INTRODUCTION

Hemophagocytic lymphohistiocytosis (HLH) is a life-threatening hyperinflammatory syndrome. Familial HLH is a genetic disease in infants with a known etiology, caused by impaired natural killer (NK) and cytotoxic T-cell function and driven by interferon-γ.1 Infections and malignancies are the primary triggers associated with HLH in adults, varying geographically.2Malignancy-associated HLH (M-HLH) is a multi-system disorder of unclear pathophysiology.3 It has a heterogeneous presentation, course, and outcome and does not have a single clinical, laboratory, pathologic, or radiological “gold standard” diagnostic feature. While the 5-year overall survival of HLH of different etiologies has improved recently (inflammatory conditions 50–88%,4, 5 infections 30–80%,4, 5 familial HLH 50–70%)6, the survival of M-HLH remains poor (10–30%).4, 5, 7 Diagnosis is based on the HLH-2004 criteria 8 that have not been validated in M-HLH.9 This creates a significant diagnostic challenge since the malignant clone itself may be responsible for developing nearly all the diagnostic features of HLH.10 We report a case of HLH in a patient with chronic lymphocytic leukemia (CLL), which highlights the challenges in diagnosis and management of M-HLH and the need for validated, specific diagnostic tools.

CASE REPORT

A 72-year-old woman with CLL presented emergently to the hematology clinic with a 10-day history of extreme weakness, fever, and night sweats. She had CLL for over 10 years, during which time she required no treatment for her disease. At diagnosis, she had lymphocytosis only—clinical-stage Binet A, no 17p chromosomal deletion—and was assessed as having an excellent prognosis. Her white blood count remained stable at approximately 80,000 lymphocytes/mm3. A year before her presentation, she developed mild splenomegaly, 2 cm below the left costal margin, but had no lymphadenopathy. Upon presentation, her physical examination was unchanged, but hemoglobin dropped to 8.1 g/dL, and her platelet count fell to 61,000/mm3 from previously normal levels, and her white cell count decreased to 22,000/mm3 with 80% lymphocytes. A bone marrow aspiration and a biopsy were performed to investigate the possibility of disease progression/transformation. A few hours later after the procedure, her temperature rose to 40°C, and she appeared ill and diaphoretic. She was admitted for further investigation. Laboratory studies revealed creatinine of 1.7 mg/dL (baseline, 0.8 mg/dL), C-reactive protein (CRP) of 3 mg/L (normal<0.5 mg/L), and normal lactate dehydrogenase (LDH). A chest X-ray(CXR) was normal.

During her hospitalization, temperature spikes above 40°C persisted and an extensive evaluation for an infectious etiology was performed. Urine and blood cultures were sterile and serologic tests for Epstein Barr virus, cytomegalovirus, hepatitis B, and C viruses were negative. Serum ferritin was 585 ng/mL (normal 10–120 ng/mL), and fibrinogen and triglyceride levels were normal. Urinalysis was consistent with pre-renal azotemia, and the sediment was bland. Total body computed axial tomography (CT) revealed splenomegaly and mild generalized lymphadenopathy. The bone marrow aspirate and biopsy performed earlier revealed moderate involvement by CLL without aplasia, hemophagocytosis, or disease transformation. After 5 days of empiric antibiotic treatment without improvement, tetracycline was added. Eight days after her admission, rickettsia serology and repeat blood cultures were negative; the patient remained febrile. An extensive autoimmune serology panel was negative, and complement levels were normal. Notably, over this same time period, her serum albumin decreased to 2.1 g/dL (normal 3.5–5.4 g/dL) and her general condition declined.

After ruling out infectious etiology, empiric treatment with prednisone 1 mg/kg was administered as a “therapeutic test,” assuming that an undiagnosed inflammatory condition was the cause of the patient’s current condition. A rapid clinical response was noted. The fever and night sweats subsided immediately, serum creatinine level normalized, and serum albumin increased to 3.2 g/dL 2 days later.

On day 11 from her first admission, 3 days after the administration of prednisone, recrudescence occurred with a fever of 40°C, night sweats, and a marked decrease in blood counts. Her hemoglobin was 6.3 g/dL, white cell count was 5230/mm3, and platelet count was 14,000/mm3. Albumin was 2.5g/dL, total bilirubin was 3.3 mg/dL (normal < 1.2 mg/dL), and creatinine was again elevated at 1.4 mg/dL. Ferritin increased significantly to 8917 ng/mL. Soluble CD25 (sCD25) levels and natural killer activity testing were ordered to investigate the possibility of HLH.

On day 12, a positron emission tomography-CT did not indicate disease transformation, and the most probable diagnosis was deemed to be HLH. Accordingly, and due to clinical deterioration, treatment for HLH with etoposide and dexamethasone was begun according to the HLH-94 protocol before the patient fulfilled 5 of the 8 requisite diagnostic criteria (Table 111). The platelet count decreased to 10,000/mm3 and fibrinogen was 117 mg/dL (normal 200–400 mg/dL).

Table 1. Diagnostic Criteria Used for the HLH-2004 Trial
Full size table

Day 14: The patient reported minor clinical improvement, but the ferritin level increased further to 28,636 ng/mL. Confirmatory test results for HLH were received and supported this diagnosis: natural killer cell activity was 0.68% (control = 79%), and sCD25 was 11,400 U/mL (normal<2000 U/mL). A repeat bone marrow biopsy revealed hemophagocytosis (Fig. 1). Treatment with pulsed methylprednisolone 1000 mg/day for 5 days and intravenous immunoglobulin (IVIG) 2 g/kg was added to etoposide.

Figure 1.
figure 1

Hemophagocytosis was present on the second bone marrow aspirate. Touch-preparation of the bone marrow biopsy showing phagocytosis by a macrophage of a lymphocyte (arrow) and platelet (arrowhead).

Full size image

Day 17: The patient’s condition was notably improved, and ferritin level decreased to 4528 mg/mL. Rituximab was administered as a treatment for CLL.

By day 30, the patient’s temperature and blood count were normal, and her HLH appeared to be controlled. However, fever recurred and was accompanied by a cough that was now productive. Physical exam revealed crackles over the left lung base. A chest CT revealed nodular infiltrates in both lungs and an evolving fungus ball (Fig. 2). She underwent bronchoscopy and a bronchial lavage sample studied under direct microscopy demonstrated branched hyphae. Treatment with voriconazole was begun and Aspergillus fumigatus subsequently grew on culture.

Figure 2.
figure 2

Lung aspergillosis further complicated the clinical course. A computed tomography image demonstrating bilateral nodular infiltrates (arrowheads) and an evolving fungus ball with a halo sign at the periphery of the right upper lobe (arrow).

Full size image

On day 40, a fever of 39.4°C was noted and was accompanied by confusion and right arm paresis. Ferritin increased to 49,000 ng/mL, fibrinogen was 100 mg/dL, the neutrophil count decreased to 840/mm3, and the platelet count was 10,000/mm3. CT of the brain ruled out intracranial hemorrhage, and the clinical estimation was that the neurological symptoms reflect HLH reactivation. On day 42, the patient received alemtuzumab as salvage therapy.

On the 45th day after her initial admission, the patient’s neurologic and general clinical condition deteriorated further. She became comatose and oliguric. Ferritin levels increased to 146,000 ng/mm3. After discussing with her two children involved in all treatment decisions throughout her disease course, active treatment was withdrawn, and comfort measures were provided. She died peacefully 2 days later.

DISCUSSION

HLH is a syndrome of immune dysregulation that is often diagnosed using the HLH-2004 diagnostic criteria. The criteria are fulfilled when a patient has a genetic lesion associated with HLH or fulfills 5 of the 8 HLH-2004 criteria (Table 18). The criteria are termed “diagnostic,” although, by definition, they are classification criteria developed for the enrollment in the international HLH-2004 therapeutic clinical trial for the genetic form of the disease, familial HLH. These criteria were based on empiric observations in familial HLH using the technology and knowledge available at the time. Their specificity and sensitivity for M-HLH are unknown. They are used pragmatically as diagnostic criteria for all forms of HLH, which may result in a delay in diagnosis or misdiagnosis.12

The process of diagnosing multi-system inflammatory disorders requires an understanding of the difference between classification and diagnostic criteria. Classification criteria are most often derived from well-defined homogenous patient cohorts established for clinical research. These are very specific by design and thus risk missing patients with the relevant diagnosis. Diagnostic criteria are more sensitive and must reflect the different features of disease to broadly identify as many individuals with the condition as possible.13 In light of this, the physician’s clinical judgment is crucial in refining the diagnostic process. Our patient was diagnosed on day 14, and ordering the sCD25 test earlier could have expedited the diagnosis (Fig. 3). Furthermore, early empiric administration of corticosteroids may delay reaching a specific diagnosis, may mask possible disease transformation, and should generally be avoided.10

Figure 3.
figure 3

Key features of the patient’s clinical course. Ferritin levels during the patient’s clinical course. The days of HLH-2004 criteria fulfillment and treatment administered are noted. HLH-94 refers to an 8-week protocol of etoposide and dexamethasone.

Full size image

Diagnosing HLH precisely is important since treatment involves the administration of potentially toxic chemotherapy. HLH-2004 features such as splenomegaly may be a direct consequence of the underlying neoplasm.14 Our patient had mild splenomegaly for at least a year as a feature of her CLL, and we were uncertain if we should consider it a diagnostic criterion. We (A.Z.L, M.J, M.E) recently examined 225 patients with hematologic malignancies HLH and demonstrated that most diagnostic features overlap among patients with or without M-HLH and cannot discriminate these clinical situations. We developed a novel diagnostic index based on the combined elevation of sCD25 (>3900 U/mL) and ferritin (>1000 ng/mL), the Optimized HLH Inflammatory (OHI) index for M-HLH diagnosis.15 Our patient had a positive OHI index which we could calculate when sCD25 test results were obtained, highlighting the importance of early sCD25 testing. This study suggests that HLH diagnosis should be more reliant on inflammatory features than on features of tissue infiltration such as cytopenia and splenomegaly.

Diagnosis may also be hampered by over-reliance on finding hemophagocytosis on tissue biopsy. HLH is frequently diagnosed by hematologists who are used to pathology-based diagnosis and may be biased toward diagnosing HLH only when tissue evidence of hemophagocytosis is present, as implied by the name of the syndrome. However, the findings of hemophagocytosis are not sensitive or specific enough to diagnose familial or reactive HLH.16 In our study of M-HLH in patients with hematologic malignancies, hemophagocytosis had a sensitivity of only 43%. Our patient did not have hemophagocytosis in her first biopsy, and there was no need to repeat the biopsy once the diagnosis was already established.

HLH has been described in association with all hematologic malignancies and due to a variety of triggers.17M-HLH is usually diagnosed at presentation or transformation of lymphoma or after chemotherapy. Pasvolsky et al. recently described a case series highlighting the diagnostic challenge of M-HLH as the presenting manifestation of lymphoma, with two of the four patients dying before a diagnosis of lymphoma was made despite HLH-directed therapy.18 HLH associated with CLL has been described with disease transformation,19, 20 (seeTable 2) infections 21,22,23,24,25,26, or ibrutinib treatment27 (Table 2). Ours is the fourth case reported in which no trigger was identified28, 29, 32 (see Table 2). Furthermore, our patient demonstrated refractory HLH and aggressive measures to exclude disease transformation such as lymph node biopsy or splenectomy33 and nucleic acid testing for EBV reactivation (serology indicated past infection) should be considered in these patients.

Table 2. Triggers for HLH in CLL Patients
Full size table

While making an accurate diagnosis is essential, persistent consideration of other processes “mimicking”10 HLH is nonetheless important. Pulmonary aspergillosis developed in our patient with a high fever as the initial clinical feature. This temporarily raised the concern for progression of HLH, but appropriate investigations soon allowed the correct infectious etiology to be identified, highlighting the importance of maintaining a high level of suspicion for infectious complications of prolonged, profound immunosuppressive therapy. Patients complicated by these infections would probably be harmed from additional immunosuppressive therapy for HLH.

While understanding of the pathogenesis of inflammatory diseases continues to improve, most cases are not diagnosed using a single test.30 Neither classification nor diagnostic criteria can accurately capture all patients, and clinical judgment is crucial in the diagnostic process. Therefore, clinical research to calibrate and validate current diagnostic criteria for M-HLH is needed. In addition, basic and translational research providing insights into the mechanisms of M-HLH may provide more specific diagnostic markers.

This case highlights the importance of early sCD25 testing, the need to apply HLH classification and diagnostic criteria while bearing in mind their limitations, and the importance of constant vigilance to detect possible HLH mimickers such as infection. Decision-making in HLH is challenging: improved understanding of the aberrant inflammation in this condition may yield tools to improve the diagnosis and management of these patients.