Leukemia Research Vol. 17, No. 5, pp. 421-427, 1993. 0145-2126/93 $6.00 + .00 Printed in Great Britain. Pergamon Press Ltd

PROPOSAL FOR A CLASSIFICATION OF ACUTE MYELOID LEUKAEMIA BASED ON PLASTIC-EMBEDDED BONE MARROW

BIOPSY SECTIONS

A N W A R U L I S L A M

Division of Hematologic Oncology, Department of Medicine, Roswell Park Cancer Institute, Elm and Carlton Streets, Buffalo, NY 14623, U.S.A.

(Received 28 July 1992. Revision accepted 5 January 1993)

Abstract--Traditional approaches to haematological diagnosis of acute myeloid leukaemia (AML) include microscopic examination of peripheral blood (PB) and bone marrow (BM) cells studied in Romanowsky-stained, dry film smears. The routine evaluation of BM aspirates, however, may be disappointing in accurately assessing marrow activity because of the sampling error inherent in the BM aspiration technique, and occassionally may fail to provide a precise diagnosis because BM aspirates cannot be obtained due to packed or fibrotic marrows. Because of the advantages it offers over the conventional method of diagnosis, we have studied Romanowsky-stained, thin sections of plastic-embedded BM biopsies from 87 newly diagnosed AML patients. Based on our study, the patients were broadly grouped into four categories: (1) those with hypocellular marrow (haemopoietic cellularity less than 50%), hypoplastic AML; (2) those with some degree of marrow fibrosis, AML with marrow fibrosis; (3) those with homogeneous infiltration of marrow with blast cells; and (4) those with inhomogenous infiltration of marrow with blast cells. Our results show that the proposed classification of AML based on the evaluation of plastic-embedded BM biopsy sections may offer benefits over those afforded by conventional morphologic smear techniques. This may provide significant prognostic information and may have a major impact on patient management. It may also provide a meaningful stratification of the patient population when comparing treatment protocols for newly diagnosed AML patients.

Key words: acute myeloid leukemia, bone marrow biopsy, plastic embedding.

INTRODUCTION

AT PRESENT, the diagnosis and classification of acute myeloid leukaemia (AML) are typically based on cytological and cytochemical studies of peripheral blood (PB) and aspirated bone marrow (BM) cells. The histological examination of a BM biopsy is usually performed only in cases with a dry tap aspir- ation or in those cases with inadequate cell numbers on the smears. It has long been recognized that BM biopsies can provide significant additional and useful information, but until improvements were made in the design of marrow biopsy needles, untraumatized tissue was difficult to obtain. In addition, the fixatives and vigorous methods of decalcification often used generated unacceptable tissue artifacts [1]. The classical haematoxylin and eosin (H&E) stain used in routine pathology is also not satisfactory for precise characterization of the various haematopoietic cells.

Correspondence to: Anwarul Islam, M.D., Ph.D., MRCPath, Division of Hematologic Oncology, Depart- ment of Medicine, Roswell Park Cancer Institute, Elm and Carlton Streets, Buffalo, NY 14623, U.S.A.

421

The introduction of newly designed BM biopsy needles [2, 3] and plastic embedding techniques [4-8] has improved the diagnostic usefulness of BM tre- phine biopsies by offering uncrushed, undistorted thin-sectioned marrow samples. Avoiding decal- cification has further enhanced their quality. Unlike H&E-stained, thick paraffin sections, thin plastic, Romanowsky-stained sections provide excellent cytomorphological detail of haemopoietic tissue. Thus, Romanowsky-stained thin, plastic-embedded sections are the choice for recognizing haemato- poietic cellular and architectural detail as it exists in vivo. This study was undertaken to use the improved methods of BM biopsy processing, embedding and staining in a population of newly diagnosed A M L patients, to explore the possibility of improving the present method of classification and prognostic eval- uation of these patients.

MATERIALS AND METHODS

Eighty-seven (87) patients with AML were investigated at the time of diagnosis. The diagnoses were based on

422 A. ISLAM

TABLE 1. DISTRIBUTION OF A M L CASES ACCORDING TO F A B CLASSIFICATION

FAB M1 M2 M3 M4 M5 M6

No. of cases 11 31 1 28 13 3

n = 87.

clinical, morphological and cytochemical criteria, accord- ing to the FAB classification [9].

Air-dried smears of peripheral blood and BM aspirates were fixed in methanol and stained with May-Griinwald and Giemsa (MGG) stain. The following cytochemical reactions were also applied: periodic acid Schiff, Sudan black, myeloperoxidase, chloroacetate esterase, non- specific esterase and acid phosphatase. A single BM biopsy specimen was taken from the posterior iliac crest of each patient using a manual trephine [2, 3], and was then pro- cessed into plastic (methyl-methacrylate), according to the methods previously described [5]. Sections were cut at a thickness of 2 #M and stained with MGG stain for cyto- logical details and by the Gomori reticulin stain.

The haemopoietic cellularity of each BM section was assessed visually under low-power microscopy and expressed as a percentage of total marrow volume. The stromal fibre content was also quantitated and graded, according to the scoring system previously described [10]. The homogeneity of leukaemic infiltrate in the marrow was also evaluated in each section. A specimen was considered homogeneously infiltrated with blast cells when it revealed a predominant population of blast cells and the other cellular components (mature or maturing erythroid and granulocytic cells, and megakaryocytes, lymphocytes, plasma cells, macrophages, and eosinophilic and basophilic granulocytes, either singly or in combination) were either absent or represented less than 10% of the total hae- mopoietic cell population. Conversely, a specimen was considered to be inhomogenously infiltrated with blast cells when it demonstrated the presence of greater than 10% non-blastic haemocellular components.

RESULTS

Table I shows the distribution of the 87 A M L cases classified according to the FAB criteria. Table 2 presents the distribution of these cases when classi- fied according to their characteristics as observed in plastic-embedded, Romanowsky-stained thin sec- tions, irrespective of their film smear morphology (Fig. l(a)--(d)). Three cases were classified as

hypoplastic AML. Their (overall) cellularity ranged from 15% to less than 50%. Fatty hypocellular mar- row with only small loci of leukaemic cells (Fig. 2(a) and (b)) was seen in two of these three cases. A homogeneous blastic infiltration of the marrow (Fig. 3) was seen in 57 cases, whereas an inhomogeneous mixture of leukaemic blasts and other haemic cells (per the described criteria) was seen in 30 patients (Fig. 4). Abnormal morphologic features charac- teristic of myeloid dysplasia were noted in all cases where mature or maturing erythroid, granulocytic or megakaryocytic precursors were present, irrespective of their number (small or large) or distribution. The connective tissue fibre content (Fig. 5) at the time of initial diagnosis of A ML in 87 cases was as follows; grade 0 in 77 (89%), grade 1 in 4 (4%), grade 2 in 1 (1%) and grade 3 in 5 cases (6%), respectively (Table 2).

DISCUSSION

There has been considerable improvement in the diagnosis and classification of A M L within the past decade. The older and widely accepted simple mor- phologic diagnostic criteria are being replaced by newer, improved methods of classification using vari- ous enzymatic and immunological markers as well as, in some cases, supplemental cytogenetic studies [11]. These more recent approaches have allowed more precise definition and identification of the dif- ferent A M L subgroups, which, in turn, has fostered conceptual harmony among the different invest- igative groups. However , there is an undesirable tendency in this era of sophisticated testing, cyto- chemistry, novel monoclonal antibody, chromosomal markers and D N A analysis, to overlook the fact that these tests are typically applied to aspirated cells, while the malignant process also involves other mar- row cells not sampled and categorized in aspirates. Conversely, the BM biopsy, which provides a more representative, integrated, intact profile of the cellu- lar and architectural characteristics of the BM has yet to emerge as a routine investigative procedure for the diagnosis and classification of AML. A number of features of potential prognostic significance such as overall marrow cellularity, extent of blast cell infil-

TABLE 2. DISTRIBUTION OF AML CASES ACCORDING TO THE PROPOSED CLASSIFICATION

Cellularity Homogeneity Myelofibrosis Grades

<50% >50% Homogeneous Inhomogenous 0 1 2 3

No. of cases 3 84 30 57 77 4 1 5

n =87.

Proposal for a classification of acute myeloid leukaemia 423

o b

C d FIG. 1. Schematic representation of the four types of AML as observed in Romanowsky-stained, thin plastic-embed- ded bone marrow sections: (a) hypoplastic AML, (b) AML with homogeneous marrow infiltration by blast cells, (c) AML with inhomogeneous marrow infiltration by blast cells (M -- megakaryocyte, E = erythroid precursors, G = granuloid precursors, PC = plasma cells, L = lymphocytes, and Eo = eosinophilic granulocytes; tissue mast cells and macrophages, either singly or in combination can also be

seen), (d) AML with marrow fibrosis.

tration, homogeneity of leukaemic infiltrate, mitotic activity, and marrow fibrosis are all best appreciated in sectioned preparations.

In our previous studies [12, 13], we have shown that AML is not a homogeneous disease. Although the FAB group has identified seven (FAB M1-M7) different morphologic varieties of AML by exam- ination of dry film smears from the aspirated marrows; concurrent studies of bone marrow sections of these diseases reveals that there is a considerable overlap among the different categories of this disease [12, 13]. We have noted, particularly in plastic sections, that the architectural and cytologic pres- entation in AML is often unequal and dipersed. Even in similar cases, there is some notable morphological and structural variation. It has become apparent, therefore, that if one were to categorize each and every difference it would be impossible to develop a meaningful classification. Therefore, to configure a logical, feasible and, perhaps, prognostically useful classification, we divided our newly diagnosed patients with AML into four categories according to morphologic appearance in thin, plastic-embedded BM biopsy section: (1) hypoplastic AML, (2) AML with marrow fibrosis, (3) AML with inhomogeneous, and (4) AML with homogeneous blastic infiltration

of the marrow. We propose that these criteria are simple, easily recognised and may prove to be useful in patient diagnosis, stratification, treatment and management.

Most pretreatment marrows in this series were very cellular with almost complete replacement of fat. Nevertheless, 3 of the 87 patients had marrow haemopoietic cellularity less than 50% (range 15 to <50%), and were classified as hypoplastic AML [14, 15].

Hypocellularity at presentation may predict a less fulminant course of this disease [14], and an accurate recognition of this status may be important. It has been proposed that intensive antileukaemic therapy should be withheld in such cases [14, 15], or that these patients may benefit from low-dose treatment with cytosine arabinoside [16, 17]. On the other hand, it has also been reported that such patients may benefit from aggressive chemotherapy in terms of improved BM function and survival [18]. Patchy marrow hypoplasia with islands of blast cells are the usual findings. These patients may be misdiagnosed either as typical AML if marrow aspirates yield blast cells only or as aplasia if the increase in blast cells is not noted due to sampling error in the aspiration technique.

424 A. ISLAM

FIG. 2. (a) Bone marrow biopsy section from a case of hypoplastic AML showing an area of fatty hypocellular marrow. (b) Another area from the same section as in (a), showing a focus of leukaemic cells. Methacrylate, MGG

stain.

The BM in A M L is usually dominated by massive infiltration of blast cells and the diagnosis is generally made from an examination of peripheral blood and BM aspirate dry film smears. With the exception of major diagnostic centres, a BM biopsy is not necessarily routinely performed in the diagnosis of

these cases. As a result, the evidence of marrow fibrosis in A M L is not adequately documented, the effect of chemotherapy on the marrow fibre content remains unclear, and the concept that marrow fibrosis in AML carries a poor prognosis still remains to be vigorously tested. This study shows that some degree

Proposal for a classification of acute myeloid leukaemia 425

FIO. 3. Bone marrow biopsy section from a case of A M L showing homogeneous infiltration of the marrow with blast cells. Note the presence of sheets of blast cells and absence

of 'other cellular elements' .

FIG. 4. Bone marrow biopsy section from a case of AML showing inhomogeneous infiltration with blast cells. Note the presence of leukaemic blast cells intermixed with 'other

cellular elements' .

426 A. ISLAM

FIG. 5. Bone marrow biopsy section from a case of AML showing considerable increase in the marrow fibre content.

Methacrylate, Gomoris stain.

of marrow fibrosis (other than cases with mega- karvoblastic leukaemia) is not uncommon in AML patients at presentation. Thus, the association of medullary fibrosis with the various types of FAB- categorized AML merits further investigation. Although the finding of increased marrow reticulin in AML patients has been typically regarded as her- aiding a poor prognosis, its occurrence may not necessarily bias the overall outcome. It is possible that appropriate treatment selection may result in resolution of the increased fibrosis and, thus, assist in the repopulation of the bone marrow with normal regenerating haematopoietic cells [10]. Studies relat- ing to the incidence of fibrosis seen in plastic-embed- ded BM biopsy sections and its correlation with FAB criteria and AML therapeusis are strongly indicated.

In this study, AML patients, excluding those with hypoplastic or fibrotic marrows which constituted only a minor population of the total AML patients, could be broadly divided into two main groups: one showing homogeneous (Fig. 3) and the other showing inhomogeneous (Fig. 4) infiltration of the marrow by blast ceils. We believe that the patients presenting with a homogeneous blastic marrow represent cases with true de n o v o AML and that this group, perhaps, responds to conventional chemotherapy better than those with inhomogeneus marrow infiltration.

We also suggest that the AML patients who dem- onstrate inhomogeneous blastic infiltration of the

marrow may be a distinct category of AML. It is possible that they may represent AML cases which are superimposed upon unrecognized pre-existing myelodysplastic disease, These patients are not adequately recognized by the conventional diagnostic methods. They seem to respond poorly to con- ventional chemotherapy and may benefit from specifically modified therapeutic regimens.

Although the FAB classification of AML, which is based on morphologic and cytochemical criteria is widely accepted, its application has not proven to be of significant value in establishing the prognosis [19, 20] or assisting in the choice of therapy in myeloid leukaemia. The failure to detect any relationship between the morphologic classification and therapeutic response maY.indicate not that this classification is invalid or peripheral but rather that it in itself is incomplete and perhaps should be con- jugated with information derived from the analysis of conventional or plastic-embedded bone marrow biopsy sections.

In most tumours, including malignant lymphomas, the major classifications are based on histological sections of biopsied tissue and are essential for diag- nosis. In human leukaemia, however, not enough attention has been given to bone marrow as a tissue, even though it is the haemopoietic organ which is primarily affected. Thorough investigation and appli- cation of the criteria outlined earlier will perhaps

Proposal for a classification of acute myeloid leukaemia 427

allow the different histologic pat terns of A M L to be recognized and their clinical and prognostic sig- nificance assessed.

Acknowledgement--Part of this work was carried out while the author was working at the Hammersmith and Royal Free Hospital in London, U.K. The author is also grateful to Mr Kevin Craig for editorial assistance.

R E F E R E N C E S

1. Wolf-Peeters C. (1991) Bone marrow trephine inter- pretation: diagnostic utility and potential pitfalls. HIS- topathology 18, 489--493.

2. Jamshidi K., Windschitl H. E. & Swaim W. R. (1971) A new biopsy needle for bone marrow. Scand. J. Haemat. 8, 69-72.

3. Islam A. (1982) A new bone marrow biopsy needle with core securing device. J. clin. Path. 35, 359-364.

4. Burkhardt R. (1971) Bone Marrow and Bone Tissue: Colour Atlas o f Clinical Histopathology. Springer, Berlin.

5. Islam A. & Frisch B. (1985) Plastic embedding in routine histology I: preparation of semi-thin sections of undecalcified marrow cores. Histopathology 9,1263- 1274.

6. Beckstead J. H. & Bainton D. (1980) Enzyme histo- chemistry on bone marrow biopsies: reactions useful in the differential diagnosis of leukemia and lymphoma applied to 2-micron plastic sections. Blood 55, 386- 394.

7. Islam A. & Henderson E. S. (1987) Glycol meth- acrylate embedding for light microscopy I: enzyme histochemistry on semithin sections of undecalcified marrow cores. J. clin. Path. 40, 1194-1200.

8. Islam A., Archimbaud, E., Henderson E. S. & Han T. (1990) Glycol methacrylate (GMA) embedding for light microscopy II: immunohistochemical analysis of semithin sections of undecalcified marrow cores. J. clin. Path. 41, 892-896.

9. Bennett J. M., Catovsky D., Daniel M. T., et al.

(1976) Proposal for the classification of the acute leuke- mias. French-American-British (FAB) Co-operative Group. Br. J. Haemat. 33, 451-458.

10. Islam A., Catovsky D., Goldman J. M. & Galton D. A. G. (1984) Bone marrow fibre content in acute myeloid leukaemia before and after treatment. J. clin. Path. 37, 1259-1263.

11. Second MIC Co-operative Study Group (1988) Mor- phologic, immunologic, and cytogenetic (MIC) work- ing classification of acute myeloid leukaemia. Cancer Genet. Cytogenet. 30, 1-15.

12. Islam A., Frisch B. & Henderson E. S. (1989) Plastic- embedded core biopsy: a complementary approach to bone marrow aspiration for diagnosing acute myeloid leukaemia. J. clin. Path. 42, 300-306.

13. Islam A., Frisch B. & Henderson E. S. (1990) Multiple- cell-lineage involvement in acute myeloid leukaemia: a morphologic study using plastic embedded bone mar- row biopsy sections. Haemat. Reo. 4, 133-140.

14. Needleman S. W., Burns P., Dick F. R. & Armitage J. O. (1981) Hypoplastic acute leukaemia. Cancer 48, 1410-1414.

15. Beard M. E. J., Bateman C. J. T., Crowther D. C., et al. (1975) Hypoplastic acute myelogenous leukaemia. Br. J. Haemat. 31, 167-176.

16. Mufti G. J., Oscier D. G., Hamilton T. J. & Bell A. J. (1983) Low dose of cytarbine in the treatment of myelodysplastic syndrome and acute myeloid leuke- mia. New Engl. J. Med. 3tl9, 1653-1654.

17. Manoharan A. (1983) Low dose cytarbine therapy in hypoplastic acute leukemia. New Engl. J. Med. 309, 1652-1654.

18. Howe R. B., Bloomfield C. D. & Mckenna R. W. (1982) Hypocellular acute leukemia. Am. J. Med. 72, 391-395.

19. Foon K. A., Naiem F., Yale C. & Gale R. P. (1979) Acute myelogenous leukemia: morphologic classifi- cation and response to therapy. Leukemia Res. 3, 171- 173.

20. Martelsmann R., Thaler H. T., et al. (1980) Morpho- logical classification, response to therapy, and survival in 263 adult patients with acute nonlymphocytic leu- kemia. Blood 56, 773-781.