Pediatr Blood Cancer 2008;51:402–404

The Role of the Initial Bone Marrow Aspirate in the Diagnosis ofHemophagocytic Lymphohistiocytosis

Abha Gupta, MD, MSc, FRCP(c),1* Pascal Tyrrell, MSc,2 Rahim Valani, MD,3 Susanne Benseler, MD,2

Sheila Weitzman, MB, BCh, FCP(SA), FRCP(C),1 and Mohamed Abdelhaleem, MD, PhD, FRCP(C)4

INTRODUCTION

HLH is a potentially life-threatening condition occurring as

a familial (inherited) disease (FHLH) or secondary to other diseases

including infection, rheumatologic disease or malignancy [1].

Almost half of patients with familial disease can be confirmed by

mutations in perforin [2], Munc-13 [3] and syntaxin-11 genes [4],

however, the diagnosis of the remaining patients is less clear, and

is based on the fulfillment of clinical diagnostic parameters [5].

Although HLH was initially defined by the cytomorphological

finding of hemophagocytosis in the bone marrow and other tissues

[6], revised diagnostic criteria state that this is not a requirement to

make the diagnosis, recognizing that HPC are often not detected at

initial presentation [5]. Serial BMA may be required to document

HPC. Despite these guidelines, when patients fail to meet required

diagnostic criteria, clinicians often rely on the finding of HPC in the

initial bone marrow aspirate to confirm the diagnosis, especially in

cases of secondary HLH.

There are no standardized reporting guidelines for BM HPC,

leading to heterogeneity in the reporting of biopsies as ‘positive’ or

‘negative’. In an early publication describing HPC in HLH, Favara

et al. suggested that careful examination of at least 3 smears should

reveal at least two HPC per slide to be significant [7]. The purpose of

this study was, therefore, to assess the frequency of HPC at the initial

BM aspirate (BMA) at presentation in patients diagnosed with HLH

and to quantify the number of HPC in order to potentially establish

guidelines for standardized reporting.

METHODS

Appropriate permission from the institutional research ethics

board was obtained. Patients diagnosed with HLH between January

2000 and December 2004 at the Hospital for Sick Children were

included in this study. Patients were diagnosed based on the revised

HLH Diagnostic Criteria [5]. Day of diagnosis of HLH was defined

as the day definitive therapy was initiated.

The slides of the initial BMA performed for each patient were

independently reviewed by each investigator (A.G. and M.A.).

During this review, although aware of the diagnosis of HLH, the

investigators were blinded to the patient’s identity, pathology report,

and clinical situation. A differential count was obtained by counting

500 nucleated cells (40� magnification). The number of histiocytes

and hemophagocytic cells were recorded separately. Adequacy,

cellularity, granulopoesis and erythropoesis were recorded from the

diagnostic report for each BMA. Agreement between investigators

was assessed by calculating the single measure intraclass correlation

(case 3).

RESULTS

Review of BMA

A total of 35 patients were diagnosed with HLH at The Hospital

for Sick Children between January 2000 and December 2004, of

whom 33 had an initial BMA performed at the time of diagnosis.

The reason listed for the BMA was to rule out HLH in 53%,

rule out leukaemia/malignancy in 40% and sepsis in 7%. The

median time from hospital admission to BMA was 3 days (0–44).

The median time from initial BMA to diagnosis of HLH, defined as

the day definitive therapy was initiated, was 0 days (�3 to 11). Three

patients were diagnosed with HLH prior to the BMA.

Five patients had missing slides and five patients had only BM

biopsy but no aspirate performed. Of the 23 aspirates for which

slides were available, 21 were of adequate quality for differential

assessment. Table I lists the description of BMA and the

total number of histiocytes and HPC counted. Twenty to 21/21

(95–100%) patients reviewed had histiocytes present morpho-

logically. The median number of histiocytes counted per

500 nucleated cells was 6 (range 0–31). Twelve to 14 (57–67%)

Background. The identification of hemophagocytosis (HPC) intissue or bone marrow (BM) represents only one of 5/8 criterianeeded for the diagnosis of hemophagocytic lymphohistiocytosis(HLH). Yet, confirmation of HPC in bone marrow aspirates (BMA)is often relied upon to make therapeutic decisions. There is nostandardized reporting criteria for the definition of ‘‘positive’’ BMA,and likely differs between institutions. The purpose of this studywas to quantify the number of HPC in the initial BMA in patientsdiagnosed with HLH at our institution. Procedure. Patient chartswere retrospectively reviewed. Numbers of HPC were counted per500 nucleated cells in initial BMA. Results. Fifty-eight percent of

patients had at least one HPC per 500 nucleated cells. Mediannumber of HPC per 500 cells was 1 (0–12). Median time from initialBMA to HLH diagnosis was 0 days (�3 to 11), suggesting that HLHdiagnosis was made regardless of the results of this initial BMA.Conclusion. The number of HPC at initial BMA is often low andvariable, confirming that a BMA lacking HPC does not rule out thediagnosis of HLH, and a negative initial BMA should not delaytherapy. We recommend that the BMA report should documentnegative as well as any positive findings of HPC. Pediatr BloodCancer 2008;51:402–404. � 2008 Wiley-Liss, Inc.

Key words: bone marrow; hemophagocytosis; HLH

� 2008 Wiley-Liss, Inc.DOI 10.1002/pbc.21564

——————1Division of Hematology/Oncology, Hospital for Sick Children,

Toronto, Ontario, Canada; 2Division of Rheumatology, Hospital for

Sick Children, Toronto, Ontario, Canada; 3Division of Emergency

Medicine, Hospital for Sick Children, Toronto, Ontario, Canada;4Division of Hematopathology, Hospital for Sick Children, Toronto,

Ontario, Canada

*Correspondence to: Abha Gupta, Division of Hematology/Oncology,

Hospital for Sick Children, Toronto, Ontario, Canada M5G 1X8.

E-mail: abha.gupta@sickkids.ca

Received 26 February 2007; Accepted 6 February 2008

patients had positive BMAwith at least 1 HPC. The median number

of HPC counted per 500 nucleated cells was 1 (range 0–12). Two

patients were noted to have HPC by reviewer #2, but not by reviewer

#1. The level of agreement in assessing HPC was obtained between

the two investigators: single measure Intraclass Correlation (ICC3)

0.897 (95% CI 0.946–0.711).

Clinical Features

The clinical records of these 21 patients were reviewed. Twelve

patients were recorded as having a BMA positive for HPC at

the time of clinical presentation as per the hospital chart. All of

these twelve patients had a BMA positive for HPC in the current

review. The number of HPC counted per 500 cells for these twelve

patients ranged from 1 to 11 (average of 2 reviewers). [Table I] Two

BMA were considered negative for HPC as per the hospital record,

on whom 1 and 2 HPC were found by reviewer #2.

Median age of the patients was 9.3 years. Three of 6 (50%)

patients less than 5 years of age had HPC in their initial BMA, and 9/

15 (60%) patients equal to or older than 5 years had positive BMA.

There was no correlation with the number of HPC counted and the

cellularity of the marrow. For example, 5/7 hypocellular BMA

had evidence of HPC, and 5/9 normo or hypercellular BMA had

evidence of HPC. The sensitivity of the initial BM HPC in the

diagnosis of HLH was 60%.

The other clinical diagnostic criteria were recorded in patients

who did and did not have HPC by the current review. Although the

patient numbers are small, there was no difference in the number or

type of other diagnostic criteria fulfilled by patients between those

who did and did not have HPC (data not shown). Patients with BMA

positive for HPC [based on current review] had a range of 1–5 other

diagnostic criteria present (median, 3) compared to patients who did

not have HPC, median 2 (range 0–4). There was no correlation

between the presence or absence of HPC and aetiology of HLH,

initial therapy prescribed, need for ICU admission or death (data not

shown).

DISCUSSION

This report describes a quantitative assessment of HPC in initial

BMA in patients diagnosed with HLH. We have demonstrated that

over half our patients diagnosed with HLH had evidence of at least

one HPC on initial BMA but that the absolute number of HPC

detected was low. The number of HPC did not reflect clinical activity

or the likelihood of other clinical criteria being present.

HPC may be absent upon initial examination and may appear

later on in the disease course, requiring serial BMA examination for

detection [5,8]. In this study, the diagnosis of HLH was made with

the initial BMA in all cases, usually the same day as the BMA was

done, irrespective of the result of the BMA. Clinicians with little

experience with HLH may rely on a positive BMA in order to

commit to definitive therapy, with relative innocuity of an initial

BMA leading to inappropriate hesitation in treatment initiation.

Revised criteria list BM HPC as only one of many possible

diagnostic criteria needed to establish the diagnosis [5], confirming

that therapy should not be delayed in acutely ill patients while

waiting for evidence of HPC to appear in the BMA

HPC may be less prevalent in familial HLH and may reflect

age and symptom duration. Ascertainment of young patients may

be relatively early based on a positive family history, prior to the

development of overt clinical symptomatology [9,10]. In the current

study, no relationship was found between initial BMA HPC

involvement and age of the patient, severity of disease or underlying

aetiology.

Pediatr Blood Cancer DOI 10.1002/pbc

TABLE I. Quantification of Histiocytes and HPC in Initial BMA in Patients With HLH

Patient Age

Histiocytes HPC

Cellularity Granulopoiesis ErythropoiesisR#1 R#2 R#1 R#2

1 9.3 1 3 0 0 Normocellular Increased eosinophilic

precursors, mild dyplasia

Moderate dyspoiesis

2 14.6 5 1 0 0 No particles Left shift Normoblastic

3 9.6 3 2 0 2 Normocellular Normal Normoblastic

4 16.3 8 5 1 1 No particles Not available Not available

5 4.3 2 2 0 1 No particles Normal Normoblastic

6 11.8 5 2 1 1 Mild hypocellular Normal Normoblastic

7 0.7 11 4 3 2 Normocellular Normal Normoblastic

8 3.3 31 35 0 0 Hypercellular Normal Normoblastic

9 7.2 0 5 3 1 Hypocellular Normal Mild dyspoiesis

10 14.6 2 3 0 0 No particles Toxic changes Not available

11 14 7 5 2 2 Normocellular Normal Normoblastic

12 0.6 3 4 4 2 Normocellular Normal Mild dyspoiesis

13 13 5 3 2 2 Mild hypocellular Normal Normoblastic

14 15.2 18 9 0 0 Normocellular Left shift Moderate dyspoiesis

15 10.3 7 5 10 8 Mild hypocellular Left shift Mild dyspoiesis

16 14.9 30 22 12 10 No particles Normal Not available

17 0.3 13 7 1 1 Hypercellular Left shift Normoblastic

18 16.6 7 3 5 2 Mild hypocellular Normal Normoblastic

19 10.3 11 6 5 5 Hypercellular Hyperplasia and left shift Normoblastic

20 1.1 3 3 0 0 Normocellular Normal Normoblastic

21 15.5 6 6 0 0 Hypercellular Normal Normoblastic

Bone Marrow Hemophagocytosis in Hemophagocytic Lymphohistiocytosis 403

The sensitivity of HPC in our study was 60%. The method for

counting HPC utilized in the current study is limited by sampling

error, possible patchy marrow involvement, as well as uneven

distribution of cells on the slide. Furthermore, only aspirates

and not biopsies were assessed for HPC as, in our institution, this is

what is immediately available for the clinician at the time of

diagnosis.

The design of our study did not allow us to determine specificity

of HPC. Risdall et al. found that 29/60 patients with non-HLH

disorders had HPC, but never of a degree similar to patients

with secondary HLH [6]. The lack of specificity of HPC in BM has

been well documented and can also be seen following blood

transfusions and surgery [7,10–13]. Patients with fever and HPC

who were not diagnosed with HLH were not included in our study,

but represent an important population to better understand the

pathophysiology and prevalence of HPC in similar disease states.

A bone marrow examination remains an important part of the

diagnostic work up in patients suspected of HLH. The well

described association of secondary HLH with leukaemia and non-

Hodgkin lymphoma makes it important to rule out the presence of

malignant disease prior to the initiation of steroid-containing

therapy. Although spleen, liver, and lymph nodes are considered to

be the most reliable sources of HPC [6,14], BM remains the most

accessible place for morphological examination.

There lacks a threshold for HPC that defines a positive marrow or

that defines HLH. Nonetheless, based on our results, we propose that

a standardized method of reporting HPC in BMA should be adopted,

and that this nomenclature should reflect a dichotomous variable,

that is marrow involvement with HPC be defined as ‘present’ or

‘absent’. The reporting of BM aspirates using qualitative descriptors

such as sparse, few, or many should be interpreted by the clinician as

present. A qualitative description of sparse HPC, should not deter

the clinician from treating the patient appropriately. Our findings

confirm that HPC on initial BMA should remain as only one of the

possible diagnostic criteria [5], and should not be used in isolation to

rule in or rule out the disease.

In summary, HLH is a disease in which a multi-disciplinary team

is often involved, where both delay to definitive therapy and over-

treatment with etoposide are problematic and despite revised

diagnostic criteria, the presence or absence of HPC in the initial BM

continues to be relied upon for diagnostic confirmation. We have

shown that approximately half of patients diagnosed with HLH may

have HPC upon initial BMA and that the absolute numbers of HPC

are often low. In order to promote standardization in the literature,

HPC should be reported as either present or absent, and qualitative

grading systems such as few and many should be interpreted equally

as present. A specific, reproducible, quick and accessible laboratory

test for HLH is much needed. As the revised diagnostic criteria

imply, while a bone marrow examination is important in the

diagnostic work-up of HLH, especially to rule out malignancy, the

results should not be relied upon exclusively making therapeutic

decisions in this life-threatening condition.

REFERENCES

1. Favara BE, Feller AC, Pauli M, et al. Contemporary classification

of histiocytic disorders. The WHO Committee On Histiocytic/

Reticulum Cell Proliferations. Reclassification Working Group of

the Histiocyte Society. Med Pediatr Oncol 1997;29:157–166.

2. Stepp SE, Dufourcq-Lagelouse R, Le Deist F, et al. Perforin gene

defects in familial hemophagocytic lymphohistiocytosis. Science

1999;286:1957–1959.

3. Feldmann J, Callebaut I, Raposo G, et al. Munc 13-4is essential for

cytolytic granules fusion and is mutated in a form of familial

hemophagocytic lymphohistiocytosis (FHL3). Cell 2003;115:

461–473.

4. zur Stadt U, Schmidt S, Kasper B, et al. Linkage of familial

hemophagocytic lymphohistiocytosis (FHL) type-4 to chromo-

some 6q24 and identification of mutations in syntaxin 11. Hum Mol

Genet 2005;14:827–834.

5. Henter JI, Horne A, Arico M, et al. HLH-2004: Diagnostic and

therapeutic guidelines for hemophagocytic lymphohistiocytosis.

Pediatr Blood Cancer 2007;48:124–131.

6. Risdall RJ, McKenna RW, Nesbit ME, et al. Virus-associated

hemophagocytic syndrome: A benign histiocytic proliferation

distinct from malignant histiocytosis. Cancer 1979;44:993–1002.

7. Favara BE. Hemophagocytic lymphohistiocytosis: A hemophago-

cytic syndrome. Semin Diagn Pathol 1992;9:63–74.

8. Janka GE. Familial hemophagocytic lymphohistiocytosis. Eur

J Pediatr 1983;140:221–230.

9. Hafsteinsdottir S, Jonmundsson GK, Kristinsson JR, et al. Findings

in familial haemophagocytic lymphohistiocytosis prior to sympto-

matic presentation. Acta Paediatr 2002;91:974–977.

10. Goldberg J, Nezelof C. Lymphohistiocytosis: A multi-factorial

syndrome of macrophagic activation clinico-pathological study of

38 cases. Hematol Oncol 1986;4:275–289.

11. Ost A, Nilsson-Ardnor S, Henter JI. Autopsy findings in 27 children

with haemophagocytic lymphohistiocytosis. Histopathology 1998;

32:310–316.

12. Listinsky CM. Common reactive erythrophagocytosis in axillary

lymph nodes. Am J Clin Pathol 1988;90:189–192.

13. Suster S, Hilsenbeck S, Rywlin AM. Reactive histiocytic hyper-

plasia with hemophagocytosis in hematopoietic organs: A re-

evaluation of the benign hemophagocytic proliferations. Hum

Pathol 1988;19:705–712.

14. Jaffe ES, Costa J, Fauci AS, et al. Malignant lymphoma and

erythrophagocytosis simulating malignant histiocytosis. Am J Med

1983;75:741–749.

Pediatr Blood Cancer DOI 10.1002/pbc

404 Gupta et al.