Jan-Inge Henter, MD, PhD, Childhood Cancer Research Unit, Karolinska Hospital Q6:05, SE-171 76 Stockholm, Sweden. E-mail: email@example.com
The familial form of haemophagocytic lymphohistiocytosis (HLH) is a fatal disease, with allogeneic stem cell transplantation (SCT) being the only curative treatment. In contrast, patients with secondary (infection-associated) HLH usually do not require SCT. Since it often is difficult to distinguish primary and secondary HLH, we wanted to identify a tool that provides guidance on whether SCT is required. The clinical outcome of 65 HLH patients was analysed in relation to the recently reported four types of defects in natural killer (NK)-cell cytotoxicity in HLH. None (0%) of the 36 patients with NK-cell deficiency type 3 attained a sustained (1-year) remission after stopping therapy without receiving SCT, in contrast to 45% (13/29) non-type 3 patients (P < 0·001). Most type 3 patients (22/36) underwent SCT (14/22, 64% are alive), whereas 11 of 14 that did not receive SCT died, and the three others had received HLH-therapy during the last year of follow-up. Of 54 patients analysed for perforin expression and/or mutation, the five with perforin deficiency were all type 3 patients. The data suggests that HLH patients with NK-cell deficiency type 3 will probably require SCT to survive. Thus, NK-cell deficiency classification may provide valuable guidance in judging whether an HLH-patient needs SCT.
Haemophagocytic lymphohistiocytosis (HLH) is a life threatening immune disorder that usually presents during infancy or early childhood (Arico et al, 1996, Henter et al, 1998, Janka, 1983). HLH is classified as either familial (primary) haemophagocytic lymphohistiocytosis (FHL), inherited as an autosomal recessive disorder, or secondary HLH, a sporadic syndrome, which is predominantly associated with viral infections, especially Epstein–Barr virus infection (Henter et al, 1998, Janka, 1983). Common characteristics of HLH include fever, hepatosplenomegaly, cytopenia, hypertriglycerideamia, hypofibrinogenaemia and haemophagocytosis in bone marrow, spleen or lymph nodes by activated macrophages (Henter et al, 1998, 1991). The most prominent histopathological feature is an accumulation of activated lymphocytes and non-Langerhans histiocytes with active phagocytosis, mainly of erythrocytes (Henter et al, 1998, Janka, 1983).
Familial HLH is a fatal disease and the median patient survival without treatment is between 1 and 2 months (Henter et al, 1998, Janka, 1983). The application of chemotherapy, immunotherapy and allogeneic stem cell transplantation (SCT), as suggested in the HLH-94 treatment protocol, has greatly improved patient survival (Henter et al, 2002). To date, SCT is considered to be the only curative treatment for familial HLH. On the contrary, most patients with secondary HLH do not need an SCT and may only need a short course of therapy or no treatment at all. However, it is often difficult to distinguish whether a patient has primary or secondary HLH, especially if there is no familial history of the disease and no evidence of gene mutations in the patient. This is a major clinical problem in relation to HLH, as it also affects the decision whether SCT needs to be performed. In an attempt to identify laboratory markers facilitating this therapeutic decision and to uncover the potential clinical significance of grouping NK cell cytotoxicity deficiency types, we analysed these NK cell cytotoxicity deficiency types in relation to certain clinical outcomes in 65 HLH patients.
Materials and methods
A total of 68 patients with HLH diagnosed during the period July 1994 to June 2002 were evaluated for NK cell cytotoxicity deficiency types. Some of the patients in this study were included in a previous report describing these cytotoxicity deficiency subtypes (Schneider et al, 2002). Each of the patients recruited was younger than 15 years of age. Among these 68 patients, three patients were lost to follow-up, leaving 65 patients for the study. The median follow-up time since diagnosis (defined as onset of HLH therapy in the treated patients) was 4·6 years (range 0·6–7·8 years) in the surviving patients. The main clinical and laboratory characteristics of the patients, as reported by the cut-off time for data analysis (July 2003), are listed in Table I. As a result, genetic studies on the Munc13–4 and syntaxin-11 gene were not included in the analysis. The majority of the patients (n = 58) were treated according to the HLH-94 protocol, five received no therapy or only corticosteroids, and two received other treatments. Analysis of NK cell cytotoxicity deficiency subtypes in most of these patients have been reported previously, and the classification of NK cell cytotoxicity deficiency subtype was performed according to the criteria previously published (Schneider et al, 2002). In the present study, central nervous system (CNS) disease was defined as having neurological alterations as well as cerebrospinal fluid (CSF) pleocytosis and/or elevated CSF protein. With regard to their HLH therapy status, the patients were classified as being ‘off-therapy’ if they had been off therapy without disease re-activation for at least 1 year after stopping therapy, and as ‘not off-therapy’ if they had either received an SCT or had been administered HLH therapy during the last follow-up year (Table II). Ethical approval for these studies was provided by the ethical review board at Karolinska Institutet.
Table I. Characteristics of the 65 HLH patients and association with NK cell cytotoxicity deficiency subtype group.
NK sub-type group 3, n (%)
NK sub-type group non-3, n (%)
Total, n (%)
N/A, not applicable.
*The definitions used were in accordance with the diagnostic criteria by the Histiocyte Society.
†Hyperferritinemia defined as >500 μg/l.
‡P-value calculated by Fisher's exact 2-sided test.
Age at registration <6 months
Central nervous system involvement at diagnosis
Fever at diagnosis*
Splenomegaly at diagnosis*
Bicytopenia at diagnosis*
Hypertriglyceridaemia or hypofibrinogenaemia at diagnosis*
Haemophagocytosis at diagnosis
Hyperferritinaemia at diagnosis†
known perforin defect
Table II. Outcome of the 65 HLH patients and association with NK cell cytotoxicity deficiency subtype group.
NK sub-type group 3, n (%)
NK sub-type group non-3, n (%)
Total n (%)
*No clinical signs of disease and no cytopenia.
† Stem cell transplantation.
‡One patient has affected siblings with HLH and is scheduled for SCT, all three have received therapy during the last year of follow-up.
§Remission lasting for more than 1 year after stopping HLH therapy or never started HLH therapy, without having had SCT at the date of the last follow-up.
Dead at 2 months after start of treatment
Alive at 2 months after start of treatment
With active disease
With inactive disease*
Missing information on disease activity
Dead prior to SCT
Alive at last follow-up
SCT not performed
Alive at last follow-up
Alive at last follow-up
Cytotoxicity data were in part included in a previous report describing the four cytotoxicity deficiency subtypes in HLH, in which the standard 51-chromium (Cr) release assay with modifications were described in detail (Schneider et al, 2002). Briefly, non-adherent lymphocytes were generated from peripheral blood of HLH patients. In the assays, lymphokine-activated killer (LAK) cells were generated by culturing peripheral blood mononuclear cells of patients in the presence of high-dose (103 IU/ml) recombinant human interleukin-2 (rhIL-2) for 72 h. Phytohaemagglutinin (PHA) was added to resident non-adherent lymphocytes to detect functional, most probably allo-restricted, cytotoxic T cells. In the 51-Cr release assay, un-stimulated and PHA-activated peripheral blood lymphocytes, as well as LAK cells, were used as effector cells. The human leucocyte antigen-class I and -II negative K562 leukaemic cell line was applied as a sensitive target cell throughout (Schneider et al, 2002).
Classifications of cytotoxicity deficiency type
Definitions of the cytotoxic deficiency types have previously been described in detail (Schneider et al, 2002). Briefly: Type 1: NK cells lacked lytic activity against K562 cells in 4–h 51-Cr release assay; cytolytic function was reconstituted in the presence of PHA but not by the rhIL-2 LAK protocol; lysis at 16 h was normal. Type 2: lymphocytes with and without PHA stimulation in vitro mediated-lysis at 4 h and 16 h showed low values, but LAK cells generated in vitro showed normal lysis rates of K562 cells in 4- and 16-h killing assays. Type 3: cellular cytotoxicity was totally absent, and neither PHA or rhIL-2 stimulation nor prolongation of the incubation time of effector and target cells could restore the deficient cytolytic activity. Type 4: cytolytic activity of the lymphocytes with and without stimulation of PHA and rhIL-2 was low or absent as determined in the 4-h killing assay, but normal in the 16-h assay. As described above, the NK cell cytotoxicity against K562 could be restored in all types except type 3. Hence, for analysing the association of NK cell cytotoxicity deficiency types with clinical outcomes, types 1, 2 and 4 were pooled together and defined as non-type 3 in the present study.
Differences in distribution were compared by using the chi-square test, or where frequencies were small, two-tailed Fisher's exact test. Mann–Whitney U-test was used to compare the difference in the median age at diagnosis between the type 3 and non-type 3 patients The survival rates were analysed using the Kaplan–Meier life table method and univariate comparison of survival using the log rank test. Subsequently, multivariate analysis using logistic regression was performed with disease activity after induction therapy as dependent variable. The covariates used were: sex, consanguinity, age at start of therapy and NK sub-type group. It was not possible to include infection status in the analysis since clinical and, in particular, laboratory data on infections were not available except in a small minority of the patients studied. The Statistical Package for the Social Sciences (SPSS) version 11·5 software (Chicago, IL, USA) was used for all statistical analyses. Differences were considered to be statistically significant where the two-sided P-value was less than 0·05.
NK cell cytotoxicity deficiency subtypes in 65 HLH patients
Peripheral blood non-adherent lymphocytes from 65 patients with HLH showed heterogeneity in the defects in in vitro cytotoxicity against K562 cells, as determined by both the standard 4-h and our modified 51-Cr release assays. According to the characteristics of their defects (Schneider et al, 2002), these patients were classified into four groups, types 1, 2, 3 and 4, with 17, 6, 36 and 6 patients in each group, respectively. Thus, type 3 was predominant among the four types, accounting for 55% of all patients. In the following analyses, types 1, 2 and 4 have been pooled together into one group, non-type 3, with 29 patients. The clinical characteristics of the type 3 and the non-type 3 patients are presented in Table I.
Parental consanguinity was more frequently present in type 3 patients [14/36 (39%)] than in non-type 3 patients [3/28 (11%)] (P = 0·011). Moreover, the frequency of the patients who had an affected sibling was higher in type 3 [7/33 (21%)] than in non-type 3 patients [2/27 (7%)], but this was not statistically significant (Table I). The age at diagnosis in the 65 patients varied from 1 month to 12·8 years. The median age at diagnosis was significantly lower in type 3 patients than in non-type 3 patients (4·8 months vs. 15·3 months, P < 0·05 by Mann–Whitney U-test). In addition, when analysing the distribution of cytotoxic activity deficiency types in patients according to age at the time of diagnosis (with a cut-off age of 6 months), the frequency of the patients who were 6 months old or younger at diagnosis was significantly increased in the type 3 group [21/36 (58%)] compared with the non-type 3 group [7/29 (24%)] (P = 0·006). There were also statistically significantly more female patients in the type 3 group [21/36 (58%)] compared with the non-type 3 group [8/29 (28%)] (P = 0·013) (Table I).
A recent study showed that deficient NK function was more frequently seen among HLH patients with central nervous system (CNS) disease than in patients who were negative for CNS disease (Imashuku et al, 2002). To determine whether there is also a relation between the cytolytic deficiency types and CNS involvement, we compared the frequency of HLH patients with CNS disease in the various NK cell types. The results in the present study did not reveal any association between NK cell cytotoxicity deficiency types and CNS disease (Table I). We also studied another variable with prognostic interest in FHL, i.e. ferritin (Esumi et al, 1989) and found that more type 3 patients had ferritin levels >500 μg/l at diagnosis than non-type 3 patients, but the difference was not statistically significant in our material (Table I).
Therapeutic guidance: HLH patients with type 3 NK cell deficiency are likely to require SCT for prolonged survival
Possibility of getting off therapy without a transplant When grouped as ‘off-therapy’ and ‘not off-therapy’ with regard to HLH therapy status, none of the 36 patients (0%) with type 3 deficiency reached ‘off-therapy’ status, i.e. none of these patients had a 1-year remission after therapy without receiving an SCT, while 45% of the patients with other deficiency types (1, 2 and 4) or non-type 3 achieved ‘off-therapy’ status (Table II). The frequency of ‘not off-therapy’ in patients with type 3 was significantly higher than that for patients with non-type 3 (100% vs. 55%, P < 0·001 by Fisher's exact test). The relationship between the probability of being ‘off-therapy’ and the NK cell cytotoxicity deficiency subtype groups in 65 patients with HLH is shown in Fig 1, as analysed by the Kaplan–Meier estimated method using time after start of therapy to off date or last follow-up date as the endpoint (Fig 1). In the non-type 3 group, the probability of being ‘off-therapy’ 3 years after start of treatment was 50% [95% confidence interval (CI) ± 20%], whereas none of the type 3 patients came off therapy (P < 0·001, Log rank test).
Survival in non-transplanted patients Among the 31 patients that did not undergo an SCT, a statistically significantly higher death rate was observed in type 3 as compared with non-type 3 patients [11/14 (79%) vs. 4/17 (23%), P < 0·005] (Table II). Among the three surviving patients with type 3 deficiency, one is scheduled for an SCT, one has been on therapy for more than 1·5 years and one has been treated intermittently. The relationship between probability of survival and NK cell cytotoxicity deficiency type in HLH patients not having a SCT was analysed with the Kaplan–Meier estimated method, using time after start of therapy to last follow-up date as the endpoint (Fig 2). The 3-year probability of survival in non-transplanted patients was statistically significantly better in non-type 3 patients (75%, 95% CI ± 21%), as compared with type 3 patients (21%, 95% CI ± 21%) (P < 0·001, Log rank test). There were three patients that received no therapy; one with type 3 deficiency (that died), and two with non-type 3 deficiency (both alive).
Prognostic marker: NK cell cytotoxicity deficiency subtypes and clinical outcome
Probability of survival Among the 36 HLH patients with type 3 NK cell cytotoxicity deficiency, 19 died. In contrast, seven of 29 patients with non-type 3 died. The 3-year probability of survival was 46% (95% CI ± 17%) for type 3 patients and 75% (95% CI ± 16%) for non-type 3 patients, respectively (P = 0·012, Log rank test).
Disease activity after 2 months To analyse if there was an association between NK cell cytotoxicity deficiency subtype group and disease activity after 2 months, we performed logistic regression analysis including co-variates with significant difference between the two groups of NK subtypes (Table I). The dependent variable was disease activity after 2 months, i.e. the duration of the initial therapy. Two patients who already had been transplanted during the first 2 months were not included in the analysis because once an SCT had been performed the status of being ‘off-therapy’ no longer existed. Information on the co-variables included was missing in four patients, leaving 59 cases available for multivariate analyses. Patients who died of the disease or who were reported as having active disease were grouped in one group ‘active at 2 months’ (n = 35) whereas patients with non-active disease were in the other group ‘non-active at 2 months’ group (n = 24). The NK cell deficiency subtype 3 was associated with a significantly increased risk of having active disease after 2 months, as indicated by an unadjusted odds ratio of 5·51 (95% CI 1·78–15·04) (Table III). Adjustment for the potential confounding factors altered these odds to 4·80 (95% CI 1·38–16·66, P = 0·013) (Table III).
Table III. Association of NK subtype group and risk of disease activity after 2 months Results of logistic regression analysis with disease activity at two months after diagnosis as the dependent variable, adjusted for potential confounding factors in 59 patients with HLH.
NK sub-type group 3
Age >6 months at diagnosis
Perforin deficiency in relation to NK cell subtypes
Perforin analyses were performed in 54 patients, in 31 by flow cytometry analysing perforin protein expression, in 15 by DNA sequencing analysing perforin mutations, and in eight patients by both these methods (Goransdotter Ericson et al, 2001, Kogawa et al, 2002, zur Stadt et al, 2003). Of these 54 patients, 31 were type 3 patients and 23 were non-type 3 patients. Altogether five patients (9%) with perforin gene mutations and/or lack of perforin expression in CD56+ cells were identified; two by DNA analyses of the perforin gene (Goransdotter Ericson et al, 2001, zur Stadt et al, 2003), two by flow cytometry of perforin expression (Kogawa et al, 2002), and one by both methods (unpublished data). These five patients were all type 3 patients (P = 0·064). Thus, among the 54 patients analysed for perforin deficiency, 26 (48%) were defined as type-3 patients without having evidence of perforin-deficient expression or mutations in the coding exons 2 and 3.
An important finding in the present study is that none of the patients with NK cell cytotoxic deficiency type 3 came ‘off-therapy’ (defined as a continuous remission for more than 1 year after stopping HLH chemo-immunotherapy, without SCT). Of 14 patients with type 3 treated with chemo-immunotherapy alone without receiving an SCT, 11 (79%) have died and the three survivors are still receiving therapy. In contrast, the probability of being ‘off-therapy’ in patients with NK cell cytotoxic deficiency type non 3 was 50% (95% CI 30–70%) 3 years after start of therapy. In addition, patients with type 3 deficiency showed a reduction in overall survival and had more active disease after 2 months of therapy. The results indicate that cytotoxicity deficiency subtyping may be of clinical value and provide evidence that there may be no, or only a limited, chance for type 3 patients to achieve a long-term treatment-free remission if treated by chemotherapy and/or immunotherapy alone, and that these patients also require an SCT to attain long-term survival. The analysis does not provide firm therapeutic guidance for non-type 3 patients.
Clinically, it may be difficult to determine whether a HLH patient requires an SCT when there is no data of familial history and/or no evidence of gene mutations available. Based on our current data, classification of the cytotoxic deficiency type could be of assistance when making such a decision. The majority of the type 3 patients had neither family history nor evidence of perforin mutation but needed an SCT. For routine clinical purposes complete subtyping into the four sub-types does not appear necessary, since in types 1 and 4 the lysis is normal after prolonged incubation. For routine use, testing cells that are un-stimulated or rhIL-2-stimulated by the 51-Cr release assay for the standard 4 h, as well as with a prolonged incubation time of 16 h, may thus be enough for discriminating a patient group that cannot be cured by chemo-immunotherapy alone, i.e. the patients in whom cytotoxic activity cannot be restored by any of these measures.
The molecular mechanism(s) underlying the four different sub-types of NK cell cytotoxic deficiencies are unclear, although different molecular defects in addition to virus infections may explain the observed effects (Russell & Ley, 2002, Schneider et al, 2002). The NK data presented here were obtained using a protocol described in detail previously (Schneider et al, 2002). Type 3 is characterized by the total absence of lymphocyte cytotoxicity that cannot be normalized either by rhIL-2, by PHA stimulation or by prolongation of the incubation time of effector (lymphocytes) and target cells (K562). Lymphocytes analysed were obtained from non-adherent peripheral blood mononuclear cells, which mainly comprise T lymphocytes and NK cells. The addition of mitogen may overcome the deficiency of target cell recognition receptors, as well as co-stimulatory molecules that stabilize immune recognition events of NK cells, but also of CD8 positive cytotoxic cells. NK cells are important cells of the innate immune system and provide a first line of defence against infections, especially viral infections and malignancies (Moretta et al, 2001). Low NK activity in HLH patients has been widely observed (Kogawa et al, 2002, Schneider et al, 2002, Sullivan et al, 1998), and appears to be one immunological marker of the disease. NK cells lyse targets primarily through the perforin-mediated cytotoxicity pathway (Podack, 1995), but may eventually use also granulysin as well as CD95-L for target cell lysis (Russell & Ley, 2002). A role of NK cell involvement in the maintenance of human immune homeostasis via perforin-mediated target lysis has been considered and demonstrated in animal models (Trapani & Smyth, 2002). During differentiation, NK cells, sequentially and differentially, use distinct members of the tumour necrosis factor (TNF) family, including TNF-related apoptosis-inducing ligand (TRAIL) and Fas ligand, or granule exocytosis to mediate target cell death (Zamai et al, 1998). Additionally, the perforin-mediated pathway is likely to be involved in down regulation of T-cell responses during chronic viral infection and autoimmunity (Matloubian et al, 1999, Stepp et al, 2000, Jordan et al, 2004). Taken together, these data suggest that the severe defect in NK cell function might provide a condition that favours the accumulation of activated macrophages and T lymphocytes in HLH (Emminger et al, 2001, Yoshida et al, 2003), which seems to be involved with the pathogenesis of the disease.
Primary HLH is an autosomal recessive disease that typically develops in infancy or early childhood. Secondary HLH may however develop at any age from infancy to adulthood, and is often associated with infections (usually viral) or malignancies (Gagnaire et al, 2000, Henter et al, 1998, 1993, Imashuku et al, 2002). The outcome for patients with secondary HLH is better than that for patients with primary HLH (Henter et al, 2002, Imashuku et al, 2001). Distinguishing primary and secondary HLH is therefore important for judging the prognosis of this disease and, in particular, for making the decision whether to perform an SCT. Our results indicate that the type 3 deficiency patients need an SCT in order to survive. In support of these results were the findings that the type 3 patients were significantly younger than other types at the time of diagnosis and, consistently, increased frequencies of the presence of parental consanguinity were also observed in these patients, highlighting the involvement of genetic factors in type 3 patients.
Three patients with known perforin gene mutations and another two with deficient perforin expression by flow cytometry were all type 3 patients, whereas no patient with perforin mutation or deficient perforin expression was observed in non-type 3 patients. Similarly, Ishii et al (2005) reported that FHL patients with perforin and hMunc13–4 gene mutations also had persistent NK cell activity deficiency, suggesting that most, if not all, type 3 patients have identified and/or unidentified gene defects. Although most patients with type 3 were found to have normal gene expression of granzyme A and B, and perforin, molecules involved in the induction of apoptosis by NK cells (Schneider et al, 2002), the possibility of mutations in other genes regulating (perforin-dependent) apoptosis can not be excluded in HLH. Obviously, gene mutations can be used as indicators for the requirement of SCT, but despite the recent findings of hMunc13–4 and syntaxin-11 mutations in HLH patients (not studied for the purpose of this publication) (Feldmann et al, 2003, zur Stadt et al, 2005), there are still many patients with unknown mutations for whom tests that do not involve gene mutations are valuable. There could be several possible compensatory bypass mechanisms used by non-type 3 patients. Obviously, the mere fact that the NK cell activity is restorable could play important role in explaining the better prognosis of the non-type 3 patients. The set of compensatory mechanisms used by non-type 3 patients are as yet unknown in detail, but it can be speculated that restoration of NK cell activity via upregulation of cytotoxic granule and adhesion molecule expression, as well as enhancement of Fas/FasL pathway-mediated cytolysis may also contribute to the better prognosis of non-type 3 patients.
All our results in this study point to worse outcomes and greater disease activity in patients with NK cell cytotoxic deficiency type 3, as indicated by analysis of the rate of patients ‘off-therapy’, the overall survival and the degree of disease activity at 2 months after start of therapy. The number of patients included in this study does not possess sufficient statistical power to conduct multivariate survival analysis with ‘off-therapy’ as an endpoint. However, we do have the power to adjust for potential confounding factors in a logistic regression of disease activity at 2 months. This reported a robust association between greater disease severity after induction therapy and NK subtype group 3. We used disease activity at 2 months after start of treatment as the dependent variable because it is standardized for time in international clinical studies and almost unaffected by SCT; also, importantly, we know that it was the best independent predictor of adverse outcomes in a larger group of patients (Henter et al, 2002, Horne et al, 2005). The difference in NK subtype group could be one of the explanations to our earlier findings that patients with active disease at 2 months after start of treatment have a lower probability of survival (Horne et al, 2005). Thus, the multivariate analysis also is in support of NK cell deficiency subtype group being an independent factor associated with adverse outcome (Table III).
In summary, this study indicated that subtyping of the NK cell deficiency may serve as an important prognostic marker in HLH. More specifically, in our study not one single HLH patient with type 3 NK cell cytotoxicity deficiency reached remission for ≥1 year after therapy with chemo-immunotherapy alone. Accordingly, performing an SCT was associated with an increased survival rate for patients with type 3 NK cell cytotoxicity deficiency. Thus, subtyping of NK cell deficiency may also provide therapeutic guidance, in that HLH patients with the NK cell cytotoxicity deficiency type 3 are likely to require an SCT in order to survive. We conclude that classification of NK cell cytotoxicity deficiency types may be a valuable tool for clinicians in order to decide whether a patient with HLH needs to undergo an SCT. However, this finding needs confirmation by further studies.
We thank Dr Bengt Fadeel for critical reading of the manuscript. Supported by grants provided by the Children's Cancer Foundation of Sweden; the Cancer and Allergy Foundation of Sweden; the Swedish Research Council (no. B0661); the Tobias Foundation; the Ronald McDonald Foundation; the Märta and Gunnar V. Philipson Foundation; the German Histiozytose Gesellschaft e.V., and the Histiocytosis Association of America.