Table 2 Limitations of the HLH- 2004 Criteria

Fever

A common manifestation of immune/inflammatory responses of all etiologies. [10]

Cytopenias

May be a consequence of any profoundly inflammatory state. Among patients hospitalized in ICUs, 50–70% will have anemia to < 9 g/dL, up to 60% thrombocytopenia, and 7% or more will leukopenia. [11]

Hypofibrinogenemia

May result from liver injury or DIC, both common consequences of any/all fulminant inflammatory states, and may occur in over a third of patients admitted to ICUs for any indication. [12]

Hypertriglyceridemia

Non-specific acute phase reactants (not unlike ESR or CRP). The cutoff used in the HLH- 2004 criteria is low enough to be within range of baseline levels among a significant proportion of adults in developed countries. [13, 14]

Splenomegaly

A frequent finding in many of the entities which share a differential diagnosis with HLH including hematologic malignancy, viral infection, and a number of rheumatologic disorders. [15]

Hyperferritinemia

The HLH- 2004 ferritin cutoff demonstrated a specificity of merely 0.3% for HLH (in a critical care population wherein pretest probability may be higher than in lower acuity settings) [16]. Among a cohort of 1055 adult patients with serum ferritin > 5000 ng/mL the prevalence of diagnosed HLH was 6.5%, with prevalence only reaching 50% as serum ferritin approached 90,000 ng/mL [17]. In these cohorts and others, a wide variety of common non-HLH conditions have been associated with profound hyperferritinemia including sepsis, hematologic malignancy, rheumatologic disease, liver injury, and kidney failure

Soluble IL- 2 receptor

May be increased in any process involving T-cell activation (including sepsis, hematologic malignancy, rheumatologic disease, sarcoidosis, and inflammatory bowel disease) [20,21,22,23,24]. Among 132 patients with soluble IL- 2 receptor levels checked for evaluation of HLH, the specificity of the HLH- 2004 cutoff value was 38.8%, with an AUC for the corresponding ROC of 0.69, and no significant difference in levels when comparing patients with HLH, and non-HLH patients with sepsis, hematologic malignancy, or rheumatologic disease [25]

NK-Cell Activity

A cohort of 34 secondary HLH patients demonstrated an “activated NK phenotype profile” similar to inflammatory conditions such as sepsis or rheumatologic disease [27]. Among a cohort of 311 HLH patients, those with primary disease had significantly lower NK cell activity than those with secondary disease, with many secondary HLH patients exhibiting NK activities within the normal range [28]. In primary HLH, the NK-cell cytotoxicity assay has displayed low reliability with a poor AUC of 0.69 at the diagnostic ROC [29]

Bone Marrow Hemophagocytosis

A non-specific finding which may be encountered in a wide array of critically ill patients, including as many as 65% of autopsied ICU deaths, and 44% of autopsied inpatients [30, 31]