Haemophagocytic lymphohistiocytosis secondary to disseminated tuberculosis in a young adult with Crohn's disease

A 25-year-old male patient reported four weeks of chills, night sweats, weight loss and dyspnoea following one week in Bali, Indonesia, seven weeks prior. His past medical history included Crohn's disease managed for ten years with infliximab, a monoclonal antibody against tumour necrosis factor-α (TNF-α). In vitro interferon-γ release assay (IGRA) performed at age 14 years, which was conducted years after he had travelled to Egypt for two months in childhood, was negative. Chest x-ray demonstrated perihilar lymphadenopathy and bilateral pulmonary infiltrates, and blood examination revealed pancytopaenia in association with splenomegaly. He was treated for community-acquired pneumonia and referred for outpatient follow-up. A fluorodeoxyglucose positron emission tomography (FDG-PET) scan was requested to assess for haematological malignancy and revealed multiple pulmonary nodules, widespread consolidation, and moderately intense (maximum standardised uptake value [SUVmax], 9.0) adenopathy in the neck, mediastinum, and upper abdomen. Within a week, he was readmitted with worsening hypoxia and fevers over 40°C, requiring intensive care unit (ICU) admission. Intravenous broad-spectrum antibiotics were commenced, and he required intubation, mechanical ventilation, and vasopressor support by day 3. A chest computed tomography scan revealed widespread pulmonary nodules with confluent airspace consolidation and bilateral pleural effusions (Box 1). The patient was seen by haematology and met the diagnostic criteria for haemophagocytic lymphohistiocytosis (HLH) due to fever, splenomegaly, pancytopaenia (haemoglobin, 87 g/L [reference interval (RI), 130–180 g/L]; white cell count, 0.8 × 109/L [RI, 4–11 × 109/L]; platelets 73 × 109/L [RI, 150–450 × 109/L]), hypertriglyceridaemia (fasting triglycerides, 3.2 mmol/L; RI, < 1.5 mmol/L), hyperferritinaemia (97 461 ng/mL; RI, 30–500 ng/mL), and elevated soluble CD25 (9993 pg/mL; RI, < 2678 pg/mL). Bone marrow biopsy evaluating for malignancy, infection and HLH confirmed haemophagocytosis, with no evidence of underlying lymphoma (Box 1). A transbronchial biopsy of the peribronchial lymph node revealed necrotic debris without well formed granulomas or evidence of malignancy; however, acid-fast bacilli were identified, and Mycobacterium tuberculosis complex DNA was detected by GeneXpert (Cepheid) nucleic amplification testing (polymerase chain reaction; PCR) in bone marrow and lymph node samples. Rifampicin resistance was not detected by PCR. Central nervous system (CNS) involvement was excluded with lumbar puncture and magnetic resonance imaging (MRI) scan of the brain and spine. Antituberculous therapy was commenced on day 9 with rifabutin, isoniazid, pyrazinamide, ethambutol and moxifloxacin. Rifabutin was favoured over rifampicin as the degree of liver enzyme induction, and thus the interaction with etoposide and systemic therapy, was expected to be less with the former. A fifth agent, moxifloxacin, was included because the patient was critically ill with a life-threatening haematological disease with a poor prognosis and the potential for drug resistance existed with recent travel to Indonesia. Subsequent drug susceptibility results became available eight weeks later, and the isolate was isoniazid-resistant and susceptible to other first line agents. The clinical picture of overwhelming HLH and sepsis was managed with 1 g/kg intravenous gammaglobulin, daily dexamethasone and weekly etoposide (from day 10) in accordance with the HLH-94 protocol.1 Before the commencement of antituberculous therapy, his liver function tests were abnormal (alanine transaminase, 219 U/L [RI, 5–40 U/L]; aspartate transaminase, 510 U/L [RI, 10–40 U/L]; bilirubin 77 μmol/L [RI, 4–20 μmol/L]), which was attributed to HLH and systemic sepsis after other causes were excluded. His liver function tests were monitored carefully with the addition of potentially hepatotoxic tuberculosis therapy, and he improved after seven days. He was extubated on day 13 of ICU admission and discharged to the ward on day 15. He made gradual clinical improvement on discharge and his Crohn's disease remained quiescent. The HLH-94 protocol was ceased after one cycle due to normalisation of his blood counts and improvement of ferritin and symptoms. HLH is a rare hyperinflammatory syndrome resulting from abnormal activation of macrophages and cytotoxic T cells. Many patients present with a syndrome of recurrent fever, cytopaenia and splenomegaly that may be clinically indistinguishable from sepsis or haematological malignancy.2 Mortality of adults with HLH is high, ranging from 20% to 88%.2 Tuberculosis-associated HLH (TB-HLH) should be suspected in patients with tuberculosis and pancytopaenia;3 HLH is one of the known causes of pancytopaenia with serum ferritin > 7000 ng/mL,2 which itself is a poor prognostic feature in HLH. Primary HLH most commonly presents in children and results from mutations affecting lymphocyte cytotoxicity and immune regulation. Secondary HLH is triggered by malignancies, autoimmune disorders, and infections, which are most commonly viral (Epstein–Barr virus, cytomegalovirus), bacterial (including mycobacteria), parasitic or fungal. Although the diagnostic criteria for HLH have been published in 2004 (Box 2), the timely diagnosis of HLH is challenging due to its rarity, variable presentation, and delays to diagnostic testing.2 Cytopaenias (affecting ≥ 2 of 3 lineages in the peripheral blood) Hypertriglyceridaemia and/or hypofibrinogenaemia Management of HLH involves chemotherapy in accordance with the HLH-94 protocol, haematopoietic cell transplant in selected cases (primary HLH, HLH secondary to haematologic malignancy, refractory cases, CNS involvement), as well as prompt identification and management of the underlying trigger.1 TB-HLH is an uncommon cause of fever of unknown origin in Australia. This patient's presenting symptoms, pancytopaenia and FDG-PET appearance were also suggestive of lymphoma, which remained a diagnostic possibility until ruled out by bone marrow and lung biopsy. TB-HLH has poorer prognosis than tuberculosis without HLH, carrying an increased risk of haematogenous tuberculosis and tuberculous meningitis. The mortality rate of patients with TB-HLH who do not receive antituberculous treatment is 100%.3 Although the most recent publication from the Histiocyte Society states that TB-HLH usually does not require HLH-94-like treatment but responds to specific antimicrobial treatment (strong consensus),2 in this case, HLH was diagnosed first in the context of a man with significant risk of underlying lymphoma and the treatment was truncated to one cycle once true haematological remission had been attained. The association between tuberculosis and TNF-α inhibition has been previously described.4 TNF-α promotes macrophage activation and immune cell recruitment, and its inhibition is associated with an elevated risk of tuberculosis.4 In this case, it may have resulted in impaired initiation and maintenance of granulomas in response to tuberculosis given the lack of well formed granulomas on histopathology. All patients should undergo latent tuberculosis screening before TNF-α therapy. Antibiotic chemoprophylaxis may reduce reactivation in individuals with latent infection.4 This patient may have been exposed to tuberculosis in childhood when he visited Egypt for two months and it is plausible that his past negative IGRA was a false negative. It is also possible that he was exposed recently in Bali. Tuberculosis risk is low with short term travel, but pre-travel counselling is important for patients taking TNF inhibitors, who are at elevated risk of acquiring active tuberculosis. The BCG vaccine is rarely offered to adult travellers and would be contraindicated in someone receiving immunosuppressive therapy.5 People at higher risk of contact with tuberculosis include health care or aid workers and those visiting friends or family, who often spend longer periods abroad.5 In immunosuppressed travellers with known or high risk contact with tuberculosis, consideration should be given to IGRA testing eight to ten weeks after return if they had a known negative pre-travel IGRA.6 However, we would not routinely consider IGRA testing after a one-week holiday to Bali, but would advocate clinicians be alert to the possibility in an individual receiving anti-TNF therapy with a respiratory illness who has a travel history. M. tuberculosis infection in immunosuppressed individuals is more likely to be extrapulmonary or disseminated and to carry greater risk of resistance to first line antituberculous agents compared with patients without immunosuppression.7 Given the risk of CNS involvement in both disseminated tuberculosis and HLH, assessment with lumbar puncture and MRI scan of the brain and spine is vital.2 Clinicians must maintain a high index of suspicion for HLH in patients with progression of a sepsis-like syndrome despite broad-spectrum empiric antibiotic therapy. Tuberculosis should be considered as a potential trigger for HLH in adult patients managed with TNF inhibitors, who have risk factors for infection such as travel to a tuberculosis-endemic country. The patient gave written consent for publication. We thank Dr Ellen Maxwell, from Melbourne Pathology, for providing the pathology figure. No relevant disclosures. Not commissioned; externally peer reviewed.