Correspondence

Collagenous spherulosis versus shadow celldifferentiation in endometrioidadenocarcinoma

Sir: Treilleux and associates1 recently described endo-metrioid adenocarcinoma with spheruloid structures,which they interpreted as collagenous spherulosis. Inour opinion, these structures represent the so-calledshadow cells, i.e. squamous cells with lysed nuclei andwith corni®cation of the hair matrix type.2 Shadow celldifferentiation is well known in pilomatrixoma and skintumours.2 However, this differentiation may occur insome visceral adenocarcinomas.3,4 In our experience,of all visceral carcinomas, shadow cell differentiationcan be observed most frequently in endometrioidadenocarcinomas.

The reasons for our different interpretation of thestructures described by Treilleux et al. are:(1) The collagenous nature of the structures was not

demonstrated persuasively enough. The PAS posi-tivity and the presence of the collagen IV on theperiphery of the structures may re¯ect the presenceof basement membrane components only or theirremnants as can be seen around many necrobioticcells. The collagen IV inside the structures stainedonly weakly and rather unconvincingly. Simplemethods for highlighting collagenous spherulesare van Gieson and Masson's trichrome stain,which should stain the ®brillar component of thespherule because of its collagenous contents.5,6 Areticulin stain can outline the ®brillary architectureof the spherule, too. These stains were notmentioned in the paper.

(2) A weak, eosinophilic-to-pale orange (saffron) colourof the structures is more consistent with shadowcells than collagenous spherules,2±4 because col-lagenous spherules are usually more eosinophilic asa consequence of their collagenous contents.5,6

Fibrillar appearance is often apparent in thecytoplasm of shadow cells. It can be explained bythe ultrastructural ®nding of strands and bundles ofalpha-keratin ®brils surrounding the empty nucleararea of the shadow cell.7

(3) The centres of the collagenous spherules usuallystain well with routine staining methods, which iscaused by a presence of amorphous or ®brillarmaterial at the ultrastructural level.8 The structuresshown in the paper show, however, quite emptycentres (especially in Figure 1b), which is a featurevery typical of shadow cells.2±4

(4) We are not aware of any report in the literaturewhere collagenous spherulosis has ever beendescribed to occur in association with squamouscells.5,6 On the contrary, such association withsquamous cells is quite typical for shadow celldifferentiation.3,4 Treilleux et al. pictured cellswhich had nuclei with pyknotic and degenerativeappearance merging with the squamous cells(Figures 1a and 2 of the report). In our opinion,these pyknotic cells at the vicinity of the squamouscells represent a transitional stage between squam-ous and shadow cells. Well-formed shadow cellshave an empty nuclear area as shown in thepictures of Treilleux et al. which evolve aftercomplete lysis of the nuclei of pyknotic cells in thevicinity of the squamous cells.

(5) Foreign body giant cell reaction is often seenin tumours with shadow cell differentiation2±4

and it has never been described in collagenousspherulosis.5,6

(6) Immunohistochernical negativity for KLI, cyto-keratin 19 and EMA is quite commonly seen inshadow cells in our experience, because the shadowcells represent `burned-out' structures lacking anyimmunoreactivity. The immunohistochernicalnegativity for KL1, cytokeratin 19 and EMA doesnot exclude an epithelial origin of these structures.

(7) Actin immunoreactivity around the discussedstructures may indicate a stromal reaction ofmyo®broblasts and not a collagenous spherulosis-associated myoepithelial differentiation. Focal reac-tivity for S100 protein is often seen in epithelialneoplasms and it represents dendritic cells.9 Othermarker of myoepithelial differentiation, such asGFAP, give negative results.

We therefore believe that the structures described byTreilleux et al. represent shadow cell differentiation inendometrioid adenocarcinoma, and that they should bedesignated `collagenous spherulosis-like'.

M ZamecnikM Michal1

P Mukensnabl1

Department of Pathology,Slovak Postgraduate Academy of Medicine,

Bratislava, Slovak Republic

1Sikl's Department of Pathology,Faculty Hospital,

Pilsen, Czech Republic

Histopathology 2000, 36, 470±478

q 2000 Blackwell Science Limited.

1. Treilleux I, Godeneche J, Duvillard P, Clement-Chassagne C,

Suignard Y, Bailly C. Collagenous spherulosis mimicking keratiniz-

ing squamous metaplasia in a borderline tumour of the ovary.Histopathology 1999; 35; 271±276.

2. Ackerinan AB, De Viragh PA, Chongchitnant N. Neoplasms with

Follicular Differentiation. Lea & Febiger, Philadelphia, 1993.

3. Zamecnik M, Michal M. Shadow cell differentiation in tumours ofthe colon and uterus. Zentralbl Pathol. 1994/95; 140; 421±426.

4. Zamecnik M, Michal M, Mukensnabl P. Pilomatrixoma-like visceral

carcinomas (letter). Histopathology 1998; 33; 395.

5. Clement PB, Young RH, Azzopardi JG. Collagenous spherulosis ofthe breast. Am. J. Surg. Pathol. 1987; 11; 411±417.

6. Skalova A, Leivo I. Extracellular collagenous spherules in salivary

gland tumors. Immunohistochernical analysis of Imninin and varioustypes of collagen. Arch. Pathol. Lab. Med. 1992; 116; 649±653.

7. Hashimoto K, Nelson RG, Lever WF. Calcifying epithelioma of

Malherbe. Histochemical and electron microscopic studies. J. Invest.

Dermatol. 1966; 46; 391±408.8. Maluf HM, Koemer FC, Dickersin GR. Collagenous spherulosis: an

ultrastructural study. Otrastruct. Pathol. 1998; 22; 239±248.

9. Oka K, Nakano T, Arai T. Adenocarcinoma of the cervix treated

with radiation alone: prognostic signi®cance of S-100 protein andvimentin immunostaining. Obstet. Gynecol. 1992; 79; 347±350.

Authors' reply

Sir: We still believe that theses spherules representcollagenous spherulosis rather than shadow cell differ-entiation. The reasons for our interpretation are:(1) The collagenous nature of the spherules has been

clearly demonstrated using immunohistochemistrywith antibodies directed against type IV collagen asshown on Figure 3.1 Furthermore, as it has beenpublished previously,2 these spherules also stainedpositive for collagen types I and III (data not shown).Moreover, they were orange because of the presenceof saffron that stains only collagen in orange but notkeratin which remains eosinophilic. Because allthese data argue for the collagenous nature of thespherules, van Gieson and Masson's trichromestains could be omitted.

(2) As illustrated in Figure 1b, we also observed collagen-ous spherulosis in endometrioid glands that did notcontain squamous metaplasia. However, for thepurpose of the paper (drawing attention to a pitfall),we choose to focus on the glands exhibiting bothsquamous metaplasia and collagenous spherulosis.

(3) Giant cell reaction is not speci®c for shadow celldifferentiation and may be seen anywhere as soon asa `foreign body' is released.

(4) The immunophenotype of the cells surrounding thespherules was highly suggestive of myoepithelialdifferentiation since they were positive for smoothmuscle actin, S100 and vimentin. The negativestaining with GFAP did not exclude a myoepithelialdifferentiation. Since the cells were not degenerate,the negativity for KL1, cytokeratin 19 and EMA was

consistent with the former hypothesis. Furthermore,if a myo®broblastic reaction could be seen in thestroma, it is dif®cult to assume that such a reactionwould occur within glandular lumens.

In conclusion, we think that shadow cell differentia-tion might be added to the list of differential diagnosesfor collagenous spherulosis but we still believe that thespherules seen in our ovarian tumour representcollagenous spherulosis rather than shadow celldifferentiation.

I TreilleuxJ GodenecheP Duvillard1

C Clement-ChassagneY Suignard

C Bailly

DeÂpartement d'Anatomie et de Cytologie Pathologiques,Centre LeÂon BeÂrard,

Lyon, and1Histopathologie B,

Institut Gustave Roussy,Villejuif, France

1. Treilleux I, Godeneche J, Duvillard P, Clement-Chassagne C,

Suignard Y, Bailly C. Collagenous spherulosis mimicking keratiniz-

ing squamous metaplasis in a borderline tumour of the ovary.Histopathology 1999; 35; 271±276.

2. Skalova A, Leivo I. Extracellular collagenous spherules in salivary

gland tumors. Immunohistochemical analysis of laminin

and various types of collagen. Arch. Pathol. Lab. Med. 1992; 116;649±653.

Clear cell change in metastatic colonicadenocarcinoma

Sir: Domoto et al. describe three cases of clear cellchange in colonic adenomas.1 Furman and Lauwersexpand these observations to describe and discuss a casewith clear cell change in a colonic adenocarcinoma.2

We illustrate an example of metastatic carcinoma in theliver with prominent clear cell change (Figure 1a).Review of the previously resected rectal adenocarcinomarevealed focal clear cell change (Figure 1b). The recogni-tion of this variant may be important in the identi®cationof a source of metastatic clear cell carcinoma.

Clear cell change in metastases from colonic adeno-carcinoma has been reported twice.3,4 In these casesand in ours the clear cells contained centrally placednuclei and stained weakly or not at all for mucin ordiastase-digestible, PAS-positive material. In all threeinstances the clear cells reacted strongly with antibodiesto carcinoembryonic antigen (CEA).

Correspondence 471

q 2000 Blackwell Science Ltd, Histopathology, 36, 470±478.

Furman and Lauwers stress that recognition of clearcell variants of colonic adenocarcinoma is important inthe distinction of primary colonic carcinoma fromtumours metastatic to the colon, notably renal cellcarcinoma. We would add that a colonic primary sitemay be a consideration in a case of metastatic clear cellcarcinoma. In our case extensive regions of themetastatic tumour mimic renal clear cell carcinomaquite well (Figure 1a). The clear cell change observed inthe primary lesion, however, consisted of clear cyto-plasm in both apical and subnuclear positions incolumnar cells (Figure 1b) as described in colonicadenomas1,3 and adenocarcinomas.4 The differentialdiagnosis posed by such a metastatic clear cell tumourshould be resolved usually by the clinical history andimmunostaining, especially for CEA.

P StenzelD Sauer

Department of Pathology and Laboratory Medicine,Oregon Health Sciences University,

Portland,Oregon, USA

1. Domoto H, Terahata S, Senoh A, Sato K, Aida S, Tamai S. Clear cell

change in colorectal adenomas: its incidence and histological

characteristics. Histopathology 1999; 34; 250±256.2. Furman J, Lauwers GY. Clear cell change in colonic adenocarci-

noma: an equally unusual ®nding. Histopathology 1999; 35; 476±

477.

3. Reed RJ, Love GL, Harkin JC. Consultation case. Am. J. Surg. Pathol.1983; 6; 597±601.

4. Rubio CA. Clear cell adenocarcinoma of the colon. J. Clin. Pathol.

1995; 48; 1142±1144.

Spindle cell tumours of the breast

Sir: The recent review article on spindle cell tumours ofthe breast1 should have highlighted in the description ofphyllodes tumour the dif®culty of classi®cation into thethree categories, as well as treatment by wide marginresection in all such cases including benign tumours.The dif®culty in classifying into the three categories hasbeen repeatedly emphasized.2±5

All phyllodes tumours, be they benign or malignant,have a propensity for local recurrence. This is particu-larly likely to occur if the lesion is inadequately excisedby enucleation without a surrounding rim of normaltissue. Large tumours, those with an irregular marginand those showing marked stromal overgrowth areparticularly likely to recur.2

Some investigators advocate subdivision into histolo-gically benign and malignant categories based oncriteria such as cellular atypia, mitotic activity, thetype of margin (pushing or in®ltrating), and the ratio ofmesenchymal over epithelial tissue. Others have chal-lenged the validity of this practice and recommended aborderline or intermediate category, while yet othershave suggested the use of proliferative indices obtainedby such techniques as ¯ow cytometry or immunostain-ing for proliferating cell nuclear antigen. Because thehistological grading of `cystosarcomas' has not provedto be a totally reliable prediction of the clinicalbehaviour, all these lesions should be consideredpotentially aggressive tumours.3

For the phyllodes tumours that do not fall easily intoone of these two extreme categories, the prognosticprediction and therapeutic recommendations have tobe made on the basis of size, pushing vs. peripheralmargins, cellular atypia, and mitotic count. There issome indication that DNA ploidy a S-phase fractionanalysis may be useful adjuncts to the assessment of thistumour.4

Predicting the behaviour of phyllodes tumour is amajor challenge for the pathologist and many studieson this topic have been published.5 As the microscopicappearance of phyllodes tumours may vary consider-ably from area to area, wide sampling is important in

472 Correspondence

q 2000 Blackwell Science Ltd, Histopathology, 36, 470±478.

Figure 1. a, Metastatic carcinoma in the liver (H & E). b, Invasive

adenocarcinoma in muscularis propria of colon (H & E).

order that the malignant potential of the stromalcomponent can be assessed.2

K Kapur

Department of Pathology,Government Medical College,

Sector 32 A,Chandigarh, India

1. Al-Nafussi A. Spindle cell tumours of the breast: practical approachto diagnosis. Histopathology 1999; 35; 1±13.

2. Millis RR, Hanby AM, Sirling AC. The Breast. In: Sternberg SS ed.

Diagnostic Surgical Pathology, 2nd edn. Philadelphia: Lippincott-

Raven, 1996: 384±386.3. Silverberg SG, Masood S. The Breast. In Silverberg SG, De Lellis RA,

Frable WJ eds. Principles, Practice of Surgical Pathology, Cytopathology,

3rd edn. New York: Churchill Livingstone Inc, 1997: 612±615.

4. Rosai J, ed. Breast. Ackerman's Surgical Pathology, 8th edn. St Louis,Missouri: Mosby, 1996: 1628±1630.

5. Sharkey FE, Alired DC, Valente PT. Breast. In: Damjanov I, Linder J,

eds. Anderson's Pathology, 10th edn. St. Louis Missouri: Mosby Inc,

1996: 2363.

Author's reply

Sir: I thank Dr Kapur for his interest in my recent reviewarticle regarding the differential diagnosis of spindle celltumours of the breast.1 I appreciate the dif®culty in thehistological classi®cation of phyllodes tumours. How-ever, one of the reasons that I have omitted thisinformation was the fact that my review article wasmainly concerned with a practical approach in dealingwith those spindle cell lesions that may be misinter-preted or underdiagnosed. A detailed description ofevery entity was perhaps beyond the scope of thisreview. Nevertheless I do believe that the commentsmade by Dr Kapur regarding the histological categoriza-tion of phyllodes tumour are quite useful. They arecertainly of clinical relevance

A Al-Nafussi

Edinburgh, Scotland, UK

1. Al-Nafussi A. Spindle cell tumour of the breast: practical approach

to diagnosis. Histopathology 1999; 35; 1±13.

Hybrid neuro®broma/schwannoma versusschwannoma with Antoni B areas

Sir: Regarding the description of hybrid neuro®broma/schwannoma by Feany et al.,1 I would appreciate anexplanation from the authors on their differentialdiagnosis between the neuro®broma component of thehybrid tumour and Antoni B areas of ordinaryschwannoma. The Antoni B areas are composed of

spindle-to-oval cells arranged haphazardly within loosematrix with delicate collagen ®bres. This pattern issimilar to that of neuro®broma. Therefore the inter-pretation of such structures and subsequently theclassi®cation of the lesion, i.e. whether it is a hybridneuro®broma/schwannoma or a schwannoma withAntoni B areas, may be dif®cult.

M Zamecnik

Department of Pathology,General Hospital,

Trencin, Slovak Republic

1. Feany MB, Anthony DC, Fletcher CDM. Nerve sheath tumours withhybrid features of neuro®broma and schwannoma: a conceptual

challenge. Histopathology 1998; 32; 405±410.

Author's reply

Sir: Dr Zamecnik wants to know why the unusualhybrid lesions which we described1 are not simplyschwannomas and he suggests that we may havemisinterpreted the Antoni 0 component as neuro®bro-matous tissue. Happily this distinction is straightforward:(1) As explained and illustrated in our paper, the

neuro®bromatous component did not at all resem-ble Antoni B tissue Ð speci®cally it lacked bothhyaline vessels and microcystic stroma and showedinstead a vaguely fascicular architecture. As antici-pated in neuro®bromatous tissue (and in contrast tothe virtual 100% positivity in schwamomatoustissue), only 30±40% of the lesional cells wereS100 protein positive in this component. Even moreconclusively, we clearly demonstrated (and illu-strated) small axons and nerve ®bres (as expected)in the neuro®bromatous component; axons andnerve ®bres are not present within re¯ecting thepathogenetically different nature of these tumourtypes).

(b) Five of the nine cases which we described had aplexifom architecture closely resembling that ofplexiform neuro®broma (see our Figure 4). Thisappearance is entirely different from plexiformschwannoma, which lacks any Antoni B componentand instead consists of multiple nodules of variablycellular tissue of Antoni A type.

C D M Fletcher

Department of Pathology,Brigham and Women's Hospital

and Harvard Medical School,Boston, MA, USA

Correspondence 473

q 2000 Blackwell Science Ltd, Histopathology, 36, 470±478.

1. Feany MB, Anthony DC, Fletcher CDM. Nerve sheath tumours with

hybrid features of neuro®broma and schwannoma: a conceptual

challenge. Histopathology 1998; 32; 405±410.

Absence of human herpesvirus-8 DNA inKaposi's sarcoma following postmastectomylymphoedema

Sir: A variety of tumours, the most common examplebeing angiosarcoma (Stewart±Treves syndrome), havebeen described in extremities affected by chroniclymphoedema.1 Lymphoedema may follow, coincide orprecede classic or HIV-associated Kaposi's sarcoma(KS).2,3 However, the development of KS in limbs withprimary or secondary chronic lymphoedema in patientswith no risk factors for KS is very rare

Only two cases of KS arising in a chronic lymph-oedematous arm following radical mastectomy havebeen reported.3 We present the clinicopathologicalfeatures of a third example, including an investigationinto the presence of associated human herpesvirus-8(HHV-8) in this situation. A 67-year-old woman with-out predisposing factors for KS underwent a radicalmastectomy and axillary lymph node resection forin®ltrating duct carcinoma of the right breast, followedby a course of local radiotherapy. She remained tumourfree with moderate, persistent lymphoedema of the rightarm for 16 years until she presented with a smallintradermal nodule in the right elbow region. Histologyrevealed a spindle cell tumour consistent with nodularKS (Figure 1). Immunohistochemistry showed positivestaining of tumour cells for CD34 and CD31 using aroutine strepavidin±biotin complex (ABC) method andmonoclonal antibodies obtained from Dako. Three yearslater `eczematous' lesions were noted in the same area.A biopsy showed features consistent with patch phaseKS (Figure 2). Apart from transient, limited exacerba-tion during a course of chemotherapy for biopsy-provenmetastatic breast carcinoma, the KS lesions haveremained static over a 14-years follow-up period,some even showing regression.

Sections from formalin-®xed, paraf®n-embeddedtissue representative of the original breast carcinoma,normal breast, subsequent metastases, and all biopsiesof KS lesions were submitted for the detection of HHV-8DNA sequences. KS330233 primers were used to amplifyan HHV-8 speci®c 233-base pair sequence of openreading frame 26. The polymerase chain reaction (PCR)products were subsequently used with nested primers toamplify a 172-base pair sequence. Dilutions of theplasmid KS330Bam were included as positive controlsand to determine the number of HHV-8 target moleculespresent. All the conditions and primers were used as

described previously.4 HHV-8 speci®c sequences werenot detected in any of the specimens.

Although HHV-8 is strongly associated with KS insystemic immune-suppressed patients, a similar rela-tionship between lymphoedema with regional immune-suppression and HHV-8 remains to be proven. RecentlyCerri et al. failed to demonstrate HHV-8 DNA in alymphangiosarcoma of the pubic region that developedin lymphoedematous tissue due to congenital lymphoe-dema.5 The absence of HHV-8 DNA in occasional KSlesions is well documented. This observation may be dueto technical factors, incorrect diagnosis, extremely lowcopy numbers of HHV-8 in the KS lesions, or viral

474 Correspondence

q 2000 Blackwell Science Ltd, Histopathology, 36, 470±478.

Figure 1. The initial biopsy showed nodular KS consisting of short

fascicles of spindle cells interrupted by slit-like spaces containing redblood cells (H & E).

Figure 2. Patch-phase KS lesion biopsied 3 years later. CD31 immu-

nostaining highlights irregular, angular vascular spaces lined by a

single layer of endothelial cells without signi®cant nuclear atypiadissecting dermal collagen bundles and surrounding native dermal

vessels and adnexal structures.

mutations that escape detection by the KS330233

primer.6 However, it remains possible that factorsother than HHV-8 may stimulate endothelial cellproliferation and induce anti-apoptotic factors in KS.

Rare cases of localized KS that develop in chronicallylymphoedematous limbs of patients who do not haveother clinico-epidemiological risk factors for KS, suggestthat chronic lymphoedema may be an additionalcondition predisposing to the development of KS.The demonstration of a reduced delayed type hyper-sensitivity response in such limbs, has prompted thehypothesis that a regional immune defect limited toareas of localized lymphoedema might also involve themechanisms of antineoplastic immune surveillance,thereby facilitating the development of certain tumoursin those areas.2 Additional studies are necessary todetermine whether KS in lymphoedematous tissue is lessfrequently associated with HHV-8 than other KS variants.

D G du PlessisJ W Schneider

F K Treurnicht1

S EngelbrechtEJ van Rensburg

Departments of Anatomical Pathology and1Virology,

University of Stellenbosch and Tygerberg Hospital,Tygerberg, South Africa

1. D'Amore ESG, Wick MR, Geisinger KR, Frizzera G. Primarymalignant lymphoma arising in postmastectomy lymphedema.

Another facet of the Stewart±Treves syndrome. Am. J. Surg. Pathol.

1990; 14; 456±463.2. Ruocco V, Astarita C, Guerrera V et al. Kaposi's sarcoma on a

lymphedematous immunocompromised limb. Int. J. Dermatol.

1984; 23; 56±60.

3. Ron IG, Amir G, Marmur S, Chaitchik S, Inbar MJ. Kaposi'ssarcoma on a lymphedematous arm after mastectomy. Am. J. Clin.

Oncol. 1996; 19; 87±90.

4. Engelbrecht S, Treurnicht FK, Schneider JW et al. Detection of

Human Herpes virus 8 DNA and sequence polymorphism inclassical, epidemic, and iatrogenic Kaposi's sarcoma in South

Africa. J. Med. Virol. 1997; 52; 168±172.

5. Cerri A, Gianni C, Pizzuto M, Moneghini L, Crosti C. Lymphangio-

sarcoma of the pubic region: a rare complication arising in congenitalnon-hereditary lymphedema. Eur. J. Dermatol. 1998; 8; 511±514.

6. Noel J-C, Hermans P, Andre J et al. Herpesvirus-like DNA sequences

and Kaposi's sarcoma. Relationship with epidemiology, clinicalspectrum, and histological features. Cancer 1996; 77; 2132±2136.

Pleomorphic lipoma with pseudopapillarystructures: a pleomorphic counterpart ofpseudoangiomatous spindle cell lipoma

Sir: The term pleomorphic lipoma was proposed in the

early 1980s for a benign lipomatous tumour withatypical features reminiscent of pleomorphic liposarc-oma.1 The tumour is composed of mature fat cellsalternating with some atypical mononuclear andmultinucleated cells of ¯oret-type in a stroma thatvaries from ®brous to ®bro-myxoid. We present a case ofpleomorphic lipoma with pseudopapillary structures.Similar ®ndings have also been observed in some casesof spindle cell lipoma (SCL) and described as pseudo-angiomatous spaces,2 but have never been reported inpleomorphic lipoma.

A 59-year-old man presented with a painless mass onthe shoulder for at least 8 years. The tumour appearedas a well circumscribed lobulated mass of 60 mmdiameter. A complete excision with free surgicalmargins was performed. Four-and-a-half years aftersurgery no recurrences or metastases have beendetected.

Macroscopical examination revealed a well-circum-scribed, lobulated tumour with a gelatinous appearancedevoid of necrosis. Microscopically, at scanning magni-®cation, the tumour consisted of an unencapsulatednodule located in the subcutaneous tissue. The lesionshowed areas of typical pleomorphic lipoma with amixture of mature fat cells and variably thick collagenbundles containing sparse cells with large hyper-chromatic nuclei and multinucleated cells of ¯oret-liketype. In addition, there were areas composed of villiformstructures containing mononuclear bizarre and multi-nucleated ¯oret-like cells in large number and maturefat cells (Figures 1 and 2). They were embedded in a®bromyxoid stroma containing mast cells. No mitotic®gures were identi®ed. Lipoblasts were not present.Immunohistochemically, bizarre mononuclear cells and¯oret-like cells showed positive immunostaining forCD34 (HPCA 1, 1 : 20, Becton-Dickinson, MountainView, CA), whereas S100 protein (1 : 200, Dako),Lysozyme (1 : 100, Dako), EMA (1 : 100, Dako), andSMA (prediluted, Progen, Heidelberg, Germany) wereconsistently negative. A moderately dense population ofdendritic cells within the stroma expressed Factor XIIIa(1 : 100, Centeon Pharma, Marburg, Germany).

In 1981 Shmookler & Enzinger described a lipoma-tous tumour that, despite atypical histopathologicalfeatures suggestive of liposarcoma, invariably showed abenign clinical behaviour. The lesion was calledpleomorphic lipoma.1 Since then, numerous cases ofpleomorphic lipoma have been reported and thistumour has become a recognized entity.3,4 Cases ofSCL with pseudopapillary structures similar to thosereported herein have been described as pseudoangio-matous SCL because of the striking angiomatoidappearance of the neoplasm.2 Because of the close

Correspondence 475

q 2000 Blackwell Science Ltd, Histopathology, 36, 470±478.

relationship between pleomorphic lipoma and SCL3 webelieve that pleomorphic lipoma with pseudopapillary(or pseudoangiomatous) structures represents thepleomorphic counterpart of pseudoangiomatous SCL.

Tumours to be considered in the differential diagnosisof pleomorphic lipoma with pseudopapillary structuresare pleomorphic, myxoid and some cases of welldifferentiated liposarcoma with bizarre stromal cells,myxoid malignant ®brous histiocytoma and giant cell®broblastoma. Liposarcoma is composed of fat cells thatvary consistently in size and shape, contain largehyperchromatic nuclei and show an increased mitoticactivity not seen in pleomorphic lipoma. Lipoblasts arevery rare in pleomorphic lipoma but a common ®ndingin liposarcoma. Myxoid liposarcoma often lacks theprominent cellular atypia observed in other lipo-sarcomas and in PL. Typically, neoplastic cells areembedded in a myxoid stroma containing a network of

capillary-sized blood vessels. Mitotic ®gures may besparse or not present at all. Myxoid malignant ®broushistiocytoma is characterized by poorly cellular myxoidareas in association with dense cellular areas. Cellularareas are composed of large pleomorphic cells withhyperchromatic nuclei and often foamy cytoplasm.Typical and atypical mitoses are common. Myxoidareas contain numerous blood vessels with numerousslit-like lumina forming arcs but lacking pseudo-papillary structures. Giant cell ®broblastoma combinesdense cellular areas, reminiscent of dermato®brosarc-oma protuberans, and poorly cellular ones containing avariable number of hyperchromatic mononuclear andmultinucleated cells. The most characteristic feature ofthe tumour, however, is the presence of angiomatoidspaces lined by a discontinuous row of multinucleatedcells.

In conclusion, a case of PL with pseudopapillarystructures is reported. Because of these unusualmicroscopical ®ndings, the tumour may cause diag-nostic problems and represents a potential pitfall forhistopathologists.

C Diaz-CascajoS Borghi

W Weyers

Center for Dermatopathology,Freiburg, Germany

1. Shmookler BM, Enzinger FM. Pleomorphic lipoma: a benign tumorsimulating liposarcoma. A clinicopathologic analysis of 48 cases.

Cancer 1981; 47; 126±133.

2. Hawley IC, Krausz T, Evans DJ, Fletcher CDM. Spindle cell lipoma ±a pseudoangiomatous variant. Histopathology 1994; 24; 565±569.

3. Azzopardi JG, Iocco J, Salm R. Pleomorphic lipoma: a tumour

simulating liposarcoma. Histopathology 1983; 7; 511±523.

4. Fischer HP, Stambolis C. Pleomorphes lipom. Ein kasuisticher unddifferentialdiagnostischer Beitrag. Pathologe 1983; 4; 103±106.

5. Grif®n TD, Goldstein J, Johnson WC. Pleomorphic lipoma. Case

report and discussion of `atypical' lipomatous tumors. J. Cutan.

Pathol. 1992; 19; 330±333.

Microscopic focal cryptitis associated withoral sodium phosphate bowel preparation

Sir: The colorectum is a well-recognized site ofiatrogenic disease.1 Oral sodium phosphate is usedincreasingly as a form of bowel preparation because ofgood patient tolerance and its effective bowel cleansingproperties.2,3 There is now good evidence that, whenused for this purpose, oral sodium phosphate mayinduce colorectal aphthous ulceration.4 The micro-scopic effects of oral sodium phosphate on colorectalmucosa are, however, less well recognized. We report

476 Correspondence

q 2000 Blackwell Science Ltd, Histopathology, 36, 470±478.

Figure 2. Pseudopapillary structures composed of mononuclearbizarre cells with hyperchromatic nuclei and multinucleated ¯oret-

like cells in a ®bro-myxoid stroma.

Figure 1. Pseudopapillary structures varying in size and shape are

distributed in large areas of the neoplasm.

four cases of oral sodium phosphate-associated micro-scopic focal cryptitis to increase awareness of this entityand to emphasize its differences from in¯ammatorybowel disease.

Attention was ®rst drawn to these cases during thereview of a colorectal biopsy series which was reportedas showing in¯ammatory bowel disease. The reviewwas requested by the patient's clinician who felt thediagnosis was discrepant from the clinical impressionwhich included a normal colonoscopic examination atthe time the biopsies were taken. Similar histologicalfeatures were subsequently noted in colorectal biopsyseries from three other patients taken within 3 monthsof this ®rst case. The ages of the patients concerned(three females, one male) were 31, 37, 54 and 56 years,respectively. All four patients had presented withsymptoms suggestive of irritable bowel syndrome:intermittent diarrhoea, bloating and abdominalcramps. None of the patients had used any antibiotics,immunosuppressive agents or non-steroidal anti-in¯ammatory drugs (NSAIDs) shortly before onset oftheir symptoms, nor showed any microbiologicalevidence of an infective aetiology. In addition, fullblood count, ESR, blood electrolyte concentrations andinvestigations for malabsorption, including coeliacdisease serology, were normal in all four patients.Colonoscopy in the four cases was normal and, inparticular, no mucosal features of in¯ammatory boweldisease were noted.

The histological pattern in all four cases was similar,as illustrated in Figures 1 and 2. Low-power scanning ofthe biopsies often gave an initial impression of normalmucosa. At higher power, however, an abnormality wasevident in the form of a focal cryptitis restricted to thelower half of the gland crypts. Most of the in®ltratingcells were neutrophils though occasional eosinophilswere often present. High-power examination revealed asecond notable feature Ð an increased number ofapoptotic bodies within the epithelium of the cryptbases; the frequency of these bodies in relation to thedegree of in¯ammation varied from biopsy to biopsy. Incases where the in¯ammation was most marked, thecrypt bases appeared tapered and showed a mild degreeof mucin depletion.

These changes were often restricted to a proportion ofcrypts in any one biopsy and to a proportion of biopsiesfrom any one biopsy site. The changes also tended to bemore pronounced in, although not restricted to, the leftside of the large bowel. Despite the prominent focalcryptitis, there was minimal extension of the in¯amma-tion into the lamina propria. Further, in each of the fourcases, the crypt architecture was relatively wellpreserved, and the surface epithelium normal. None of

the biopsies showed any features to suggest an infectiveor an ischaemic aetiology; intraepithelial lymphocyteswere not a signi®cant feature. Careful examination ofthe four patients' histories failed to show a link betweenthem aside from the fact that all four patients had beenbiopsied after the recent change of bowel preparationused in our endoscopy unit to an oral sodium phosphatepreparation. Thus far, clinical follow-up (median dura-tion: 15 weeks; range: 10±19 weeks) has failed to revealany evidence to refute a diagnosis of irritable bowelsyndrome in any of the four cases.

Focal cryptitis has been described in association withinfective, drug-related (NSAIDs, antibiotics or steroids)and ischaemic colitides,5 whereas HIV infection, in¯am-matory bowel disease, graft vs. host disease, andcytotoxic and nonsteroidal anti-in¯ammatory druguse are the best known causes of prominent cryptapoptosis.6 Having excluded these conditions in all fourpatients, we suggest the changes observed in their

Correspondence 477

q 2000 Blackwell Science Ltd, Histopathology, 36, 470±478.

Figure 1. Medium-power image showing a focal, predominantly

basal cryptitis with normal crypt architecture and minimal laminapropria in¯ammation (´ 250).

Figure 2. High-power image of the bases of the central colonic

crypts shown in Figure 1, demonstrating a cryptitis, involving neu-trophils and eosinophils, and increased numbers of apoptotic bodies

(arrowed) with mild mucin depletion (´ 630).

biopsies are directly attributable to the use of oralsodium phosphate. While the clinicopathological effectsof sodium phosphate enemas are well recognized,7 onlythree previous studies have speci®cally examined thehistological changes seen in colorectal mucosa follow-ing use of oral sodium phosphate.2,4,8 Aphthousulceration seen endoscopically is characterized bysurface epithelial erosion and surrounding oedemaand acute in¯ammation.4,8 Arguably of more interestis the description of focal cryptitis in the presence ofnormal endoscopic ®ndings. We are aware of only onepublished report of these changes in association withoral sodium phosphate use.8 In this paper, Driman et al.found the microscopic features seen in our four cases, in11 out of 316 biopsies (3.5%).8 A retrospective andprospective study is currently being undertaken tocalculate the frequency of these speci®c changesamongst colorectal biopsies taken at our endoscopy unitsince the introduction of the new bowel preparation.

There appear to be several features which helpdistinguish between microscopic oral sodium phosphate-associated focal cryptitis (mSPFC) and in¯ammatorybowel disease. In mSPFC, the surface epithelium isnormal with no ulceration, there is minimal increase inthe mononuclear cell population of the lamina propria,and the crypt architecture is well preserved. Anotherstriking feature of mSPFC is an increased number ofapoptotic bodies which, together with the cryptitis, ismost prominent at the base of the crypts (Figure 1). Thereason for basal, as opposed to surface changes, in thepresence of a luminal insult is uncertain though it mayrepresent a metabolic effect rather than a directphysico-chemical injury. Indeed, having shown thatexposure to sodium phosphate increases both cellproliferation and apoptosis, Driman et al. speculatethat this phosphate salt is acting, in part, by chelatingfree calcium ions, which are known to suppress colonicepithelial turnover.8

When confronted with a case of microscopic focalcryptitis of uncertain aetiology, the possibility that thechanges are due to use of oral sodium phosphate asbowel preparation should be considered. Awareness ofthe existence of mSPFC is important in light of the

increasing use of this preparation and, particularly so,in the investigation of patients with chronic diarrhoea ifmisdiagnoses of in¯ammatory bowel disease are to beavoided. As a means of preventing such errors, we agreewith the suggestion that oral sodium phosphate bowelpreparations be avoided in patients being investigatedfor in¯ammatory bowel disease.5

AC K N O W L E D G E M E N T S

We thank Dr HM Gilmour for his comments.

N A C S WongI D Penman1

S Campbell1

A M Lessells

Department of Pathology and1Gastrointestinal Unit,

Western General Hospital,Edinburgh, UK

1. Warren BF, Shepherd NA. Iatrogenic pathology of the gastro-intestinal tract. In: Kirkham N, Hall P, eds. Progress in Pathology.

Edinburgh: Churchill Livingstone, 1995: 31±54.

2. Vanner SJ, MacDonald PH, Paterson WG, Prentice RSA, Da Costa

LR, Beck IT. A randomized prospective trial comparing oral sodiumphosphate with standard polyethylene glycol-based lavage solution

(Golytely) in the preparation of patients for colonoscopy. Am.

J. Gastroenterol. 1990; 85; 422±427.

3. Hsu C-W, Imperiale TF. Meta-analysis and cost comparison ofpolyethylene glycol lavage versus sodium phosphate for colono-

scopy preparation. Gastrointest. Endosc. 1998; 48; 276±282.

4. Zwas FR, Cirillo NW, El-Serag HB, Eisen RN. Colonic mucosalabnormalities associated with oral sodium phosphate solution.

Gastrointest. Endosc. 1996; 43; 463±466.

5. Greenson JK, Stern RA, Carpenter SL, Barnett JL. The clinical

signi®cance of focal active colitis. Hum. Pathol. 1997; 28; 729±733.6. Lee FD. Drug-related pathological lesions of the intestinal tract.

Histopathology 1994; 25; 303±308.

7. Meisel JL, Bergman D, Graney D, Saunders DR, Rubin CE. Human

rectal mucosa: proctoscopic and morphological changes caused bylaxatives. Gastroenterology 1997; 72; 1274±1279.

8. Driman DK, Preiksaitis HG. Colorectal in¯ammation and increased

cell proliferation associated with oral sodium phosphate bowelpreparation solution. Hum. Pathol. 1998; 29; 972±978.

478 Correspondence

q 2000 Blackwell Science Ltd, Histopathology, 36, 470±478.