Management of pyrexia of unknown origin in HIV-positive patients

INTRODUCTION

We were inspired to conduct a review of the literature around pyrexia of unknown origin (PUO) in HIV-positive individuals following a case at our unit that we all found difficult and at times frustrating to manage. Here we present those findings and also discuss strategies for managing such patients. A 33-year-old gentleman was admitted to our hospital with pyrexia and constitutional symptoms of several days duration. Apart from pyrexia and some cardiovascular compromise, there were no positive findings on examination. He was known to be HIV positive and had previously been managed at a different unit, having only recently moved to our area. His CD4 count was 240 cells × 10⁶/L and he was naïve to antiretrovirals. Over subsequent weeks numerous investigations were performed, which did not reveal the cause of his illness; his PUO persisted and his clinical state deteriorated.

DEFINITIONS OF PUO

Petersdorf and Beeson ¹ classically defined PUO in 1961 as a temperature of greater than 38.3°C on multiple occasions over a period of more than three weeks, with failure to reach a diagnosis after one week of inpatient investigations. Although HIV-related conditions are an important cause of prolonged fever and HIV testing should be performed at an early stage in all cases of PUO where HIV status is unknown, patients with known HIV can also present with a PUO during follow-up. Despite the availability of new and sophisticated diagnostic tools, PUO still poses a challenge in clinical practice. In 1991, Durak and Street ² re-devised a definition for HIV-related PUO as a pyrexia greater than 38.3°C occurring on multiple occasions over a period of more than four weeks for outpatients or three days for inpatients, with negative microbiological results after at least two days incubation.

CAUSES OF PUO

Within the general population studies have indicated that infections account for approximately 30% of causes of PUO (neoplasms 18%, collagen vascular diseases 12%, miscellaneous causes 14% and undiagnosed 26%), ³ whereas in HIV-positive individuals PUO is usually caused by a disseminated opportunistic infection (OI) accounting for over 70% of such presentations; with the single most frequently identifiable cause being mycobacterial infection. Furthermore, while a single pathology is the norm in the general population with PUO, two or more simultaneous causes can be identified in approximately one-fifth of HIV-related cases. Factors that influence the nature of infections include the patient's CD4 count, use of prophylactic agents and geographical setting (including country of residence, country of origin and travel history). ^4–12 Occult malignancies, although much less common than infections, are an important, consistent and at times coexistent cause of PUO in reported case series: most frequently lymphoma (Hodgkin's or non-Hodgkin's lymphoma), which may be complicated by haemophagocytic syndrome. ^13,14 As in HIV-negative patients, bacterial infections including occult collections, dental sepsis and syphilis ¹⁵ should be considered as well as drug reactions, ¹⁶ autoimmune conditions such as thyrioditis ¹⁷ or collagen vascular diseases. ¹⁸ Case reports of less frequently occurring infections, for example Babesiosis, demonstrate the importance of a through travel history. ¹⁹ Table 1 summarizes the commonly identified causes of PUO in HIV-positive individuals (adapted from the references given above). Given that in the vast majority of cases an underlying cause for the PUO is found, it is important not to attribute the pyrexia to the HIV virus itself until other pathologies are ruled out.

REACHING A DIAGNOSIS

Establishing the cause(s) of a PUO can be a difficult and prolonged process and one that does not always result in a clear answer. Before commencing a long list of investigations, a careful and detailed history should be taken followed by a thorough clinical examination that should be regularly revisited and will often provide important clues to the underlying diagnoses. The history should include a careful documentation of all symptoms, with particular attention to any localizing symptoms, sweats, weight loss, rashes, lumps or itching. Other important specific points to cover include documentation of any travel, all concomitant and previous medications including recreational and herbal drug use, a detailed sexual history, immunization history, questions about any trauma, bites or injuries sustained, past surgical and medical history, and contact with animals. Clinical examination should include a full general and systemic examination, especially looking for any lymphadenopathy, hepatosplenomegaly, abnormal nail and skin lesions or rashes, ear, nose and throat examination, dental examination, thyroid palpation, genital (including rectal) examination, and any subtle neurological or eye signs (including dilated fundoscopy). ³

INVESTIGATIONS

Investigations suggested are summarized in Table 2. There are baseline investigations suggested for all HIV-positive patients, many of which will already be performed in most pyrexial patients before the criteria for PUO are met. Targeted investigations should then be undertaken guided by the history, examination and any consistent abnormalities of the baseline investigations. It is crucial to maintain a measured and logical approach to investigating a PUO and to regularly re-visit the patient's history and examination as relevant details may be recalled by the patient or family and clinical signs may evolve. Despite this approach some PUOs may be difficult to diagnose, or may be ongoing despite a positive diagnosis and appropriate treatment, which may indicate multiple pathologies and the need for further or repeated investigations. It is important to remember that even in patients taking antimicrobial prophylaxis against certain pathogens (usually for Pneumocystits jirovecii pneumonia [PJP] or Mycobacterium avium complex), these causes cannot be excluded without appropriate investigations.

The clinical utility of bone marrow (BM) biopsy for PUO in HIV-positive patients is controversial. Kilby et al. ²¹ compared the utility of BM biopsy with that of blood cultures for mycobacterial and fungal infections and found no difference in turnaround time between the two, and concluded that the combined use of BM biopsy and culture as well as blood cultures provided the highest diagnostic yield. In a retrospective study of 72 HIV patients with PUO, BM examination was found to be a useful diagnostic procedure when an infectious process was suspected, but it was not useful to establish the diagnosis of lymphoma. ²² The low utility of a BM biopsy in diagnosing lymphoma, presenting as PUO, can be explained by the fact that most HIV-positive patients have intermediate/high-grade lymphomas, with a BM involvement of only 30%. In another study from Spain, 182 episodes of PUO in 160 HIV patients were retrospectively studied. ²³ Fifty-four of these episodes were diagnosed by a BM examination, and in 36 of these this was the only diagnostic tool. In this study, the presence of thrombocytopenia and an elevated serum aspartate aminotransferase level were the factors associated with a high probability of obtaining the diagnosis through a BM study.

A percutaneous liver biopsy (PLB) may also be useful for the diagnosis of PUO in HIV-positive patients and has a diagnostic yield of 45%. ²⁴ In a study of 58 HIV patients who underwent a liver biopsy for the evaluation of PUO, PLB was diagnostic in 25 cases, helpful in 13 and normal or non-specific in the remaining 20. ²⁵ The presence of hepatomegaly or splenomegaly was the most useful factor in predicting the usefulness of the PLB. Although liver and BM biopsies may frequently provide a method of diagnosis especially in the case of mycobacterial infection, ²⁶ they are invasive and have associated risks; therefore, before commencing such invasive investigations, the history, examination and results from other less invasive tests should be reviewed.

EMPIRICAL TREATMENT: WHAT AND WHEN TO CONSIDER, AND THE PROBLEM OF ‘IMMUNE RECONSTITUTION INFLAMMATORY SYNDROME’

The most effective treatment of a PUO is obviously the identification of and then the treatment for the causative condition(s); however, as discussed earlier, making a definitive diagnosis may be a lengthy and challenging process and a diagnosis may prove elusive in up to 40% of patients. ^5,10 As many investigations as possible should be conducted prior to initiation of any treatment and whether to then initiate empirical antimicrobial treatment depends on the clinical picture, which may deteriorate and necessitate treatment prior to investigation results being available or tests being conducted. Specific features in the history or examination may point to a likely diagnosis, and HIV parameters such as CD4 count may also guide the choice of empirical therapy.

Although mycobacterial infection is most common, infection with other bacteria may account for up to 15% of cases of PUO. ⁴ Hence, broad-spectrum empirical antibacterial therapy may be indicated. In this case an agent such as piperacillin/tazobactam may be appropriate, although local epidemiology and sensitivity patterns should be considered when choosing an empirical agent. Antibiotics with antimycobacterial activity such as ciprofloxacin, clarithromycin and rifampicin should be used with caution as they can hinder mycobacterial culture or generate resistance, making subsequent therapy problematic. Empirical antifungal therapy with an agent such as liposomal amphotericin may also be indicated, in particular if there is a history of travel to a region where endemic mycoses may be encountered.

As discussed earlier, mycobacterial infection is the commonest cause of PUO in HIV, with non-tuberculous mycobacteria more often seen in those with CD4 counts below 100. Mycobacterial infection in the context of HIV can be more difficult to confirm due to atypical presentations and fewer positive sputum smears, cultures and granulomas seen on biopsy specimens. ²⁷ In an HIV-negative cohort of patients with pulmonary disease treated empirically for tuberculosis (TB), 84.4% had subsequent clinical and radiological improvement. ²⁸ Treatment should be initiated with reference to the British HIV Association Guidelines for TB and HIV co-infection. ²⁹

Further complicating the picture is the possibility of immune reconstitution inflammatory syndrome (IRIS), which, although documented with many OIs after the initiation of highly active antiretroviral treatment (HAART), is most frequently a problem during mycobacterial treatment, when it has been found to occur more often in those with HIV infection (28%) than in those without (10%), ³⁰ and in HIV-positive patients is more frequent in those receiving HAART. ³¹ This can produce a further diagnostic dilemma if a patient with a PUO is started empirically on TB treatment and consequently develops fever with clinical and/or radiological deterioration, and it is unclear whether co-pathologies or IRIS is the culprit. Steroids are the usual treatment for IRIS but this may also have the disadvantage of producing a clinical response masking in part or full the underlying cause of the PUO, which may remain undiagnosed only to relapse at a later date.

A recently presented study has suggested that antiretroviral treatment started earlier (a median of 12 days) after initiation of treatment of an OI reduced the risk of death or progression to AIDS. ³² Although this study included OIs that may cause a PUO such as PJP and cyptococcal infections, it excluded TB, thus making the significance of this in mycobacterial infection uncertain. Early initiation of HAART has also been shown to improve survival in those with lymphoma. ³³

CONCLUSIONS

PUO in HIV-positive individuals remains a condition that can present a diagnostic dilemma for clinicians and patients. Compared to PUO occurring in HIV-negative patients, infectious aetiologies and concurrent multiple pathologies occur much more frequently, with the commonest being mycobacterial infections. The likelihood of other infections is dependent on geographical location and travel history, and underlying malignancies are important to consider. As an underlying pathology in addition to the HIV diagnosis is usually found, it is important not to attribute PUO to HIV infection itself until all appropriate diagnostic strategies have been employed.

In the case of our patient who prompted this review, he had a nine-week hospital stay during which almost all of the diagnostic procedures discussed were employed and his PUO continued. His condition deteriorated and a number of empirical treatments were tried and antiretroviral therapy commenced. This proved to be a difficult time for the patient, his family and the health-care workers involved, and we found that an open, honest and regular dialogue between the medical and nursing teams and the patient and his family were crucial. An initial BM biopsy, which had been unhelpful, was repeated and reported as demonstrating haemophagocytic syndrome, and he died very shortly afterwards in the intensive care unit. The final reviewed report of the repeated BM examination taken a few days prior to death revealed Hodgkin's lymphoma.