Background: Hemophagocytic lymphohistiocytosis (HLH) syndrome is due to an abnormal and sustained activation of T lymphocytes and macrophages releasing pro-inflammatory cytokines in the microenvironment. HLH is characterized by clinical and biological features that includes fever, hepatomegaly, splenomegaly, high ferritin level, pancytopenia and liver biology abnormalities. HLH is commonly divided into genetic/hereditary cases (HLHg) and non genetic/ sporadic (HLHs) with no hints for an inherited causes. However, similarity of both clinical symptoms and biological abnormalities between HLHg and HLHs raised, however, the hypothesis that HLHs could have a genetic component that may be related to HLHg. However contribution of genetic factors in both, pathophysiology and prognosis of sporadic hemophagocytic lymphohistiocytosis (HLHs) remains unknown. To explore the hypothesis of a genetic susceptibility in HLHs, from 2010 to 2017, we constituted a cohort of 205 HLHs patients (adults and children) and analysed genes involved in genetic HLH (HLHg).

Methods: Patients fulfillied modified HLH-2004 diagnostic criteria. Data collected included (i) clinical and biological features, (ii) associated diseases (AID), (iii) infectious events, (iv) treatment, and (v) clinical outcome during one year of follow up. HLHg related genes LYST, PRF1, UNC13-D, STX11, STXBP2, RAB27A, SH2D1A, XIAP were NGS sequenced in 172 patients. Selected variants were (i) unknown but met at least two criteria among MAF<0.05, CADD >10, and/or Polyphen HDIV (>0.485) (n=50), (ii) previously related to HLHg (n=2), PRF1 A755G/N252S (MAF <0.05) and PRF1 C272T/A91V (MAF <0.05, CADD >10 and Polyphen HDIV). Two control cohorts were selected, the first one using the variant files from the European subpopulation of the 1000 Genomes (1000G cohort, n=502) phase 3 project and the second one from the local in-house variants data base (in-house cohort, n=5092).

Results: Patients presented with neoplasia (n=62), AID (n=46) in which infection was mainly Herpes virus (EBV) (> 80%), or no etiology (idiopathic) (n=64) with a greater infectious diversity. Fifty-two variants met the selection criteria, they were distributed overall in 115 alleles in 84/172 patients. Variants were globally heterozygous. They were recurrent with occurrence in 2 patients for STX11, RAB27A and LYST genes to 10 patients for the N252S and 18 patients for the p.Ala91Val in PRF1.

Overall the frequency of individuals exhibiting at least one variant was significantly in excess (48.8%, n=84) in the HLHs versus in in-house (35.9%, n=1826) (p=0.001, CAST test) or versus in 1000G population patients (38.6%, n=194) (p=0.02, CAST test).

Importantly, the distribution of patients carrying either 0, 1, 2, 3 or 4 mutated alleles in HLHs cohort was statistically different from the two control populations (Trend test: HLH vs in-house p<0.0001 and HLH vs 1000G p<0.001; SUM test: HLH vs in-house p<0.001 and HLH vs 1000G p=0.001). We observed an excess of patients carrying one or two mutated in the HLHs cohort vs in-house OR (1 allele vs 0 allele) =1.49 [IC 1.05-2.11]; (2 alleles vs 0 allele) =2.52 [IC 1.54-4.00] and vs 1000G OR 1 allele vs 0 allele =1.27 [IC 0.85-1.88]; 2 alleles vs 0 allele =2.76 [IC 1.50-5.05]. Finally, PRF1/RAB27Aa and PRF1/STXBP2 were found to be positively associated with a highier frequency in HLHs group than expected from a random distribution (p<0.01 and p=0.05, respectively).

Importantly, severe patients carried a significantly higher number of variants (1, 2 or 3) than the one with a better outcome (p<0.001, Fisher test exact). A multivariate analysis including sex, age, associated diseases, and infectious status as confounding factors, confirm that the number of HLH genes variants is correlated to the severity of the syndrome independently of these others factors OR(S/NS; 1 variant/0 variant)=3.03 [1.45-6.32] (p=0.003) and OR(S/NS; 2 variants/0 variant) =4.55[1.61-12.84] (p=0.004).

Conclusion: The results presented in this study strongly support the hypothesis that HLHs has a genetic component when tested with genes of HLH. Overall, this results established clearly a genotype/phenotype correlation between presence of one or more filtered HLH genes variants and severity of the HLH syndrome in patients. Finally, we proposed a di/tri genic model as an expression of genetic component in sporadic HLH.


Hermine:Novartis: Research Funding; AB science: Consultancy, Equity Ownership, Honoraria, Research Funding; Celgene: Research Funding.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution