- Email: [email protected]
Morphometric analysis of chronic b-cell leukemias—An aid to the classification of lymphoid cell types
Leukemia Research Vol. 13, No. 5, pp. 357-365, 1989.
0145-2126/89 $3.00 + .00 Pergamon Press plc
Printed in Great Britain.
M O R P H O M E T R I C ANALYSIS OF C H R O N I C B-CELL L E U K E M I A S - - - A N A I D TO THE CLASSIFICATION OF L Y M P H O I D CELL TYPES DANIELLE ROBINSON, PETER LACKIE, VICTOR ABER and DANIEL CATOVSKY MRC Leukaemia Unit, Histochemistry Department and Computer Centre, Royal Postgraduate Medical School, London, U.K.
(Received 2 November 1988. Accepted 28 November 1988) Abstract--In order to evaluate the ultrastructural features which may be of relevance in distinguishing cells from the various chronic B-cell leukemias, a morphometric analysis was performed on a large number of cells from each disease group. The parameters selected were: cell size, nucleo: cytoplasmic ratio, chromatin condensation, size of the nucleolus and degree of irregularity of both the nucleus and the cytoplasmic outlines. The mean values obtained for each parameter for each disease group were compared statistically. In disorders in which the cells have villous cytoplasmic projections, the quantitative analysis of the cellular features was helpful to characterise the different types of B cells involved. Thus, cells from cases of splenic lymphoma were found to be different from those of hairy cell leukemia, and a variant form of HCL was also identified by its distinct ultrastructural features. Similarly cells from chronic lymphocytic, prolymphocytic leukemia and an intermediate group CLL/ PL were identified by the size of the nucleolus and the degree of chromatin condensation. The morphometric findings provided an objective morphological basis for the differential diagnosis between these closely related B-cell leukemias which was further supported by differences in the cells immunophenotype.
Key words: Lymphoid leukemia, ultrastructure, lymphocyte lymphoma, hairy cell leukemia, morphometry.
INTRODUCTION
reduced heterochromatin condensation [7-11]. However, in certain cases it is difficult to assign a cell population to one of these well recognised groups. For example, in "prolymphocytoid" transformation of CLL ( C L L / P L ) [12] where there is a gradual appearance within CLL of a clone of cells with the morphology of prolymphocytes as well as some transitional cells [13, 14]. Secondly, individual cells from some leukemias appear to have characteristic features of two or more cell types. For example in a variant form of H C L (HCL-V) the cells appear to be villous like H C L cells, but they are also nucleolated as prolymphocytes [15, 16]. Thus the distinction of such a variant hairy cell from a prolymphocyte may be based on its degree of villosity. Another example of the need of precision in morphological diagnosis is given by the villous lymphocytes of the splenic Bcell lymphoma (SLVL) [17]. In SLVL the circulating cells resemble CLL lymphocytes but have, in addition, cytoplasmic villi, like in hairy cells. This is the reason why these cases are often misdiagnosed as C L L or H C L [18, 19]. The defining criteria of most of these disorders are based on a subjective appraisal of the relevant
ULTRASTRUCTURAL analysis of the peripheral blood lymphocytes often aids, and in some cases is essential, for the differential diagnosis of leukemia [1, 2]. Certain morphological features have been regarded as characteristic of some cell types. For example, lymphocytes from B-chronic lymphocytic leukemia are small, with a high nuclear to cytoplasmic ( N : C ) ratio, well condensed chromatin and little organelle development [3-6], prolymphocytes from Bprolymphocytic leukemia are larger, have less chromatin condensation than C L L lymphocytes and, characteristically contain a large prominent nucleolus [1]. Lymphocytes from hairy cell leukemia, on the other hand, are recognised primarily by their numerous cytoplasmic villi, low N : C ratio and
Abbreviations: B-PLL, B-prolymphocytic leukemia; CCI, cytoplasmic contour index; CLL, chronic lymphocytic leukemia; EM, electron microscopy; HCL, hairy cell leukemia; NCI, nuclear contour index; SLVL, splenic lymphoma with circulating villous lymphocytes. Correspondence to: Professor D. Catovsky, Academic Department of Haematology and Cytogenetics, The Royal Marsden Hospital, London SW3 6JJ, U.K. 357
358
DANIELLEROBINSONet
m o r p h o l o g i c a l features. H o w e v e r , for borderline or difficult cases the existence of objective criteria for the categorisation of cells m a y be useful. In o r d e r to provide quantitative data to allow this categorisation, we p e r f o r m e d a m o r p h o m e t r i c analysis o f cells f r o m different types of chronic B-cell leukemia with the following aims: (i) to establish a quantitative n o r m a l range for each feature e x a m i n e d in each of the cell types investigated, (ii) to ascertain w h e t h e r the m o r phological features of S L V L and H C L - V cells represent extremes within the n o r m a l range f o u n d in typical H C L , or w h e t h e r the differences b e t w e e n cell types are sufficient to allow a distinction b e t w e e n t h e m to be m a d e , (iii) to e x a m i n e w h e t h e r the p r o l y m p h o c y t o i d cells of C L L / P L are identical to the p r o l y m p h o c y t e s seen in P L L , or represent, in some way, a transitional cell b e t w e e n the small l y m p h o c y t e of C L L and the p r o l y m p h o c y t e of P L L .
MATERIALS
AND METHODS
Peripheral blood samples from the following disorders were studied: B-CLL: five cases; B-PLL: seven cases; CLL/ PL: six cases; HCL: seven cases; SLVL: six cases and HCL-V: one case. In each case, the diagnosis was based on morphological, clinical and immunological data. The CLL/PL group was distinguished from B-CLL and B-PLL according to the percentage of peripheral blood prolymphocytes (11-55%) [12]. The mononuclear layer was prepared by density centrifugation on Lymphoprep (Nyegaard) and fixed for electron microscopy as described previously [20]. The analysis of the cell types of each case was based on the EM examination of between 100 and 300 cells. Some 25--40 representative cells from each sample were photographed and used for the morphometric analysis. An IBAS 2 computerised image analysis system (Kontron Bildanalyse GmBh, F.R.G.) was used for the morphometric analysis. The IBAS 2 system used a standard television input which is digited as an array 512 × 512 pixels with a range of 256 grey levels for each point. Television pictures from EM negatives at a magnification of 6300 or 8000 were stored for measurement and projected onto a television screen. Three additional image memories were used to allow image enhancement and selection of features of the image which were to be measured. Selection of the pertinent areas was either performed manually using the interactive drawing system to outline the features of interest, or automatically on the basis of their relative levels of greyness compared to the rest of the image. Greyness thresholds were set interactively taking into account the differences in particular images. Figure 1 illustrates the processes involved in obtaining measurements of particular features. In each case, values for the area and perimeter of the feature under investigation were obtained. These were corrected after calibration of the machine and expressed in ~tm2 or ~tm respectively. Film negatives were used for the measurements, therefore the colors in Fig. 1 appear reversed, i.e. the heterochromatin masses are light and the euchromatin appears dark.
al.
The outline of the nucleolus was traced using the interactive drawing system and the inappropriate area excluded. The area of the nucleolus was measured automatically. The heterochromatin condensation was distinguished by the degree of differential greyness of the heterochromatin masses within the nucleus with respect to the remainder of the cell. Any non-nuclear masses within the cytoplasm which had the same level of greyness as the heterochromatin were selectively excluded. Similarly, in order to ensure that the measurement obtained represented only that percentage of the nucleus occupied by heterochromatin, the nucleolus was excluded from the image. The perimeter and area of the nucleus were estimated by tracing its outline and automatically editing the remainder of the cell. In order to assess the area and perimeter of the cell, its outline was traced and care was taken to include all the villi which were attached to the cell at the plane under investigation. The area and perimeter of the cell body were estimated manually by tracing a line around the cell at the origin of the villi which were then automatically edited from the image. From the corrected values, it was possible to assess: (i) the cross-sectional area of the cell outline including the villi (~tm2), (ii) the cross-sectional area of the cell body, excluding the villi (~tm2), (iii) the nuclear to cytoplasmic ratio (N: C) as measured by the percentage of the cytoplasm occupied by the nucleus, (iv) the degree of heterochromatin condensation, as measured by the percentage of the nucleus occupied by the sum of the heterochromatin masses, (v) the "prominence" of the nucleolus, as measured by the percentage of the nucleus occupied by the nucleolus, (vi) the cytoplasmic contour index (CCI), or the degree of cytoplasmic outline irregularity, this being a measure of the degree of villosity of the cell. The following formula was used Perimeter of the cell outline (P) CCI = Area of the cell outline ~ (A~)" This is a size independent measurement [21, 22] whose minimum value is 3.545, and which increases as the cell becomes more elongated or to a greater extent as the numbers and/or length of the villi increase, and (vii) nuclear contour index (NCI) or the irregularity of the nucleus. This is calculated from the equation: NCI =
Perimeter of the nucleus Area of the nucleus i
FIG. 1. Quantitative measurements of an HCL-V cell using the IBAS 2 image analyser. (a) Original appearance of cell from EM negative. Colours are inverted thus heterochromatin is pale and euchromatin dark (× 6300); (b) the nucleolar boundary was traced and the remainder of the cell excluded; (c) the chromatin condensation was distinguished by the differential greyness of the heterochromatin masses within the nucleus with respect to the remainder of the cell; (d) the nuclear outline and perimeter were estimated by tracing the outline of the nucleus and selectively editing the remainder of the cell; (e) to estimate the cytoplasmic area, the outline of the cell was traced including all villi which were attached to the cell at that section; (f) to measure the area of the cell body, a line was traced around the cell at the origin of the villi, which were then excluded.
~ILII D ~
Morphometry of B-cell leukemias Initially, the morphometric data from each case were compared within each disease group and an average value for each feature was then calculated for the whole group. Analysis of variance showed that differences between disease groups were significantly greater than between patients within the same group. The group means were compared by one way analysis of variance followed by the least significant difference method. Where necessary, the data were transformed by taking square roots in order to have approximately equal variances in different groups.
361
phology of the various cell types is shown schematically in Fig. 2. As statistically significant differences were noted for four parameters between the small lymphocytes and large (prolymphocytoid) cells within C L L / P L , namely cross-sectional area of the cell, the nuclear to cytoplasmic ratio, the heterochromatin condensation and the prominence of the nucleolus, these two subpopulations were therefore treated as separate for the purpose of the morphometric analysis.
RESULTS Cells from six disease groups were analysed: CLL, C L L / P L , H C L , H C L - V and SLVL. The E M mor-
CLL
PLL
Cross-sectional area of the cell (Fig. 3) Lymphocytes from CLL and from C L L / P L had the smaller areas when compared to all other groups. The prolymphocytes of C L L / P L were slightly smaller than their P L L counterparts, but this difference was not statistically significant. H C L cells were significantly larger than cells from all other groups, including those of SLVL (p < 0.001) (Fig. 3a). When the cross-sectional area of the cell in the absence of the villi, i.e. the area of the cell body, was measured (Fig. 3b) H C L cells still remained significantly larger than those from all other groups (p < 0.01). HCLV cells had a value intermediate between, but not significantly different from either H C L or PLL cells. (o)
HCL-V
~-
5o
~
40
g
3o
+
20
:::k
NCL
:.= "N
~6 o o
~ o
FIG. 2. Schematic representation of peripheral blood lymphocytes typical of the various disease groups studied. Chronic lymphocytic leukemia (CLL) cells are small with a high nuclear to cytoplasmic ratio (N : C), well condensed heterochromatin and little organelle development; prolymphocytes from prolymphocytic leukemia (PLL) are larger than the lymphocytes of CLL, have a lower N:C ratio, greater development of cellular organelles, and a prominent nucleolus; cells from CLL with more than 10% circulating prolymphocytes (CLL/PL) are of two types, one resembling the lymphocytes of CLL, the other the prolymphocytes of PLL; cells from hairy cell leukemia (HCL) have a villous outline, a low to medium N: C ratio, an eccentrically placed nucleus, a visible Golgi region and perinuclear microfibrils. Hairy cell variant (HCL-V) cells are villous and possess a prominent nucleolus; splenic lymphoma with circulating villous lymphocytes (SLVL) cells also have numerous villi which are polar in distribution, a nucleolus is often present.
o
Cb) 40 ao •
20
-6 10 <
0
..a o
FIG. 3. Histogram showing mean values of cross-sectional area of cell outline including (a) and excluding villi (b) from cells of the B-lymphoproliferative disorders studied.
362
DANIELLE ROBINSON et al.
,g o
60 8
E
40
.~.
20
z
6
0 O
Z
{.
1
w
|
,
5
5
O
o
o
o
FIG. 4. Histogram showing the nuclear to cytoplasmicratio of cells from each group. The values are expressed as the percentage of the cytoplasm occupied by the nucleus.
FIG. 6. Histogram showing the percentage of the nucleus occupied by the nucleolus. The percentages of cells from each group which displayed a nucleolus can be found in the text.
Nuclear to cytoplasmic ratio (N : C) (Fig. 4)
of chromatin condensation than did HCL and PLL cells (p < 0.01).
HCL cells had a significantly lower N: C ratio than cells from all other groups, including HCL-V and SLVL (p < 0.001). Small lymphocytes of CLL on the other hand, had a significantly larger N : C ratio than both the small lymphocytes of CLL/PL (p < 0.01) and cells from all other disorders (p < 0.001). There was no significant difference in the values for both the small lymphocytes of CLL/PL and the prolymphocytes of the same disease. SLVL cells had a significantly higher N: C ratio than both HCL cells (p < 0.001) and HCL-V cells (p < 0.01).
Chromatin condensation (Fig. 5) The higher values were recorded in the small lymphocytes of CLL and CLL/PL. There was no significant difference between the mean values for the degree of heterochromatin condensation of HCL cells and the prolymphocytes from PLL and CLL/ PL. Lymphocytes from SLVL had a greater degree
o
5O
30
.~ 20 E
10
o
0
A variable proportion of cells with nucleolus was found in all the disease groups. In PLL and HCL-V 60-70% of cells had a visible nucleolus within a given plane of section; in CLL and HCL this proportion was 20-30% of cells and in SLVL, 40-50%. Since the nucleolus was not unique to prolymphocytes, it was necessary to establish whether or not differences in nucleolus size could be measured between the various cell types. This was estimated by calculating the area occupied within the nucleus by the nucleolus (Fig. 6). Cells from PLL had the largest nucleolus, although this was not significantly greater than that of cells from SLVL and HCL-V, or from the prolymphocytes of CLL/PL. The nucleolar size from all these groups were significantlygreater (p < 0.001) than those from the small lymphocytes of CLL, CLL/ PL and hairy cells.
Degree of irregularity of cytoplasmic outline (CCI) (Fig. 7)
4o a,
Size of the nucleolus
O
o O
FIG. 5. Histogram showing the chromatin condensation of cells from each group. The values are expressed as the percentage of the nucleus occupied by heterochromatin masses.
HCL cells had the largest value for this parameter which indicated that they were significantly more villous than cells from all other groups, including SLVL and HCL-V (p < 0.001). Both HCL-V and SLVL cells were in turn more villous than cells from the remaining groups (p < 0.001). Cells from CLL and CLL/PL were the least villous of all the groups studied. Included within the PLL group was one patient in whom the cells were significantly more villous (p < 0.01) than those from the remainder of the group, but still less villous than SLVL or HCLV cells (p < 0.01). Cells from this patient were no different from those of other PLL patients with respect of other measured parameters.
Morphometryof B-cell leukemias 0
o
g *j
"~===
.9.0 "O
~-
~. .,_1 °
FIG. 7. Histogram showing the cytoplasmic contour index of cells from each disease group studied. This size-independent parameter indicates the villosity of the cell and is proportional to the length and/or number of villi at the cell surface.
Degree of nuclear irregularity (NC1) (Fig. 8) The nuclei of CLL cells were significantly more regular than those of all other conditions, except for the HCL-V cells which also had a round nucleus with a low NCI. DISCUSSION In an analysis of leukemic cells, particularly for the purpose of diagnosis, there are several morphological parameters which need to be taken into consideration. Many of the features which are typical of a particular cell type may be seen, albeit to a lesser degree, in other cells. For example, nucleoli can be seen in cells from CLL or HCL, although those displaying organelle are fewer than in a similar section through population of prolymphocytes [3, 8]. Similarly, although villi are characteristic of hairy cells, they may be found with equal frequency in cells from HCL-V and SLVL. Furthermore, some prolymphocytes also display short villi, indicating
4.0 3.0 o
2.0
o o
1.0 z
0.0
_,,
g
o
FIG. 8. Histogram showing the nuclear contour index or the degree of irregularity of nuclear outline.
363
that villosity per se cannot be regarded as characteristic of hairy cells. It is thus apparent that morphological descriptions alone may not always distinguish cell types when the differences between them are quantitative rather than qualitative. For this reason, we undertook a quantitative analysis of cells from different B-cell leukemias, using an IBAS 2 image analyser system. The measurements obtained were found to be reproducible, indicating that the system was reliable for performing this analysis. The measurements were used to give numerical information as to what is the normal range of values expected for each parameter in each cell type and to indicate which features may be important in distinguishing one cell type from another. This study was originally devised to allow a distinction between cells from SLVL and HCL, and also between PLL, HCL and HCL-V cells where a purely qualitative analysis may fail to differentiate between cell type. Cells from SLVL, like those of HCL have a villous outline, a feature which may, in the past, have led to the misdiagnosis of this disease [17-19, 23, 24]. However, the morphometric data indicate that the degree of villosity of SLVL cells was lower than that of HCL cells. Furthermore, other features such as the cell size, N : C ratio and chromatin condensation could distinguish between these cell types. SLVL involves the white and red pulps of the spleen and this also distinguishes this disease from both HCL and HCL-V, where the red pulp is the primary site of involvement [17, 18, 23, 24]. In addition IgM/D is the main Ig class expressed at the cell surface of SLVL and PLL cells, indicating an earlier stage of maturation than in HCL [17]. These data in addition to the morphometric analysis confirms that SLVL may be regarded as a separate entity from HCL. HCL-V cells have been regarded as an intermediate type between PLL and HCL cells, because they display a nucleolus like PLL cells, and are highly villous like cells. Two questions were considered of interest with respect of HCL-V cells: (i) is their degree of villosity outside the normal range for PLL? and (ii) is the size of their nucleolus greater than that of typical HCL cells? The morphometric analysis confirmed that the nucleolus of HCL-V cells is as large as that of PLL cells, and that in both conditions it is significantly greater than in HCL cells. HCL-V cells were also similar to prolymphocytes in terms of the N : C ratio, which was higher than in HCL cells in both cases. Although HCL-V cells were significantly less villous than HCL cells their degree of cytoplasmic irregularity measured by the CCI was greater than that obtained for PLL cells. This leads us to conclude that the villosity of HCL-V cells is
364
DANIELLE ROBINSON et al.
outside the normal range of values obtained for PLL, and support morphologically the concept that the cells of HCL-V have features of both prolymphocytes and hairy cells of HCL, but which cannot satisfactorily be classified as either of those cell types. HCL-V cells were also found here to have a significantly lower NCI than PLL and HCL cells and a greater degree of chromatin condensation. In addition HCL-V cells can be distinguished by other ultrastructural features not quantified here, such as the amount of endoplasmic reticulum, position of the nucleus and type of lysosomal granules [16]. Further evidence for the intermediate nature of HCL-V may be found in the immunological profile of this cell. HCL-V cells are RFB1 positive, a monoclonal antibody which is also positive in HCL [25], but not so consistently in PLL, and like those of PLL, HCL-V cells are negative with the anti-hairy cell antibody HC2 [26]. Clinically HCL-V unlike HCL presents with a high WBC, usually greater than 50 x 109/1. The present analysis has confirmed the existence of significant differences in at least three parameters between the two subpopulations seen in CLL/PL; namely cell size, heterochromatin condensation and the N: C ratio, in addition to the more easily discernable presence of a prominent nucleolus. Work from this laboratory has shown that CLL, PLL and CLL/PL may be distinguished from each other according to the proportion of circulating prolymphocytes, less than 10% in CLL, 11-55% in CLL/PL and over 55% in PLL [12]. However, since even in CLL, there is a small proportion of prolymphocytoid cells [12, 13, 27-29] it is uncertain whether CLL/PL forms part of a continuous spectrum between CLL and PLL, or whether the three diseases are separate entities. Immunological data support the view that CLL/PL cells retain the characteristics of CLL, viz a low density surface immunoglobulin, low FMC7 and high mouse erythrocyte rosette binding capacity [13, 14, 30, 31]. The prolymphocytes of CLL/PL were morphologically more heterogeneous than those of PLL as demonstrated by a large standard deviation observed in many parameters. Nevertheless, with the exception of the N : C ratio, the mean values for all other parameters were similar to the prolymphocytes of PLL groups. This indicates that between CLL and PLL CLL/PL cells may indeed represent intermediate stages and that the acquisition of membrane phenotype characteristics of PLL occurs as a late event during this transformation. However, and as shown by Melo et al. [32] C L L / P L rarely progresses to typical PLL. A morphometric analysis, such as that described here, appears to be useful in determining which ultra-
structural features are of importance in distinguishing one cell type from another. This information may prove useful, not only in permitting future characterisation of cell types, but also in helping to study the relationships between closely related cell types from different leukemias.
REFERENCES 1. Costello C., Catovsky D., O'Brien M. & Galton D. A. G. (1980) Prolymphocytic leukaemia: an ultrastructural study of 22 cases. Br. J. Haemat. 44, 389. 2. Parreira A., Robinson D. S. F., Melo J. V., Ayliffe M., Ball S., Hegde U., Baughan A., Fairhead S., Talavera J. G., Katzmann J. A. & Catovsky D. (1985) Primary plasma cell leukaemia: immunological and ultrastructural studies in 6 cases. Scand. J. Haemat. 35, 570. 3. Schumacher H. R., Maugel T. K. & Davis K. D. (1970) The lymphocyte of chronic lymphaticleukemia. I. Electron microscopy--onset. Cancer 26, 895. 4. Schrek R. (1972) Ultrastructure of blood lymphocytes from chronic lymphocytic and lymphosarcoma cell leukemia. J. natn. Cancer Inst. 48, 51. 5. Cawley J. C. & Hayhoe F. G. J. (1973) Ultrastructure of haemic cells. In A Cytological Atlas of Normal and Leukaemic Blood and Bone Marrow. W. B. Saunders, London. 6. Greaves M. F., Habeshaw J. A. & Stansfeld A. G. (1981) Lymphoproliferative disorders. In Atlas of Blood Cells: Function and Pathology (Zucker F. D., Greaves M. F., Grossi C. E. & Marmont A. M., Eds), p. 409. Edi Eimes, Milan. 7. Katayama I., Li C. Y. & Yam L. T. (1972) Ultrastructural characteristics of "hairy cells" of leukemic reticuloendotheliosis. Am. J. Path. 67, 361. 8. Flandrin G., Daniel M. T., Fourcade M. & Chellow N. (1973) Leuctmie/~ "tricholeucyte" (hairy cell leukemia); 6tude clinique et cytologique de 55 observations. Nouv. Revue fr. H~mat. 13, 609. 9. Daniel M. T. & Flandrin G. (1974) Fine structure of abnormal cells in hairy cell (tricholeukocytic) leukemia with special reference to their in vitro phagocytic capacity. Lab. Invest. 30, 1. 10. Bouroncle B. (1979) Leukemic reticulodendotheliosis (hairy cell leukemia). Blood 53, 412. 11. Cawley J. C., Burns G. F. & Hayhoe F. G. J. (1980) Hairy cell leukaemia. In Recent Results in Cancer Research, p. 72. Springer, Berlin. 12. Melo J. V., Catovsky D. & Galton D. A. G. (1986) The relationship between chronic lymphocytic leukaemia and prolymphocytic leukaemia. I. Clinical and laboratory features of 300 patients and characterization of an intermediate group. Br. J. Haemat. 63, 77. 13. Enno A., Catovsky D., O'Brine M., Cherchi M., Kumaran T. O. & Galton D. A. G. (1979)"Prolymphocytoid" transformation of chronic lymphocytic leukaemia. Br. J. Haemat. 41, 9. 14. Catovsky D., Matutes E., Crockard A., O'Brien M. & Costeilo C. (1984) Prolymphocytic leukemia of Band T-cell types. In Human Leukemias (Polliack A., Ed.), p. 251. Martinus Nijhoff, Boston.
Morphometry of B-cell leukemias 15. Cawley J. C., Burns G. F. & Hayhoe F. G. J. (1980) A chronic lymphoproliferative disorder with distinctive features: a distinct variant of hairy cell leukaemia. Leukemia Res. 4, 547. 16. Catovsky D., O'Brien M., Melo J. V., Wardle J. & Brozovic M. (1984) Hairy cell leukemia (HCL) variant: an intermediate disease between HCL and Bprolymphocytic leukemia. Sere. Onc. 11, 362. 17. Melo J. V., Robinson D. S. F. & Catovsky D. (1987) Splenic B-cell lymphoma with circulating "villous" lymphocytes in the peripheral blood: a disorder distinct from HCL. Leukemia 1, 294. 18. Palutke M., Tabacza P., Mirchandani I. & Goldfarb S. (1981) Lymphocytic lymphoma simulating hairy cell leukemia: a consideration of reliable and unreliable diagnostic features. Cancer 48, 2047. 19. Spriano P., Barosi G., Invernizzi R., Ippoliti G., Fortunato A., Rosso R. & Magrini U. (1977) Splenomegalic immunocytoma with circulating hairy cells. Report of eight cases and revision of literature. Haematologica 71, 217. 20. Robinson D. S. F., Melo J. V., Andrews C., Schey S. A. & Catovsky D. (1985) Intracytoplasmic inclusions in B-prolymphocytic leukaemia: ultrastructural, cytochemical and immunological studies. J. clin. Path. 38, 897. 21. Schrek R., Mayron L. W. & Knospe W. H. (1971) Quantitative electron microscopy of normal and leukaemia lymphocytes. Lancet i, 348. 22. Meijer C. J. L. M., Van der Valk P. & Jansen J. (1984) Hairy cell leukemia: an immunohistochemic and morphometric study. Sere. Oncol. 11, 347. 23. Neiman R. S., Sullivan A. L. & Jaffe R. (1979) Malignant lymphoma simulating leukemic reticuloendo-
24.
25. 26. 27.
28. 29.
30.
31. 32.
365
theliosis. A clinicopathologic study of ten cases. Cancer 43, 329. Fohlmeister I., Schaefer H-E., Modder B., Hellriegel K-P. & Fischer R. (1981) Chronische lymphoproliferative erkrankung unter dem bild einer haarzell-leukamie. Blut 42, 367. Melo J. V., Catovsky D. & Bodger M. (1984) Reactivity of the monoclonal antibody RFB-1 in chronic Bcell leukaemias. Scand. J. Haemat. 32, 417. Posnett D. N., Chiorazzi N. & Kunkel H. G. (1982) Monoclonal antibodies with specificity for hairy cell leukemia cells. J. clin. Invest. 70, 254. Peterson L. C., Bloomfield C. D., Sundberg R. D., Gajl-Peczalska K. J. & Brunning R. D. (1975) Morphology of chronic leukemia and its relationship to survival. A m . J. Med. 59, 316. Kjeldsberg C. R. & Marty J. (1981) Prolymphocytic transformation of chronic lymphocytic leukemia. Cancer 48, 2447. Economopoulos T., Fotopoulos S., Hatzioannou J. & Garkikas C. (1982) "Prolymphocytoid" cells in chronic lymphocytic leukemia and their prognostic significance. Scand. J. Haemat. 28, 238. Catovsky D., Cherchi M., Brooks D., Bradley J. & Zola H. (1981) Heterogeneity of B-cell leukemias demonstrated by the monoclonal antibody FMC7. Blood 58, 406. Catovsky D., Wechsler A. & Cherchi M. (1981) Mrosettes in B-cell leukemias (Letter). Blood 58, 410. Melo J. V., Catovsky D. & Galton D. A. G. (1986) The relationship between chronic lymphocytic leukaemia and prolymphocytic leukaemia. II. Patterns of evolution of "prolymphocytoid" transformation. Br. J. Haemat. 64, 77.
0145-2126/89 $3.00 + .00 Pergamon Press plc
Printed in Great Britain.
M O R P H O M E T R I C ANALYSIS OF C H R O N I C B-CELL L E U K E M I A S - - - A N A I D TO THE CLASSIFICATION OF L Y M P H O I D CELL TYPES DANIELLE ROBINSON, PETER LACKIE, VICTOR ABER and DANIEL CATOVSKY MRC Leukaemia Unit, Histochemistry Department and Computer Centre, Royal Postgraduate Medical School, London, U.K.
(Received 2 November 1988. Accepted 28 November 1988) Abstract--In order to evaluate the ultrastructural features which may be of relevance in distinguishing cells from the various chronic B-cell leukemias, a morphometric analysis was performed on a large number of cells from each disease group. The parameters selected were: cell size, nucleo: cytoplasmic ratio, chromatin condensation, size of the nucleolus and degree of irregularity of both the nucleus and the cytoplasmic outlines. The mean values obtained for each parameter for each disease group were compared statistically. In disorders in which the cells have villous cytoplasmic projections, the quantitative analysis of the cellular features was helpful to characterise the different types of B cells involved. Thus, cells from cases of splenic lymphoma were found to be different from those of hairy cell leukemia, and a variant form of HCL was also identified by its distinct ultrastructural features. Similarly cells from chronic lymphocytic, prolymphocytic leukemia and an intermediate group CLL/ PL were identified by the size of the nucleolus and the degree of chromatin condensation. The morphometric findings provided an objective morphological basis for the differential diagnosis between these closely related B-cell leukemias which was further supported by differences in the cells immunophenotype.
Key words: Lymphoid leukemia, ultrastructure, lymphocyte lymphoma, hairy cell leukemia, morphometry.
INTRODUCTION
reduced heterochromatin condensation [7-11]. However, in certain cases it is difficult to assign a cell population to one of these well recognised groups. For example, in "prolymphocytoid" transformation of CLL ( C L L / P L ) [12] where there is a gradual appearance within CLL of a clone of cells with the morphology of prolymphocytes as well as some transitional cells [13, 14]. Secondly, individual cells from some leukemias appear to have characteristic features of two or more cell types. For example in a variant form of H C L (HCL-V) the cells appear to be villous like H C L cells, but they are also nucleolated as prolymphocytes [15, 16]. Thus the distinction of such a variant hairy cell from a prolymphocyte may be based on its degree of villosity. Another example of the need of precision in morphological diagnosis is given by the villous lymphocytes of the splenic Bcell lymphoma (SLVL) [17]. In SLVL the circulating cells resemble CLL lymphocytes but have, in addition, cytoplasmic villi, like in hairy cells. This is the reason why these cases are often misdiagnosed as C L L or H C L [18, 19]. The defining criteria of most of these disorders are based on a subjective appraisal of the relevant
ULTRASTRUCTURAL analysis of the peripheral blood lymphocytes often aids, and in some cases is essential, for the differential diagnosis of leukemia [1, 2]. Certain morphological features have been regarded as characteristic of some cell types. For example, lymphocytes from B-chronic lymphocytic leukemia are small, with a high nuclear to cytoplasmic ( N : C ) ratio, well condensed chromatin and little organelle development [3-6], prolymphocytes from Bprolymphocytic leukemia are larger, have less chromatin condensation than C L L lymphocytes and, characteristically contain a large prominent nucleolus [1]. Lymphocytes from hairy cell leukemia, on the other hand, are recognised primarily by their numerous cytoplasmic villi, low N : C ratio and
Abbreviations: B-PLL, B-prolymphocytic leukemia; CCI, cytoplasmic contour index; CLL, chronic lymphocytic leukemia; EM, electron microscopy; HCL, hairy cell leukemia; NCI, nuclear contour index; SLVL, splenic lymphoma with circulating villous lymphocytes. Correspondence to: Professor D. Catovsky, Academic Department of Haematology and Cytogenetics, The Royal Marsden Hospital, London SW3 6JJ, U.K. 357
358
DANIELLEROBINSONet
m o r p h o l o g i c a l features. H o w e v e r , for borderline or difficult cases the existence of objective criteria for the categorisation of cells m a y be useful. In o r d e r to provide quantitative data to allow this categorisation, we p e r f o r m e d a m o r p h o m e t r i c analysis o f cells f r o m different types of chronic B-cell leukemia with the following aims: (i) to establish a quantitative n o r m a l range for each feature e x a m i n e d in each of the cell types investigated, (ii) to ascertain w h e t h e r the m o r phological features of S L V L and H C L - V cells represent extremes within the n o r m a l range f o u n d in typical H C L , or w h e t h e r the differences b e t w e e n cell types are sufficient to allow a distinction b e t w e e n t h e m to be m a d e , (iii) to e x a m i n e w h e t h e r the p r o l y m p h o c y t o i d cells of C L L / P L are identical to the p r o l y m p h o c y t e s seen in P L L , or represent, in some way, a transitional cell b e t w e e n the small l y m p h o c y t e of C L L and the p r o l y m p h o c y t e of P L L .
MATERIALS
AND METHODS
Peripheral blood samples from the following disorders were studied: B-CLL: five cases; B-PLL: seven cases; CLL/ PL: six cases; HCL: seven cases; SLVL: six cases and HCL-V: one case. In each case, the diagnosis was based on morphological, clinical and immunological data. The CLL/PL group was distinguished from B-CLL and B-PLL according to the percentage of peripheral blood prolymphocytes (11-55%) [12]. The mononuclear layer was prepared by density centrifugation on Lymphoprep (Nyegaard) and fixed for electron microscopy as described previously [20]. The analysis of the cell types of each case was based on the EM examination of between 100 and 300 cells. Some 25--40 representative cells from each sample were photographed and used for the morphometric analysis. An IBAS 2 computerised image analysis system (Kontron Bildanalyse GmBh, F.R.G.) was used for the morphometric analysis. The IBAS 2 system used a standard television input which is digited as an array 512 × 512 pixels with a range of 256 grey levels for each point. Television pictures from EM negatives at a magnification of 6300 or 8000 were stored for measurement and projected onto a television screen. Three additional image memories were used to allow image enhancement and selection of features of the image which were to be measured. Selection of the pertinent areas was either performed manually using the interactive drawing system to outline the features of interest, or automatically on the basis of their relative levels of greyness compared to the rest of the image. Greyness thresholds were set interactively taking into account the differences in particular images. Figure 1 illustrates the processes involved in obtaining measurements of particular features. In each case, values for the area and perimeter of the feature under investigation were obtained. These were corrected after calibration of the machine and expressed in ~tm2 or ~tm respectively. Film negatives were used for the measurements, therefore the colors in Fig. 1 appear reversed, i.e. the heterochromatin masses are light and the euchromatin appears dark.
al.
The outline of the nucleolus was traced using the interactive drawing system and the inappropriate area excluded. The area of the nucleolus was measured automatically. The heterochromatin condensation was distinguished by the degree of differential greyness of the heterochromatin masses within the nucleus with respect to the remainder of the cell. Any non-nuclear masses within the cytoplasm which had the same level of greyness as the heterochromatin were selectively excluded. Similarly, in order to ensure that the measurement obtained represented only that percentage of the nucleus occupied by heterochromatin, the nucleolus was excluded from the image. The perimeter and area of the nucleus were estimated by tracing its outline and automatically editing the remainder of the cell. In order to assess the area and perimeter of the cell, its outline was traced and care was taken to include all the villi which were attached to the cell at the plane under investigation. The area and perimeter of the cell body were estimated manually by tracing a line around the cell at the origin of the villi which were then automatically edited from the image. From the corrected values, it was possible to assess: (i) the cross-sectional area of the cell outline including the villi (~tm2), (ii) the cross-sectional area of the cell body, excluding the villi (~tm2), (iii) the nuclear to cytoplasmic ratio (N: C) as measured by the percentage of the cytoplasm occupied by the nucleus, (iv) the degree of heterochromatin condensation, as measured by the percentage of the nucleus occupied by the sum of the heterochromatin masses, (v) the "prominence" of the nucleolus, as measured by the percentage of the nucleus occupied by the nucleolus, (vi) the cytoplasmic contour index (CCI), or the degree of cytoplasmic outline irregularity, this being a measure of the degree of villosity of the cell. The following formula was used Perimeter of the cell outline (P) CCI = Area of the cell outline ~ (A~)" This is a size independent measurement [21, 22] whose minimum value is 3.545, and which increases as the cell becomes more elongated or to a greater extent as the numbers and/or length of the villi increase, and (vii) nuclear contour index (NCI) or the irregularity of the nucleus. This is calculated from the equation: NCI =
Perimeter of the nucleus Area of the nucleus i
FIG. 1. Quantitative measurements of an HCL-V cell using the IBAS 2 image analyser. (a) Original appearance of cell from EM negative. Colours are inverted thus heterochromatin is pale and euchromatin dark (× 6300); (b) the nucleolar boundary was traced and the remainder of the cell excluded; (c) the chromatin condensation was distinguished by the differential greyness of the heterochromatin masses within the nucleus with respect to the remainder of the cell; (d) the nuclear outline and perimeter were estimated by tracing the outline of the nucleus and selectively editing the remainder of the cell; (e) to estimate the cytoplasmic area, the outline of the cell was traced including all villi which were attached to the cell at that section; (f) to measure the area of the cell body, a line was traced around the cell at the origin of the villi, which were then excluded.
~ILII D ~
Morphometry of B-cell leukemias Initially, the morphometric data from each case were compared within each disease group and an average value for each feature was then calculated for the whole group. Analysis of variance showed that differences between disease groups were significantly greater than between patients within the same group. The group means were compared by one way analysis of variance followed by the least significant difference method. Where necessary, the data were transformed by taking square roots in order to have approximately equal variances in different groups.
361
phology of the various cell types is shown schematically in Fig. 2. As statistically significant differences were noted for four parameters between the small lymphocytes and large (prolymphocytoid) cells within C L L / P L , namely cross-sectional area of the cell, the nuclear to cytoplasmic ratio, the heterochromatin condensation and the prominence of the nucleolus, these two subpopulations were therefore treated as separate for the purpose of the morphometric analysis.
RESULTS Cells from six disease groups were analysed: CLL, C L L / P L , H C L , H C L - V and SLVL. The E M mor-
CLL
PLL
Cross-sectional area of the cell (Fig. 3) Lymphocytes from CLL and from C L L / P L had the smaller areas when compared to all other groups. The prolymphocytes of C L L / P L were slightly smaller than their P L L counterparts, but this difference was not statistically significant. H C L cells were significantly larger than cells from all other groups, including those of SLVL (p < 0.001) (Fig. 3a). When the cross-sectional area of the cell in the absence of the villi, i.e. the area of the cell body, was measured (Fig. 3b) H C L cells still remained significantly larger than those from all other groups (p < 0.01). HCLV cells had a value intermediate between, but not significantly different from either H C L or PLL cells. (o)
HCL-V
~-
5o
~
40
g
3o
+
20
:::k
NCL
:.= "N
~6 o o
~ o
FIG. 2. Schematic representation of peripheral blood lymphocytes typical of the various disease groups studied. Chronic lymphocytic leukemia (CLL) cells are small with a high nuclear to cytoplasmic ratio (N : C), well condensed heterochromatin and little organelle development; prolymphocytes from prolymphocytic leukemia (PLL) are larger than the lymphocytes of CLL, have a lower N:C ratio, greater development of cellular organelles, and a prominent nucleolus; cells from CLL with more than 10% circulating prolymphocytes (CLL/PL) are of two types, one resembling the lymphocytes of CLL, the other the prolymphocytes of PLL; cells from hairy cell leukemia (HCL) have a villous outline, a low to medium N: C ratio, an eccentrically placed nucleus, a visible Golgi region and perinuclear microfibrils. Hairy cell variant (HCL-V) cells are villous and possess a prominent nucleolus; splenic lymphoma with circulating villous lymphocytes (SLVL) cells also have numerous villi which are polar in distribution, a nucleolus is often present.
o
Cb) 40 ao •
20
-6 10 <
0
..a o
FIG. 3. Histogram showing mean values of cross-sectional area of cell outline including (a) and excluding villi (b) from cells of the B-lymphoproliferative disorders studied.
362
DANIELLE ROBINSON et al.
,g o
60 8
E
40
.~.
20
z
6
0 O
Z
{.
1
w
|
,
5
5
O
o
o
o
FIG. 4. Histogram showing the nuclear to cytoplasmicratio of cells from each group. The values are expressed as the percentage of the cytoplasm occupied by the nucleus.
FIG. 6. Histogram showing the percentage of the nucleus occupied by the nucleolus. The percentages of cells from each group which displayed a nucleolus can be found in the text.
Nuclear to cytoplasmic ratio (N : C) (Fig. 4)
of chromatin condensation than did HCL and PLL cells (p < 0.01).
HCL cells had a significantly lower N: C ratio than cells from all other groups, including HCL-V and SLVL (p < 0.001). Small lymphocytes of CLL on the other hand, had a significantly larger N : C ratio than both the small lymphocytes of CLL/PL (p < 0.01) and cells from all other disorders (p < 0.001). There was no significant difference in the values for both the small lymphocytes of CLL/PL and the prolymphocytes of the same disease. SLVL cells had a significantly higher N: C ratio than both HCL cells (p < 0.001) and HCL-V cells (p < 0.01).
Chromatin condensation (Fig. 5) The higher values were recorded in the small lymphocytes of CLL and CLL/PL. There was no significant difference between the mean values for the degree of heterochromatin condensation of HCL cells and the prolymphocytes from PLL and CLL/ PL. Lymphocytes from SLVL had a greater degree
o
5O
30
.~ 20 E
10
o
0
A variable proportion of cells with nucleolus was found in all the disease groups. In PLL and HCL-V 60-70% of cells had a visible nucleolus within a given plane of section; in CLL and HCL this proportion was 20-30% of cells and in SLVL, 40-50%. Since the nucleolus was not unique to prolymphocytes, it was necessary to establish whether or not differences in nucleolus size could be measured between the various cell types. This was estimated by calculating the area occupied within the nucleus by the nucleolus (Fig. 6). Cells from PLL had the largest nucleolus, although this was not significantly greater than that of cells from SLVL and HCL-V, or from the prolymphocytes of CLL/PL. The nucleolar size from all these groups were significantlygreater (p < 0.001) than those from the small lymphocytes of CLL, CLL/ PL and hairy cells.
Degree of irregularity of cytoplasmic outline (CCI) (Fig. 7)
4o a,
Size of the nucleolus
O
o O
FIG. 5. Histogram showing the chromatin condensation of cells from each group. The values are expressed as the percentage of the nucleus occupied by heterochromatin masses.
HCL cells had the largest value for this parameter which indicated that they were significantly more villous than cells from all other groups, including SLVL and HCL-V (p < 0.001). Both HCL-V and SLVL cells were in turn more villous than cells from the remaining groups (p < 0.001). Cells from CLL and CLL/PL were the least villous of all the groups studied. Included within the PLL group was one patient in whom the cells were significantly more villous (p < 0.01) than those from the remainder of the group, but still less villous than SLVL or HCLV cells (p < 0.01). Cells from this patient were no different from those of other PLL patients with respect of other measured parameters.
Morphometryof B-cell leukemias 0
o
g *j
"~===
.9.0 "O
~-
~. .,_1 °
FIG. 7. Histogram showing the cytoplasmic contour index of cells from each disease group studied. This size-independent parameter indicates the villosity of the cell and is proportional to the length and/or number of villi at the cell surface.
Degree of nuclear irregularity (NC1) (Fig. 8) The nuclei of CLL cells were significantly more regular than those of all other conditions, except for the HCL-V cells which also had a round nucleus with a low NCI. DISCUSSION In an analysis of leukemic cells, particularly for the purpose of diagnosis, there are several morphological parameters which need to be taken into consideration. Many of the features which are typical of a particular cell type may be seen, albeit to a lesser degree, in other cells. For example, nucleoli can be seen in cells from CLL or HCL, although those displaying organelle are fewer than in a similar section through population of prolymphocytes [3, 8]. Similarly, although villi are characteristic of hairy cells, they may be found with equal frequency in cells from HCL-V and SLVL. Furthermore, some prolymphocytes also display short villi, indicating
4.0 3.0 o
2.0
o o
1.0 z
0.0
_,,
g
o
FIG. 8. Histogram showing the nuclear contour index or the degree of irregularity of nuclear outline.
363
that villosity per se cannot be regarded as characteristic of hairy cells. It is thus apparent that morphological descriptions alone may not always distinguish cell types when the differences between them are quantitative rather than qualitative. For this reason, we undertook a quantitative analysis of cells from different B-cell leukemias, using an IBAS 2 image analyser system. The measurements obtained were found to be reproducible, indicating that the system was reliable for performing this analysis. The measurements were used to give numerical information as to what is the normal range of values expected for each parameter in each cell type and to indicate which features may be important in distinguishing one cell type from another. This study was originally devised to allow a distinction between cells from SLVL and HCL, and also between PLL, HCL and HCL-V cells where a purely qualitative analysis may fail to differentiate between cell type. Cells from SLVL, like those of HCL have a villous outline, a feature which may, in the past, have led to the misdiagnosis of this disease [17-19, 23, 24]. However, the morphometric data indicate that the degree of villosity of SLVL cells was lower than that of HCL cells. Furthermore, other features such as the cell size, N : C ratio and chromatin condensation could distinguish between these cell types. SLVL involves the white and red pulps of the spleen and this also distinguishes this disease from both HCL and HCL-V, where the red pulp is the primary site of involvement [17, 18, 23, 24]. In addition IgM/D is the main Ig class expressed at the cell surface of SLVL and PLL cells, indicating an earlier stage of maturation than in HCL [17]. These data in addition to the morphometric analysis confirms that SLVL may be regarded as a separate entity from HCL. HCL-V cells have been regarded as an intermediate type between PLL and HCL cells, because they display a nucleolus like PLL cells, and are highly villous like cells. Two questions were considered of interest with respect of HCL-V cells: (i) is their degree of villosity outside the normal range for PLL? and (ii) is the size of their nucleolus greater than that of typical HCL cells? The morphometric analysis confirmed that the nucleolus of HCL-V cells is as large as that of PLL cells, and that in both conditions it is significantly greater than in HCL cells. HCL-V cells were also similar to prolymphocytes in terms of the N : C ratio, which was higher than in HCL cells in both cases. Although HCL-V cells were significantly less villous than HCL cells their degree of cytoplasmic irregularity measured by the CCI was greater than that obtained for PLL cells. This leads us to conclude that the villosity of HCL-V cells is
364
DANIELLE ROBINSON et al.
outside the normal range of values obtained for PLL, and support morphologically the concept that the cells of HCL-V have features of both prolymphocytes and hairy cells of HCL, but which cannot satisfactorily be classified as either of those cell types. HCL-V cells were also found here to have a significantly lower NCI than PLL and HCL cells and a greater degree of chromatin condensation. In addition HCL-V cells can be distinguished by other ultrastructural features not quantified here, such as the amount of endoplasmic reticulum, position of the nucleus and type of lysosomal granules [16]. Further evidence for the intermediate nature of HCL-V may be found in the immunological profile of this cell. HCL-V cells are RFB1 positive, a monoclonal antibody which is also positive in HCL [25], but not so consistently in PLL, and like those of PLL, HCL-V cells are negative with the anti-hairy cell antibody HC2 [26]. Clinically HCL-V unlike HCL presents with a high WBC, usually greater than 50 x 109/1. The present analysis has confirmed the existence of significant differences in at least three parameters between the two subpopulations seen in CLL/PL; namely cell size, heterochromatin condensation and the N: C ratio, in addition to the more easily discernable presence of a prominent nucleolus. Work from this laboratory has shown that CLL, PLL and CLL/PL may be distinguished from each other according to the proportion of circulating prolymphocytes, less than 10% in CLL, 11-55% in CLL/PL and over 55% in PLL [12]. However, since even in CLL, there is a small proportion of prolymphocytoid cells [12, 13, 27-29] it is uncertain whether CLL/PL forms part of a continuous spectrum between CLL and PLL, or whether the three diseases are separate entities. Immunological data support the view that CLL/PL cells retain the characteristics of CLL, viz a low density surface immunoglobulin, low FMC7 and high mouse erythrocyte rosette binding capacity [13, 14, 30, 31]. The prolymphocytes of CLL/PL were morphologically more heterogeneous than those of PLL as demonstrated by a large standard deviation observed in many parameters. Nevertheless, with the exception of the N : C ratio, the mean values for all other parameters were similar to the prolymphocytes of PLL groups. This indicates that between CLL and PLL CLL/PL cells may indeed represent intermediate stages and that the acquisition of membrane phenotype characteristics of PLL occurs as a late event during this transformation. However, and as shown by Melo et al. [32] C L L / P L rarely progresses to typical PLL. A morphometric analysis, such as that described here, appears to be useful in determining which ultra-
structural features are of importance in distinguishing one cell type from another. This information may prove useful, not only in permitting future characterisation of cell types, but also in helping to study the relationships between closely related cell types from different leukemias.
REFERENCES 1. Costello C., Catovsky D., O'Brien M. & Galton D. A. G. (1980) Prolymphocytic leukaemia: an ultrastructural study of 22 cases. Br. J. Haemat. 44, 389. 2. Parreira A., Robinson D. S. F., Melo J. V., Ayliffe M., Ball S., Hegde U., Baughan A., Fairhead S., Talavera J. G., Katzmann J. A. & Catovsky D. (1985) Primary plasma cell leukaemia: immunological and ultrastructural studies in 6 cases. Scand. J. Haemat. 35, 570. 3. Schumacher H. R., Maugel T. K. & Davis K. D. (1970) The lymphocyte of chronic lymphaticleukemia. I. Electron microscopy--onset. Cancer 26, 895. 4. Schrek R. (1972) Ultrastructure of blood lymphocytes from chronic lymphocytic and lymphosarcoma cell leukemia. J. natn. Cancer Inst. 48, 51. 5. Cawley J. C. & Hayhoe F. G. J. (1973) Ultrastructure of haemic cells. In A Cytological Atlas of Normal and Leukaemic Blood and Bone Marrow. W. B. Saunders, London. 6. Greaves M. F., Habeshaw J. A. & Stansfeld A. G. (1981) Lymphoproliferative disorders. In Atlas of Blood Cells: Function and Pathology (Zucker F. D., Greaves M. F., Grossi C. E. & Marmont A. M., Eds), p. 409. Edi Eimes, Milan. 7. Katayama I., Li C. Y. & Yam L. T. (1972) Ultrastructural characteristics of "hairy cells" of leukemic reticuloendotheliosis. Am. J. Path. 67, 361. 8. Flandrin G., Daniel M. T., Fourcade M. & Chellow N. (1973) Leuctmie/~ "tricholeucyte" (hairy cell leukemia); 6tude clinique et cytologique de 55 observations. Nouv. Revue fr. H~mat. 13, 609. 9. Daniel M. T. & Flandrin G. (1974) Fine structure of abnormal cells in hairy cell (tricholeukocytic) leukemia with special reference to their in vitro phagocytic capacity. Lab. Invest. 30, 1. 10. Bouroncle B. (1979) Leukemic reticulodendotheliosis (hairy cell leukemia). Blood 53, 412. 11. Cawley J. C., Burns G. F. & Hayhoe F. G. J. (1980) Hairy cell leukaemia. In Recent Results in Cancer Research, p. 72. Springer, Berlin. 12. Melo J. V., Catovsky D. & Galton D. A. G. (1986) The relationship between chronic lymphocytic leukaemia and prolymphocytic leukaemia. I. Clinical and laboratory features of 300 patients and characterization of an intermediate group. Br. J. Haemat. 63, 77. 13. Enno A., Catovsky D., O'Brine M., Cherchi M., Kumaran T. O. & Galton D. A. G. (1979)"Prolymphocytoid" transformation of chronic lymphocytic leukaemia. Br. J. Haemat. 41, 9. 14. Catovsky D., Matutes E., Crockard A., O'Brien M. & Costeilo C. (1984) Prolymphocytic leukemia of Band T-cell types. In Human Leukemias (Polliack A., Ed.), p. 251. Martinus Nijhoff, Boston.
Morphometry of B-cell leukemias 15. Cawley J. C., Burns G. F. & Hayhoe F. G. J. (1980) A chronic lymphoproliferative disorder with distinctive features: a distinct variant of hairy cell leukaemia. Leukemia Res. 4, 547. 16. Catovsky D., O'Brien M., Melo J. V., Wardle J. & Brozovic M. (1984) Hairy cell leukemia (HCL) variant: an intermediate disease between HCL and Bprolymphocytic leukemia. Sere. Onc. 11, 362. 17. Melo J. V., Robinson D. S. F. & Catovsky D. (1987) Splenic B-cell lymphoma with circulating "villous" lymphocytes in the peripheral blood: a disorder distinct from HCL. Leukemia 1, 294. 18. Palutke M., Tabacza P., Mirchandani I. & Goldfarb S. (1981) Lymphocytic lymphoma simulating hairy cell leukemia: a consideration of reliable and unreliable diagnostic features. Cancer 48, 2047. 19. Spriano P., Barosi G., Invernizzi R., Ippoliti G., Fortunato A., Rosso R. & Magrini U. (1977) Splenomegalic immunocytoma with circulating hairy cells. Report of eight cases and revision of literature. Haematologica 71, 217. 20. Robinson D. S. F., Melo J. V., Andrews C., Schey S. A. & Catovsky D. (1985) Intracytoplasmic inclusions in B-prolymphocytic leukaemia: ultrastructural, cytochemical and immunological studies. J. clin. Path. 38, 897. 21. Schrek R., Mayron L. W. & Knospe W. H. (1971) Quantitative electron microscopy of normal and leukaemia lymphocytes. Lancet i, 348. 22. Meijer C. J. L. M., Van der Valk P. & Jansen J. (1984) Hairy cell leukemia: an immunohistochemic and morphometric study. Sere. Oncol. 11, 347. 23. Neiman R. S., Sullivan A. L. & Jaffe R. (1979) Malignant lymphoma simulating leukemic reticuloendo-
24.
25. 26. 27.
28. 29.
30.
31. 32.
365
theliosis. A clinicopathologic study of ten cases. Cancer 43, 329. Fohlmeister I., Schaefer H-E., Modder B., Hellriegel K-P. & Fischer R. (1981) Chronische lymphoproliferative erkrankung unter dem bild einer haarzell-leukamie. Blut 42, 367. Melo J. V., Catovsky D. & Bodger M. (1984) Reactivity of the monoclonal antibody RFB-1 in chronic Bcell leukaemias. Scand. J. Haemat. 32, 417. Posnett D. N., Chiorazzi N. & Kunkel H. G. (1982) Monoclonal antibodies with specificity for hairy cell leukemia cells. J. clin. Invest. 70, 254. Peterson L. C., Bloomfield C. D., Sundberg R. D., Gajl-Peczalska K. J. & Brunning R. D. (1975) Morphology of chronic leukemia and its relationship to survival. A m . J. Med. 59, 316. Kjeldsberg C. R. & Marty J. (1981) Prolymphocytic transformation of chronic lymphocytic leukemia. Cancer 48, 2447. Economopoulos T., Fotopoulos S., Hatzioannou J. & Garkikas C. (1982) "Prolymphocytoid" cells in chronic lymphocytic leukemia and their prognostic significance. Scand. J. Haemat. 28, 238. Catovsky D., Cherchi M., Brooks D., Bradley J. & Zola H. (1981) Heterogeneity of B-cell leukemias demonstrated by the monoclonal antibody FMC7. Blood 58, 406. Catovsky D., Wechsler A. & Cherchi M. (1981) Mrosettes in B-cell leukemias (Letter). Blood 58, 410. Melo J. V., Catovsky D. & Galton D. A. G. (1986) The relationship between chronic lymphocytic leukaemia and prolymphocytic leukaemia. II. Patterns of evolution of "prolymphocytoid" transformation. Br. J. Haemat. 64, 77.