Quantitative exfoliative cytology of normal oral squames:

Transcription

1 Journal ofthe Royal Society ofmedicine Volume 78 December Quantitative exfoliative cytology of normal oral squames: an age, site and sex-related survey' J G Cowpe BDS PhD R B Longmore BDS PhD M W Green BA PhD Departments ofdental Surgery, Anatomy and Mathematical Sciences, University ofdundee Summary: This study describes the development of quantitative cytological techniques and their application to oral smears. Nuclear and cell size has been measured and matched with age, sex and site in an attempt to produce a baseline for comparison with identical measurements carried out on pathological smears. The results displayed a significant variation in nuclear and cytoplasmic area between different sites. Nuclear size varied significantly with advancing age; however, this was not the case for cytoplasmic area. There was no significant variation in either criterion between males and females. ntroduction Since Papanicolaou & Traut (1941) developed their staining techniques, the effective application of cytological techniques in the detection of asymptomatic cancers is now widely and successfully used in the prevention of uterine cervical cancer. Thus, together with its use in the detection of early cancer in other sites, it has been suggested that similar techniques might be equally effective in the detection of asymptomatic oral malignancy (Montgomery 1951, Staats & Goldsby 1963). Although a number of studies have been carried out on precancerous and cancerous lesions presenting in the oral cavity (Bernstein & Miller 1978, Sandler et al. 1960, Selbach & von Haam 1963), few studies have been performed on clinically normal oral mucosa matched for age, sex and site (Montgomery & von Haam 1951, Brown & Young 1970). Beyond that, quantitative techniques have been used sparingly (Pappelis et al. 1973, Lee etal. 1973). Morrison et al. (1949) stressed the importance of being aware of the normal cytological appearance prior to the examination of pathological squames. Montgomery (1951), studying normal oral squames collected from the soft palate, buccal mucosa, lower lip, tongue and gingivae, observed a variation in the staining of the cell cytoplasm for the five sites and described three specific cell types designated by the colours blue, red and yellow. He concluded that although there was variation in the shape and size of the individual cells and their nuclei within a single smear, there was no relationship between cell colour and sex, age or the menstrual cycle. However, like Hillman et al. (1972), he did observe a variation in the number of cytoplasmic granules with advancing age. The mouth is an ideal site for observing the manifestations of ageing (Hillman et al. 1972); however, few of the associated signs and symptoms of age lend themselves to quantification. This is unusual, as there is continuous replacement of the oral epithelium, suggesting that this accessible tissue should provide objective indications of senescence. One of the major values of exfoliative cytology is the noninvasive nature of the technique, which allows a simple and pain-free collection of intact cells from different layers within the epithelium for microscopic examination (Folsom et al. 1972), since the most characteristic histological and cytological changes of oral premalignancy are observed in the epithelium (Silverman et al. 1977, Squier 1980). 'Accepted 23 July /85/ /$02.00/ The Royal Society of Medicine

2 996 Journal of the Royal Society ofmedicine Volume 78 December 1985 One method of quantifying cytological smears is to apply the technique of DNA cytophotometry. This technique has been used extensively in the past to characterize the proliferative activity of cell populations (Fossa & Kaalhus 1977, Grove 1979). Normal cell populations display diploid DNA distributions and malignant cell populations display a variety of DNA profiles. However, it is the occurrence of aberrant diploid DNA distributions in malignancy which has reduced the value of DNA measurement as a single diagnostic criterion of malignancy (Bohm & Sandritter 1975, Sprenger & Witte 1979). Pfitzer & Pape (1973, 1975) expressed a need for the measurement of more variables than just DNA content, in order to increase the accuracy of cytology in the diagnosis of oral malignancy. n 1964, Goldsby et al. concluded that the measurement of nuclear and cell size and nuclear/cytoplasmic ratio were the three most important factors to consider when producing a baseline for normal oral squames. n 1981, Cowpe & Longmore carried out a small survey on clinically normal buccal squames collected from a young adult sample; DNA content and an additional criterion known to vary in malignancy, namely nuclear area, were measured. All the specimens displayed diploid (normal) DNA distributions -indicative of non-replicating cell populations. There was no statistically significant variation in mean nuclear size amongst the sample as a whole. A further study on normal and abnormal oral squames, revealed identical findings for the series of young adult normal buccal smears (Longmore & Cowpe 1982). For the abnormal smears, the malignant lesions displayed DNA polyploidy and a marked elevation in mean nuclear size. However, some of the non-malignant lesions, although displaying diploid DNA distributions, also displayed a significant elevation in their mean nuclear area values. As well as this discrepancy one must question the validity of comparing normal young adult buccal smears with smears collected from lesions presenting in a variety of sites in the oral cavity and from patients who are generally in an older age bracket. Thus it was felt that a more extensive study on the effects of age, sex and site was required. Previous age-related surveys have, in the main, investigated the degree of keratinization of oral smears and its relationship to age. The results are somewhat contradictory. Some authors observed a decrease in the degree of keratinization with advancing age (Papic & Glickman 1950, Stone 1953), others observed no change (Montgomery 1951, Miller et al. 1951), while Jacobs (1959) reported an increase with age. n 1965, Zimmerman & Zimmerman observed a decrease in keratinization in palatal and gingival mucosa but no change in the buccal mucosa and the tongue. Since oral cancer is primarily a disease of adults and especially males who smoke, Brown & Young (1970) set out to determine a normal mucosal maturation pattern for males and females and ascertain any age or smoking influences. They applied a karyopyknotic index (K) to oral smears, concluding that this index was more reliable than the cornification index used by other authors, as cell colour alone was not a reliable indicator of epithelial maturity and not all areas of the oral mucosa underwent full maturation to anucleate cells. They detected an increase in the K for the tongue and a decrease in K for the palate with advancing age. The most comprehensive quantitative study on exfoliative cytology of buccal squames was carried out by Pappelis et al. (1973) using interference microscopy. They could detect no significant variation in nuclear and cell size for buccal squames with advancing age. However, this involved only a small sample of 11 males aged 6 to 80 years of age. Lee et al. (1973), using similar techniques, observed no significant variation in nuclear and cell size between buccal squames and lower labial squames. There have been few quantitative histological analyses of the effects of ageing on normal human oral epithelium. Studies have largely been restricted to postmortem examination. Scott et al. (1983), working on lingual epithelium, detected thicker epithelium in males than females but similar rates of ageing changes between the sexes. There was a reduction in nuclear/cytoplasmic ratio with advancing age. n view of the recent -developments of image analysis techniques, which are both more rapid and accurate than previous planimetric methods (Fraher 1980), the present study describes the application of a semiautomatic image analysis technique to normal oral mucosal smears. The aim of this definitive study was to

3 Journal of the Royal Society ofmedicine Volume 78 December detect any changes in nuclear area (NA), cytoplasmic area (CA) and nuclear/cytoplasmic ratio (NC ratio) values, for clinically normal oral mucosal smears in relation to age, site and sex, as previously outlined in a preliminary communication (Cowpe 1984). Thus an attempt was made to establish a baseline for these three variables with which to compare identical measurements made on pathological smears. Previous studies (Cowpe & Longmore 1981, Longmore & Cowpe 1982) displayed diploid DNA distributions for normal smears. Further confirmation of this was sought by carrying out DNA cytophotometry on a selection of the oral smears in this study, and it was hoped that the results would further aid the diagnostic sensitivity of quantitative cytological techniques in the management of early oral malignancy above and beyond subjective interpretation. Methods Smears were collected from the clinically normal oral mucosa of 105 patients, using wooden tongue spatulas. The spatula was scraped firmly across the site under investigation and then transferred to dry glass slides. nitially three sites, in 40 patients were investigated: the buccal mucosa, the floor of-the mouth and the junction of the hard and soft palate. Later the study was extended to include the lower labial mucosa (24 patients; 4 sites) and the dorsum of the tongue (41 patients; 5 sites). There were 55 females and 50 males, aged between 12 and 87 years. Each patient was either under review or an inpatient awaiting routine dental extractions. Name, number, age, occupation and relevant medical history were recorded, as well as the site investigated and the coded slide number for each smear. Each patient underwent routine venepuncture to determine haemoglobin levels and exclude cases of anaemia. Following fixation and staining of the smear, the quantitative techniques of DNA cytophotometry and cytomorphology were applied. DNA cytophotometry: For every tenth patient, DNA measurements were carried out for each site in that patient. This included a selection of patients over a wide age range. The slides were fixed immediately in methanol; formalin; acetic acid (85:10:5) and underwent Feulgen hydrolysis as previously described (Cowpe & Longmore 1981, Longmore & Cowpe 1982). DNA measurements were carried out using a Vicker's M85 scanning and integrating microdensitometer. For each smear the Feulgen DNA content of 50 randomly selected nuclei was measured and the values were fed into a computer to obtain a DNA distribution histogram. Cytomorphology: For cytomorphology (measurement of NA, CA, NC ratio) each smear was fixed immediately in equal parts diethylether and 95% ethanol for one hour, followed by washing in running tap water for a further hour. Smears were then stained with Papanicoloau stain. Each slide was placed on a motor-driven stage (mechanical stage EK17-316, Micro nstruments - Oxford Ltd) attached to the microscope. For each smear the stage was moved in a roster fashion across the slide. Cell and nuclear boundaries of 50 randomly selected cells were traced. The tracings were carried out using a Leitz drawing tube attached to the microscope at x 100 under oil immersion (Figure 1). t was important to trace clearly defined cells and to avoid clumped or visibly folded cells. From these tracings the NA, CA and NC ratio for each cell was measured using a Reichert digiplan image analyser (Figure 2) (Heidelberg, West Germany) interfaced to a Pet 8032 computer (Commodore Business Machines, Santa Clara, California). Previously, computer programs had been designed to collect the data, perform the calculations and print out each value for NA, CA and NC ratio, mean values and a histogram for each of the three variables, for each specimen. Results Clinically normal oral smears were collected from 5 selected sites in the oral cavity of 105 patients: 105 smears were collected from the buccal mucosa and the floor of the mouth; 103 from the junction of the hard and soft palate; 65 from the lower labial mucosa; and 41 from

4 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.....w Journal of the Royal Society ofmedicine Volume 78 December 1985 Figure 1. Leitz drawing tube and motor-driven stage attached to the microscope, with tracing in situ Fge.ehrdinia aas Figure 2. Reichert digiplan image analyser Q the dorsum of the tongue. Only 8 smears- all confined to palatal mucosa and lower labial mucosa - produced insufficient cells for nuclear and cell measurement. Smears collected from the buccal mucosa and floor of mouth produced an abundant crop of cells, although clumping was present in some areas. n other areas the cells were homogeneously dispersed with clearly stained nuclei and cytoplasm and well demarcated cell boundaries. Lower labial smears were difficult to collect due mainly to the dry nature of the mucosal surface once the lip was everted. This tended to produce clumping and a reduced cell harvest. The palatal smears displayed a large number of anucleate squames. However, the distribution was homogeneous, with adequate numbers of cells to measure. This site required firmer application of the spatula during cell collection. The dorsum of the tongue was an easy site to smear, producing more than adequate numbers of cells for measurement. t was especially important to clean this site prior to smearing to avoid debris within the fissures of the tongue and to avoid collecting squames exfoliated from other sites in the mouth.

5 Journal of the Royal Society ofmedicine Volume 78 December NA, UCCAL MUCOSA NA /m: FLOOR OF MOUTH 40 ;i5si : ; 110 a* : * * 0 00 _ a -0. a* 60W :1. F X b ; M a a ;. r AGE L' _~~~~~~ i ~~~~~~~~ M F M F 1S F 11 F ~60 AGE NA M.' 100 LOWER LP NA,.-' 100- PALATE NA zm: 100- TONGUE go nm a0 0 9 n a. a. *. 80- : l: p L a i L. *L : 40 * so -!; m a _ -_ i _ F-_ - l M F M F M F M F F F M F X F M F F > 60 < >60 < >-0 vs0 AGE AGE AGE Figure 3. Scattergrams of mean NA values for the five sites under investigation within the four age groups DNA measurements: For each specimen, undergoing DNA cytophotometry, the clinically normal oral squames displayed diploid (normal) DNA distributions. Cytomorphology: Patients were divided into four age groups for statistical analysis, these being <21, 21-40, and >60 years. For each site, within each age range, there was a wide variation in the mean values for all three variables, as illustrated in Figures 3, 4 and 5, in which scattergrams of mean NA, mean CA and mean NC ratio values have been plotted for the five sites over the four age ranges. Despite the variation within each site, there were broad similarities between some of the sites as was revealed by the statistical analysis. Statistical analysis was carried out and % frequency distribution histograms constructed for the mean NA and CA values, with regard to site and age. The buccal mucosa, floor of mouth and lower lip displayed broadly similar distributions and overall mean values both for NA and CA. The tongue and palate displayed similar distributions and overall mean values, with essentially a lower range of values compared with the other three sites. The NA distributions were broadly similar for the four age groups with a shift of the histograms to the left over the age of 40. For CA the distributions were broadly similar for all age groups. A two-way analysis of variance for mean NA and mean CA, broken down by site and sex revealed a highly significant variation in mean NA (F = 38.2) and mean CA (F = 38.5) amongst the five sites. However the mean NA and mean CA values displayed no significant variation

6 1000 Journal of the Royal Society ofmedicine Volume 78 December 1985 CA HmO BUCCAL MUCOS CA,..-' FLOOR OF MOUTH < H 0 O L * H 0 AGE AGE CA..: LOWER LlF CA m" FALATH CAr TONGUE 2= seoo HOHO - sooo 50HO H *H LLH < > 60 < so" < 21 1 so0> 40 1 AGE~~~~~~~~G AGEGG 0 *M F M F M F M F M F F F 11 F 11 Hoo HOo * * o H_* CA >P>0.0 LOWE (>60). mean CA There AF=3.-' waspaate also significant e varithiestogesthrwaasinfct CA,..-' Hma N A00 2~ bewe mae HOd feae F=09 o A.5fr A.Atowyaayi fvrac 4h 2 gegopad l -rus he other O.>P ) <0 (16) 21H0 and H10aegop 00>>.2,btn infcn aito ewe h and ag grup > 0.2) but no siniicn vaato etentho16 Fgr4.SaegrmofmaCAvalues Fgr4.SatrrmofmaCAvalues betweenesiesude foree nvsigtallihn thaiestsunelnetgtinwtilh hfour age groups. ariabetween fromstheoand therafou sitoes (P3.)For meanccaa values, the buccal mucosaris- betweent' malest and fealeso( =se 0.9 fompretema NA:,16oCA) antwo-wayoanalysisofearanee broendownvb sites(1 andr wageaginevaeohgl significant variation formean NAvausbtenheucl (Fcsa =lo 358)an meanuat(=n 34.0) vlpues ameongstte five siltes Therelase a significant variation trokngu displayed inomeanenahevalues t n geaanrveldhgl atithes 5%< significant level (F2.6), variation meno but uoa froohrucl sigifianus,thvariatini loro otn loern mcoan lip Cipaydsgiiatyeetd vaipio<000annavaue;a with adanin values agee(fu=ie0.44).; h pltead h Usingu dstudyentstts,dhr there% significant evl (was63, u no a significant ine elevationfro thca variation vuoa,lues f hbeweihnlo formeuhand ot n NAbeweenli h loerali 0.05> P002) (>0)there was alo significant variation inwentelo mea NAe values betee theloeli played significantly elevated NC ratios from floor of mouth, lower lip and palate (P<0.001) and from the tongue (0.1 >P>0.05); the tongue displayed significant variation from the palate

7 Journal ofthe Royal Society ofmedicine Volume 78 December NC Rato 8UCCAL MUCOSA NC Rao FLOOR OF MOJTH r _ S *, t k M F M F M F M F >0 < >60 AGE AGE hl F M F F NC R0.io LOWER LP NC Ra.o PALATE NC Raoe TONGUE ri 06 * 06 F B@ * * i JF M F 6 F- 6 F M F 6 F 6 F ia6 F F F N 6 F >606< >0-o.<21 M >06 AGE AGE AGE Figure 5. Scattergrams of mean NC ratio values for the five sites under investigation within the four age groups and lower lip (0 05 > P>0.02); but there was no significant variation between floor of mouth, palate and lower lip. Whilst the mean NA and mean CA values displayed normal distributions, the mean NC ratio values did not. Comparison of the mean values for NA with those for CA showed evidence of a slight positive correlation (0.36), as was also the case when comparing the mean NA and NC ratio values. However, there was a strongly negative correlation between the mean CA and mean NC ratio values (0.72). Although this indicates that it is impossible to predict the size of a cell from the NA value, or the size of a nucleus from the CA value, from the statistical point of view it was established that both NA and CA were suitable criteria to measure, but NC ratio displayed such marked variation that it was considered to be of no statistical value. Twenty-six of the 105 patients were smokers (defined as such if smoking at least one packet of cigarettes, three cigars or three pipes per day: Brown & Young 1970). They were evenly distributed between males and females, the majority being in the age group. There was no clear deviation of the NA, CA and NC ratio values between smokers and non-smokers. None of the haemoglobin levels indicated a possible state of anaemia. Discussion n this study quantitative techniques have been applied to smears collected from clinically normal oral mucosa, matched for age, site and sex, in an attempt to define a baseline for comparison with pathological smears. t was hoped that if this was achieved, the diagnostic

8 1002 Journal of the Royal Society ofmedicine Volume 78 December 1985 potential of exfoliative cytology in the diagnosis of early oral malignancy would be greatly enhanced. At present the diagnosis of oral malignancy relies primarily on subjective histopathological interpretation of the degree of epithelial dysplasia present in tissue sections (WHO 1978, Kramer 1980, Pindborg 1980). Attempts have been made to quantify the changes occurring in premalignant and malignant tissue sections using stereology and computer-aided analysis (Kramer et al. 1970a, b, Franklin & Smith 1980). From previous pilot studies (Cowpe & Longmore 1981, Longmore & Cowpe 1982) it was felt that quantitative techniques might improve the diagnostic sensitivity of exfoliative cytology in the detection of early oral malignancy. n the latter study the results obtained for a variety of lesions from different sites in older patients were compared with the results obtained for young adult buccal smears. However, premalignant lesions such as leukoplakia and erythroplakia present primarily in middle aged and older patients. n addition, lesions which progress to malignancy may present in a selection of different sites within the oral cavity (Shafer & Waldron 1975, Waldron & Shafer 1975, Kramer 1980, Pindborg 1980). Thus it is important to determine any variation in the quantitative measurements carried out on clinically normal oral smears prior to comparison with measurements made on- pathological smears. The advantage of a histological section is that it displays the cells of a lesion in relation to one another, to the underlying connective tissue and preferably in relation to the surrounding normal tissue. This is not the case in cytology which can only be of value when a sufficient number of cells are collected in each smear. These cells must be representative of the distinct variations which occur in all layers of the epithelium (Staats & Goldsby 1963, Hayes et al. 1969, Blozis 1972, Folsom et al. 1972). n this study only 8 smears out of a total of 411 produced an insufficient number of cells for measurement. The scraping action varies each time a smear is collected and there is a degree of variation in the susceptibility of individual mouths to smearing. There is also a variation in the susceptibility to smearing of individual sites within a single mouth. The use of a single operator would appear to ensure maximum reproducibility. Previous false-negative results for pathological smears may be due, in the majority of cases, to the technique of smear collection (Allegra et al. 1973). For nearly all the specimens in this study, the method of smear collection produced a more than adequate harvest of cells but keratinized areas, such as the palate, required a firmer scraping action than the unkeratinized areas like the buccal mucosa. As in the previous studies, all the clinically normal specimens which underwent DNA cytophotometry displayed diploid (normal) DNA distribution histograms, thus confirming that the cells collected from each site in the oral cavity, irrespective of age or sex, represented non-replicating cell populations. Nuclear size varied significantly with advancing age, but there was only slight variation amongst the younger age groups which increased as age progressed. There was no significant variation in total cell area with advancing age. ndividual smears confirmed previous reports of the wide variation in size, shape and staining characteristics of the cells under investigation (Montgomery 1951, Sandler et al. 1960). Nuclear and cell size varied significantly between the five sites under investigation; however there was close correlation between the buccal mucosa, floor of mouth and lower lip and between the palate and the tongue. Schroeder (1981) has described three types of oral mucosa whose individual structure and composition are related to their functional requirements: lining, masticatory and specialized. Lining epithelium is normally unkeratinized, whereas the other types are usually keratinized. This may account for the similarity in values between the non-keratinized sites (buccal mucosa, floor of mouth, lower lip) and between the keratinized sites (palate and dorsum of tongue). There was less variation in the NA, CA and NC ratio values between the five sites within each individual, than amongst the sample as a whole. These results, from a larger number of individuals, contradict the findings of Pappelis et al. (1973) and Lee et al. (1973), in that they display a significant variation in nuclear and cell size between the five sites and significant variation in nuclear size over a wide age range. From a statistical point of view, although there was a slight positive correlation between the

9 Journal of the Royal Society ofmedicine Volume 78 December mean NA and mean CA values, it was still not possible to predict cell size from nuclear size and vice versa. NC ratio values displayed a significant variation with advancing age and between the five sites under investigation, and thus statistically NC ratio has proved to be an unsatisfactory criterion to consider. There was no significant-variation in NA, CA or NC ratio between males and females. n 1951, Montgomery detected no relationship between the degree of keratinization of oral squames and the menstrual cycle. However, more recent reports are somewhat contradictory. Some authors observed that oral squames were not a reliable indicator of the stage of the menstrual cycle (Jacobs 1959, Cohen 1966, Brown & Young 1970, Dokumov & Spasov 1970) while others detected some correlation between oral and vaginal smears during the menstrual cycle (Ziskin & Moulton 1948, Main & Ritchie 1967, Anderson et al. 1969, Murthy & Thomas 1977, Quaranta et al. 1980). n unpublished observations we have detected no relationship between nuclear and cell size of buccal smears and the stage of the menstrual cycle using the techniques outlined above. n this study, evidence of anaemia was sought using haemoglobin estimations; these may, however, be insufficient for assessing the presence or absence of anaemia. t may be necessary to carry out a more extensive blood screen, measuring factors such as serum iron, serum transferrin, serum B12 and folate and red cell folate, and this is underway at the present time. Previous studies on the effects of anaemia on oral squames are by no means conclusive. The majority of authors agree that B12 deficiency results in an increase in nuclear size which returns to normal following treatment therapy (Boen 1957, Nieburgs et al. 1962, Boddington & Spriggs 1959, Boddington 1959). Finch (1971) detected no change in the nuclear size of buccal squames in patients on the antifoliate drug methotrexate. The effects of iron deficiency on nuclear and cell size are contradictory (Boen 1957, Boddington 1959, Monto et al. 1961) and further detailed studies on the effects of anaemia on oral smears are required. Brown & Young (1970) and Baric et al. (1982) stated that leukoplakias and oral carcinomas were prevalent in people who smoked. The former authors, using a karyopyknotic index (K), detected an increase in K for the palatal, lingual and buccal mucosa of patients who smoked. There were few smokers in the present study, and their NA, CA and NC ratio values displayed no obvious variation from the sample as a whole. Again a further detailed investigation is indicated. n conclusion, the use of the semi-automated image analysis technique outlined in this study has shown that a cytomorphological baseline for clinically normal oral squames matched for age and site is unattainable. Previous studies have suggested that comparison of clinically normal oral mucosa with clinically suspicious mucosa from the same patient, presenting with oral cancer, would be of more diagnostic value (Peters 1958, Doyle & Manhold 1975). Already the quantitative techniques described in this study have been applied to suspicious lesions presenting in the oral cavity. n these cases an 'intrapatient control' smear of clinically normal oral mucosa has proved very reliable and the results are sufficiently encouraging to warrant further investigation. Despite the drawback of a normal baseline being unattainable for oral mucosal smears, the quantitative techniques in exfoliative cytology described here display high accuracy and reproducibility.,the accessibility of the oral cavity lends itself to the technique of cytology, a procedure which is noninvasive and readily acceptable to patients. The simplicity and the reproducibility of these quantitative procedures have already, in a continuing investigation of pathological smears, demonstrated the value of exfoliative cytology in the monitoring of clinically suspicious lesions and in their follow up after surgery, radiotherapy or chemotherapy. Although still primarily acting as an adjunct to biopsy, standardization of these cytological techniques may reduce the need for multiple biopsies in patients on long-term follow up of clinically innocuous but suspicious oral lesions. References Aliegra S R, Broderick P A & Corvese N (1973) Acta Cytologica 17, Anderson W R, Belding J & Pixley E (1969) Acta Cytologica 13, Baric J M, Alman J E, Feldman R S & Chauncey H H (1982) Oral Surgery, Oral Medicine, Oral Pathology 54, Bernstein M L & Miller R L (1978) Journal of the American Dental Association 96,

10 1004 Journal of the Royal Society ofmedicine Volume 78 December 1985 Blozis G G (1972) nternational Dental Journal 22, Boddington M M (1959) Journal ofclinical Pathology 12, Boddington M M & Spriggs A (1959) Journal ofclinical Pathology 12, Boen S T (1957) Acta Medica Scandinavica 159, Bohm N & Sandritter W (1975) Current Topics in Pathology 60, Brown A M & Young A (1970) Acta Cytologica 14, Cohen L (1966) Oral Surgery, Oral Medicine, Oral Pathology 21, Cowpe J G (1984) Journal of the Royal Society ofmedicine 77, Cowpe J G & Longmore R B (1981) Journal oforal Pathology 10, Dokumov S & Spasov S A (1970) Acta Cytologica 14, Doyle J L & Manhold J H (1975) Journal ofdental Research 54, Finch R R (1971) Acta Cytologica 15, Folsom T C, White C P, Bromer L, Canby H F & Garrington G E (1972) Oral Surgery, Oral Medicine, Oral Pathology 33,61-74 Fossa S D & Kaalhus 0 (1977) Acta Pathologica et Microbiblogica Scandinavica 85, Fraher J P (1980) Journal ofanatomy 128, Franklin C D & Snith C J (1980) Journal ofpathology 130, Goldsby J W, Newton G L & Staats 0 J (1964) Acta Cytologica 8, Grove G L (1979) Journal ofhistochemistry and Cytochemistry 27, Hayes L H, Berg G W & Ross W L (1969) Journal of the American Dental Association 79, Hillman R W, Smith H S & Levine J S (1972) Journal of Gerontology 27, Jacobs A (1959) Journal of Clinical Pathology 12, Kramer R H (1980) n: Oral Premalignancy. Ed. C Mackenzie et al. University of owa Press; p 34 Kramer R H, Lucas R B, El-Labban N & Lister L (1970a) British Journal of Cancer 24, Kramer R H, Lucas R B, El-Labban N & Lister L (1970b) British Journal of Cancer 24, Lee L-H, Pappelis A J, Pappelis G A & Kaplan H M (1973) Acta Cytologica 17, Longmore R B & Cowpe J G (1982) Analytical and Quantitative Cytology 4, Main D M G & Ritchie G M (1967) British Journal ofdermatology 75, Miller S C, Sobernan A & Stahl S S (1951) Journal of Dental Research 30, 4-11 Montgomery P W (1951) Journal ofdental Research 30, Montgomery P W & Von Haam E (1951) Journal ofdental Research 30, Monto R W, Rizek R A & Fine G (1961) Oral Surgery, Oral Medicine, Oral Pathology 14, Morrison L F, Hopp E S & Wu R (1949) Annals of Otolaryngology 58, Murthy U K & Thomas J A (1977) ndian Journal ofmedical Research 66, Nieburgs H E, Herman B E & Reisman H (1962) Laboratory nvestigations 11, Papanicolaou G N & Traut H F (1941) American Journal ofobstetrics & Gynaecology 42, Papic M & Glickman (1950) Oral Surgery, Oral Medicine, Oral Pathology 3, Pappelis G A, Pappelis A J & Courtis W S (1973) Acta Cytologica 17, Peters H (1958) American Journal ofclinical Pathology 29, Pfitzer P & Pape H-D (1973) Acta Cytologica 17, Pfitzer P & Pape H-D (1975) Journal ofmaxillo-facial Surgery 3, Pindborg J J (1980) Oral Cancer and Precancer. John Wright & Sons, Bristol; ch 14 pp ; ch 2 p 16; ch 13 p 21 Quaranta C, Pica A & Galgano A (1980) Archives Stomatologica (Napoli) 21, Sandler H C, Stahl S S, Cahn L R & Freund H R (1960) Oral Surgery, Oral Medicine, Oral Pathology 13, Schroeder H F (1981) Differentiation of Human Oral Stratified Epithelia. S Karger, Basel, Miinchen; ch 1 Scott J, Valentine J A, Hill C A & Balasooriya B A W (1983) Journal de Biologie Buccale 11, Selbach G J & Von Haam E (1963) Acta Cytologica 7, Shafer W G & Waldron C A (1975) Cancer 36, Silverman S jr, Bilmoria K F, Bhargava K, Mani N J & Shah R A (1977) Acta Cytologica 21, Sprenger E & Witte S (1979) Journal ofhistochemistry & Cytochemistry 27, Squier C A (1980) n: Oral Premalignancy. Ed. C Mackenzie et al. University of owa Press; pp Staats 0 J & Goldsby J W (1963) Acta Cytologica 7, Stone A (1953) Journal ofdental Medicine 8, Waldron C A & Shafer W G (1975) Cancer 36, WHO Collaborating Centre for Oral Precancerous Lesions (1978) Oral Surgery, Oral Medicine, Oral Pathology 46, Zimmerman E R & Zimmerman A L (1965) Journal ofdental Research 44, Ziskin D E & Moulton R (1948) Journal of Clinical Endocrinology 8,