organisms. Special stains for mycobacteria and fungus, and immunohistostaining for lymphoma should be requested. If other diagnoses such as Castleman disease, visceral leishmaniasis and Penicillium marneffei are under consideration then discuss with a histopathologist and, if the patient is not under the care of an HIV or infectious disease specialist, then contact your local HIV or Infectious disease department for advice.

9.3.2 Fine needle aspirate (FNA) of lymph nodes

FNA is a worthwhile procedure in patients with adenopathy and fever. Sampling is quick and easy to perform and may obviate the need for more invasive sampling and enable immediate treatment of specific infections [25,26]. In one large reported series, which included more than 650 samples, a diagnosis was reached in 80% of cases with malignancy accounting for 13% and infection 14% (mainly mycobacterial). A definitive diagnosis was associated with deep aspirate location and lesion size >2 cm [27].

The procedure is associated with a low rate of uninterpretable slides/inadequate sample or false-negative results. In these situations consideration should be given to either lymph node sampling or surgical excision of the lymph node.

9.3.3 Lymph node sampling

Lymph node biopsy is a useful alternative to FNA. It has been shown to have a good diagnostic yield in patients with smear-negative TB [28]. If Castleman disease is suspected biopsy or excision of the lymph node may be preferable to FNA due to the focal nature of Castleman disease within lymph nodes.

9.3.4 Percutaneous liver biopsy (PLB)

The presence of hepatomegaly or splenomegaly have been reported to be the most important factors in predicting the usefulness of the PLB, with a positive predictive value (PPV) of 86.1% and negative predictive value (NPV) of 68.2%. In patients with tuberculosis, an increased alkaline phosphatase and hepatomegaly had a PPV of 86.4% and NPV of 71.4% [29].

9.3.5 Imaging

Imaging plays a key role in the diagnostic work up in PUO. It assists in identification of focal pathology that may be amenable to biopsy in order to get a tissue diagnosis.

A chest X-ray film should be part of the routine investigations. More detailed investigations should be based on clinical symptoms and signs and may include CT/MRI of chest, abdomen and pelvis. There is emerging evidence that 18-fluorodeoxyglucose positron emission tomography (FDG-PET)/CT scanning can help identify the source of disease when earlier investigations have been unsuccessful [30,31].