Solving the fever mystery: Fever and tonsillar ulcer in an elderly patient

Abstract

Pyrexia of unknown origin in the elderly is not an uncommon clinical problem. In day to day clinical practice, it has been observed that antibiotics are often empirically escalated to ‘treat fever’ in an otherwise well patient. However, this may not be warranted. It is more important that a comprehensive history, detailed physical examination and a systematic evaluation be undertaken to identify the cause so that definitive therapy may be administered. Apart from infection, neoplastic and connective tissue disorders are other common causes of pyrexia of unknown origin which should be considered. We describe our approach for evaluating an 80-year-old man who presented with fever and sore throat.

Introduction

The causes of pyrexia of unknown origin (PUO) in the elderly are different in that they are often identifiable in the older patient. Part of the diagnostic algorithm must include comprehensive history taking and physical examination, including a review of the patient’s past and present medical history, drug history, travel history, social history (including occupation, hobbies and animal exposure), and laboratory and radiology investigations. ¹

Infective, neoplastic and connective tissue disorders are common causes of fever of unknown origin in the elderly which should be considered. ¹ In particular, malignancies may be a more common cause of PUO in the elderly.^1,2

We describe an 80-year-old man who presented to the Singapore General Hospital (SGH) for evaluation of a one-month history of fever and sore throat.

Case report

Mr AL, an 80-year-old retired architect, was admitted for evaluation of one month’s history of fever and sore throat. Prior to his admission, he had visited his family doctor and private ear nose and throat (ENT) surgeon. An outpatient nasoendoscopic examination performed by the private ENT surgeon showed bilateral tonsillar exudates. A biopsy was not taken then. He was given symptomatic outpatient treatment as well as two courses of antibiotics, oral amoxicillin-clavulanic acid followed by oral moxifloxacin, with no improvement. Instead, his throat pain progressed to the point that he had difficulty in swallowing.

His medical history included chronic lymphocytic leukaemia (CLL), diabetes mellitus, hypertension, hyperlipidaemia, bilateral renal cysts and a history of left traumatic femoral fracture which had been conservatively managed. The CLL was Rai stage III at time of diagnosis and he had completed four cycles of rituximab and bendamustine. He was in remission and had been off immunosuppression for two years. There had been no recent change in his medication, which included enalapril 2.5mg o.m., metformin 500mg t.d.s., neurobion one tablet daily, omeprazole 20mg b.d. and calcium and vitamin D one tablet b.d.

He was married with three children and lived with his wife and daughter. He had no recent travel history, no tuberculous contact, no pets and did not take part in outdoor activities. He had not taken traditional medicines.

At presentation, he was febrile with a temperature of 38.8°C. The pulse rate was 80. The blood pressure was 123/63 mmHg. He was alert and non-toxic. The main findings were: the presence of bilateral tonsillar exudates (left more than right) and palpable subcentimetre left cervical lymph nodes. See Figure 1.

Figure 1.

Bilateral tonsillar exudates seen on nasoendoscopic evaluation.

The white cell count was 11.03 × 10⁹/l (neutrophils 70.4%); atypical lymphocytes and blast cells were not identified on full blood examination. Procalcitonin was <0.12µg/l. Other admission investigations include: haemoglobin 11.9 g/dl, platelet count 258 × 10⁹/l, creatinine 69 µmol/l, protein 76g/l, albumin 44 g/l, bilirubin 18 µmol/l, alkaline phosphatise 186 U/l, alanine transaminase 17 U/l, aspartate transaminase 213 U/l. Chest radiograph was normal. His human immunodeficiency virus (HIV) screen was negative. He was tested negative for anti-double-stranded DNA antibodies and antinuclear antibodies.

He was persistently febrile for 14 days during his admission with a peak temperature of 39.4°C. Serial blood cultures (a total of eight sets) collected over this period were negative. Computed tomography (CT) of the neck, chest, abdomen and pelvis performed on the second hospital day showed bilateral cervical lymphadenopathy, scattered bilateral subcentimetre pulmonary nodules and multiple ill-defined splenic nodules. The intra-abdominal lymph nodes were smaller than when he was diagnosed with CLL.

He was started on intravenous amoxicillin-clavulanic acid on the first day of admission. This was escalated to intravenous piperacillin-tazobactam the next day due to persistent fever. A superficial swab of his tonsillar exudates taken on the first hospital day grew Pseudomonas aeruginosa, sensitive to piperacillin-tazobactam, cefepime, aztreonam, gentamicin, meropenem and imipenem. He received intravenous piperacillin-tazobactam for eight days with no fever response. Antibiotics were later escalated to intravenous meropenem. He was clinically stable throughout the period of his admission, with no haemodynamic instability. His white cell count and procalcitonin during this period of time were not elevated.

A definitive left tonsillar excisional biopsy performed on the eighth hospital day confirmed the diagnosis of Epstein–Barr virus (EBV) associated B-cell lymphoproliferative disorder (LPD). In addition, bacterial cultures from the tonsillar biopsy grew sensitive Citrobacter koseri. Blood EBV quantitative polymerase chain reaction (PCR) level was elevated at 26,257 copies/ml.

Upon commencement of hydrocortisone for the treatment of EBV-related LPD, his fever lysed. Mr AL was given four cycles of rituximab and steroids with response. His constitutional symptoms resolved, the tonsillar ulcers healed and his cervical lymph nodes were no longer palpable. He was also treated with a six-week course of intravenous ceftriaxone for possible splenic abscesses, based on the presumption that the bacteria could have translocated from the mucocutaneous membranes and later seeded the spleen. A repeat CT scan showed resolution of the splenic hypodensities.

Discussion

Sustained, unexplained fever despite a comprehensive workup is called PUO. ² Recognizing the problem allows us to approach it methodically. PUO may be classified into classic PUO, neutropenic PUO, nosocomial PUO and HIV-associated PUO. ²

The case presented was a classic PUO. The evaluation of PUO usually revolves around a search for ‘potentially diagnostic clues’.^3,4 In our patient, the clues were the tonsillar ulcers, and the hypodensities in the spleen, detected on CT.

Focusing first on infectious causes, a simple way to put together these two clues would be to consider the ulcer as a portal of entry for oral pathogens, with metastatic infection in the spleen. If the splenic hypodensities were abscesses, then a study of the causes of the splenic abscesses might yield an answer for the patient’s clinical syndrome. In a 10-year review of splenic abscesses in a local hospital, 21 cases were found and melioidosis was the cause in 15 of these cases. ⁵ A review of the literature on splenic abscesses found that Escherichia coli, Salmonella and Staphylococcus aureus were the most common organisms implicated; in addition, more than one-third of these patients were intravenous drug users or were HIV positive. ⁶ Our patient was neither of these.

As mentioned previously, melioidosis is one of the most common causes of splenic abscesses locally. ⁵ Melioidosis presenting as a tonsillar ulcer would be very rare indeed. Nasal-associated lymphoid tissue in mice can be infected with Burkholderia pseudomallei. ⁷ Experiments on have shown that this equivalent of Waldeyer’s ring can be a portal of entry for the causative agent of melioidosis. Clinically, however, melioidosis in the head and neck region is uncommon. A Malaysian report describes four cases, none of them involving the tonsil. ⁸

When one receives a consultation and the patient has been worked up and even treated (which happened to be the case in this instance), one can use clues other than physical signs to rule in or rule out diagnostic possibilities. In this patient, two such clues stand out: the persistently normal procalcitonin and the lack of a temperature response to meropenem procalcitonin, an acute phase reactant, a biochemical marker used to monitor both the progression and prognosis of bacterial infections in adults and to guide antibiotic use.^9,10 A meta-analysis has shown that the use of procalcitonin is not inferior in the elderly patient. In addition, normal procalcitonin levels coupled with reassuring clinical findings suggest a low probability of systemic bacterial infection. ¹¹ This points to a non-infectious cause of PUO in this patient. The lack of response to meropenem lowers the likelihood of melioidosis.

In an elderly person, an ulcer in the oral cavity is always worrying as a possible malignancy. If so, could the splenic lesions be metastases? Splenic metastases from solid tumours are considered exceptional. ¹² Metastases to the spleen usually occur in the context of multi-visceral, end-stage cancer. ¹³ In the large series of PUO described by de Kleijn et al., ³ neoplasms constituted 12.6% of PUO cases, the most common of which was haematological malignancies. Nevertheless a handful of solid tumours were found as a cause of PUO.

Non-healing tonsillar ulcer in a patient with PUO is always worrying as possible lymphoma. One-quarter of extra-nodal lymphomas occur in the head and neck. ¹⁴ More than half of head and neck lymphomas occur in Waldeyer’s ring. ¹⁵ It is well known that lymphoma is the most common malignant neoplasm affecting the spleen, and splenic involvement is usually secondary. ¹³

The definitive diagnosis of PUO in this particular case was clinched by obtaining excisional tonsillar biopsy and the diagnosis of age-related LPD with extranodal polymorphic LPD was confirmed by histopathology. This could be driven by loss of control of EBV proliferation due to immunosenescence.^16,17 This case highlights the importance of early tissue biopsy of the diseased site to obtain definitive diagnosis. In so doing, we could have potentially started definitive treatment earlier while avoiding empiric escalation of antibiotics with no additional benefit.

The mucocutaneous manifestations of EBV occur in both primary and chronic infections; it is also reported in EBV-associated LPD and nasopharyngeal carcinoma. ¹⁸ Specifically, pharyngitis associated with EBV is often described during primary EBV infection and classically described as the ‘worst sore throat ever experienced’. ¹⁶ In the presence of systemic symptoms, EBV-related pharyngeal ulcers in the elderly are a distinct clinicopathologic entity, more accurately labelled as age-related EBV-related LPD, and respond well to systemic chemotherapy. ¹⁷ This was the case for the patient described in this report.

An inflammatory disorder which is associated with PUO and tonsillar ulcers is Behcet’s disease. However, this is the patient’s first presentation of tonsillar ulcer, in the absence of pathergy phenomenon, eye, skin and genital lesions. Based on the international study group (ISG) criteria, Behcet’s disease was deemed less likely. ¹⁹

Although splenic lesions are seen in EBV-LPD, the possibility that the splenic lesions were abscesses could not be excluded. As the treatment of his EBV LPD would be immunosuppressive, the decision was taken to treat the splenic lesions as if they were abscesses. A splenic biopsy was considered risky. The bacterial aetiology was hard to ascertain, but translocation from the tonsillar mucosa was a distinct possibility. It was decided that a reasonably broad-spectrum antibiotic would be used, in particular one which would cover most oral flora, as well as the C. koseri isolated from his tonsillar biopsy. The resolution of the splenic lesions on repeat CT of the abdomen may suggest one of two things: either successful treatment of EBV-related LPD, or successful treatment of C. koseri splenic abscesses.

In retrospect, the extremely painful tonsillar ulcers were a clinical clue to suggest a possible EBV-related disorder that might have prompted one to biopsy the tonsils earlier to clinch the diagnosis. This might have obviated the serial escalation of antibiotics.

Conclusion

In an elderly patient presenting with PUO and mucocutaneous lesions, apart from ruling out infectious causes, connective tissue disorders and malignant causes including EBV-associated LPD have to be considered. A thorough investigation, early identification and biopsy of the diseased site for definitive tissue diagnosis is crucial in the workup and would allow one to achieve an early diagnosis and start definitive treatment. Empiric escalation of antimicrobials should not be encouraged in patients with PUO.