cytological diagnosis of lymphoma in serous effusions · cases an almost pure population of...

J Clin Pathol 1981 ;34:1311-1325

Cytological diagnosis of lymphoma in serouseffusionsAl SPRIGGS, RI VANHEGAN*

From the Laboratory of Clinical Cvtology, Churchill Hospital, Oxford and the Department ofHistopathology, John Radcliffe Hospital, Oxford

SUMMARY A consecutive series is presented of 100 cases of malignant lymphoma in which a correla-tion could be made between the histological findings and the cytology of pleural or peritoneal fluidor both.

Using the Kiel classification, 80 cases could be given a histological label, while 20 remain un-

classified.In those cases in which lymphoma cells were identifiable in the serous fluid, their morphology was

studied and was found in general to correlate well with the cell-type expected from the histology.It is not claimed that the type of lymphoma can be reliably diagnosed from the cytology of pleural

or peritoneal fluid, but an opinion can be provided as to the grade (high or low), and in some cases

it is also possible to give an indication of the cell-type involved.

Pleural and peritoneal effusions are a well-knowncomplication of the lymphomas, and a cytologicalexamination of serous fluid often clarifies a diagnosiswhich would otherwise be obscure.

In Hodgkin's disease the cytological picture ismost commonly non-specific, and a diagnosis canbe made only occasionally by the finding of Stern-berg-Reed cells and the associated cellular exudate.In the rest of the malignant lymphomas tumour cellsmay survive and multiply in the fluid, and in manycases an almost pure population of lymphomacells is present in the pleural or peritoneal cavity.

In the past, many of these cases were labelledsimply as "lymphosarcoma" or "reticulosarcoma,"and with experience such a cytological diagnosiscould be highly reliable. More recently, with in-creased understanding concerning the lymphocyteand the immune system, a more sophisticatedclassification has become possible. Even if nomen-clature has not yet become standardised, patholo-gists working in this field can certainly understandeach other, and the time seems ripe for a re-examina-tion of lymphoma cells in serous fluids, in the lightof the new knowledge.

In what follows, we shall use the Kiel classificationas described by Lennert et al.1 2

*Present address: Department of Pathology, PrincessMargaret Hospital, Swindon.Accepted for publication 7 January 1981

Material and methods

Since 1949, stained smears of pleural and peritonealdeposits have all been mounted and permanentlyfiled in the Laboratory of Clinical Cytology atOxford. Since the preparations used (dried smearsstained by May-Grunwald-Giemsa and mountedbefore use) show little deterioration over severaldecades, it is possible to review almost all thematerial over a 30 year period. Most of those seensince 1960 have also been wet-fixed and stained byPapanicolaou's method.

Similarly, embedded histological material isavailable over the same period, in the files of theDepartment of Histopathology, John RadcliffeHospital, Oxford. The cases used were all of thosefor which adequate cytological and histologicalspecimens were available; out of 140 cases, 100provided material which could be compared. Weestimate that pleural, peritoneal or pericardial fluidwas submitted from about one out of every 20 casesof malignant lymphoma diagnosed in Oxford in theperiod 1949-79 (excluding acute leukaemia).

Since most of the smears were already stained andmounted, few further methods could be applied tothese, but in all possible cases a periodic acid-Schiffstain was done, either on a destained smear or, ifavailable, on a spare unstained one. In a few of themore recent cases an acid phosphatase stain was alsoperformed, using the method of Li et al.3

1311

copyright. on M

arch 13, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.34.12.1311 on 1 D

ecember 1981. D

ownloaded from

http://jcp.bmj.com/

Spriggs, Vanhegan

In many cases, new sections were cut from theblocks, and (where appropriate) immunoperoxidasestains were done for light chains and for muramidase.Surface markers were only studied in very few cases.In 11 cases cell deposits were processed for electronmicroscopy.Smears and sections were reviewed together by

both the authors. The histological classificationrepresents the opinion of RIV.

Results

The diagnoses were distributed as shown in Table 1.As regards the non-Hodgkin's lymphomas, the

distribution of those with effusions compared withall diagnosed cases can be seen in Table 2.

The cytological features found in the differentcategories were as follows:

Table 1 Diagnoses in 100 cases of malignant lymphoma

Type No of cases

Low-grade NHLLymphocytic 6lmmunocytic(a) Lymphoplasmacytoid 3(b) Plasmacytic 2Centroblastic/centrocytic 16Centrocytic 9

High-grade NHLCentroblastic 8Lymphoblastic(a) non-T 7(b) T 5Immunoblastic 9

True histiocytic 2Hodgkin's diseaseLymphocytic predominance 4Nodular sclerosing 4Mixed cellularity 3Lymphocyte depleted 2

Unclassified 20Total 100

MALIGNANT LYMPHOMA, LYMPHOCYTICThere are six cases in the series (other effusions havebeen seen complicating chronic lymphocytic leu-kaemia, but since there was no histological con-firmation, they have been excluded). All but one ofthe six were recorded as leukaemic. Five were maleand one female, and the youngest age was 58 yr.

All the effusions were pleural, and containednumerous lymphocytes. In two cases there were alsoabundant red cells. Apart from an apparently raisednucleo-cytoplasmic ratio, no morphological ab-normality was observed in the lymphocytes in fourcases. In two cases they showed the well knownappearance of "cellules grumelees."4 This expres-sion refers to the markedly clumped chromatin,giving a clock-face or tessellated appearance (Fig. 1).We have not observed this in the lymphocytes inother conditions.

MALIGNANT LYMPHOMA, LYMPHOPLASMA-

CYTOID (IMMUNOCYTIC)There were three cases; a woman aged 68 yr present-ing with lymphomatous involvement of the pleura;a woman aged 40 yr with a tumour arising in thepelvis, peritoneal effusion, and involvement of kid-neys, lungs and lymph nodes; and a man aged 71 yrwith a tumour of the ileum and widespread mesen-teric and retroperitoneal involvement, and pleuraland peritoneal effusions.

All three show a distinctive cytological picturewith numerous lymphoplasmacytoid (LP) cells. Inthe case with a pelvic tumour these formed nearly100% of the population in the ascitic fluid, butdiffered little from cells which can be seen in benignconditions, except that there was a deficit of the moremature forms. In the other woman, there wasrather more maturation, and there were associatedsmall lymphocytes (Plate la). In the man withlymphoma of the small bowel, LP cells accounted

Table 2 Distribution ofnon-Hodgkin's lymphoma

Type Present series with Surgical node biopsy (Lennert 1978)' Oxford patients July 75-Feb 79effusions-87 cases 2215 cases* 211 cases

Low-gradeLymphocytic 6-9 18 6 15 6Immunocytic(a) Lymphoplasmacytoid 34 5-7 16 7 8-9(b) Plasmacytic 23fCentroblastic/centrocytic 18 4 28-7 29 5 36-9Centrocytic 103f

High-gradeCentroblastic 9-2 5-0 3-4Lymphoblastic(a) non-T 8-O4(b)T 57f 7 110 45Immunoblastic 10-3 9 3 10-1

True histiocytic 2-3 (n.i.) 5-0Unclassified 23-0 9-8 15 6

*Derived from Lennert's Table 13.'

1312

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/

Cvtological diagnosis of lymphoma in serous effusions

Fig. 1 Malignant lymphoma, lymphocytic. Cells frompleural fluid. Note the abnormally discrete blocks ofchromatin. May-Grunwald-Giemsa x 1300.

for only about 40% of the cells, and included somemature forms; these were not originally reported as

being neoplastic, though "immature" forms wereunusually numerous.A feature which might be useful in diagnosis is

that mitotic activity was very low in all three cases.This is in contrast to the behaviour of LP cells inbenign effusions, where they usually show frequentmitotic figures (Plate Ib; also see Spriggs & Bod-dington5 Fig. 20). Numerous LP cells were alsofound in pleural and peritoneal effusions of anothercase in our series, with a centrocytic lymphoma ofthe stomach; and here too the (presumably reactive)LP cells showed numerous mitoses.

PLASMACYTOMAThere were two cases of plasmacytoma with pleuraleffusion. In one of these, a woman of 45 yr admittedfor dialysis on account of myeloma kidney, thepredominant cells in pleural fluid were myelomacells (Fig. 2). In the other, a man of 70 yr with known

myelomatosis and an encysted pleural effusion,myeloma cells were found in small numbers amongabundant red cells and other nucleated cells (inclu-ding a few myelocytes and normoblasts); but thediagnosis could not have been made on cytologicalgrounds.

MALIGNANT LYMPHOMA, CENTROBLASTIC/CENTROCYTICAs expected, the low grade follicle centre tumourswere a common cause of effusions, exceeding thosedue to Hodgkin's disease. There were 16 in the series,seven men (aged 45-66 yr) and nine women (aged32-73 yr). Twelve had pleural and three peritonealeffusions only, and one had both. One peritoneal andfour pleural fluids were chylous.

In four cases no cells could be found which couldbe confidently identified as neoplastic centrocytes orcentroblasts. In one of these, the fluid was full of cho-lesterol crystals, and the cells were unrecognisable.In the others the predominant cells were small lym-phocytes, and one had many mesothelial cells,macrophages and eosinophils as well.

All but one of the remaining cases showed to agreater or lesser extent a picture which we havelearnt to recognise, and which in typical form ischaracteristic of centroblastic/centrocytic lymphoma.The identifiable lymphoma cells did not account fora high proportion of the cell population; usuallythere was a predominance of small lymphocytes, andin six cases there were also substantial numbers ofmesothelial cells and macrophages. Eosinophils werenot a feature.The usual picture is as follows, as seen in Roman-

owsky-stained smears (Plate Ic, d). Besides normal-looking small lymphocytes, there are slightly largerones (10-14 ,um diameter) with condensed nuclearchromatin, not fine as in a "blast cell," and nucleoliare invisible or quite small. The nuclear outline isirregular or cleft. The cytoplasm forms a verynarrow rim, and is no more basophilic than that ofan ordinary small lymphocyte.

These smaller forms correspond to the neoplasticgerminal centre cells illustrated by Moeschlin,6 or tothe "haematogones" of Rosenthal,7 and are evidentlyneoplastic centrocytes.8 Their recognition depends onthe quality of the smear, and since the lymphocytesof effusions are in any case quite variable, they areseldom obvious enough to be diagnostic by them-selves. The presence in addition of larger forms, withdiameters up to 20 ,tm, but with similar nuclear andcytoplasmic features, makes the diagnosis highlyprobable. All these cells may show slight positivegranulation with PAS.The cells described above are not the same as the

normal centrocytes seen in imprints from reactive

1313

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/

Spriggs, Vanhegan

i 11''I

IC Id

1314

I

p-

Aial& .1

.0,;:

I.. -1 ."I.-mft..:

cg. I

L& 1. .."';.ar

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/

Cytological diagnosis of lymphoma in serous effusions

Plate la Malignant lymphomna, lymphoplasn-acytoid(immunocytic). Cells from pleural effusion. A mixedpopulation of lymphocytes and lymphoplasmacytoid (LP)cells, some of the larger ones almost classifiable asimmunoblasts. May-Gruinwald-Giemsa x 750.Plate lb Benign lymphoplasmacytoid reaction in pkeuralfluid, in a case ofsquamous carcinoma of the lung. Nomalignant cells were seen. The two largest cells in lowerframe are immunoblasts. May-Grunwald-Giemsa x 750.Plate Ic Malignant lymphoma, centroblastic/centrocytic.Cells in peritoneal fluid. Some of the small "lymphocytes"have cleft nuclei and reduced cytoplasm. Two largelymphoid cells are clearly neoplastic; one has extremelyreduced cytoplasm, and both show large irregularnucleoli; they may be considered intermediate betweencentroblast and centrocyte. May-Grunwald-Giemsa x750.Plate Id Malignant lymphoma, centroblastic/centrocytic.Cells in pleural fluid. The small lymphocytes areindistinguishable from normal. Larger cells with identicalnuclear chromatin have irregularly cleft nuclei and littlecytoplasm (neoplastic centrocytes). The largest cell is acentroblast. Macrophages and neutrophils also present.May-Grunwald-Giemsa x 750.

Fig. 2 Malignant lymphoma, plasmacytic. Pleuralfluid, in which a high proportion of the cells presentwere myeloma cells. May-Grunwald-Giemsa x 1300.

lymph nodes. We have called them "neoplasticcentrocytes" to include the small "haematogones"and the larger cells sometimes called "lymphosar-coma cells" in the context of "lymphosarcoma-cellleukaemia."

Neoplastic centroblasts have also been seen inthese smears, but they were only frequent in onecase. They differed from large neoplastic centro-cytes in having finer chromatin with obvious nucleoli,and some of them had more profuse cytoplasm(Plate Id).One case differed clearly from the rest; a woman

aged 35 yr, who two years before had had lymphnode biopsies showing centroblastic/centrocyticlymphoma. The disease evidently entered a moreaggressive phase, and a rapidly developing peritonealeffusion showed a monotonous picture of lymphomacells, many of them in varying stages of necrosis.They mostly measured 14 ,um or less in diameter,and the nuclei were slightly irregular or notched. Thecytoplasm was scanty and pale staining, and withPAS showed fine granular positivity. These cells canbe accepted as malignant centrocytes, and presum-ably the ultimate diagnosis would have been centro-cytic lymphoma, but no biopsy was taken at thatstage. A concomitant pleural effusion contained smallnumbers of the same cells, but associated withnumerous mesothelial cells, macrophages, neutro-phils and lymphocytes.

In five cases a remarkable cytophagocyticphenomenon was seen. Compound structures,which at low magnification looked like multi-nucleate giant cells, in fact consisted of lymphomacells (or lymphocytes) crammed into the cytoplasmof another cell, presumably a macrophage. (Thisphenomenon has been described previously fromserous effusions by Murad9 and is illustrated byHajdu & Hajdu10.) It is common for macrophages ineffusions to contain the remains of digested cells,but in these lymphoma cases many, or most, of theingested lymphoid cells appear well preserved.

There are various descriptions in the experimentalliterature of the presence, real or apparent, of livinglymphocytes within the cytoplasm of culturedcells,"-13 and where macrophages are concerned itseems likely that some interaction of immunologicalsignificance is taking place. Histiocytes full of lympho-cytes are also a distinctive feature of the lymph nodesin the benign condition known as "sinus histio-cytosis with massive lymphadenopathy.'"14

Besides these four cases of centroblastic/centro-cytic lymphoma, the same phenomenon was seen inserous fluid from one case of diffuse centrocyticlymphoma (Fig. 3, & Plate hIa), one of T lympho-blastic lymphoma, and one of an unclassifiedlymphoma (possibly Lennert's lymphoepithelioid cell

1315

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/

Spriggs, Vanhegan

aI

Nrgw..W.,A........I.F wo -W-

*w5s wt~~~~~._.gll_~~AW

6;..

XLs~

L..AmA.'

i.kil ..r'.

*.S

...I

pp

.9

A.M *S.

IlaaI

a a-

*:i

t'-

'IVa

t ._

I$" -

A:

:A4

:tS.

,A.j ::

I.p* pIi

a.4

W.S

IF*4It

IIc

1316

AkL

A ...

* .4

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/


Fig. 3 Malignant lymphoma,centrocytic. Electronmicrograph showing part ofamacrophage with ingestedlymphoid cells. Compare PlatelIa x 9600.

Pt,6 It A W i

[email protected].. # \;was-n1B~~~~~~~~AL- S S ,¢ ¢ f * ;g 4 . ix axv X j4e Xd.

lymphoma). We have not observed it in any serouseffusion apart from lymphoma and it is therefore ofpositive diagnostic usefulness. The above findingsshow that it is not specific for any single type, evenif it occurs most commonly in follicle centre lym-phomas.

Plate IIa Malignant lymphoma, centrocytic. Thispleural fluid contained many compound structuresconsisting of lymphoid cells within macrophage cytoplasm.Same case as Fig. 3. Papanicolaou x 750.Plate Ilb Malignant lymphoma, centroblastic. Pleuralfluid deposit showing a high proportion ofmalignantcentroblasts. This was the terminal stage ofacentroblastic-centrocytic lymphoma. May-Grunwald-Giemsa x 750.Plate IIc T lymphoblastic lymphoma. The cells inpleural fluid have the usual features of lymphoblasts asseen in the blood in acute lymphoblastic leukaemia.Their "convoluted" nuclei are packaged into more orless spherical shape by the thin skin of cytoplasm.May-Grunwald-Giemsa x 750.Plate Ild T lymphoblastic lymphoma. Same materialas IIc, stainedfor acid phosphatase. The cells withdiffuse cytoplasmic staining are eosinophil granulocytes.The lymphoma cells show a localised positive reaction.May-Grunwald-Giemsa x 750.

MALIGNANT LYMPHOMA, CENTROCYTICAlthough this is included in Lennert's2 classificationas a low grade lymphoma, he admits that the centro-cytes are "somewhat more anaplastic" than incentroblastic/centrocytic lymphoma, and the prog-nosis poorer; its leukaemic variant is the so-called"lymphosarcoma cell leukaemia."There were nine cases in our series, six women

(37-85 yr) and three men (59-78 yr). Six had pleuraleffusion alone (two of them chylous), two hadascites alone (one chylous), and one had pleural andperitoneal effusions.

Judging by the cytology of the fluids they are aheterogeneous group, perhaps depending on whetheror not the serosal surface has been invaded byneoplastic cells.

In four cases there was nothing cytologicallydistinctive of lymphoma. Three of these showednumerous lymphocytes, and the fourth, a woman of31 yr with a lymphoma of the stomach, had inaddition large numbers of LP cells in both pleuraland peritoneal fluid. These were of normal mor-phology and had a high mitotic index-a findingwhich is seen in benign effusions where immuno-cytes are numerous, but was not seen in thethree genuine cases of immunocytic lymphoma.

1317

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/

Spriggs, Vanhegan

Although at the time (1954) they were thought to belymphoma cells, we do not think so now.The remaining five cases showed recognisable

lymphoma cells, but they were not entirely consistentin type. In one of them, a chylous pleural effusiondeveloping in a 59-year-old man, almost the onlynucleated cells found were haematogone-like cellshardly larger than normal lymphocytes-that is,about 10 Ztm in diameter. The same cells wereabundant in his bone marrow. In another casethere were ordinary lymphocytes about 6-8 ,um indiameter, and lymphoma cells measuring 10-14 pLm,of the type described above. In another there was asimilar picture, but with the addition of manynormal LP cells. In another with chylous ascitesfrom a lymphoma of the bowel, there was a mixedcellular picture including LP cells, and small num-bers of cells similar to lymphocytes but about 16,um across, presumed to be neoplastic centrocytes.

The last case was mentioned before in connectionwith the phenomenon of phagocytosis of lymphomacells in centroblastic/centrocytic lymphoma. Awoman aged 85 yr had an illness with a course ofonly four months, with a mass in the left hilum andhepatosplenomegaly. At necropsy there was diffuseinfiltration of lymph nodes and other organs by smalllymphoma cells thought probably to be centrocytes.The pleural fluid showed a picture like that describedabove for centroblastic/centrocytic lymphoma; therewere normal lymphocytes, macrophages and meso-thelial cells, neoplastic centrocytes varying in sizefrom "haematogones" 8 ,um across up to formsmeasuring 20 ,um; and there were numerous clustersof lymphoma cells or lymphocytes, or both, packagedwithin the cytoplasm of a macrophage as describedabove (Plate Ila and Fig. 3).From this series of centrocytic lymphomas it

would appear that a diagnosis of lymphoma frompleural fluid cytology is possible in some but not allcases, and that the cells could probably be recognisedas of follicle centre origin; but evidently the sub-classification into "centrocytic" lymphoma of inter-mediate aggressiveness, as distinct from the low-grade "centroblastic/centrocytic" lymphoma, wouldbe unreliable on cytological grounds, except perhapsin the leukaemic form.

MALIGNANT LYMPHOMA, CENTROBLASTICThis high-grade lymphoma is often the end-stage ofa chronic low-grade lymphoma of centroblastic/centrocytic type. It is represented by eight cases inthis series-two men (44 and 69 yr) and six women(36 to 72 yr). Seven had pleural effusions only(one chylous), and one had chylous peritoneal andpleural effusions.

In all of these cases the cellular deposit containeda high proportion of neoplastic cells and all werecorrectly called malignant lymphoma in the originalreport. Four of them showed cells which, in thelight of this study, might be recognised again asneoplastic centroblasts (Plate IIb).The cell diameter is 18-30 ,um. The nucleus is more

or less irregular in shape, with fine chromatin anddistinct nucleoli (usually two or more). The cyto-plasm is moderately or only slightly basophilic, lessthan that of most immunoblasts, and does not showthe "Hof" or pale Golgi area at the cell centre whichdistinguishes plasma cells and many immunoblasts.Small lymphoid cells (12 ,um diam) which could beneoplastic centrocytes were sometimes present aswell.

In two cases there was variation from this picture.One had mainly small lymphocyte-like cells, 10-18,um in diameter but mostly 12-14 ,um, resemblingthe cells of centrocytic lymphoma except in thepossession of visible nucleoli and rather morecytoplasm. In the other case there was a "pureculture" of cells measuring 12-22 ,um, mostlyoccupied by a nucleus with very fine chromatin andprominent nucleoli; the cytoplasm deeply basophilicand full of fat granules, as in classical Burkittlymphoma cells.15 This was the end stage of afollicular lymphoma of nine years duration.

MALIGNANT LYMPHOMA, LYMPHOBLASTIC(NON-T)This category includes seven cases, four women(57-73 yr) and three men aged 28, 42 and 76 yr. Fivehad pleural effusions and three peritoneal. Two ofthese patients had acute lymphoblastic leukaemia,but the rest were not recorded as leukaemic.

In all cases the lymphoma cells accounted for ahigh proportion of the nucleated cells present. Intwo cases, both with peritoneal effusions (one chy-lous), a large number of the cells were necrotic,although no cytotoxic treatment had been given.This phenomenon was also observed in theimmunoblastic lymphomas.

In six of the cases the lymphoma cells were small,10-16 ,um in diameter, accounted for almost entirelyby nucleus, the cytoplasm being extremely reduced.Usually the nucleus had clefts or folds; the chromatinwas fine, but in some cases less fine than in classical"blast cells." Nucleoli were only visible in two casesfrom this group.One case differed from the rest in having larger

cells, mostly 14-20 ,um in diameter, with a coarsernuclear chromatin, prominent nucleoli, and baso-philic cytoplasm. In these features they resembledimmunocytes or small immunoblasts. Histologicallyalso, there were areas which looked immunocytic

1318

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/


rather than lymphoblastic, and this may be a B cellneoplasm of intermediate type.

In all cases the cytoplasm was PAS negative.

MALIGNANT LYMPHOMA, T LYMPHOBLASTICFive cases with pleural effusions can be assigned tothis group. One was a man of 68 yr with a tumourarising in the nasopharynx, one a woman of 24 yrwith pleural effusion as the main and presentingsymptom, and three were women, aged 13, 18 and29 yr, with mediastinal lymphomas. Four of thesecases had marrow involvement; one of these is alivefive years after beginning successful chemotherapyand has borne a healthy child.

In two cases the great majority of the nucleatedcells present were lymphoma cells but in one therewere scattered eosinophils. In a third case lympho-blasts were mixed with smaller numbers of neutro-phils and macrophages. In a fourth, the survivor,lymphoblasts were scattered among much largernumbers of lymphocytes, macrophages, mesothelialcells and neutrophils. The fifth case, the girl of 13 yr,showed to a striking degree the phenomenon ofJymphophagocytosis, many of the ingested lymphoidcells being in various stages of digestion.

In all cases the lymphoma cells were of similartype. The diameter ranged from 8 to 20 ,um, accoun-ted for almost entirely by nucleus, the cytoplasmexisting as a very narrow skin visible mainly atpoints where the nucleus was indented. The chroma-tin was usually less fine than in a classical "lympho-blast" and nucleoli were invisible (Plate lIc). The"convoluted" shape of the nuclei was not obviousin air-dried smears, but could be seen in wet-fixedPapanicolaou smears; nevertheless lymphoma cellsof other types so often have irregularly shapednuclei that this feature is quite non-specific.Acid phosphatase stain was done in two of the

five cases, and showed the classical strongly positivespot (Plate Ild).'6 17

MALIGNANT LYMPHOMA, IMMUNOBLASTICNine cases were placed in this category-two men(49 and 54 yr), and seven women (41 to 68 yr). Sixhad pleural effusions and three peritoneal; none ofthem was chylous.Two of these effusions presented a mixed cyto-

ogical picture with lymphocytes, neutrophils,macrophages, mesothelial cells and in one caseeosinophils, but no neoplastic cells were recognised.In one of these there were frequent LP cells but theyappeared normal.The remaining seven cases all had recognisable

malignant cells. In three they were outnumbered byother cell types, while in four they formed a veryhigh percentage of the cells present (Plate lila). In

1319

three of the latter there were numerous dead tumourcells. This picture, with a few surviving lymphomacells amongabundant necrotic ones, is very character-istic and has been previously described.5 10 18 19 Inone of these cases immunoblastic lymphomasupervened after two years of chemotherapy formyelomatosis (Plate IlIb).Neoplastic immunoblasts in the above cases hadcertain features in common as follows:Most cells measure 15-22 .tm in diameter. The

nucleus has a fine chromatin pattern. Its outline isusually rounded but can be of very irregular shape,and it contains one or more large irregular nucleoli.The cytoplasm varies greatly in amount but isnearly always basophilic, sometimes intensely so,and may be full of fat granules. A rather distinctivefeature is the pale Golgi area at the cell centre, andin forms with well developed cytoplasm this causesa distinct resemblance to an "immature" plasmacell (Plate Illa). In two cases this feature was notpresent, and the distinction from centroblastic lym-phoma was not possible on cytological grounds.PAS was variable, sometimes negative but in somecases quite strongly granular positive.

It must be emphasised that each of these highlymalignant neoplasms is unique, and the appearanceof the cells is distinctive for that case only. Resem-blances between the cells of different cases are muchvaguer and there is no "typical" neoplastic immuno-blast.

MALIGNANT HISTIOCYTOSISThere were two cases of true histiocytic neoplasm.One was a woman of 77 yr who developed purpura,developing into multiple cutaneous nodules; withina few months she developed pleural effusion, anddied three months later. The other was a man of 27 yrwith dysphagia and Horner's syndrome, due tolymphadenopathy of the neck and mediastinum.He developed a massive bloody pleural effusion anddied soon after.

In the first case, the pleural fluid contained redcells with many neoplastic cells and a few lympho-cytes. The neoplastic cells ranged from monoblast-like forms 22-26 ,tm across, occupied mainly bynucleus, to macrophage-like types with profusefoamy cytoplasm, measuring 30-40 ,tm in diameter.The former had multiple irregular nucleoli, usuallyindistinct. Some of them contained inclusionsderived from cell debris, and occasionally a recognis-able leucocyte, but ingested red cells were rarely seen(Plate Illc). Acid phosphatase stain was diffuselypositive.

In the second case, besides spherocytic red cellsthere were scattered malignant cells, most of themnecrotic. They varied greatly in size, shape and

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/

1320 Spriggs, Vanhegan

i..lj

Illb

*... rSM

Illd

#:. :e* i

-F

lllc

I 0Astak

..I- j

.: I

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/


Plate Illa Malignant lymphoma, immunoblastic. Cellsin pleural fluid. Note the one or two large nucleoli, thelarge pale Golgi area in an otherwise basophiliccytoplasm, and the presence offorms with eccentricnuclei. May-Grurnwald-Giemsa x 750.Plate Illb Malignant lymphoma, immunoblastic. Inthis case the majority of the cells are dead, withkaryorrhexis and pyknosis. This appearance is verycharacteristic of high-grade lymphoma. x 750.Plate IlIc Malignant histiocytosis. This pleural fluidshows immature-looking histiocytes together withrelatedforms containing phagocytic vacuoles. May-Grunwald-Giemsa x 750.Plate Illd Lymphocyte-depleted Hodgkin's disease. Thepleural fluid contained a mixture of benign cells as well asnumerous cells of Sternberg-Reed type as shown hereat centre. May-Grunwald-Giemsa x 750.

nucleo-cytoplasmic ratio, but typically were 25 ,umin diameter with an irregular nucleus containingmultiple small nucleoli. The cytoplasm was fairlyprofuse and foamy, staining grey-blue. A few ofthem were phagocytic to debris. Acid phosphatasestaining was not done in this case.

In both cases the cytoplasm contained fat andsome granular PAS-positive material

HODGKIN'S DISEASEThere were 13 cases of Hodgkin's disease. Thebreakdown of these is as follows:Lymphocytic predominance 4; nodular sclerosing

4; mixed cellularity 3: lymphocyte depleted 2.There were six men and seven women among the

cases of Hodgkin's disease, and they ranged in agefrom 18 to 71 yr; 10 had pleural effusion alone, oneperitoneal alone, and two had both.

Lymphocytic predominanceTwo patients had empyema, one due to Streptococcusfaecalis (following a bowel perforation), and theother to a probable bronchopleural fistula. Thelater case showed some cells of Sternberg-Reed typeamong many pus cells.Two patients had a mixed cellular picture in

pleural fluid, with a predominance of macrophages.One showed no lymphoma cells, but the other had afew large "Hodgkin cells" in two successive pleuralsamples, as well as some eosinophils. A month laterthe cell picture had changed; there was a predomi-nance of small lymphocytes together with cellslooking like small centrocytes and "blast cells," andfairly frequent cells interpreted as Hodgkin cells;they were about 20 ,um in diameter, occupied mainlyby a pale nucleus with fine chromatin and sometimesthe outline of a nucleolus; the cytoplasm forming amoderately basophilic rim. There were also frequenteosinophils. No Sternberg-Reed cells were seen.

1321

Noduilar sclerosingIn all four pleural effusions the predominant cellswere lymphocytes, and all had LP cells, macro-phages and mesothelial cells present as well. Twoof them also contained a very few cells which mightbe Sternberg-Reed cells, but not definite enough fordiagnosis; they were up to 30 ,m across, bilobed andwithout cytoplasmic vacuolation, and consequentlydid not resemble "lacunar cells."

Mixed cellularityOf these three pleural effusions, none contained anyrecognisable lymphoma cells. All showed a mixedcell picture with lymphocytes, mesothelial cells,macrophages, and neutrophils.

Lymphocyte depletedOf two cases, one had no recognisable lymphomacells in two separate samples of pleural fluid. Theother, a woman of 63 yr, had clearly neoplastic cellsin pericardial as well as pleural effusions.The pericardial fluid contained abundant eosino-

phils, together with lymphoma cells varying indiameter from 16 ,um up to 50 ,um in the largeSternberg-Reed types. The majority measured about25 ,um, with an oval nucleus showing fine or spongychromatin, and a very large nucleolus-for example,10 ,m across. The cytoplasm was fairly profuse,slightly to deeply basophilic and without any paleGolgi area. The Sternberg-Reed cells were of theabove type, but with divided nuclei showing amirror-image or butterfly configuration (Plate IlId).The pleural fluid contained numerous similar

cells, associated with macrophages, lymphocytes,neutrophils, and a few mesothelial cells and eosino-phils." FALSE POSITIVEIn the 30-year period covered, there was only onecase in which lymphoma was diagnosed on cyto-logical grounds, but in which confirmation waslacking and the patient recovered. This case has beenbriefly described and illustrated.5 The patient wasan 80-year-old woman with a pleural effusion whichdid not recur after two aspirations, and required noother treatment. When she died 5 years later,necropsy showed no evidence of lymphoma or othermalignant disease. On review, the slides would stillbe reported to show a high-grade malignant lym-phoma; there was an almost uniform populationof abnormal lymphoid cells resembling thosedescribed from Burkitt's lymphoma,15 together withpyknotic remains of cells as seen frequently in high-grade lymphomas. Two samples showed this p;cturetwelve days apart. This might be an example ofspontaneous regression, as it is difficult to think ofanother rational explanation.

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/

Spriggs, Vanhegan

Discussion

Previous authors writing about lymphoma cells inserous effusions have belonged to one of threedifferent schools, each with its own methodology.First there are those who have used only the histo-logical method, and have examined sections ofcell-blocks made from fluid deposit.20-23 Others haveseen this as an extension of the cytological study ofblood, marrow, and puncture material from lymphnodes and spleen. They have used dried smearsfixed and stained with Giemsa or its analogues.5 19Finally, the influence of Papanicolaou after thesecond world war led to the development of a wholespeciality of "exfoliative cytology," almost exclu-sively using wet-fixed smears stained with haema-toxylin and Papanicolaou's counterstains, and thismethod has been applied to the identification oflymphoma cells in effusions' 8 24-26 Eosin has some-times been used as an alternative cytoplasmic stain.27The lymphoma cells occurring in effusions are,

of course, the same as those which would be foundin the same case in lymph node imprints or punc-tures, and sometimes in bone marrow and blood.Even for these it is impossible to refer to any singlework since the early sixties, offering a comprehensivecorrelation between cytology and histology. Jllustra-tions are scattered through various books andpapers.2 15 28-36The names given to the different cell types in all

of the currently recognised classifications are thoseused, and often introduced, by histopathologists,and the corresponding cells as seen in smears havein many cases not been clearly defined. There is afurther difficulty inherent in the very nature of malig-nant disease, that each individual neoplasm appears tobe unique. The cells in any given case differ amongthemselves "like the leaves on a tree" (Lewis),37 butall have a family resemblance specific to that clone.They can be rationally classified according to thedegree and direction in which they tend to differen-tiate, as shown by simple morphology or by varioushistochemical or immunological tests; but it is notto be expected that they will fall into exact categorieslike postage stamps.The cytopathologist is faced in practice with two

questions before the classification of lymphomasever arises. Firstly, are the cells in question benignlymphoid cells, and not neoplastic at all? Secondly,if the cells are neoplastic, are they from a lymphomaor from some other neoplastic condition (carcinoma,sarcoma, melanoma etc)?

BENIGN LYMPHOID CELLS V LYMPHOMAFirst, the cytopathologist has to be familiar with the

Fig. 4 Infectious mononucleosis. Cells in the depositfrom peritoneal fluid aspirated during laparoscopy.May-Grunwald-Giemsa x 650.

various appearances of reacting lymphoid cells suchas LP cells and immunoblasts. Apart from serousfluids, it is helpful to have experience of the cellsfound in blood smears in viral infections and inallergic reactions. Another excellent source ofhuman lymphoid cells is the cerebrospinal fluid inviral meningitis.

Stimulated lymphoid cells seldom account formore than 10% of the population in benign serousfluids. Hence neoplasia may be strongly suspectedif such cells account for a majority of the cellspresent. A rare exception to this was seen in a case ofinfectious mononucleosis with a peritoneal effusion.The patient was a woman of 27 yr, presenting withabdominal pain attributed to endometriosis. Peri-toneal fluid was aspirated during laparoscopy, andabout 90% of the cells present were atypical lympho-cytes ("glandular fever cells") with frequent mitoticfigures. The diagnosis was confirmed by bloodexamination and serology, the patient was well amonth later, and there has been no relapse. (Fig. 4).Transformed B cells are not uncommon in serous

1322

_....

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/


fluids, and when they are numerous they normallyshow striking mitotic activity. This is not a criterionof malignancy. If adequate direct chromosomalpreparations can be made, this can provide objectiveevidence whether the dividing cells are malignant ornot; cells with normal karyotypes are most probablybenign, and a clone with an abnormal karyotype is,in this context, presumed to be malignant. Thismethod is inapplicable if the cells are not dividing;in chronic lymphocytic leukaemia for instance,spontaneous mitosis is not seen in the "smalllymphocytes," and they do not respond to PHA.(This fact has even been used to distinguish themfrom normal lymphocytes.)38

Benign-looking stimulated lymphoid cells are acommon feature of effusions in Hodgkin's disease,and recognisable neoplastic cells were only found ina minority of our series (although Simecek39 sawneoplastic cells in 14 of 23 effusions).Most cytogenetic studies on lymph nodes in

Hodgkin's disease support the concept that theSternberg-Reed cells and mononuclear "Hodgkincells" are neoplastic, and that the other leucocytesare reactive. The morphology of the cells in effusionssuggests this too. However, there is some evidencethat this impression may not be correct. Hossfeldand Schmidt40 made direct chromosome preparationsfrom six effusions complicating Hodgkin's disease,of which only one contained identifiable Hodgkincells or Sternberg-Reed cells; but all of themshowed a majority of dividing cells to have grosslyabnormal karyotypes, which were clearly clonal.One had a modal count of 60, but the rest were nearto diploid (45-48) with extensive rearrangements andmarkers. If the dividing cells in these cases weremorphologically like benign LP cells (and this isnot stated), the straightforward inference is thatthese form the neoplastic stemline in Hodgkin'sdisease.

All these were advanced cases which had beentreated, and a conceivable alternative explanation isthat cytotoxic treatment had resulted in the emer-gence of new abnormal clones (as occurs in the casesof iatrogenic leukaemia), but this is not the inter-pretation of the authors. This is an interesting areafor further studies.

LYMPHOMA V CARCINOMA AND OTHER

NEOPLASMSMost cases of neoplastic effusion arise from epi-thelial tumours, and in these the malignant cells areusually coherent. Difficulty only arises when all themalignant cells are free, because lymphoma cellsdo not form cell to cell attachments.

In the occasional free-cell carcinoma, the distinc-tion from lymphoma may arise. Round nuclei with

one or two round nucleoli are in favour of carcinoma,while cleft or irregular nuclei with multiple irregularnucleoli favour lymphoma. Oat cells, when free andpresent in small numbers, may be very difficult todistinguish from malignant lymphoblasts or centro-cytes. One looks for clusters, and even one suchcluster with mutual nuclear moulding would ruleout lymphoma. Free carcinoma cells sometimesshow evidence of mucus secretion, or may showdense surface coverings of microvilli; both of theserule out lymphoma.Sarcomas are nearly always distinguished by the

clinical history. Amelanotic melanoma has occasion-ally been a problem, but these cells usually have avery large round nucleolus, which would be mostuncommon in a lymphoma cell.The presence of large numbers of dead tumour

cells showing pyknosis and karyorrhexis is strongevidence of malignant lymphoma rather than anyother type of neoplasm.26

CYTOLOGICAL IDENTIFICATION OF

SEPARATE LYMPHOMA TYPESThe subdivisions used here are believed to have abasis in reality, in that they represent neoplasmswhich are separable by immunological or histo-chemical methods, or both. However, the necessarybattery of tests (including those on cells in suspen-sion) was not applied to the cases in this series, andthe diagnosis has been made on histological mor-phology backed up in some cases by immuno-peroxidase staining. Quite apart from the limitationsresulting from this indirect identification, nobodywould claim at the present time that lymphomacells in every case can be fitted into existing classifi-cations, even using all the techniques currentlyavailable. Additional evidence from tissue architec-ture and cell distribution is certainly required, andeven then there may be disagreement about thehistological classification. We would not thereforesuggest that a diagnosis of lymphoma type canregularly be made on cytological evidence alone.

Having said this, we do claim that with fewexceptions the cytological picture has fitted well withthe histological one, and that a cytological report ona lymphoma effusion can reasonably include aprovisional opinion as to the grade of lymphoma(high or low) and a suggestion as to the direction ofdifferentiation ("cell of origin").

Well differentiated lymphomas of lymphocytic,immunocytic, and plasmacytic types can be readilydistinguished, provided that one can be confidentthat the relevant cells are neoplastic at all. Follicularlymphomas of centroblastic/centrocytic type some-times show a characteristic picture, depending on thepresence of neoplastic centrocytes with enough

1323

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/

1324

distinctive features to separate them from smalllymphocytes; the diagnosis is made easier if somecentroblasts are present too. Centrocytic lymphomacan only be separately distinguished if there is a

rather uniform population of centrocytes, and thisis most likely to be associated with a leukaemicpicture. Of the high-grade lymphomas, centroblasticand immunoblastic lymphomas consist of relativelylarge cells, but are probably not separable from eachother except in cases of immunoblastic lymphomawith some differentiation towards the plasma cell.Lymphoblastic lymphomas are separable as a group,

provided that the neoplastic cells are in the lowerpart of their size range. The T lymphoblasticlymphomas are histochemically distinct. Malignanthistiocytosis can be diagnosed if the cells are suffi-ciently differentiated to show phagocytosis, and togive reactions for lysosomal enzymes.

An effusion containing high-grade lymphomacells sometimes supervenes in a case being treated as

a low-grade lymphoma on the grounds of a previousbiopsy, and in such cases the cytological findingprovides essential supplementary evidence.

We are grateful to Mr MM Boddington, whocollaborated with Dr Al Spriggs during most of theperiod over which the cytological material wascollected.

References

Lennert K, Mohri N, Stein H, Kaiserling E. The histo-pathology of malignant lymphoma. Br J Haematol1975;31suppl: 193-203.

2 Lennert K. Malignant lymphomas other than Hodgkin'sdisease. Berlin: Springer, 1978.

3 Li CY, Yam LT, Lam KW. Acid phosphatase isoenzymein human leukocytes in normal and pathologic con-ditions. J Histochem Cytochem 1970;18:473-81.

4Strunge T. La ponction des ganglions lymphatiques. Copen-hagen: Munksgaard, 1944.

Spriggs Al, Boddington MM. The cytology of effusions,pleural, pericardial and peritoneal, and of cerebrospinalfluid. 2nd ed. London: W Heinemann, 1968.

Moeschlin S. Phasenkontrastuntersuchungen in derHaimatologie. Acta Haematol (Basel) 1949;2:399-426.

7Rosenthal N, Dreskin OH, Vural IL, Zak FG. Thesignificance of hematogones in blood, bone marrow andlymph node aspiration in giant follicular lympho-blastoma. Acta Haematol (Basel) 1952;8:368-77.

McKenna RW, Bloomfield CD, Brunning RD. Nodularlymphoma: bone marrow and blood manifestations.Cancer 1975;36:428-40.

9 Murad TM. Electron microscopic studies of cells in pleuraland peritoneal effusions. Acta Cytol 1973 ;17:401-9.

Hajdu SI, Hajdu EO. Cvtopathology ofsarcomas and othernonepithelial malignant tumors. Philadelphia: Saunders,1976.

Humble JG, Jayne WHW, Pulvertaft RJV. Biologicalinteraction between lymphocytes and other cells. Br JHaematol 1956;2:283-94.

Spriggs, Vanhegan12 loachim H, Furth J. lntrareticular cell multiplication of

leukemic lymphoblasts in thymic tissue cultures. J NatlCancer Inst 1964;32:339-59.

13 Lipsky PE, Rosenthal AS. Macrophage-lymphocyteinteraction. I Characteristics of the antigen-independentbinding of guinea pig thymocytes to syngeneic macro-phages. J Exp Med 1973 ;138:900-24.

14 Rosai J, Dorfman RF. Sinus histiocytosis with massivelymphadenopathy: a pseudolymphomatous benigndisorder. Cancer 1972 ;30:1 174-88.

15 Berard C, O'Conor GT, Thomas LB, Torloni H. Histo-pathological definition of Burkitt's tumour. Bull WHO1969;40:601-7.

16 Catovsky D, Galetto J, Okos A, Miliani E, Galton DAG.Cytochemical profile of B and T leukaemic lymphocyteswith special reference to acute lymphoblastic leukaemia.J Clin Pathol 1974;27:767-71.

17 McKenna RW, Brynes RK, Nesbit ME et al. Cytochemicalprofiles in acute lymphoblastic leukemia. Am J PediatHematol/Oncol 1979;1 :263-75.

18 Melamed MR. The cytological presentation of malignantlymphomas and related diseases in effusions. Cancer1963 ;16:41 3-3 1.

19 Lopes Cardozo P. Atlas of clinical cytology. s'Herto-genbosch: Lopes Cardozo, nd, 1976.

20 Wihman G. A contribution to the knowledge of the cellularcontent in exudates and transudates. Acta Med Scand1948;suppl 205.

21 Dawe CJ, Woolner LB, Parkhill EM, McDonald JR.Cytologic studies of sputum, secretions and serousfluids in malignant lymphoma. AmJ ClinPathol 1955;25:480-8.

22 Luse SA, Reagan JW. A histocytological study of effusions.II Effusions associated with malignant tumors. Cancer1954;7:1167-81.

23 Luse SA, Reagan JW. A histocytologic and electron-microscopic study of effusions associated with malignantdisease. Ann N YAcad Sci 1955-6;63:1331-47.

24 Johnson WD. The cytological diagnosis of cancer inserous effusions. Acta Cytol 1966;10:161-72.

25 Billingham ME, Rawlinson DG, Berry PF, Kempson RL.The cytodiagnosis of malignant lymphomas andHodgkin's disease in cerebrospinal, pleural and asciticfluids. Acta Cytol 1975;19:547-56.

26 Koss L. Diagnostic cytology and its histopathologic bases.3rd ed. Philadelphia: Lippincott, 1979.

2 Woyke S, Domagala W. Cytological evaluation of fluidsfrom body cavities. III Malignant lymphoma cells(in Polish). Nowotwory 1971 ;21 :31-6.

28 Lennert K. Ober die Erkennung von Keimzentrumszellenim Lymphknotenausstrich. Klin Wochenschr 1957 ;35:1130-2.

29 Leiber B. Der menschliche Lymphknoten. Munich: Urban& Schwarzenberg, 1961.

30 Lennert K. Lymphknoten. Diagnostik in Schnitt undAusstrich. In: Lubarsch 0, Henke F, Rossle R, UehlingerE, eds. Handbuch der speziellen pathologischen Anatomieund Histologie. Vol Al, part 3. Berlin: Springer, 1961.

31 Math6 G, Seman G. Aspects histologiques et cytologiquesdes leuc6mies et hOmatosarcomes. Paris: Maloine, 1963.

32 Dick F, Bloomfield CD, Brunning RD. Incidence, cytologyand histopathology of norn-Hodgkin's lymphomas in thebone marrow. Cancer 1974 ;33:1382-98.

33 Math6 G, Pouillart P, Schlumberger JR, Paintrand M.Cytology in the classification of diffuse non-leukaemicmalignant lymphomata (lympho- and reticulosarcomata).BrJ Cancer 1975 ;31suppl 11:53-9.

34 Pinkel D, Johnson W, Aur RJA. Non-Hodgkins lymphomain children. Br J Cancer 1975; 31 suppl. II: 298-323.

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/


35 Rilke F, Pilotti S, Carbone A, Lombardi L. Morphology oflymphatic cells and of their derived tumours. J ClinPathol 1978 ;31:1009-56.

36 Feinberg MR, Bhaskar AG, Bourne P. Differentialdiagnosis of malignant lymphomas by imprint cytology.Acta Cytol 1980;24:16-25.

3 Lewis WH. Malignant cells. Harvey lectures, 16 April1936:214-34.

3d Boccato P, Saran B, Pasini L, Pasini P. Immunology oflymphocytes in pleurisy and in effusions due to pleuralinfiltration by chronic lymphocytic leukemia cells.

1325

Acta Cytol 1978;22:284-5.31gimecek C. Zytologie der Pleuraergusse bei den Sarkomen

des lymphoretikularen Gewebes. Arch Geschuvulstforsch1977;47:197-203.

40 Hossfeld DK, Schmidt CG. Chromosome findings ineffusions from patients with Hodgkin's disease. Int JCancer 1978 ;21 :147-56.

Requests for reprints to: Dr Al Spriggs, Laboratory ofClinical Cytology, The Churchill Hospital, Headington.Oxford OX3 7LJ, England.

copyright. on M


http://jcp.bmj.com

/J C


ecember 1981. D

ownloaded from

http://jcp.bmj.com/

cytological diagnosis of lymphoma in serous effusions · cases an almost pure population of...

Documents