HLH Diagnosis: Criteria, HScore, and Biomarkers

Hemophagocytic lymphohistiocytosis, or HLH, is diagnosed by meeting a combination of clinical and laboratory criteria rather than any single test. The most widely used framework requires at least five of eight specific findings, ranging from persistent high fever and an enlarged spleen to extremely elevated ferritin and signs of immune system overactivation. Despite those formal criteria, HLH remains one of the most difficult diagnoses in medicine because its symptoms mimic sepsis, aggressive cancers, and other critical illnesses, and no individual lab value confirms it on its own.

The HLH-2004 Criteria

The standard diagnostic framework comes from the Histiocyte Society’s HLH-2004 guidelines, which updated the earlier HLH-94 criteria. The original criteria used five findings; HLH-2004 expanded the list to eight and still requires that five be met for a clinical diagnosis.

¹ The eight criteria are:

Fever: temperature above 38.5 °C (101.3 °F)
Splenomegaly: an enlarged spleen, defined as more than 2 cm below the rib margin
Low blood counts: at least two of three cell lines affected, meaning low hemoglobin, low platelets, or low neutrophils
High triglycerides or low fibrinogen: fasting triglycerides above a certain threshold, or fibrinogen below 1.5 g/L
Hemophagocytosis: immune cells seen engulfing blood cells on a tissue biopsy, usually from bone marrow
High ferritin: above 500 µg/L
Low or absent NK-cell activity: natural killer cells that are not functioning normally
Elevated soluble CD25: a marker of T-cell activation, above 2,400 U/mL

An important exception exists: if a patient has a confirmed family history or genetic mutation known to cause HLH, they can be diagnosed without meeting five of the eight criteria.² This matters because some patients present so rapidly that waiting for every lab result would cost dangerous time.

A revised version of these criteria, sometimes called HLH-2024, was recently proposed by the Histiocyte Society. The update drops NK-cell activity as a standalone criterion and instead emphasizes genetic testing and lymphocyte cytotoxicity assays as two separate diagnostic pathways alongside the remaining seven clinical criteria.³

The HScore as an Alternative

The HLH-2004 criteria were originally designed for pediatric patients enrolled in a clinical trial, which makes their application to adults somewhat awkward. The HScore, developed in 2014, takes a different approach: instead of a yes-or-no checklist, it assigns weighted points for each clinical and laboratory finding and produces a probability of HLH on a continuous scale. Variables include temperature, organ size, blood counts, ferritin, triglycerides, fibrinogen, liver enzyme levels, and whether hemophagocytosis is seen on biopsy.

Studies comparing the two tools have found that an HScore of roughly 168 to 169 predicts HLH with high sensitivity. One multicenter ICU study found that an HScore cutoff of 168 achieved perfect sensitivity and about 94% specificity.⁴ A separate multicenter validation found that an HScore of 169 or above had 96% sensitivity but only 71% specificity, meaning it casts a wider net and catches more cases but also flags some patients who turn out not to have HLH.⁵ In practice, the HScore tends to be more useful than HLH-2004 when the trigger is autoimmune or inflammatory, while HLH-2004 outperforms it in most other settings.⁶

Neither tool is perfect, and many clinicians use both. The HScore is particularly valued because it does not require NK-cell activity testing or soluble CD25, both of which take days to come back from specialized labs, making it a faster bedside calculation.⁷

What Ferritin Actually Tells You

Ferritin is the lab value most associated with HLH in the popular clinical imagination, and extremely high levels do raise a red flag. In children, a ferritin above 10,000 µg/L has been reported to be about 90% sensitive and 96% specific for HLH, which is a strong performance for a single test.⁸ That finding, from a pediatric study, gets quoted frequently. The trouble is that it does not hold up as well in adults.

A study of over a thousand adult patients with ferritin above 5,000 found that only about 6.5% actually had HLH. The most common reasons for an extremely high ferritin were bacterial infections, blood cancers, kidney failure, and liver injury. HLH only became the most likely diagnosis once ferritin approached 90,000. The study identified an optimal adult cutoff of 16,000, which caught about 79% of HLH cases while correctly ruling it out in another 79% of non-HLH patients.⁹ A separate ICU-focused study suggested that even 3,951 might be a better cutoff than the 500 in the HLH-2004 guidelines when screening critically ill adults.¹⁰

The upshot: a ferritin of 500 as used in HLH-2004 is deliberately set low to avoid missing cases, but it will flag many patients who have something else entirely. Very high ferritin should always prompt a clinician to consider HLH, but the diagnosis still depends on the full clinical picture.

Soluble IL-2 Receptor and Immune Activation Markers

Soluble interleukin-2 receptor, also called soluble CD25 or sIL-2r, is one of the HLH-2004 criteria and reflects how activated a patient’s T cells are. When T cells go into overdrive, they shed this receptor into the blood, and levels spike. In one adult study, sIL-2r had an area under the curve of 0.90 for distinguishing HLH from other conditions, outperforming ferritin’s 0.78 in the same cohort. At a threshold around 2,515 U/mL, it was 100% sensitive, meaning it caught every case of HLH, though its specificity was 72.5%, so a positive result still required clinical judgment.¹¹

A systematic scoping review confirmed that sIL-2r is an important diagnostic and disease marker but also noted a lack of large-scale primary data on how it performs in adults specifically.¹² The practical limitation is availability: many hospitals cannot run this test in-house, so results may take several days. When it is available, though, sIL-2r is one of the strongest single markers a clinician can use.

Why a Bone Marrow Biopsy Is Not Definitive

Many people assume that seeing hemophagocytosis, the literal engulfing of blood cells by macrophages, on a bone marrow biopsy clinches the diagnosis. That is a misconception. One study found that hemophagocytosis in the marrow had a sensitivity of about 83% but a specificity of only 60%, meaning it missed some genuine HLH cases and appeared in many patients who did not have HLH at all.¹³ A separate analysis confirmed that the amount of hemophagocytosis on a marrow sample did not correlate with how likely a patient was to actually have HLH.¹⁴

There is a more nuanced version of this finding. Not all hemophagocytosis is equal. Engulfment of non-nucleated red blood cells alone is relatively nonspecific and can appear in patients with infections, transfusions, or other inflammatory states. What is much more diagnostic is seeing macrophages that have swallowed granulocytes, nucleated red blood cells, or multiple nucleated cells at once. A study that trained pathologists to quantify these specific types found they could distinguish HLH from non-HLH with an area under the curve around 0.90.¹⁵ So the biopsy can be genuinely informative when read with a trained eye, but a casual “yes, there’s some hemophagocytosis” is not enough to confirm or rule out HLH.

Cytotoxic Function Testing and Genetic Workup

NK-cell function testing has been part of the diagnostic criteria since HLH-2004, but its role has evolved. Traditional NK-cell cytotoxicity assays measure how well a patient’s natural killer cells can kill target cells in the lab. These tests have shown high sensitivity in some settings, but a study comparing multiple approaches found that NK-cell cytotoxicity actually had only about 60% sensitivity for detecting patients with confirmed genetic HLH mutations. By contrast, measuring perforin protein levels by flow cytometry had 97% sensitivity, and CD107a degranulation testing had 94% sensitivity.¹⁶ This is one of the reasons the updated 2024 criteria de-emphasized the traditional NK-cell activity test in favor of these more targeted assays.

Perforin and CD107a tests are particularly useful for identifying familial (genetic) HLH, where the underlying problem is a defect in the machinery that immune cells use to kill infected or abnormal cells. In adults, genetic HLH turns out to be more common than previously thought, though adults tend to have milder mutations that may not fully knock out function until some additional stressor, like an infection, pushes the system over the edge.¹⁷

Telling HLH Apart from Sepsis

The single hardest diagnostic challenge in HLH is distinguishing it from severe sepsis, because the two conditions share many features: high fever, organ dysfunction, dropping blood counts, and a critically ill patient. Several features tilt the balance toward HLH rather than sepsis. Extremely high ferritin is probably the most useful single clue. A low fibrinogen level also points toward HLH, since sepsis typically drives fibrinogen up rather than down. Marked splenomegaly, severe cytopenias involving multiple cell lines, and high triglycerides in adult patients all favor HLH.¹⁸ A low C-reactive protein in the setting of an otherwise critically ill patient is another red flag, since CRP is usually sky-high in bacterial sepsis.

Cytokine patterns can also help. A pattern of markedly elevated interferon-gamma and IL-10 with only mildly elevated IL-6 is more consistent with HLH than with straightforward infection.¹⁹ The complication, of course, is that HLH and sepsis are not mutually exclusive. Infection is one of the most common triggers of HLH, so a patient can have both at the same time.

Common Triggers and Why They Matter for Diagnosis

HLH is broadly split into primary (genetic) and secondary (acquired) forms. In secondary HLH, the underlying trigger directly shapes the diagnostic approach, because treating HLH without addressing its cause leads to relapse.

Infections, particularly viral ones, are the most common trigger. Epstein-Barr virus occupies a complicated role: it can trigger familial HLH in someone with an underlying genetic defect, cause HLH on its own in otherwise healthy people through an overwhelming immune reaction, or drive HLH through an associated lymphoma. The spectrum ranges from highly curable non-cancerous forms to rapidly fatal T-cell lymphomas, so pinning down exactly how EBV is involved is critical.²⁰

Malignancy-associated HLH, particularly from lymphomas, creates a diagnostic challenge because cancer symptoms and HLH symptoms overlap heavily. A patient with an undiagnosed lymphoma may present with fevers, cytopenias, and organ enlargement that could be attributed to either the cancer or the HLH it is triggering.²¹ Missing the underlying malignancy while treating the HLH alone is a well-recognized pitfall.

Autoimmune-driven HLH, often called macrophage activation syndrome (MAS), is most commonly seen in children with systemic juvenile idiopathic arthritis but can occur with lupus and other rheumatic diseases. MAS has its own diagnostic criteria, developed separately in 2016, because the baseline lab abnormalities of active autoimmune disease change what “abnormal” looks like. For instance, the ferritin-to-ESR ratio has been studied as a quick screening tool: a ratio above 21.5 was 100% sensitive for identifying MAS in one cohort of children with systemic JIA.²² Disease-specific criteria help clinicians recognize MAS earlier than HLH-2004 criteria alone would.²³

When the Brain Is Involved

Central nervous system involvement in HLH, sometimes called neuro-HLH, is more common than many clinicians realize and carries a worse prognosis. Some experts recommend that all patients with confirmed or suspected HLH undergo a lumbar puncture and brain MRI, even if they have no neurological symptoms.²⁴ In one pediatric study, half the patients had abnormal spinal fluid findings but only a third had abnormal MRI results, suggesting that imaging alone will miss many cases of CNS disease.²⁵

When MRI is abnormal, HLH tends to produce a characteristic pattern of symmetric white matter lesions near the brain’s ventricles, which differs from the more scattered pattern seen in conditions like acute disseminated encephalomyelitis.²⁶ Blood markers including low platelets, low fibrinogen, and high liver enzymes correlate with the severity of brain involvement on MRI.²⁷ Spinal fluid analysis should include standard cell counts and protein, plus a cytospin preparation examined specifically for hemophagocytosis.²⁸

Why Diagnostic Speed Matters

HLH is often fatal without treatment, and every day of delay translates to worse outcomes. A large study comparing patients who received treatment within six days versus those treated later found that the late-treatment group had roughly double the odds of dying in the hospital. They also had higher rates of shock, need for mechanical ventilation, blood clots, kidney injury, and serious infections.²⁹ Troublingly, the same study found no significant trend toward faster treatment over the years, suggesting the medical system is not getting much better at recognizing HLH promptly.

This is why many experts advocate starting treatment on strong clinical suspicion rather than waiting for every criterion to be met. A patient with high fever, dropping cell counts, an enormous ferritin, and no clear alternative explanation should receive HLH-directed therapy while the soluble CD25 and genetic tests are still pending.

Emerging Biomarkers on the Horizon

Two newer biomarkers are generating excitement as potential additions to the diagnostic toolkit. CXCL9, a chemokine produced downstream of interferon-gamma signaling, has emerged as a promising marker for confirming HLH, tracking treatment response, and catching relapses early.³⁰ A recent head-to-head comparison of CXCL9 and interferon-gamma found that CXCL9 was elevated in nearly all HLH patients tested (about 98%), while interferon-gamma was normal in about 17% of confirmed cases. Both were strong at distinguishing HLH from non-EBV infections, but interferon-gamma was far better at telling EBV-driven HLH apart from ordinary infectious mononucleosis.³¹

The two markers also behaved differently over time. Interferon-gamma dropped to normal within a week in most treated patients, making it useful as a quick check on whether treatment is working. CXCL9, on the other hand, stayed elevated longer, and persistent elevation correlated with disease relapse.³² This suggests a complementary strategy: use both to confirm the diagnosis initially, then lean on CXCL9 for longer-term surveillance. Neither is yet part of formal diagnostic criteria, but their incorporation seems likely as more validation data accumulate.

HLH After CAR-T Cell Therapy

A relatively new clinical scenario is HLH triggered by chimeric antigen receptor T-cell (CAR-T) immunotherapy. CAR-T cells are engineered immune cells infused to fight certain cancers, and they can provoke an intense inflammatory reaction. HLH occurring after CAR-T infusion was initially lumped in with cytokine release syndrome, the expected inflammatory side effect. It has since become clear that CAR-T-associated HLH is a distinct entity with a higher mortality rate and requires different management.³³ For oncologists managing CAR-T patients, recognizing this overlap early is essential, because standard cytokine release syndrome protocols alone may not be sufficient.

Neonates Present Differently

Diagnosing HLH in newborns poses its own set of problems. Neonates with HLH often present with multi-organ dysfunction that looks indistinguishable from neonatal sepsis. Herpes simplex virus is the most common infection associated with neonatal HLH, and the condition carries a very high mortality rate in this age group.³⁴ HLH should be on the differential for any neonate with unexplained multi-organ failure, particularly when standard sepsis treatment is not producing improvement. Genetic testing is especially important in this group, since neonatal onset is a strong indicator that a familial form may be at play.