Primary hemophagocytic lymphohistiocytosis in adults: the utility of family surveys in a single-center study from China

Background This study investigated the clinical characteristics of primary hemophagocytic lymphohistiocytosis (HLH) in adults, including immunological markers, pedigree findings, and conditions of allogeneic hematopoietic stem cell transplantation (Allo-HSCT). Methods The study included clinical data of 18 adult patients with primary HLH treated in our center from June 2010 to January 2017. Results Of these 18 cases, pathogenic variants were found in the following genes: PRF1 (n = 11), UNC13D (n = 5), SH2D1A (n = 2), RAB27a (n = 1), and LYST (n = 2). One patient had pathogenic variants in both PRF1 and UNC13D genes, one patient had pathogenic variants in both LYST and UNC13D genes and another patient had pathogenic variants in both PRF1 and SH2D1A genes. Additionally, 3 of the 18 cases involved homozygous pathogenic variants, while 2 cases involved hemizygous pathogenic variants. The remaining 13 cases involved compound heterozygous pathogenic variants. The natural killer (NK) cell activity test was conducted in all 18 cases where 14(77.8%)patients showed reduction in NK cell activity. Furthermore, this article presents 3 representative results of the pedigree findings from 12 patients who underwent family surveys. The 8 patients who underwent Allo-HSCT had a median survival of 27.2 months, as compared with the median survival of 7 months for the10 patients who did not undergo Allo-HSCT, a significant difference between the two groups of patients (p = 0.006). Conclusion PRF1 was one of the most commonly mutated gene in adult patients with primary HLH. Family surveys and immunological markers were important for the HLH diagnosis and the selection of an appropriate donor. Allo-HSCT was an effective therapy for adult primary HLH.


Background
Primary HLH is an autosomal and/or X-linked recessive inheritance characterized by immune disorders. In the past, diagnosis of patients with primary HLH was based on a disease onset at an early age and a positive family history. In 1999, a research group in Paris reported the first primary HLH-related gene map [1]. The "HLH-2004: diagnostic criteria for hemophagocytic lymphohistiocytosis" proposed by the Histiocyte Society clearly reported that genetic defects are the gold standard of primary HLH. At least 12 relevant genes associated with primary HLH have been reported, include: PRF1, UNC13D, STX11, STXBP2, SH2D1A, BIRC4, RAB27A, LYST, ADTB3A, ITK, CD27 and MAGT1 [2]. With the gradual understanding of adult primary HLH, age is no longer the basis for the diagnosis. Currently, delayed onset of adult primary HLH has been considered to relate to the gene pathogenic variants, the pathogenic variant patterns, and the presence of a trigger.
Immunological markers are important for the early diagnosis of primary HLH. Janka et al. [3] showed that NK cell activity was reduced in almost all patients at the early stage of primary HLH. Thus, timely detection of NK cell activity is important for the early diagnosis of the disease. CD107a detection can be used as a highly sensitive method for the identification of primary HLH [4]. Timely detection of the immunological markers has been considered to be significantly better than genetic testing and has been used as an effective and rapid screening for primary HLH in the diagnostic processes of international studies [4,5].
Early diagnosis of primary HLH may require a positive family history gained from family surveys of individuals with abnormal immunological markers or suspected primary HLH as a supportive basis to help confirm the diagnosis. Carrier testing of family members through gene sequencing, cytotoxicity analysis and protein measurements can result not only in the identification of yet healthy homozygous carriers but also aid the selection of suitable donors in the family.
The long-term strategy of HLH treatment is to correct the immunodeficiency, so patients with primary HLH require Allo-HSCT to correct the potential genetic defects. Allo-HSCT is recommended by HLH-94 for FHL and refractory recurrent HLH patients, using the traditional myeloablative pretreatment regimen [6]. Combined chemotherapy only temporarily controls the incidence of FHL effectively. Patients who do not receive Allo-HSCT cannot achieve long-term survival [7]. Henter et al. [8] reported the first multicenter prospective study in 2002 and showed that remission occurred in 119 primary HLH patients after combined chemotherapy and subsequent HSCT. Clear recovery was observed with an overall survival rate of 55% in the 43-year follow up where the majority of deaths occurred after the early diagnosis or before the transplantation. More and more clinical studies have shown that Allo-HSCT is the only curative regimen for adult primary HLH. Currently, publications regarding adult primary HLH diagnosis and treatment are few and most of them are case reports. We conducted a related study to identify the clinical characteristics in the adult patients with primary HLH.

Research subject
Eighteen adult patients with primary HLH, diagnosed and treated in Beijing Friendship Hospital, Capital Medical University from June 2010 to January 2017, were included in this study. All patients (≥ 18 years old) underwent genetic testing and were found to have primary HLH-related genetic defects. (Supplement: Zhang K et al. considered that "In patients with combination defects involving 2 genes in the degranulation pathway, CD107a degranulation was decreased, comparable to patients with biallelic mutations in one of the genes in the degranulation pathway. This suggests a potential digenic mode of inheritance of FHL as a result of a synergistic function effect within genes involved in cytotoxic lymphocyte degranulation." [9] and Sepulveda et al. considered that an additive effect of HLH mutations in different genes has been shown in mouse models [10]. So the patient who has defects involving 2 genes in the degranulation pathway and CD107a degranulation decreased (P14) and the patient who has polygenic inheritance additive effect(P15) were diagnosed with presumably primary HLH in our center.

Immunological markers and family surveys of adult HLH patients
Immunological markers of the 18 adult patients with primary HLH were assessed using a cytotoxicity assay, CD107a degranulation assay, perforin, granzyme, signaling lymphocytic activation molecule (SLAM)-associated protein(SAP), and x-linked inhibitor of apoptosis protein (XIAP) detections. Gene sequencing of family members of the patients were performed. Cytotoxicity and involved protein expression of some family members were detected.

Gene sequencing
Using specific-primer design and PCR on DNA extracted from mononuclear cells, the exon and related cleavage products of HLH-related genes were obtained. This was followed by bi-directional Sanger sequencing.
The pathogenic variants and types were identified by Esembl genomic databases. To see if a pathogenic variant has been previously reported before, some information were identified by the Exome Aggregation Consortium(ExAC)and Human Gene Mutation Database(HGMD). The pathogenicity of single nucleotide polymorphisms was predicted by Polyphen2 and SIFT databases. Polyphen2 > 0.95 or SIFT < 0.05 predicted that the amino acid change caused by pathogenic variant may affect protein expression or function.

HSCT
Pre-transplant treatment was conducted based on an HLH-94/04 regimen (i.e., etoposide, glucocorticoid, and cyclosporine) and a DEP regimen (i.e., methylprednisolone, liposomal adriamycin, and etoposide). For donor selection of the HSCT, complete matching and gene screening of patients' parents and siblings were conducted, followed by complete family surveys. NK cell activity and CD107a detections, as well as primary hemophagocytic gene screening, were performed in all donors to exclude the possibility of primary hemophagocytosis.

Statistical methods
SPSS version 22.0 was used for statistical analysis. All the data that conformed to the normal distribution was represented by x±SD and paired t test was used, while non-normal-distributed data was presented using the median and extreme values, Wilcoxon's rank-sum test was used. Statistical difference was defined as P < 0.05, significant statistical difference was defined as P < 0.01.

General conditions of adult patients with primary HLH
Of the 18 adult HLH patients, there were 10 males(55.6%)and 8 females(44.4%), with the median age of onset at 25.5 years (18,54). Of these 18 cases, pathogenic variants were found in the following genes: PRF1 (n = 11), UNC13D (n = 5), SH2D1A (n = 2), RAB27a (n = 1), and LYST (n = 2). One patient had pathogenic variants in both PRF1 and UNC13D genes, one patient had pathogenic variants in both LYST and UNC13D genes and another patient had pathogenic variants in both PRF1 and SH2D1A genes. Three of the 18 cases involved homozygous pathogenic variants, while 2 cases involved hemizygous pathogenic variants. The remaining 13 cases involved compound heterozygous pathogenic variants. Pathogenic variants patterns included missense pathogenic variant, nonsense pathogenic variant, and frameshift pathogenic variant. Adult primary HLH gene pathogenic variants are listed in Table 1.
Immunological marker and family surveys of adult patients with primary HLH Immunological marker of adult patients with primary HLH Detection of immunological markers, including cytotoxicity, CD107a, perforin, granzyme, SAP, and XIAP were performed in the 18 primary HLH patients ( Table 2). The NK cell activity test was conducted in all 18 cases where 14(77.8%)patients showed reduction in NK cell activity. The 8 patients who completed the CD107a degranulation assay, P08, P09, and P10 showed normal CD107a expression whereas P11, P12, P13, P14, and P17 showed reduced CD107a expression (In 6 patients with UNC13D or RAB27a or LYST pathogenic variants who also completed the CD107a degranulation assay, 5 patients have decreased CD107a expression, P10 showed normal CD107a expression) Among the 8 patients who completed the perforin assay, P03, P06, P08, and P09 had reduction of perforin levels, while P10, P11, P13, and P17 had normal perforin levels. (In 5 patients with PRF1 pathogenic variants who also completed the perforin assay, 4 patients have decreased perforin expression) No significant abnormality was noted among the 3 patients who completed the SAP and XIAP protein assays, and their gene detections showed no involvement of SH2D1A and XIAP gene.

Family surveys of adult patients with primary HLH
Among the 12 patients who underwent family surveys, some of their relatives completed gene sequencing and detection of immunological markers. This study was selected to further interpret the representative pedigree findings of 3 cases, including P11 (with UNC13D homozygous missense pathogenic variant), P03 (with PRF1 compound heterozygous pathogenic variant), and P16 (with SH2D1A hemizygous missense pathogenic variant).
Investigation of families of individuals with homozygous pathogenic variants Investigation of the family of P11 showed that: proband, male, 52 years old. UNC13D had a homozygous missense pathogenic variant at c.2588G > A. Family surveys of P11 were confirmed by Sanger sequencing, which showed that the patient's father and offspring also had heterozygous pathogenic variants. Patient's mother did not complete the family survey. However, given the homozygous pathogenic variant of P11, patient's parents should have heterozygous pathogenic variant at the same locus. In addition, P11's sibling also had the same homozygous missense pathogenic variant but did not suffer from HLH. Cytotoxic degranulation (CD107a expression) of P11 and his sibling was reduced (Figs. 1 and 2).
Investigation of families of individuals with compound heterozygous pathogenic variants Investigation of the family of P03 showed that: proband, male, 18 years old. PRF1 had compound heterozygous pathogenic variants, including a heterozygous missense pathogenic variant at c.172 T > C and a non-frameshift pathogenic variant at c.1083_1094del. Family surveys of P03 showed that P03 had suspected a positive family history of HLH, as two elder siblings of P03 died from unexplained fevers in their childhoods. Through Sanger sequencing, the two pathogenic variants of P03 originated from P03's parents. However, P03 also had two healthy siblings who carried different pathogenic variants (Figs. 3 and 4).
Investigation of families of individuals with Hemizygous pathogenic variants Investigation of the family of P16 showed that: proband, male, 32 years old. SH2D1A had a hemizygous pathogenic variant at c.32 T > G, and the family surveys of P16 showed that P16's maternal grandfather died at 30+ years old with an unknown cause of death. P16's nephew (the offspring of P16's elder sibling,1-year-old) simultaneously had the same onset with similar symptoms as P16 and died within a week after hospital admission. Through Sanger sequencing, the pathogenic variant of P16 originated from P16's mother, and three siblings of P16 also carried the same pathogenic variant,P16's mother and three siblings did     Table 3 shows the general information of the patients with follow-ups until May 2017. Among the 8 patients who received Allo-HSCT, one HLH patient died while the remaining seven HLH patients survived. Of the 10 patients who did not receive Allo-HSCT, 7 patients died, 1 was lost to follow-up, the other 2 patients survived. The 8 patients who underwent Allo-HSCT had a median survival of 27.2 months, as compared with the median survival of 7 months for the10 patients who did not undergo Allo-HSCT, a significant difference between the two groups of patients (p = 0.006).

Discussion
Primary HLH is a rapidly progressive and lifethreatening disease characterized by a reduction of NK cell and cytotoxic T lymphocyte function caused by genetic defects, resulting in excessive immune activation. Patients with an onset before 2 years of age account for more than 90% of all patients [3]. At least 12 relevant genes associated with primary HLH have been reported, changes in gene sequences were reported in adult HLH patients in the subsequent studies. Delayed onset of adult primary HLH has been considered to relate to the gene pathogenic variants, the pathogenic variant patterns, and the presence of a trigger. For example, a study by Ueda et al. [11] analyzed the gene pathogenic variants in pediatric HLH patients and showed that nonsense and frameshift pathogenic variants that occurred in infants mostly happened in the classical onset of HLH while missense pathogenic variants occurred in the later onset of the disease at older ages. Pagel et al. [12] found that patients with STXBP2 splice-site pathogenic variants, as compared with patients carrying nonsense pathogenic variants, had a higher age of onset with a median age of 4.1 years and 2 months, respectively. The age of onset of compound heterozygous pathogenic variants involved only the perforin or the degranulation pathway was later than the age of onset of the homozygous pathogenic variants [9]. The presence of a trigger also plays an important role in the incidence of primary HLH. For example, A91V is a milder class I perforin mutation with partial maturation and reduced but  detectable perforin and NK function [13]. A91V confers genetic susceptibility for the development of FHL, but is not enough to trigger the disease on its own, the M. tuberculosis infection as synergistic factors play a role in the development of FHL [14]. In our data, in the family surveys of P11, the sibling of the proband P11 had the same homozygous pathogenic variant as P11. In addition, the cytotoxic degranulation (CD107a expression) was reduced in P11's sibling but with no incidence of HLH. The plausible reason was that P11 had positive EBV-DNA(8.2 × 10 5 copies/ml)while P11's sibling had negative EBV-DNA. In the family surveys of P16, the proband had positive EBV-DNA(8.8 × 10 4 copies/ml)detected when he suffered from HLH in 32 years old, P16's maternal grandfather and nephew died with an unknown cause, we speculate that the possibility of the occurrence of HLH can not be excluded. P16's mother and three siblings had no evidence of EBV infection or other triggering factor(s), although they carried the same pathogenic variant, they did not suffer from HLH. This would suggest that in addition to inherent immune deficiency, other triggering factor(s) might be involved in the incidence of primary HLH. Age was no longer the diagnostic basis of primary HLH. Our data also showed that approximately 72% of the 18 adult patients with primary HLH had compound heterozygous pathogenic variants. Some of the pathogenic variants(such as c.1349C > T)were reported before in other patients with late onset