Stem cell transplantation for children with hemophagocytic lymphohistiocytosis: results from the HLH-2004 study

Abstract

We report the largest prospective study thus far on hematopoietic stem cell transplantation (HSCT) in hemophagocytic lymphohistiocytosis (HLH), a life-threatening hyperinflammatory syndrome comprising familial/genetic HLH (FHL) and secondary HLH. Although all patients with HLH typically need intensive anti-inflammatory therapy, patients with FHL also need HSCT to be cured. In the international HLH-2004 study, 187 children aged <18 years fulfilling the study inclusion criteria (5 of 8 diagnostic criteria, affected sibling, or molecular diagnosis in FHL-causative genes) underwent 209 transplants (2004-2012), defined as indicated in patients with familial/genetic, relapsing, or severe/persistent disease. Five-year overall survival (OS) post-HSCT was 66% (95% confidence interval [CI], 59-72); event-free survival (EFS) was 60% (95% CI, 52-67). Five-year OS was 81% (95% CI, 65-90) for children with a complete response and 59% (95% CI, 48-69) for those with a partial response (hazard ratio [HR], 2.12; 95% CI, 1.06-4.27; P = .035). For children with verified FHL (family history/genetically verified, n = 134), 5-year OS was 71% (95% CI, 62-78) and EFS was 62% (95% CI, 54-70); 5-year OS for children without verified FHL (n = 53) was significantly lower (52%; 95% CI, 38-65) (P = .040; HR, 1.69; 95% CI, 1.03-2.77); they were also significantly older. Notably, 20 (38%) of 53 patients without verified FHL had natural killer cell activity reported as normal at diagnosis, after 2 months, or at HSCT, suggestive of secondary HLH; and in addition 14 (26%) of these 53 children had no evidence of biallelic mutations despite having 3 or 4 FHL genes analyzed (natural killer cell activity not analyzed after 2 months or at HSCT). We conclude that post-HSCT survival in FHL remains suboptimal, and that the FHL diagnosis should be carefully investigated before HSCT. Pretransplant complete remission is beneficial but not mandatory to achieve post-HSCT survival. This trial was registered at www.clinicaltrials.gov as #NCT00426101.

Conflict of interest statement

Conflict-of-interest disclosure: I.A., A.H., K.L., K.L.M., E.S., and J.-I.H. serve as consultants for SOBI. A.H. serves as a speaker for Novartis. The remaining authors declare no competing financial interests.

© 2020 by The American Society of Hematology.

Figures

Graphical abstract

Figure 1.

Kaplan-Meier estimates of survival after HSCT in the HLH-2004 study. Five-year probabilities of survival (pSu) are indicated with a 95% CI. OS or EFS is displayed, where death, and death or second HSCT, were defined as events, respectively. P values from Cox proportional hazards models. (A) OS for the entire HLH-2004 cohort (n = 187, events n = 67). (B) EFS for the entire HLH-2004 cohort (n = 187, events n = 78). (C) OS for children with verified FHL (n = 134, events n = 43, blue line) and children without verified FHL (n = 53, events n = 24, red line dashed); P = .040. (D) EFS for children with verified FHL (n = 134, events n = 54, blue line) and children without verified FHL (n = 53, events n = 24, red line, dashed); P = .27.

Figure 2.

Kaplan-Meier estimates of EFS based on HSCT donor and conditioning in the HLH-2004 study. Five-year probabilities of survival (pSu) are indicated with a 95% CI. Death or second HSCT were defined as events. P values from Cox proportional hazards models. (A) EFS for the entire cohort based on donor: MRD (n = 44, event n = 21, blue line); MUD (n = 60, event n = 24, red line, dashed); UCB (n = 53, event n = 23, black line, dotted)], P = .52. [MUD vs MRD, P = .26; UCB vs MRD, P = .61]. (B) EFS for children with verified FHL based on donor (MRD, n = 31, event n = 14, blue line; MUD, n = 45, event n = 18, red line, dashed; UCB, n = 39, event n = 14, black line, dotted; P = .65 (MUD vs MRD, P = .40; UCB vs MRD, P = .44). (C) EFS for the entire cohort based on conditioning (busulfan-based, n = 99, events n = 42, blue line; fludarabine based, n = 39, events n = 18, red line dashed; treosulfan based, n = 20, events n = 5, black line, dotted; P = .31 (fludarabine vs busulfan, P = .74; treosulfan vs busulfan, P = .16). (D) EFS for children with verified FHL based on conditioning (busulfan based, n = 69, events n = 27, blue line; fludarabine based, n = 26, events n = 12, red line, dashed; treosulfan based, n = 18, events n = 5, black line, dotted; P = .49 (fludarabine vs busulfan, P = .49; treosulfan vs busulfan, P = .43).

Figure 3.

Kaplan-Meier estimates of OS based on complete or partial response at HSCT in the HLH-2004 study. Displayed are OS where death was defined as event. Five-year probabilities of survival (pSu) are indicated with a 95% CI. P values from Cox proportional hazards models. Complete response (CR) was defined as no fever, no splenomegaly, ANC ≥1.0 × 109/L, platelets ≥100 × 109/L, fibrinogen >1.5 g/L, and a ferritin cutoff of either <500 µg/L or <2000 µg/L (allowing missing information on one of the these parameters). Partial response (PR) was defined as having 3 to 5 normal parameters. (A) OS for the entire cohort based on CR (n = 42, events n = 10, blue line) or PR (n = 87, events n = 37, red line dashed); P = .035, using the ferritin cutoff of <500 µg/L. (B) OS for the entire cohort based on CR (n = 55, events n = 15, blue line) or PR (n = 76, events n = 31, red line dashed); P = .080, using the ferritin cutoff of <2000 µg/L.