Section 4 -

White Lesions of the Oral Mucosa with Emphasis on Dysplasia EW Odell King's College London Guy's Hospital

White lesions in the mouth grouped by aetiology Fordyce spots/granules Leukoedema Erythema migrans White sponge naevus and Cannon's syndrome Pachyonychia congenita Dyskeratosis congenita Hereditary benign intraepithelial dyskeratosis Frictional keratosis Cheek and tongue biting Stomatitis nicotina Smokeless tobacco lesions Chemical burns Candidal infection Chronic hyperplastic type Chronic mucocutaneous types hrush Hairy leukoplakia Syphilitic leukoplakia Lichen planus, reticular and plaque types Lichenoid reactions systemic topical Lupus erythematosus "Lichenoid dysplasia" Psoriasis Idiopathic keratosis ("leukoplakia") with or without dysplasia homogeneous and speckled leukoplakia sublingual keratosis proliferative verrucous leukoplakia Koilocytic dysplasia White lesions of renal failure Verruciform xanthoma Skin grafts Diagnosis requires good clinical information, as an absolute minimum the patient's age, sex, ethnic origin and the appearance, extent and exact site of the lesion(s), particularly whether any red or ulcerated areas are associated with it, and any risk factors for oral malignancy including tobacco habits, alcohol intake or betel quid use. Knowledge of presence or absence of a rash is also helpful. Could this be a developmental condition? The developmental conditions are relatively rare. Commonest is white sponge naevus caused by mutations in the genes for keratin 4 or 13. Diagnosis requires both clinical and histological information. A family history may aid diagnosis and the pattern of inheritance is usually autosomal dominant. However, sporadic cases appear relatively frequently and it is not uncommon for clinicians to request exclusion of white sponge naevus on biopsy specimens. The most diagnostic histological appearances are seen after puberty and oral lesions maybe minimal or absent during the first decade. Not all cases show the classical condensation of the cytoplasm around the nucleus. Is a lesion normal, tongue biting or hairy leukoplakia? Tongue biting, hairy leukoplakia and white sponge naevus share several features. The key diagnostic feature of hairy leukoplakia in routine preparations is the presence of large koilocyte-like cells and cytological evidence of viral infection in prickle cells. These cells are identified on the basis of their enlarged pale staining finely granular eosinophilic cytoplasm, smooth cell borders and perinuclear haloes. Nuclei usually show pyknotic changes but these are little different from the features of cells in the normal upper prickle cell layer. Unfortunately clusters of similar cells are a normal feature of the oral mucosa of the lateral tongue, the commonest site for hairy leukoplakia. HAART has caused florid presentations of hairy leukoplakia to become rare and the histological features are markedly suppressed by treatment. This has made confident diagnosis of hairy leukoplakia considerably more difficult and Immunocytochemistry or in situ hybridization are required to differentiate trivial lesions including tongue biting on occasion. Can a lichenoid lesion be more fully characterised? The lichenoid histological features of interface stomatitis with keratinocyte destruction maybe the end point of several different pathological processes. The individual features are relatively non-specific and it is often not possible to differentiate the underlying cause with any degree of certainty. Eosinophils are rare in classical oral lichenoid reactions and when present are usually relatively sparse. Many lichenoid reactions are indistinguishable from lichen planus and it is our practice to report lichen planus and lichenoid reactions as a single entity. However, a number of histological pointers may favour a lichenoid drug reaction. These include a mismatch between the density of the infiltrate and the amount of epithelial destruction, lymphocytic infiltration focused at the suprabasal or prickle cell levels, a plasma cell component to the deeper infiltrate (in the absence of fungal infection) and deep perivascular inflammatory infiltrates separated from the subepithelial band. Topical lichenoid reactions triggered by dental restorations are almost always strongly suggested by the clinical presentation, particularly ulceration or atrophy close to the restoration. Typically these reactions show marked atrophy, superficial germinal centres (in the absence of fungal infection), prominent apoptoses in a localized band and very dense sharply demarcated perivascular infiltrates around. Is dysplasia present in a lichenoid lesion? It is usual for dysplasia of the oral epithelium to elicit a lymphocytic host response. The processes of lymphocytic infiltration and killing of epithelial cells are essentially lichenoid and indistinguishable from the features of lichen planus, lichenoid reactions and lupus erythematosus. Clinical information may be a helpful pointer but it is important to maintain a high index of suspicion. There have been no detailed histopathological studies comparing the features of the host response in dysplasia and lichen planus and the following diagnostic tips have only an anecdotal evidence base. Destruction of the basal cells in lichen planus can give rise to a range of atypical features. When basal cell destruction is limited, there is usually reactive basal cell hyperplasia. This may account for a degree of basal cell crowding and the presence of scattered hyperchromatic cells along the basement membrane but almost never causes significant generalized expansion of the basaloid compartment. The rete process pattern of lichen planus will vary from site to site in the mouth. On the dorsum and lateral tongue the rete processes are often large and irregularly shaped and retain the residue of the pattern normally associated with the filiform papillae, which are lost early in disease. Thus, regularly spaced acanthotic rete processes are not worrying, particularly when the lichenoid infiltrate is associated with only the tips of the rete processes, a common feature of lichenoid lesions on the dorsal/lateral tongue. More rounded short rete processes are worrying and indicate dysplasia if accompanied by a generalized enlargement of the basaloid cell compartment. Dysplasia and Malignant Transformation Dysplasia remains the best indicator of a risk of malignant change but the data are old and this may be about to change. Several classical studies have demonstrated the relationship between dysplasia and malignant transformation: total lesions no. with dysplasia no. excised transformation rate % Silverman et al. 1984 257 17 Mincer et al 1972 45 31 11 Banoczy and Csiba 1976 * 500 120/68** 66 13.2 Pindborg et al. 1977 61 6.6 Gupta et al. 1980 90 0 7 Lumerman et al. 1995 44 14 * included lip lesions ** 68 followed up. Interestingly, the value of grading is unproven and importance should be attached to even slight dysplasia. It is known that grade alone is not directly proportional to the risk (Lind 1987) and also that many dysplastic lesions can regress. The distribution of dysplasia grades in idiopathic white patches in various studies is shown in the following table. % mild % moderate % severe Waldon and Shafer 1975 12 5 Platkajs 1979 7 5 3 Seifert and Burkhardt 1979 45 9 Grossel and Hornstein 1982 10 5 3 Banoczy 1982 5 14 5 How should dysplasia be graded? I know of no peer-reviewed published data explaining the detail of how dysplasia should be scored and there appears to be almost no published data on which valid scoring systems could be based. Almost all papers on dysplasia grading identify the same list of features of dysplasia without definitions and no explanation as to how their severity should be recorded and no indication of the weighting which should be applied to each. The WHO 'blue book' published in 2004 made significant changes. It now recommends using a "thirds affected" step in grading dysplasia, but allows the pathologist to up or downgrade lesions based on mitotic activity and atypia. However, there is still no defined scheme and many of the criteria used mean very different things to different people. Size of cells, nuclei and total epithelial thickness are routinely ignored but probably significant. Given the difficulty in grading reproducibly, it is perhaps unfortunate that the new WHO system continues to validate the concept of carcinoma in situ. This is intended to imply that malignant transformation has occurred but that invasion is not present. This is more of a concept than a definable histological grade but it is suggested that full thickness cytological atypia with atypical mitoses would be seen. Many pathologists perceive that the problems with dysplasia grading systems lie at the level of hyperplasia and mild dysplasia. Lesions arise relatively frequently in which microinvasive carcinoma is covered by an intact layer of well-stratified epithelium showing relatively good maturation. In many well- and moderately-well differentiated carcinomas there may not be significant expansion of the basaloid compartment and only a mild degree of uniform cytological atypia. Grading systems that recognise this type of risk lesion would be a great advantage. Other risk assessment methods To date, no immunocytochemical or molecular analytical method has proved a better independent predictor of malignant transformation than dysplasia grading. Immunocytochemistry for a proliferation marker (we prefer mib1) and simple keratin (8 or 18) and increased levels of p53 form a panel that has some value but in general the staining patterns reflect changes visible histologically. Ploidy analysis (large scale genomic status) is measurement of individual cell variation from diploid state and reflects chromosomal instability, a fundamental and consistently early step in malignant transformation. Image-based systems have advantages for routine laboratories. No fresh tissue is required and paraffin-embedded samples can almost always be processed to provide a result, making it a more reproducible technique than flow cytometry. In a major retrospective series performed at the Radium Hospital, Oslo, Norway (Sudbo et al 2001), of 105 diploid leukoplakias only 3 carcinomas developed (at 35, 46 and 76 months), a 97% disease free survival. Conversely, in 20 tetraploid lesions 12 carcinomas developed and in 27 aneuploid lesions 21 carcinomas developed with a mean delay of 35 months. In this referral population the positive predictive value of ploidy analysis (84% for an aneuploid result) far exceeds the published figure for dysplasia grading. Unfortunately, there is no good data to indicate the positive predictive value of dysplasia grading for comparison. We started a ploidy analysis service early in 2004 and coupled this to a major retrospective analysis of malignant transformation rates in the UK. The database includes 2800 biopsy specimens of white and other suspicious lesions from 2000 patients, each of which was graded by agreement of two pathologists. 926 specimens showed dysplasia and follow up (5-15 years) was available from National Cancer Registries for 657 lesions in 273 patients. When our data is analysed by patient, assuming a "field change" risk, the positive predictive value of dysplasia grade for malignant transformation was: mild dysplasia 4.9%, moderate dysplasia 10.4% and severe dysplasia 16.4%. When ploidy analysis was performed on dysplastic lesions, the positive predictive values were aneuploid 46.2%, (95%ci 0.34-0.56), tetraploid 5.5%. Thus ploidy analysis outperforms dysplasia grading alone. However, as with all tests, the predictive value falls when applied to a population with a low prevalence. Ploidy analysis performed indiscriminately on all white lesions has a predictive value about the same as conventional dysplasia grading. When the relatively few lesions in which it could be confidently said that biopsy was performed at, or adjacent to, the site of malignant transformation, were analysed, an abnormal ploidy result predicted transformation in 14 of 21 lesions regardless of dysplasia grade, a strikingly successful result. These results will change as further patients are analysed. References Banoczy J, Monograph, Oral Leukoplakia,1982 , M Nijhof, The Hague. Banoczy J, Csiba A Oral Surg 1976, 42:766-74. Grossel, Hornstein Mund Kiefer Gesichts Chir 1982 6:343. Gupta PC et al., Commun Dent Oral Epidemiol 1980, 8:287-333. Lind P, J Dent Res 1987, 95:449-55. Lumerman H, Freedman P and Kerpel S, Oral Surg 1995, 79:321-9. Mincer HH, Coleman SA, Hopkins KP, Oral Surg 1972, 33:389-99. Pindborg JJ, Daftary JK, Mehta FS, Oral Surg 1977, 43:383-90. Platkajs MA, J Can Dent Assoc 1979, 45:107-13. Seifert G, Burkhardt A, Verh Dtsch Ges Pathol 1979 63:74-82. Silverman S, Gorsky M, Lozada F, Cancer 1984, 53:563-8. Sudbo J, Kildal W, Risberg B et al.,N Engl J Med 2001, 344:1270-8. Waldon CA , Shafer WG, Cancer 1975, 36:1386-92.