The inherited forms of HLH result from mutations that disable the granule-mediated killing pathway of natural killer (NK) cells and cytotoxic T cells; the first to be identified was in the perforin gene (Stepp et al., 1999). A lymphocyte that cannot kill an infected or antigen-presenting cell goes on being stimulated by it, and the resulting cytokine output activates macrophages. Secondary, or acquired, HLH reaches the same end state without a known single-gene cause. The division is less sharp than the names suggest. Zhang et al. (2011) found hypomorphic variants in the genes responsible for familial disease in a proportion of adults with apparently acquired HLH, which implies that some young adults carry a partial defect that declares itself only under the stress of a strong infectious trigger.

Virus-associated hemophagocytic syndrome was described as a distinct entity by Risdall et al. (1979). Herpesviruses dominate the list of triggers, Epstein-Barr virus (EBV) above all, followed by cytomegalovirus (CMV); HIV, influenza, dengue and parvovirus B19 are also well documented (Rouphael et al., 2007). EBV-associated HLH ranges from a self-limiting illness to a fulminant one and has been studied most closely in East Asia (Imashuku, 2002). In young men, fulminant EBV infection with HLH may be the first sign of X-linked lymphoproliferative disease. The hyperinflammation seen in some people with severe COVID-19 renewed general interest in screening for the syndrome (Mehta et al., 2020).

No single feature of HLH is specific. Fever, cytopenias, raised ferritin, abnormal liver function and coagulopathy all occur in severe sepsis, and in HLH they accumulate over days, so that the full pattern may not be present at any one assessment. Hemophagocytosis in the marrow, the finding that gives the syndrome its name, is often absent on the first aspirate (Gupta et al., 2008) and can be seen in other conditions. The diagnosis therefore depends on someone deciding to look for a combination of findings and to keep looking.

The prompt for the work-up is a clinical pattern and not a single result. HLH should enter the differential diagnosis when fever continues beyond the expected course of a viral illness, or returns after initial improvement, together with any two of the following: a fall in two or more blood cell lines, a palpable or imaged spleen, liver enzymes rising without another explanation, unexplained coagulopathy, or new neurological symptoms. Failure to respond to appropriate antibiotics strengthens the suspicion. The threshold is deliberately low at this point, since what follows is a set of inexpensive blood tests.

The first-line screen consists of a full blood count with film, ferritin, fasting triglycerides, fibrinogen with a coagulation profile, liver enzymes, lactate dehydrogenase and renal function. Ferritin is the test most often omitted and among the most informative. In children a very high ferritin is strongly suggestive of HLH (Allen et al., 2008). In adults the same cannot be assumed: Schram et al. (2015) reviewed adults with extreme hyperferritinemia at a general hospital and found that most had renal failure, liver injury, infection or haematological malignancy, with HLH accounting for a minority. A normal ferritin makes HLH much less likely; a raised one is a reason to proceed and not a diagnosis. Because the abnormalities evolve, the screen is repeated every day or two while suspicion persists, and the direction of travel is noted. A falling fibrinogen in a febrile person is a particular warning, since fibrinogen ordinarily rises with inflammation.

Two instruments should be completed in writing. The HLH-2004 criteria, devised for a paediatric treatment trial, are met by a molecular diagnosis or by five of eight findings: fever; splenomegaly; cytopenias in at least two lineages; raised fasting triglycerides or low fibrinogen; hemophagocytosis in tissue; low or absent NK-cell activity; ferritin of 500 μg/L or more; and raised soluble CD25 (Henter et al., 2007). The HScore was derived and validated in adults and converts nine weighted variables into a probability of reactive hemophagocytic syndrome (Fardet et al., 2014). The two overlap, as Table 1 shows, but they differ in purpose. The criteria classify; the score estimates likelihood and can be calculated before specialised tests return. Consensus recommendations for adults endorse the use of both and stress that failure to meet five criteria on a given day does not exclude the diagnosis (La Rosée et al., 2019).

If Steps 2 and 3 leave HLH probable, a bone marrow aspirate and trephine are indicated. Their purpose is twofold: to look for hemophagocytosis and, more importantly, to exclude lymphoma, leukaemia and marrow infection. The absence of hemophagocytosis should not stop the work-up. Soluble CD25, the shed α chain of the interleukin-2 receptor, reflects T-cell activation and should be requested where a laboratory offers it, as should NK-cell function. These results take days, and experienced authors advise against waiting for them when the clinical picture is deteriorating (Jordan et al., 2011).

The antecedent viral illness should be defined by testing and not presumed. Quantitative polymerase chain reaction (PCR) for EBV and CMV in blood is the core of this step, since serology can mislead when immunity is disturbed. It is accompanied by HIV testing and, according to exposure and geography, tests for dengue, influenza, parvovirus B19, the hepatitis viruses and others (Rouphael et al., 2007). Finding a virus does not close the search. Lymphoma, of T-cell and NK-cell types especially, can present as HLH with EBV in the blood, so cross-sectional imaging, with biopsy of any accessible abnormal tissue, forms part of the same step. An autoimmune screen and a review of recent medication complete it. In people with a known rheumatic disease the same syndrome is termed macrophage activation syndrome and has its own classification criteria (Ravelli et al., 2016).

In a young adult the possibility of a late-presenting genetic form deserves explicit thought, particularly when the episode is severe, when EBV is the trigger, when it recurs, or when there is parental consanguinity or a family history of early death from infection. Flow cytometric screening for perforin expression, for degranulation measured by surface CD107a and, in males, for the SAP and XIAP proteins can be followed by sequencing of the relevant genes (Canna and Marsh, 2020). The result matters beyond the acute episode, because confirmed primary disease is an indication to consider haematopoietic stem cell transplantation and has implications for relatives.

Haematology should be involved from the moment HLH becomes probable, and critical care early if organ function is declining. Published regimens derive from the HLH-94 paediatric protocol, built on etoposide and dexamethasone (Henter et al., 2002), and adult practice adapts them to the trigger and to the condition of the individual (Schram and Berliner, 2015; La Rosée et al., 2019). The choice and intensity of treatment lie outside this guide. What the diagnostic approach contributes is a dated record of which criteria were met and when, a search for the trigger that continues after treatment starts, and a planned reassessment of the diagnosis if the expected response does not follow.