J Cutan Pathol 2001: 28: 174–183 Copyright C Munksgaard 2001

Printed in Denmark ¡ All rights reserved Journal ofCutaneous Pathology

ISSN 0303-6987

Histologic criteria for the diagnosis ofmycosis fungoides: proposal for a gradingsystem to standardize pathology reportingBackground: The histological diagnosis of early lesions of mycosis Joan Guitart1, John Kennedy2,fungoides (MF) is often difficult for dermatopathologists and prior Salve Ronan†,2, Joan S. Chmiel3,studies have shown a low agreement rate among pathologists. An im- Yi-Ching Hsiegh3 and Dainaportant reason for such difficulty may be the lack of specific histological Variakojis4

criteria. 1Department of Dermatology, NorthwesternMethods: We tested a new method to interpret and report biopsies University Medical School, Chicago, Illinois,

USA, 2Department of Pathology University ofsuspicious for MF. The method is based on a grading system reflectingIllinois, Chicago, Illinois, USA, 3Department ofthe pathologist’s degree of diagnostic certainty. A panel of four pathol-Preventive Medicine, Northwestern Universityogists independently assessed a set of 50 biopsies suspicious for MF Medical School, Chicago, Illinois, USA,

first without (Phase I) and subsequently with specific histological cri- 4Department of Pathology, Northwesternteria (Phase II). A third Phase was carried out after a training session, University Medical School, Chicago, †Deceasedusing a new set of cases with corresponding immunophenotyping andgene rearrangement analysis. Weighted and unweighted kappa statisticswere used to assess inter- and intra-pathologist variation.Results: The consensus rate among pathologists improved from 48%in Phase I to 76% in Phase III. Overall precision weighted kappasincreased from 0.630 in Phase I to 0.854 in Phase III, indicating excel-lent inter-pathologist agreement by Phase III. There was a significantassociation between the presence of an abnormal phenotype and/orT-cell clonality and a higher diagnostic score.Conclusions: The use of uniform criteria and training sessions can Joan Guitart, M.D., Department of Dermatology,substantially improve the consensus rate among pathologists in the Northwestern University Medical School,

695 North St. Clair Suite 19–150,diagnosis of MF.Chicago IL 60611, USATel π1 312 695 4174

Guitart J, Kennedy J, Ronan S, Chmiel JS, Hsiegh YC, Variakojis D. Fax: π1 312 695 0664Histologic criteria for the diagnosis of mycosis fungoides: proposal for e-mail: j-guitart/nwu.edua grading system to standardize pathology reporting.

Accepted October 2, 2000J Cutan Pathol 2001; 28: 174–183. C Munksgaard 2001.

Plato maintained a strong distrust of knowledgegained by sense perceptions, since such knowledge isconstantly changing, relative and private to each indi-vidual.1 Plato’s observation transcends to our presentday histopathology practice. Variability in diagnosticinterpretation is common even among ‘experts’.2 Animprovement in inter-observer agreement has beenshown in panels of pathologists when diagnostic cri-teria and educational material were provided.3 Thisis a concern to the medical community and ultimatelyto patients that can raise questions regarding pathol-ogy results and treatment intervention.

174

One of the most challenging situations for derma-topathologists is the initial diagnosis of mycosis fung-oides (MF). Since MF is a malignant lymphoma andthe diagnostic gold standard still relies on a solidclinicopathological correlation, a prompt and decisivediagnosis is often expected. Yet the concordance andreproducibility rates among pathologists are particu-larly low in the diagnosis of early MF.4,5 The difficultyin the diagnosis of MF is partly due to the nature ofthe disease. MF tends to evolve slowly and patientsoften carry a long-standing diagnosis of eczema orparapsoriasis before a definitive diagnosis can be

Histologic criteria for mycosis fungoides

reached.6,7 Inconclusive biopsy results can be ex-plained by the frequently slow evolution of MF witha predominance of reactive lymphocytes in the earlystages. Secondly, the disease can have a protean pres-entation with a variety of clinical and morphologicappearances. A third problem, the lack of specific cri-teria for the diagnosis of early MF, is addressed in thisstudy.

Despite major advances in the understanding of thedisease in recent years, there is still confusion andconflicting interpretation about how to approachequivocal cases. We believe there is a need for uni-form diagnostic criteria and a standard reporting sys-tem for early lesions suspicious for MF.

The purpose of our study is to define and test pro-posed criteria for grading skin biopsies from lesionssuspicious for MF. By providing grades of diagnosticcertainty, we recognize that frequently a definitive di-agnosis of MF cannot be rendered. Hematopatholog-ists and dermatopathologists have been included inour panel since the problem of early diagnosis of MFis shared by both subspecialties.

Material and methodsFour pathologists, two with special training and ex-pertise in hematopathology (D.V., J.K.) and two simi-larly qualified dermatopathologists (S.R., J.G.) carriedout the study. All members of the panel have hadroutine exposure to skin biopsies of cutaneous T-celllymphoma (CTCL).

Phase IFifty hematoxylin and eosin (H&E)-stained slideswere selected for study. Each slide was from a bisected4-mm skin punch biopsy obtained from one of 50different patients who presented with clinical findingssuspicious for MF. All slides were processed at North-western University’s Dermatopathology Laboratoryto assure uniformity in staining and thickness. Thecases were assessed independently by the four pathol-ogists using the same diagnostic categories, but with-out specific criteria.

Phase IIThe same set of slides used in Phase I was rearrangedarbitrarily and relabeled by an individual not in-volved in the study. Two months later the slides werereassessed independently by the four pathologistsusing the proposed criteria.

Phase IIIFifty new skin biopsies from different patients withlesions suspicious for MF were selected. Prior to the

175

independent slide evaluation, the criteria were dis-cussed and reviewed by the group and examples ofeach histological feature and grade were providedand discussed (Table 1). A T-cell clonality test hadbeen performed in all the new cases, but the resultswere not provided to the panel of pathologists.

Major criteriaAt low power the density of the infiltrate was assessedand scored as follows: 1Ωmild perivascular infiltrate(Fig. 1). ScoreΩ2 was a moderately dense perivas-cular or band-like lymphoid infiltrate without thicken-ing of the papillary dermis (Fig. 2). ScoreΩ3 was adense, superficial infiltrate with widening of the papil-lary dermis or with extension of the infiltrate into thedeep reticular dermis (Fig. 3).

At medium power the epidermotropism was as-sessed and scored as follows: 1Ωfocal basal epidermo-tropism and/or a single Pautrier’s microabscess or afew duplets/triplets or scattered lymphocytes withoutspongiosis. Basal epidermotropism was defined asfour or more lymphocytes aligned along the basal celllayer (Fig. 4). Pautrier’s microabscess was defined asa cluster of 4 or more lymphocytes with a surroundingclear halo. Duplets and triplets are intraepidermalclusters of 2 or 3 lymphocytes respectively. ScoreΩ2was extensive basal epidermotropism and/or two ormore Pautrier’s microabscesses (Fig. 5). ScoreΩ3 wasfor extensive epidermotropism with more lympho-cytes than keratinocytes in the epidermis (Fig. 6).

At high power lymphocyte atypia was evaluatedand scored. ScoreΩ1 was for mild atypia with smalland intermediate size lymphocytes with irregular nu-clear contour or cerebriform features (Fig. 7). ScoreΩ2 was for moderate atypia with small, intermediateand some large atypical cells (Fig. 8). ScoreΩ3 wasfor a uniformly atypical or transformed lymphoid in-filtrate with numerous mitoses and few reactive smalllymphocytes. Lymphocytes are mostly intermediateor large (Fig. 9).

Minor criteriaAdditional points were added to the score when theatypical infiltrate was largely confined to the intraepi-dermal compartment (Fig. 10). One point was addedfor low-grade atypia (atypia scores 1 or 2) and 2points for high-grade atypia (atypia score 3). An in-traepidermal lymphoid infiltrate without other signsof inflammation added 1 point. Inflammatory fea-tures included: edema of the papillary dermis, mixedinflammatory infiltrate with eosinophils and neutro-phils, spongiosis and vascular margination of neutro-phils with prominent endothelial cells. One point wasalso given for reticular fibroplasia (wiry collagen) ofthe papillary dermis defined as delicate bundles of

Guitart et al.

Table 1. Criteria for the diagnosis of mycosis fungoides

I. Major criteriaLow power Density

0Ωscant infiltrate.1Ωmild perivascular superficial infiltrate.2Ωmoderately dense perivascular or band-like infiltrate without thickening of the papillary dermis.3Ωdense, confluent infiltrate with thickening of the papillary dermis and involvement of the reticular dermis.

Medium power Epidermotropism0 – none1 – focal basal epidermotropism1 and/or single Pautrier microabscesses2 or a few couplets/triplets3 or scattered lymphocytes without spongiosis.2 – extensive basal and/or two or more Pautrier microabscesses3 – extensive epidermotropism with more lymphocytes than keratinocytes.

High power Atypia0–No atypia1–Mild atypia (small and intermediate cells)2–Moderate atypia (small, intermediate plus large atypical cells)3–Uniformly atypical or pleomorphic cells, many mitotic figures, few small round reactive lymphocytes. Lymphocytes are mostly intermediate or large in size.

II. Minor criteria1–Reticular fibroplasia of the papillary dermis around single lymphocytes (wiry collagen)1 or 24–Primarily intraepidermal atypical lymphocytes1–Lymphocytic infiltrate without inflammatory features:O Edema of the papillary dermisO Mixed inflammatory infiltrate (eosinophils, PMN)O SpongiosisO Margination of neutrophils with prominent endothelial cells

1Four or more lymphocytes lined along the basal cell layer.2Cluster of four or more lymphocytes with a surrounding clear halo3Intraepidermal clusters of 2 or 3 lymphocytes4Intraepidermal atypia 1Ωlow grade, 2Ωhigh grade

collagen around single lymphocytes with thickeningof the papillary dermis (Fig. 10).

Each feature was scored with full points unless thespecific findings were between two categories, inwhich case half points were used. The diagnosisrendered under the proposed system was reached byadding the components to obtain the total score(Table 2). Specifically, I) Perivascular or interfacelymphocytic dermatitis corresponds to a total score of0–2 points; II) Atypical lymphocytic infiltrate (mycosisfungoides can not be excluded) corresponds to a totalscore of 3–4 points; III) Atypical lymphocytic infil-trate suspicious for mycosis fungoides corresponds toscores of 5–6 points; IV) Mycosis fungoides corre-sponds to scores of 7 or more points.

ImmunohistochemistryImmunohistology was performed on acetone-fixedfrozen sections of the skin biopsies taken at the same

Table 2. Proposed diagnosis terminology

Diagnostic category Total score*

I. Perivascular/interface dermatitis or other specific diagnosis 0–2II. Atypical lymphocytic infiltrate (mycosis fungoides cannot be 3–4

excluded)III. Atypical lymphocytic infiltrate suggestive of mycosis fungoides 5–6IV. Mycosis fungoides 7π

*based on scoring system proposed in this paper.

176

time as the routine biopsy. We used the avidin-bi-otinylated peroxidase (Vector Laboratories, Bur-lingame, CA, USA) method according to the manu-facturers directions. Slides were counterstained withHarris hematoxylin. Frozen normal tonsil sectionswere used as positive controls for each antibody andequivalently diluted non-immune mouse immuno-globulin was substituted for each antibody to serve asnegative control.

The panel of monoclonal antibodies used includedCD2 (Coulter, Miami, FL, USA), CD3 (C), CD4 (C),CD5 (C), CD6 (C), CD7 (Becton-Dickinson, Moun-tain View, CA, USA), CD8 (C), CD15 (BD), CD20(C), CD25 (BD), CD30 (Dako, Carpinteria, CA,USA) , CD45RB (D) and normal mouse serum (C).The percentage of mononuclear cells expressing thedifferent surface markers was estimated by one of us(J.G.). Antigen deletion was defined as lack of express-ion of a pan-T-cell marker (CD3, CD5, CD7) result-ing in less than 30% of the mononuclear cell infiltratestaining with a specific T-cell antibody.

T-cell clonality polymerase chain reaction (PCR) analysisHigh molecular weight DNA was obtained fromfrozen skin samples using the Pure Gene kit (GentraSystems, Minneapolis, MN, USA) according to themanufacturer’s directions. PCR amplification wasperformed as previously described.8,35 Consensusprimers to the V (V1.3) and J (J17) segments were

Histologic criteria for mycosis fungoides

used for PCR amplification, generating an approxi-mate 268-bp fragment which included the N region.8

The reaction mixture consisted of standard 1X PCRbuffer containing 1.5 mM MgCl2, 50 pmoles of eachprimer, 200 mM of each deoxyoligonucleotide, 1 mgDNA and 2.5 units Taq DNA polymerase in a totalvolume of 50 mL. Amplification was performed in aPerkin Elmer 9600 thermal cycler (Perkin Elmer,Norwalk, CT, USA) and consisted of an initial 3 min-ute denaturation followed by 30 cycles of 94 æC for60 s; 61 æC for 60 s, and 72 æC for 60 s. Initial productanalysis was performed via electrophoresis using an0.7% agarose minigel containing ethidium bromide.Positive (1% CEM cells mixed with polyclonallymphocytes) and negative controls (polyclonallymphocyte DNA and DNA blanks consisting of ster-ile water) were used in each experiment. The UTP/UNG system was used to control carry over contami-nation, along with spatially separate work areas anddedicated equipment at each area. SSCP analysis wasperformed using the semiautomated Phastgel system(Pharmacia). One ml of the amplified PCR reactionmixture was diluted with an equal volume of formam-ide containing 5% bromphenol blue and applied to apreformed 12.5% homogeneous acrylamide gel. Theelectrophoretic program, run at 4 æC using native buf-fer strips (Pharmacia, Uppsala, Sweden), consistent ofa prerun of 400 V, 10 mA, 2.5 W, 100 Vh, automaticsample application of 400 V, 1 mA, 2.5 W and 2 Vh,and electrophoresis of 400 V, 10 mA, 2.5 W, and 80Vh. The approximate prerun time was 30 min andelectrophoresis required 20 min. Silver staining wasperformed in the automated staining apparatus usingdirections and prepared reagents from the Phastystemsilver staining kit (Pharmacia).9

Statistical analysisConsensus agreement for each biopsy sample was de-fined to have occurred when scores differ by no morethan one diagnostic category among the four pathol-ogists. We calculated weighted and unweighted kappastatistics to summarize the level of agreement (beyondchance agreement) between pathologists (for Phases I,II and III) and within pathologists (Phase II vs. PhaseI matched biopsy samples).10 We calculated 95%confidence intervals (CI) to evaluate the precision andsignificance of these statistics. In general, kappa.0.75 denotes excellent agreement beyond chance; 0.4to 0.75 denotes fair to good agreement beyond chance,and less than 0.4 denotes poor to marginal agreementbeyond chance. Precision weighted averages of indi-vidual kappa values were calculated to obtain overallestimates of agreement. The individual kappas beingaveraged were tested for homogeneity using a chi-square test.10

Finally, for each of the four criteria (Density, Epid-

177

ermotropism, Atypia and Minor Criteria) and for thefinal diagnosis scores of Phases I, II and III, we ana-lyzed the variability unadjusted for chance among thefour pathologists’ scores for each of the 50 biopsysamples. The sample variance of the four scores foreach biopsy was calculated using usual statisticalmethods and the average of the 50 variances wascomputed. The standard error of these 50 variancesalso was calculated.

ResultsPhase IConsensus agreement among the 4 pathologists wasreached in 48% of the cases (24/50). Unanimousagreement was reached in 6% (3/50) of the cases.Kappa statistics based on our definition of agreementranged from 0.360 to 0.802 when comparing pairs ofpathologists with an overall precision weighted aver-age kappa of 0.630 (SEΩ0.042). There was a statisti-cally significant lack of homogeneity among the indi-vidual kappas (pΩ0.04). Conventional unweightedand weighted kappas based on exact agreement sug-gested poor agreement among the pathologists (Table3).

Phase IIConsensus agreement among the 4 pathologists wasreached in 58% of the cases (29/50). Unanimousagreement was reached in 20% (10/50) of the cases.Kappa statistics based on our definition of agreementranged from 0.520 to 0.841 when comparing pairs ofpathologists with an overall precision weighted aver-age kappa of 0.741 (SEΩ0.037). The agreement rate

Table 3. Kappa statistics for agreement between and within pathologists

Overall Kappa Kappa Range Test of(standard error) homogeneity

p-value

Phase IUnweighted 0.141 (0.032) 0.079–0.221 0.72Weighted 0.271 (0.035) 0.191–0.384 0.52Guitart et al. method* 0.630 (0.042) 0.360–0.802 0.04

Phase IIUnweighted 0.273 (0.037) 0.116–0.452 0.12Weighted 0.407 (0.038) 0.276–0.523 0.16Guitart et al. method 0.741 (0.037) 0.520–0.841 0.051

Phase IIIUnweighted 0.345 (0.036) 0.212–0.466 0.22Weighted 0.536 (0.032) 0.421–0.641 0.31Guitart et al. method 0.854 (0.029) 0.680–0.920 0.19

Phase I vs. Phase IIUnweighted 0.254 (0.046) 0.001–0.391 0.01Weighted 0.393 (0.046) 0.150–0.496 0.03Guitart et al. method 0.730 (0.050) 0.657–0.823 0.57

* The Guitart et al. method defines agreement to have occurred between twoscores if they differ by no more than one diagnostic category.

Guitart et al.

between the two dermatopathologists was 78% (39/50) with kappaΩ0.572 (95% CIΩ0.363 to 0.782).The agreement rate between the two hematopatholo-gists was 92% (46/50) with kappaΩ0.841 (95% CIΩ0.693 to 0.987). There continued to be some evidenceof lack of homogeneity (pΩ0.051).

Phase IIIThe consensus agreement rate among the 4 pathol-ogists increased to 76% (34/50). Unanimous agree-ment was reached in 28% of the cases (14/50). Kappastatistics based on our definition of agreement rangedfrom 0.680 to 0.920 comparing pairs of pathologistswith an overall precision weighted average kappa of0.854 (SEΩ0.029). The agreement rate among thetwo dermatopathologists was 92% (46/50) with akappaΩ0.837 (95% CIΩ0.686 to 0.988). The agree-ment rate among the two hematopathologists was88% (44/50) with kappaΩ0.763 (95% CIΩ0.591 to0.936).

Density was the most reproducible assessment. Ona density scale from 1 (sparse infiltrate) to 3 (denseinfiltrate), the average variation (unadjusted forchance agreement) was 0.134 for Phase II and 0.105for Phase III. For epidermotropism, also on a scale of1 to 3, it was 0.184 for Phase II and 0.186 for PhaseIII. Atypia was more difficult to grade (mild, moder-ate or severe) with an average variation of 0.395 forPhase II and 0.228 for Phase III. The least depend-able factor was minor criteria with an average vari-ance of 0.693 for Phase II and 0.576 for Phase III(Table 4).

The diagnostic scores between the four panelistswere compared with kappa statistics, as noted pre-viously. The overall precision weighted average kappabased on our definition of agreement improvedthroughout the study from 0.630 in Phase I to 0.741

Table 4. Variation among Pathologists (unadjusted for chance agreement)

Phase Average variation (S.E.)*

IDiagnosis 0.72 (0.082)

IIDensity 0.134 (0.018)Epidermotropism 0.184 (0.018)Atypia 0.395 (0.057)Minor Criteria 0.693 (0.091)Diagnosis 0.545 (0.072)

IIIDensity 0.105 (0.020)Epidermotropism 0.186 (0.027)Atypia 0.228 (0.029)Minor Criteria 0.576 (0.073)Diagnosis 0.388 (0.055)

*See text for details. Average variation is the arithmetic mean of the 50 samplevariances, each of which measures the variation among the four pathologists(unadjusted for chance agreement). S.E. is the standard error of these 50 values.

178

Table 5. Average diagnostic scores plotted with normal or abnormal immuno-phenotyping and T-cell clonality analysis

in Phase II and 0.854 in Phase III (Table 3). Improve-ment was also noted when the analyses were doneusing exact agreement definitions, and both un-weighted and weighted kappas.

When the results of immunophenotyping and generearrangement (normal vs. abnormal) were plottedagainst the average histological scores for Phase III, asignificant correlation (p,0.001) was detected (Table5). The average diagnostic score for the cases withabnormal immunophenotype and/or T-cell clonalitywas 3.1 (standard deviation 0.66) while the averagescore for the cases without T-cell clonality or abnor-mal phenotype was 2.0 (standard deviation 0.91).

Since the sample materials used for Phases I andII were the same but in a different order, we matchedthe individual diagnostic case scores to compare theresults from the same panelists with or without theproposed criteria. The diagnostic correlation wasgood with an overall precision weighted averagekappa of 0.730, ranging individually from 0.657 to0.823 for individual kappas (Table 3).

DiscussionThe pathological changes of early MF can be non-specific or even closely mimic a variety of dermatoses.Darier acknowledged in 1920 that the diagnosis of

Histologic criteria for mycosis fungoides

Table 6. Phase IIIΩPatient’s data

. Age/sex Clinical presentation Duration Diagnosis and evolution Average TCR Antigenin years Score PCR deletion

1 59M MF with possible recurrence ,1 MF in CR 2.25 π ND2 65M Diffuse patches, pruritus 1 MF IIb Progression 4 π ND3 44M MF Ia, patches torso NA MF Ia, stable 4 π π4 74F Erythematous patches face/neck ,1 Dermatitis resolved 2.25 ª ND5 59F Extensive patches torso/extremities 4 Granuloma anulare, resolved 1.75 ª ND6 61M Extensive patches/plaques 3 MF Ib, stable 2.5 ND ª7 68F Generalized erythroderma ,1 LFU 2.75 ª π8 43F Papulo-squamous eruption 1 LFU 1 ª ª9 39M Generalized erythroderma, MF ,1 DOD, progression 4 π π

10 42M Thick plaques, MF Ib NA MF, progression IIb 4 π π11 37M Patches torso 2.5 Dermatitis, LFU 1.5 ª ND12 50M Patches face/neck/hands 5 Actinic reticuloid 3.25 ª ND13 72M Patches, MF Ib 5 MF, progression IIb 4 π π14 41F Guttate patches ,1 Dermatitis, resolved 1 ª ND15 47F Recurrent MF after electron beam ,1 MF Ib, stable 1.75 π ND16 68M Patches torso MF Ia, stable 2.75 π π17 52M Extensive plaques & follicular 4 MF progression IIb, stable PUVA- 3.25 ª π

mucinosis , MF Ib Interferon a18 70M Recurrent MF after mustargen ,1 MF Ia, stable on PUVA 2.75 ND ª19 76M Pustular patches torso/extremities 2 MF pustular, progression IIb 2 ª ND20 72M Extensive patches MF Ib 20 MF Ib, stable on PUVA 2.5 π π21 68M Papules on the legs ,1 MF/LyP, Ia, LFU 3.5 ª π22 70M Pruritic extensive patches/plaques 2 Probable MF Ib, LFU 3.25 ª ª23 43M Lichenified patches 1 Dermatitis, No progression 2.25 ª ª24 77M Patches torso/extremities ,1 Dermatitis, No progression 2.25 ª ª25 58M Patches buttocks/thighs/arms 20 MF, complete response PUVA 3.75 π ND26 69F Poikilodermatous patches 4 Dermatitis, no progression 1.25 ª ª27 59F Generalized erythroderma 1 LFU 2.75 ª ª28 58M Patches buttocks MF Ia 25 MF, complete response mustargen 3.25 π ª29 42F Patches on hips ,1 Dermatitis Fluoxetine-induced 1 π ª30 63M Diffuse patches neck/torso 40 LFU 2.5 ª ª31 49F Follicular patches 10 MF Ib, stable 3 π π32 75F Annular patches/plaques 1 MF Ib, progression 2.5 π π33 60M Lichenified patches, head/neck/arms ,1 Actinic reticuloid, stable 2 π ª34 63M Patches torso 6 MF Ib, stable 3.5 π ND35 32M Extensive lichenified patches 2 Dermatitis, stable 1 ª ND36 87F Extensive patches 1 LFU 1.75 ª ND37 83M Generalized erythroderma 10 Sezary syndrome, stable 1.25 π ª38 47F Patches on buttocks NA MF Ia, stable 3.5 π π39 61M Follicular patches face/chest 1 Dermatitis/folliculitis, resolved 1.5 ª ª40 84M Patches torso/legs ,1 MF Ib, improved on PUVA, LFU 4 ª π41 74F Extensive patches/plaques 5 DOD, tumor progression 3 π π42 79F MF s/p electron beam, recurrence ,1 DOD, tumor progression 4 π π43 47M Patches buttocks/thighs 1 LFU 2.25 ª ª44 65M Pruritic lichenified patches ,1 Dermatitis, resolved 1.75 ª ª45 47M Erythematous plaques face/back ,1 Dermatitis, resolved 1.25 ª ª46 16F Patches torso/extremities 1 MF Ia, complete response 2.75 ND ª47 69M Diffuse patches 12 Dermatitis, stable 2 ª ª48 52F Extensive plaques 20 MF, progression IIb 4 π π49 68M Patches and papules 4 MF/LyP Ia, stable 3.5 ª π50 74M Extensive patches 3 Dermatitis, LFU 1 ND ª

NDΩnot done; LFUΩlost of follow-up; DODΩdead of disease; LYPΩlymphomatoid papulosis.

MF in early stages may be impossible to reach.11,12

In many occasions, neither dermatologists nor pathol-ogists know with reasonable certainty if a patient hasparapsoriasis, a non-specific dermatoses or a poten-tially lethal cutaneous lymphoma. The debate on themeaning of parapsoriasis and its relationship with MFis complex and may never be satisfactorily resolved.One could argue that the true nature of ‘‘parapsori-asis’’ can only be resolved retrospectively when thedisease has followed its natural course. However, his-torical studies are not clear about its defining par-ameters and presently most patients with ‘‘parapsori-

179

asis’’ are effectively treated with topical modalities.Despite major advances in the detection of T-cell clo-nality and immunophenotyping, these tests have notresolved this basic dilemma which remains to thepresent times.

MF is the most common form of cutaneous T-celllymphoma characterized by a neoplastic proliferationof mature post-thymic T lymphocytes with homingproperties to the skin. In early MF, the neoplastic Tlymphocytes survive in a proinflammatory micro-environment with mixed reactive lymphocytes andantigen presenting histiocytes. Therefore, the histo-

Guitart et al.

logical findings of early MF can be subtle, often mim-icking inflammatory skin diseases.13,14 Practicallyevery histological pattern encountered in inflamma-tory skin lesions has been identified in MF.15 Thereason for such a disparity of features at presentationmay be the fact that the neoplastic T cells are im-munologically mature and capable of eliciting diverseinflammatory responses. Consequently, the accuratediagnosis of MF in early stages is often difficult orimpossible to reach. A common way to convey one’sdegree of uncertainty or concern about the possibilityof MF is by using terminology like ‘‘suspicious forMF’’ or ‘‘consistent with MF’’ in the report. Thisproblem is addressed in our reporting system by intro-ducing four standard grades of diagnostic certainty.

A caveat of our system is that it cannot be useduniversally whenever MF is clinically suspected. Theheterogeneous presentation of MF makes it inappro-priate to apply the proposed grading system to rarevariants of MF. Rather the system should be usedwhen the typical presentation of MF enters the differ-ential diagnosis and only after specific dermatoseshave been excluded from the diagnosis. Because ofthe variability in the initial presentation, there are nopathognomonic histological features that categoricallyestablish or refute the diagnosis of MF in all cases.Many authors have described in detail the classicalhistological features of MF.15–19 Other reports havefocused on unusual presentations such as lichenoid,eczematous, bullous, pigmented, vasculitic, pustular,verrucous, hypopigmented and folliculotropic vari-ants.20–27 Most of these papers have emphasized vari-ous histological features. However, a reproducible,methodic and reliable system to evaluate such biop-sies had not been proposed. We have tested our sys-tem with a panel of dermatopathologists and hemato-pathologists. The results show that the system can beapplied to difficult cases providing uniformity and re-producibility in the interpretation.

The assessment of lymphocyte density at low powerexamination was the most reproducible variable inour study. Although often not present, we believe adense lymphoid infiltrate may contribute to the diag-nosis of MF. Furthermore, it has been shown thatthick plaques with a dense infiltrate correlate with dis-ease progression and therefore can have prognosticsignificance.28,29 Dense lymphoid infiltrates are notrare in inflammatory skin diseases and thereforeshould not be considered an independent diagnosticfeature of MF. However, when classic lymphoid-richdermatoses such as drug eruptions, lichen planus orlupus erythematosus have been excluded, a denselymphoid infiltrate should raise the possibility of MF.One must emphasize that the present criteria shouldbe used only after common and uncommon inflam-matory skin conditions, which could mimic MF, havebeen clinically and histologically excluded.

180

The next step, at intermediate power, is the assess-ment of intraepidermal lymphocytes. Epidermotrop-ism is a common finding in MF that has been re-ported in 75–100% of cases.13,15,16 Pautrier microab-scesses are highly specific for MF, but only reported

Fig. 1. Density 1.

Fig. 2. Density 2.

Fig. 3. Density 3.

Histologic criteria for mycosis fungoides

in 4–37% of cases.13,15–17 Exocytosis of lymphocytes,occasionally arranged in groups resembling Pautrier’smicroabscesses, can be seen in spongiotic dermatos-es.30 In such cases Langerhan’s cells and/or degener-ated keratinocytes may be seen with the lymphoidcells often assuming a vase-like shape. The detection

Fig. 4. Epidermotropism 1.

Fig. 5. Epidermotropism 2.

Fig. 6. Epidermotropism 3.

181

Fig. 7. Atypia 1.

Fig. 8. Atypia 2.

of intraepidermal lymphocytes is more relevant to thediagnosis of MF in the absence of spongiosis or otherinflammatory features such as dermal edema or mar-gination of neutrophils. Intraepidermal lymphocytes,in the absence of spongiosis, tend to show a periph-eral clear halo. Smoller et al. found that detection ofhaloed lymphocytes was the most reliable histologicalfinding to discriminate MF from non-MF cases.31

However, it should be noted that inflammatorychanges, including extensive spongiosis and even pus-tules, can be seen in MF.32 Since we believe that thelack of inflammatory features can point toward thediagnosis of MF, this observation has been includedas a minor criterion.

The assessment of nuclear atypia, done at highmagnification, was less reproducible. Although intuit-ively more important in the diagnosis of a malignantlymphoma, assessment of atypia in any cell line is no-toriously subjective and difficult to grade. In MF, thisproblem is aggravated by the presence of many reac-tive lymphocytes, histiocytes and even fibroblasts,

Guitart et al.

Fig. 9. Atypia 3.

Fig. 10. Two minor criteria are illustrated. Primary intraepidermalatypical lymphocytes and reticular fibroplasias of the papillary der-mis (wiry collagen).

which may masquerade or dilute the neoplasticlymphocytes. Histopathological studies of mycosisfungoides have reported prevalence rates of atypiaranging from less than 50% to 100% of the cases.15,16

Santucci et al. identified the presence of hyperconvo-luted medium to large lymphocytes, singly or in clus-ters, in the epidermis or in clusters in the dermis asthe most important independent criteria in the diag-nosis of MF.17 In a recent dermatopathology work-shop organized by the International Society of Cu-taneous Lymphomas in Zurich, the presence oflymphocytes with strikingly irregular nuclear contourwas found to be the most sensitive criteria for thediagnosis of MF. Our system allows higher scores forlesions with large and pleomorphic lymphocytes. Thisis important because such cases are more likely toprogress into thick plaques and tumors.15,33 Despitethe difficulties in grading nuclear atypia, the presenceof lymphocyte atypia within the intraepidermal com-partment is highly specific for MF.13,15,34 We concurwith this important histological feature, but to avoid

182

creating a more complex evaluation system, this find-ing is included as a minor criterion with one point forlow-grade atypia and two points for high-gradeatypia. In our study, the presence of conspicuous mi-totic figures was not included as an independent cri-teria, but as an additional feature to support high-grade atypia. Although one study claimed that thepresence of mitoses is a useful prognostic feature inMF, this observation was not confirmed by others.4,13

Wiry fibroplasia, a characteristic finding in MF,was also included as a minor criterion in ourstudy.16,18 This feature, characterized by thin bundlesof dense collagen arranged in a reticular pattern, isseen in the papillary dermis surrounding lymphocytesand probably implies chronicity of the superficiallymphoid infiltrate. Wiry fibroplasia should be dis-tinguished from the thick collagen bundles arrangedparallel to the dermatoepidermal junction noted inpruritic conditions like lichen simplex chronicusrather than in MF.

Besides uniformity and reproducibility, the resultsof Phase III showed a statistically significant positivecorrelation with confirmatory diagnostic testing (i.e.,detection of an abnormal immunophenotype and/ordetection of T-cell clonality) corroborating the valid-ity of the method.

In our experience the diagnosis of MF can be im-proved by optimizing the specimens. For instance, bi-secting punch biopsies, performing step sections intothe block, taking large elliptical excisions or multiplepunch biopsies can enhance the odds of reaching amore precise pathological diagnosis. Since the path-ological changes tend to become more evident overtime, repeating biopsies after a span of time and dis-continuing topical steroids or light therapy 3–4 weeksprior to biopsy can also improve the diagnostic yield.Finally, a consensus diagnosis involving more thanone pathologist may decrease the individual observerbias and improve the accuracy of the interpretation.We believe the proposed criteria could be initiallyuseful for entering patients into cutaneous T-celllymphoma clinical studies. Eventually, and if vali-dated with use and time, it may help standardize andclarify our routine dermatopathology reports, in pa-tients with early stage MF.

AcknowledgementWe are indebted to Professor Marco Santucci for his thoughtfulcomments and advise.

References1. Tarnas R. The passion of the western mind. Understanding

the ideas that have shaped our world view. New York: Ballan-tine Books, 1991.

2. Cramer SF. Interobserver variability in dermatopathology.Arch Dermatol 1997; 133: 1033.

Histologic criteria for mycosis fungoides

3. Schnitt SJ, Connolly JL, Tavassoli FA, et al. Interobserver re-producibility in the diagnosis of ductal proliferative breastlesions using standardized criteria. Am J Surg Pathol 1992;16: 1133.

4. Olerud JE, Kulin PA, Chew DE, et al. Cutaneous T-celllymphoma. Evaluation of pretreatment skin biopsy specimensby a panel of pathologists. Arch Dermatol 1992; 128: 501.

5. Santucci M, Biggeri A, Feller AC, Burg G. Accuracy, concord-ance and reproducibility of histologic diagnosis in cutaneousT-cell lymphoma. Arch Dermatol 2000; 136: 497.

6. Graham JH, Johnson WC, Helwig EB. Dermal pathology.Hagerstown, MO: Harper & Row, 1972.

7. Lazar AP, Caro WA, Roenigk HH, Pinski KS. Parapsoriasisand mycosis fungoides: the Northwestern University experi-ence, 1970 to 1985. J Am Acad Dermatol 1989; 21: 919.

8. Meyer JC, Hassam S, Dummer R, et al. A realistic approachto the sensitivity of PCR-DGGE and its application as a sensi-tive tool for the detection of clonality in cutaneous T-cell pro-liferations. Exp Dermatol 1997; 6: 122.

9. Guitart J, Kaul K. A new polymerase chain reaction-basedmethod for the detection of T-cell clonality in patients withpossible cutaneous T-cell lymphoma. Arch Dermatol 1999;135: 158.

10. Fleiss JL. Statistical methods for rates and proportions. 2ndedition. New York: John Wiley and Sons, 1981; 112.

11. Darier JA. textbook of dermatology. Philadelphia: Lea & Fa-biger, 1920.

12. Hilbrich D, Ackerman AB. Mycosis fungoides is common,rarely fatal, and diagnosable when lesions are flat. Dermatopa-thology: Practical & Conceptual 1997; 3: 24.

13. Smoller BR, Bishop K, Glusac E, et al. Reassessment of histo-logic parameters in the diagnosis of mycosis fungoides. Am JSurg Pathol 1995; 19: 1423.

14. Montgomery H. Dermatopathology. New York: Harper &Row, 1967.

15. Shapiro PE, Pinto FJ. The histologic spectrum of mycosisfungoides/Sezary syndrome (cutaneous T-cell lymphoma). Areview of 222 biopsies, including newly described patterns andthe earliest pathologic changes. Am J Surg Pathol 1994; 18:645.

16. Nickoloff BJ. Light-microscopic assessment of 100 patientswith patch/plaque-stage mycosis fungoides. Am J Dermatopa-thol 1988; 10: 469.

17. Santucci M, Biggeri A, Feller AC, Massi D, Burg G. Efficacyof histologic criteria for diagnosing early mycosis fungoides.Am J Surg Pathol 2000; 24: 40.

18. Sanchez JL, Ackerman AB. The patch stage of mycosis fung-oides. Criteria for histologic diagnosis. Am J Dermatopathol1979; 1: 5.

183

19. Smith NP. Histologic criteria for early diagnosis of cutaneousT-cell lymphoma. Dermatol Clin 1994; 12: 315.

20. Kartosin J, Brettschneider F, Weissman A, Rosen L. Mycosisfungoides bullosa. Am J Dermatopathol 1990; 12: 76.

21. Lund KA, Parker CM, Norins AL, Tejada E. Vesicular cu-taneous T-cell lymphoma presenting with gangrene. J AmAcad Dermatol 1990; 23: 1169.

22. Golden D, Rook A. Mycosis fungoides with necrotizing angi-itis. Br J Dermatol 1975; 92: 62.

23. Moreno JC, Ortega M, Conejo-Mir JS, Sanchez-Pedreno P.Palmoplantar pustulosis as a manifestation of cutaneous T-celllymphoma (mycosis fungoides). J Am Acad Dermatol 1990;23: 759.

24. LeBoit PE. Variants of mycosis fungoides and related cu-taneous T-cell lymphomas. Semin Diag Pathol 1991; 8: 73.

25. Whitmore SE, Simmons-O’Brien E, Rotter FS. Hypopig-mented mycosis fungoides. Arch Dermatol 1994; 130: 476.

26. Guitart J, Padula M, Caro WA, Roenigk H. Lichenoidchanges in mycosis fungoides. J Am Acad Dermatol 1997; 36:417.

27. Gilliam AC, Lessin SR, Wilson DM, Salhany KE. Folliculo-tropic mycosis fungoides with large cell transformation pre-senting as dissecting cellulitis of the scalp. J Cut Pathol 1997;24: 169.

28. Zackheim HS, Amin S, Kashani-Sabet M, McMillan A. Prog-nosis in cutaneous T-cell lymphoma by skin stage: Long-termsurvival in 489 patients. J Am Acad Dermatol 1999; 40: 418.

29. Marti RM, Estrach T, Reverter JC, Mascaro JM. Prognosticclinicopathological factors in cutaneous T-cell lymphoma.Arch Dermatol 1991; 127: 1511.

30. Leboit PE, Epstein BA. A vase-like shape characterizes theepidermal-mononuclear cell collections seen in spongiotic der-matitis. Am J Dermatopathol 1990; 12: 612.

31. Smoller BR, Detwiler SP, Kohler S, et al. Role of histology inproviding prognostic information in mycosis fungoides. J CutPathol 1998; 25: 311.

32. Jakob T, Tiemann M, Kuwert C, et al. Dyshidrotic cutaneousT-cell lymphoma. J Am Acad Dermatol 1996; 34: 295.

33. Vonderheid EC, Tam DW, Johnson WC, et al. Prognosticsignificance of cytomorphology in cutaneous T-cell lym-phomas. Cancer 1981; 47: 119.

34. Ackerman AB, Troy JL, Rosen LR, et al. Differential diag-nosis in dermatopathology II. Philadelphia: Lea and Febiger,1988; 66.

35. Curco N, Servitje O, Llucia M, et al. Genotypic analysis ofcutaneous T-cell lymphoma: a comparative study of Southernblot analysis with polymerase chain reaction amplification ofthe T-cell receptor-gamma gene. Br J Dermatol 1997; 137:673.