ORIGINAL ARTICLE

TREATMENT AND FOLLOW-UP OF ORAL DYSPLASIA — A

SYSTEMATIC REVIEW AND META-ANALYSIS
Hisham M. Mehanna, MBChB, FRCS,1 Tim Rattay, MBChB, MRCS,1
Joel Smith, MBChB, MRCS,1 Christopher C. McConkey, MSc2
1
Department of Head and Neck Surgery, Institute of Head and Neck Studies and Education,
Department of Head and Neck Surgery, University Hospitals Coventry and Warwickshire,
Coventry, United Kingdom. E-mail: [email protected]
2
Clinical Trials Unit, Warwick Medical School, University of Warwick, Coventry, United Kingdom

Accepted 18 February 2009

Published online 19 May 2009 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/hed.21131

This suggested the need for excision and continued

Abstract: Background. The aim of this study was to inform
surveillance. V
C 2009 Wiley Periodicals, Inc. Head Neck 31:
an evidence-based management policy for oral dysplastic
1600–1609, 2009
lesions.
Methods. Systematic review was performed with meta- Keywords: dysplasia; leukoplakia; oral lesion; malignancy;
analysis. Studies reporting follow-up of patients with histologi- follow-up
cally confirmed oral dysplasia were included. Outcome mea-
sures included malignant transformation rate (MTR) and time
to malignant transformation (TMT). Subgroup analysis was
performed by histologic grade, clinical risk factors, and
Oral dysplasia is a relatively common premalig-
treatment modality. Heterogeneity was assessed.
nant condition, which affects approximately 2.5
Results. Fourteen nonrandomized studies, reporting on 992 to 5 per 1000 of population. It is usually seen by
patients, were included. There was considerable heterogeneity general medical and dental practitioners in the
between studies: mean overall MTR ¼ 12.1% (confidence inter- form of leukoplakia (a white patch that does not
val: 8.1%, 17.9%) and mean TMT ¼ 4.3 years. Histologic rub off), which affects 1% to 2.5% of the popula-
grade significantly affected mean MTR (p < .008). Lesions that
were not excised demonstrated considerably higher MTR than
tion at any one time.1
those that were excised (p ¼ .003). Oral dysplasia can only be diagnosed histologi-
Conclusions. Oral dysplasia showed a significant rate of cally. In oral dysplasia, cells of the normal oral ep-
transformation to cancer, which was related to grade and ithelium are replaced by cells showing immature
was decreased significantly but not eliminated by excision. or inappropriate differentiation with a resem-
blance to cells usually seen in malignancy.2 The
importance of oral dysplasia lies in that a propor-
Correspondence to: H. M. Mehanna tion of these lesions can progress to cancer. How-
This study was presented at the American Head & Neck Society Annual ever, reports on the risk of progression to cancer
Meeting, San Francisco, USA, July 23, 2008, and British Association of
Head and Neck Oncologists Annual Meeting, London, United Kingdom, vary considerably from 6% to 36%.3
April 25, 2008 (poster). At present, the only effective treatment is
V
C 2009 Wiley Periodicals, Inc. surgical excision, and there are no effective

1600 Treatment and Follow-Up of Oral Dysplasia HEAD & NECK—DOI 10.1002/hed December 2009
medical treatments.4 Even surgical excision car- Table 1. Description of search strategy.
ries a high risk of recurrence (up to 35%).5 Fur-
thermore, resection of large lesions can cause No. Search term
significant morbidity and may sometimes 1 Dysplasia and oral
require extensive reconstructive techniques. 2 Dysplasia and mouth
As a result of these controversies, there is 3 1 or 2
4 3 and progression
currently no consensus over the treatment and
5 3 and follow-up
follow-up required for patients with oral dyspla- 6 3 and treatment
sia. Management strategies vary from incisional 7 3 and cohort
biopsy aiming to exclude cancer to complete 8 3 and natural history
excision with an adequate margin. Reported fol- 9 3 and recurrence
low-up strategies vary from immediate dis-
charge to life-time follow-up.6
To develop an evidence-based management for the same patient group, the data for the lon-
and surveillance policy for oral dysplasia, we gest follow-up period were used. When an
undertook a systematic review and meta-analy- abstract was published but no full study was
sis of observational cohort and case-controlled found, it was agreed that the authors would be
studies to assess the risk of and interval to pro- contacted, but no such instances were identified
gression to cancer in patients diagnosed with from the search.
oral dysplasia. We also examined the effects Studies that reported on patients with oral
that histologic grade, clinical risk factors, and leukoplakia without histologically confirmed di-
treatment strategies may have on this. agnosis of dysplasia were excluded. Cross-sec-
tional studies of the prevalence of oral dysplasia
were also excluded. Data from studies in which
PATIENTS AND METHODS patients with dysplasia were not reported sepa-
rately or could not be extracted from the provided
Search Strategy. In accord with the meta-analy- data were excluded. Patients with synchronous
sis of observational studies in epidemiology cancer at or identified within 3 months of the
(MOOSE) guidelines, the investigators searched time of diagnosis of dysplasia were also excluded
Medline (1966 to June 2008), Embase (1980 to from the meta-analysis. Studies with less than 1
June 2008), and Cochrane databases (CENTRAL, year follow-up were also excluded.
Cochrane Library, 1995 to June 2008). The last
search was performed on June 4, 2008. Our core Outcome Measures. The following outcome
search terms were oral and dysplasia, mouth and measures were extracted from the studies, as
dysplasia. Each of the searches were then com- they were clinically relevant: the rate of malig-
bined with each of the following terms: progres- nant transformation to oral cancer and the time
sion, follow-up, treatment, cohort, natural history, interval or duration to malignant transforma-
and recurrence (Table 1). Reference lists of tion. Where available, data were subclassified
obtained studies, personal reference lists, and by histologic grade of the lesion (mild, moderate,
existing reviews were also searched manually. severe dysplasia, or carcinoma in-situ [CIS]) and
by management strategy (surgical excision or
Selection Criteria. To be included in the review, incisional biopsy). The relative risks of malig-
studies had to report on patients with a histo- nant transformation for proven or potential clin-
logically confirmed diagnosis of oral dysplasia. ical risk factors of progression—gender, site of
They also had to study at least one of the out- lesion, continuation of smoking, and alcohol
come measures and one intervention method or intake—were also extracted.
clinical risk factor. In studies on oral lesions
which contained a defined subset of patients Data Abstraction. Eligibility of the studies
with oral dysplasia, this subset but not all obtained from the search was determined inde-
patients with oral lesions were included in the pendently from the abstracts by two reviewers
meta-analysis. Observational studies were (T.R., J.S.) blinded to each other’s selections.
included in the review because there was a pau- Both reviewers underwent a training period in
city of randomized controlled trials with which 10 studies were assessed together with
adequate follow-up. For multiple publications the lead author (H.M.). If a citation was deemed

Treatment and Follow-Up of Oral Dysplasia HEAD & NECK—DOI 10.1002/hed December 2009 1601
eligible by at least one reviewer, the article was Where sufficient data existed, heterogeneity was
subjected to review. Data were then extracted studied in the statistical model.
from the studies into a Microsoft Excel Transformation rates were modeled by apply-
spreadsheet. ing a random effects logistic regression model
that assumed a fixed effect of the underlying
Quality and Confounding Assessment. As there is mean log-odds of transformation rate with a ran-
no validated tool for assessing quality in system- dom study effect. This allowed estimates of the
atic reviews of observational studies, quality cri- underlying mean and of the effect of covariates
teria were agreed upon a priori, adapted from such as dysplasia grade and clinical risk factors.
the levels of evidence for prognostic studies,7 When analyzing effects of dysplasia grade,
and adapted from Greenhalgh,8 Khan et al,9 the mild and moderate subgroups were almaga-
and Laupacis et al.10 Included in the criteria mated, as were severe dysplasia and CIS. This
were attempts to address confounding factors was done because it is widely acknowledged
within individual studies, length and rate of that histological grading of oral dysplasia shows
follow-up achieved, and blinding of assessment significant intraobserver and interobserver vari-
of outcome. The quality of the studies was ability, especially for the two diagnoses in each
assessed independently by two reviewers (T.R., group.11
J.S.). When there was disagreement, a third
reviewer (H.M.) was consulted.
RESULTS
Statistical Analysis. Outcome data were abs-
tracted independently by a researcher (T.R.) and a Description of Studies. The search strategy pro-
statistician (C.M.) and checked by a third reviewer duced 2837 articles in total, of which 28 were
(H.M.). A funnel plot was generated showing indi- deemed eligible and subjected to full review
vidual MTRs in each study by study size. In this (Figure 1). After applying the selection criteria,
plot, the rates are variance-stabilized using the eight cross-sectional studies and four observatio-
arc-sine square-root transformation because most nal cohort studies on oral lesions without histo-
were below 20%. Heterogeneity was assessed logically confirmed dysplasia were excluded.12–15
graphically in the forest plot and the exact bino- There were no randomized-controlled trials or
mial confidence intervals (CIs) were calculated. case-control studies. After exclusion of two studies

FIGURE 1. Flowchart of search results. [Color figure can be viewed in the online issue, which is available at www.interscience.
wiley.com.]

1602 Treatment and Follow-Up of Oral Dysplasia HEAD & NECK—DOI 10.1002/hed December 2009
due to duplication of data,16,17 14 prospective

Nonsurgically

patients
treated
and retrospective studies were identified (Figure

158

651
0
21
41

70
55
44
0
47
22
90
23
35
45
1). In total, these reported on 992 patients who
satisfied our inclusion criteria (Table 2).
Mean or median age of the study population
ranged from 47.5 to 60.8 years and was given in

Surgical
excision
patients

0
0

0
0
166
67
50
0
0

13
0
0
0
45

341
7 studies. In the absence of access to individual
patient data, it is important to note that these
averages were derived from averages given for

dysplasia
all patients in those studies which reported on

patients
Total

166

158

992
88
91

70
55
44
13
47
22
90
68
35
45
all oral leukoplakia lesions and not just from
patients with confirmed dysplasia.
Median or mean follow-up ranged from 1.5 to

Mean or
10 years and was stated in all but 1 study. Mini-

median
age, y
47.5
60.8
54.0

57.0
59.3

57.7
54.0
NR
NR

NR

NR
NR

NR
>35
mum follow-up, reported in 7 studies, is short,
ranging between 6 and 12 months.
Malignant transformation rate (MTR) was

follow-up,
reported in all studies. Subgroup analysis by

Min.
Table 2. Description of studies included in the meta-analysis.

NR
1.0
0.6
NR
NR
0.5
0.5
1.0
NR
0.5
NR
NR
NR
1.0
y
histologic grade was reported by 9 studies.
Seven studies reported time interval to malig-
nant transformation (TMR). Surgical treatment

follow-up,
of dysplastic lesions by excision was specified in

Mean

3.6
6.0
4.0
NR
7.0
2.5
1.5
2.5
9.3
7.2
8.5
6.3
2.0
8.0
y
only 5 of 14 studies. The remaining 9 studies
reported on biopsy-confirmed lesions without
surgical excision. The association of malignant

1991–2001
1977–1997
1976–1997
1975–1994
1988–1998
1973–1997
1974–1982
1969–1984
1970–1980

1967–1971
1963–1971
enrollment
progression with clinical risk factors was

Not clear
<1968–?
Date of

1966–?
reported in 4 studies.

Quality Assessment of Studies. Three studies

were population based. There were 2 prospective

Retrospective pathology laboratory

studies from India (level 1 evidence),30,32 and 1
retrospective study from Northern Ireland (level
2).21 The remainder of studies were single-insti-
Retrospective population
Methodology
and setting

Prospective population

Prospective population
tution case series from Eastern Asia, North
Retrospective hospital
Retrospective hospital

Retrospective hospital

Retrospective hospital
Retrospective hospital

Retrospective hospital
Prospective hospital

Prospective hospital

Prospective hospital

Prospective hospital
America, and Western Europe (level 4). They
used pathology slides, medical records, or ques-
tionnaires sent to medical and dental practi-
tioners to collate data. The results of the quality
assessment of the studies are summarized in
Table 3.
The Netherlands

The Netherlands
Northern Ireland

Abbreviations: min, minimum; NR, not reported.

Malignant Transformation and Heterogeneity of

Studies. The pooled estimate for the mean
Country

MTR for all studies is 12.1% (CI: 8.1%, 17.9%),

Denmark

Hungary
Norway
Taiwan

with a wide variation between studies (0%–

Japan

India

India
USA

USA

USA

USA

36.4%). The rate is little affected by the mean

follow-up time of the study (p ¼ .26; Figure 2).
To assess possible bias in the studies and
Banoczy and Csiba29

examine heterogeneity, a funnel plot is shown

Schepman et al23
Lumerman et al24
Holmstrup et al19

Hogewind et al25

Silverman et al27

Silverman et al30

(Figure 3). In this plot, the studies’ individual

Cowan et al21

Mincer et al31
Gupta et al28

MTRs are variance-stabilized using the arc-sine

18

Saito et al20

Lee et al22

All studies
Hsue et al

square-root transformation because most were

Lind26
Study

below 20%. The meta-analysis shows a degree of

heterogeneity between studies when comparing

Treatment and Follow-Up of Oral Dysplasia HEAD & NECK—DOI 10.1002/hed December 2009 1603
 Table 3. Quality assessment of studies included in review.

1604
Method of
Assessment Sampling outcome Blinding of Adequacy of
Methodology Level of of exposure method Cohort assessment outcome follow-up
Study* and setting evidence (dysplasia) bias characteristics (cancer) assessment rate >80% Comments
18
Hsue et al Prospective 4 Pathology Yes All patients smoked and Medical records No Yes Data set checked
hospital records chewed betel; cancer against national
within 6 months excluded cancer registry to
allow for drop-outs
Holmstrup Retrospective 4 Medical Yes Concurrent cancer Medical records No No Surgical versus
et al19 hospital records excluded; all non-surgical
smokers treatment was
encouraged to not randomized
quit
Saito et al20 Retrospective 4 Medical Yes Nonsurgical Medical records No No Surgical versus
hospital records treatments included nonsurgical
cryotherapy and laser treatment was

Treatment and Follow-Up of Oral Dysplasia

not randomized
Cowan et al21 Retrospective 2 Pathology No Concurrent cancer excluded; Medical records No Yes Data for all malignant
population slides slides were reviewed and transformers
reclassified if available; 7 of 24
necessary transformed in
1 mo hence excluded
Lee et al22 Prospective 4 Biopsy Yes All patients received high-dose Biopsy No Yes Sample includes
hospital isotretinoin followed by patients with
low-dose isotretinoin versus hyperplastic lesions
beta-carotene maintenance in high-risk sites
Schepman Retrospective 4 Pathology Yes Concurrent and previous Medical records No No Not subdivided by
et al23 hospital records malignancy excluded grade of dysplasia
Lumerman Retrospective 4 Biopsy Yes Slides were reviewed and Questionnaire No No Only about 25%
et al24 pathology reclassified if necessary administered to response rate
laboratory dental practitioners to follow-up
questionnaire
Hogewind Retrospective 4 Medical Yes Concurrent cancer excluded Medical records No No Some lesions treated
et al25 hospital records by cryosurgery
Lind26 Retrospective 4 Pathology Yes Concurrent cancer excluded Medical records No Yes Data set checked
hospital records against national
cancer registry
to allow for drop-outs

HEAD & NECK—DOI 10.1002/hed

Silverman Prospective 4 Biopsy Yes Lesions > 1 cm size, Biopsy No Yes
et al27 hospital persistent for > 6 months,
concurrent
cancer excluded
Gupta et al28 Prospective 1 Biopsy No Random sample from villagers Biopsy No Yes Ten-year follow-up
population in 5 Indian districts, varying study but patients
(Continued)

December 2009
 MTR against study size, but there is little evi-

recruited during first

hence shorter mean

dence of the asymmetry of publication bias.

3 years of study

dysplasias from
If the 3 outliers in the funnel plot are

Comments

Excluded all mild

excluded, all of which are prospective stud-
ies,22,30,33 the overall MTR for all studies and CI

follow-up

follow-up
are reduced slightly (11.3%; CI: 8.4%, 15.1%).

Malignant Transformation and Histologic Grade. Ta-

Adequacy of

rate >80%

ble 4 shows a clear difference between the

follow-up

Yes
transformation rate for mild to moderate dyspla-
No

No
sia, estimated from the logistic model as 10.3%,
(CI: 6.1, 16.8%), compared with severe dysplasia
and CIS, and 24.1% (13.3, 39.5%; p < .008).
assessment
Blinding of
outcome

No

No

No
Malignant Transformation and Treatment Modality-
Patients whose lesions were not excised surgi-
Table 3. Quality assessment of studies included in review (Continued).

cally reported considerably higher overall

transformation rates when compared with
assessment

Medical records
Method of
outcome

(cancer)

patients who underwent surgical excision

(14.6% vs 5.4%). After adjusting for grade in the
logistic model, this was still statistically signifi-
Biopsy

Biopsy

cant (p ¼ .003).
tobacco habits, in excess of

Time Interval to Malignant Transformation. Seven

of 50,000 patients in an oral
Sample under study was part

within 12 months excluded

50,000 villagers examined

Large male preponderance

studies reported on the time interval to malig-

university center, cancer
workers in Gujarat, India

cytology programme at
Sample of 6718 industrial

nant transformation. In studies including all

characteristics

oral leukoplakia, this mean time interval

Cohort

applied to all lesions and not just lesions with

dysplasia. Only 4 studies provided data specifi-
cally for dysplastic lesions (mean, 4.3 years;
range, 0.5–16.0 years; data not shown). Sub-
group analysis by dysplasia grade and treat-
ment modality showed no significant differences
Sampling

in TMT.
method
bias

Yes

Yes
No

Relative Risk in Relation to Clinical Risk Factors.

Assessment
of exposure
(dysplasia)

The risk factors—gender, smoking status, alco-

records
Pathology

hol, and site of lesions—were evaluated in 4 dif-

Biopsy

Biopsy

ferent studies in total. From the limited data

available, there appears to be a trend for malig-
evidence

nant progression to be associated with lesions of

Level of

the tongue (relative risk [RR] ¼ 1.87; CI: 1.11,

4

3.17), whereas continuing smoking and alcohol

use after diagnosis appears to have no effect on
Retrospective
Methodology

population
and setting

*Studies included in meta-analysis.

Silverman et al30 Prospective

Prospective

malignant transformation (Table 5). However,

hospital

hospital

the 2 studies with data on smoking are non-

homogenous (p ¼ .008, Breslow-Day test).
Mincer et al31
Banoczy and

DISCUSSION
Csiba29
Study*

Principal Findings. Oral dysplasia carries a sig-

nificant rate of transformation to cancer

Treatment and Follow-Up of Oral Dysplasia HEAD & NECK—DOI 10.1002/hed December 2009 1605
FIGURE 2. Forest plot with calculated exact binomial confidence intervals of malignant transformation rate of oral dysplastic lesions
for studies included in the meta-analysis.

(12.3%). This increases considerably for high- study selection, and presentation of results and
grade dysplasia (severe dysplasia and carcinoma- discussion.34
in-situ). Furthermore, surgical excision appears The possibility of publication bias remained
to decrease the risk of transformation by more a problem for all systematic reviews. This re-
than half, but does not eliminate it. This suggests view was limited by the paucity of high-quality
the need for continued surveillance, especially for follow-up studies. The meta-analysis is therefore
high-grade dysplastic lesions, even after surgical of the best available evidence. This may com-
excision. There was insufficient information in pound the methodological deficiencies of the
the studies to formally assess the effect of clinical original studies, most importantly, referral bias
risk factors, such as smoking and alcohol, on pro- seen in single-center cohorts, and short follow-
gression to malignancy. up. There are differences between the included
This systematic review demonstrated signifi-
cant heterogeneity between follow-up studies. It
also showed a distinct lack of randomized con-
trolled trials examining different surgical man-
agement and follow-up protocols, especially in
view of the wide variations in clinical practice
that prevail in these areas.

Strengths and Limitations. This systematic re-

view has concentrated on the long-term risk of
malignant transformation and follow-up of oral
dysplastic lesions. This is an area of clinical
practice that showed great variability and lacks
high-quality evidence. The search strategy was
therefore extended to include nonrandomized FIGURE 3. Graph of the transformation rate (arcsine trans-
formed) against study size, a form of meta-analysis funnel plot
studies. This systematic review complies with
in which the study rates should lie mainly within the 95% confi-
the MOOSE statement, which includes detailed dence interval lines. It shows a degree of heterogeneity (they
guidelines on formulating the clinical problem, should be more clustered toward the center). The 3 outliers are
search strategy and identification of sources, Lee et al,22 Silverman et al,27 and Silverman et al.30

1606 Treatment and Follow-Up of Oral Dysplasia HEAD & NECK—DOI 10.1002/hed December 2009
 Table 4. Malignant transformation rates for all studies.

Malignant
Total transform
Study* dysplasia Mild Mod Severe CIS (No. of patients) Mild Mod Severe CIS MTR
18
Hsue et al 166 138 15 13 NR 8 6 2 0 NR 4.8%
Holmstrup et al19 88 42 26 14 6 12 5 2 1 4 11.4%
Saito et al20 91 48 36 7 NR 7 6 1 0 NR 7.7%
Cowan et al21 158 NR 17 NR 14.5%
Lee et al22 70 61 9 NR 22 16 6 NR 31.4%
Schepman et al23 55 Not subdivided 12 Not subdivided 21.8%
Lumerman et al24 44 19 18 6 1 7 3 3 1 0 16.3%
Hogewind et al25 13 5 3 5 NR 1 0 1 0 NR 7.7%
Lind26 47 16 24 7 NR 8 1 4 3 NR 17.0 %
Silverman et al27 22 NR 8 NR 36.4%
Gupta et al28 90 NR 6 NR 6.7%
Banoczy and Csiba29 68 13 43 12 NR 9 1 3 5 NR 13.2%
Silverman et al30 35 NR 0 NR 0.0%
Mincer et al31 45 Omitted 32 13 NR 5 Omitted 2 3 NR 11.1%
Abbreviations: mod, moderate; CIS, carcinoma-in-situ; MTR, malignant transformation rates; NR, not reported.

studies in their design, inclusion criteria, intraobserver and interobserver variability.11

patient population (including ethnicity), patient Therefore, it is impossible to be certain that the
management, length of follow-up, and analysis lesions included in the same grade by different
of patient risk factors, all of which may have an studies are similar. For this reason, the sub-
effect on outcomes. However, combining the rel- groups—mild and moderate, and severe dysplasia
atively scarce data that are available has made and CIS—were almagamated.
it possible to provide a better understanding of Assessment of the quality of observational
the overall behavior of oral dysplasia. studies is difficult, as there is no validated tool
The meta-analysis used published data and currently in wide use. To overcome this problem,
not original patient data, as many of the studies the factors that could introduce bias into the
were conducted in the 1970s and 1980s. This meta-analysis were evaluated using a combina-
can affect the quality of the meta-analysis. In tion of methods.7–10
particular, it has impeded the ability to perform
subgroup analyses, especially for TMT and clini- Implications for Clinical Practice.Oral dysplasia
cal risk factors. The former is likely to be con- appears to carry a significant transformation
founded by length of follow-up and the latter by rate to oral cancer, which occurs over a period of
age and grade of the dysplastic lesions. years, even when treated by surgical excision.
It was demonstrated that there are statisti- This would suggest that patients should be kept
cally significant differences in progression rates under surveillance after diagnosis for anything
between the various grades of oral dysplasia. up to 20 years. It may be feasible to tailor the
However, it is widely acknowledged that histo- duration of the surveillance period, and possibly
logic grading of oral dysplasia shows significant its frequency, according to several clinical

Table 5. Relative risk of malignant transformation of oral lesions analyzed by clinical risk factor.

Risk Risk
Studies factor Malignant factor Malignant
Risk factor reporting present transformation absent transformation RR (CI)
Smoking (patient continues to smoke) 2 81 17 79 11 n/a
Alcohol (patient continues to drink) 1 49 16 21 6 1.14 (0.52–2.51)
Site (tongue vs other) 3 53 18 129 20 1.87 (1.11–3.17)
Sex (male vs female) 3 78 16 83 22 0.77 (0.44–1.37)
Abbreviations: RR, relative risk; CI, confidence interval; n/a, not available.

Treatment and Follow-Up of Oral Dysplasia HEAD & NECK—DOI 10.1002/hed December 2009 1607
factors, including grade of dysplasia, and clini- 4. Lodi G, Sardella A, Bez C, Demarosi F, Carrassi A.
Interventions for treating oral leukoplakia. Cochrane
cal management (lesions which are not excised).
Database Syst Rev 2006;4:CD001829.
However, the limited data available for this 5. van der Waal I, Schepman KP, van der Meij EH, Smeele
review can neither confirm nor reject the hy- LE. Oral leukoplakia: a clinico-pathological review. Oral
pothesis that higher grade dysplastic lesions Oncol 1997;33:291–301.
6. Epstein JB, Gorsky M, Fischer D, Gupta A, Epstein M,
would transform earlier. Elad S. A survey of the current approaches to diagnosis
and management of oral premalignant lesions. J Am
Dent Assoc 2007;138:1555–1562.
Implications for Research. Future research is 7. Oxford Centre for Evidence-Based Medicine. Levels of
needed in 3 areas. First, research is needed in Evidence. May 2001. Available at: http://www. cebm.net/
index.aspx?o¼1047. Accessed: June 9, 2007.
the search for better prognostic markers for pro- 8. Greenhalgh T. How to read a paper. The basics of evi-
gression from dysplasia to malignancy. This dence based medicine, 2nd edition. London: BMJ Press;
would replace or augment histologic grading to 2001.
9. Khan KS, ter Riet G, Popay J, Nixon J, Kleijnen J. Study
improve the identification of patients at risk of quality assessment. Undertaking systematic reviews of
malignant transformation, so that they can be research on effectiveness. CRD’s guidance for those carry-
targeted for more aggressive treatment and ing out or commissioning reviews. CRD report no. 4, 2nd
edition. York: NHS Centre for Reviews and Dissemination;
closer surveillance. 2001.
Second, future research is needed in the 10. Laupaucis A, Wells G, Richardson WS, Tugwell P. Users’
evaluation of follow-up protocols for efficacy and guides to the medical literature. V. How to use a article
on prognosis. JAMA 1994;272:234–237.
cost-effectiveness, including determination of 11. Abbey LM, Kaugars GE, Gunsolley JC, et al. Intraexa-
surveillance duration and frequency. Further- miner and interexaminer reliability in the diagnosis of
more, an assessment is required of whether a oral epithelial dysplasia. Oral Surg Oral Med Oral
Pathol Radiol Endod 1995;80:188–191.
surveillance policy would influence eventual out- 12. Banoczy J, Sugar L. Longitudinal studies in oral leuko-
come through earlier diagnosis of recurrence plakia. J Oral Pathol 1972;1:265–272.
and/or transformation and earlier treatment. 13. Mehta FS, Shroff BC, Gupta PC, Daftary DK. Oral leu-
koplakia in relation to tobacco habits. Oral Surg 1972;
Third, future research is needed in the 34:426–433.
search to identify less invasive and more effec- 14. Pindborg JJ, Renstrup G, Jolst O, Roed-Petersen B.
tive treatments, which would prevent recur- Studies in oral leukoplakia: a preliminary report on the
period prevalence of malignant transformation in leuko-
rence and progression, without the attendant plakia based on a follow-up study of 248 patients. J Am
comorbidity of surgical resection. Dent Assoc 1968;76:767–771.
15. Silverman S, Rozen RD. Observations on the clinical
characteristics and natural history of oral leukoplakia.
J Am Dent Assoc 1968;76:772–777.
16. Pindborg JJ, Daftary DK, Mehta FS. A follow-up study
CONCLUSION of sixty-one oral dysplastic precancerous lesions in In-
Oral dysplasia carries a significant rate of trans- dian villagers. Oral Surg Oral Med Oral Pathol 1977;
43:383–389.
formation to cancer, which increases consider- 17. Banoczy J. Follow-up studies in oral leukoplakia. J Max-
ably for high-grade dysplasia. Surgical excision illofac Surg 1977;5:69–75.
appears to decrease but does not eliminate the 18. Hsue SS, Wang WC, Chen CH, Lin CC, Chen YK, Lin
LM. Malignant transformation in 1458 patients with
risk. These findings suggest the need for surgi- potentially malignant oral mucosal disorders: a follow-up
cal excision and continued surveillance, espe- study based in a Taiwanese hospital. J Oral Pathol Med.
cially for high-grade lesions. 2007;36:25–29.
19. Holmstrup P, Vedtofte P, Reibel J, Stoltze K. Long-term
treatment outcome of oral premalignant lesions. Oral
Oncol 2006;42:461–474.
REFERENCES 20. Saito T, Sugiura C, Hirai A, et al. Development of squa-
1. Axell T, Pindborg JJ, Smith CJ, van der Waal I. Oral mous cell carcinoma from pre-existent oral leukoplakia:
white lesions with special referenace to precancerous with respect to treatment modality. Int J Oral Maxillofac
and tobacco-related lesions: conclusions of an interna- Surg 2001;30:49–53.
tional symposium held in Uppsala, Sweden, May 18–21, 21. Cowan CG, Gregg TA, Napier SS, McKenna SM, Kee F.
1994. International Collaborative Group on Oral White Potentially malignant oral lesions in Northern Ireland: a
Lesions. J Oral Pathol Med 1996;25:49–54. 20-year population-based perspective of malignant trans-
2. Pindborg JJ, Reichart P, Smith CJ, von der Waal I. formation. Oral Dis 2001;7:18–24.
World Health Organization: histological typing of cancer 22. Lee JJ, Hong WK, Hittelman WN, et al. Predicting
and precancer of the oral mucosa. Berlin: Springer Ver- cancer development in oral leukoplakia: ten years of
lag; 1997. translational research. Clin Cancer Res 2000;6:1702–
3. Reibel J. Prognosis of oral pre-malignant lesions: signifi- 1710.
cance of clinical, histopathological, and molecular biolog- 23. Schepman KP, van der Meij EH, Smeele LE, van der
ical characteristics. Crit Rev Oral Biol Med 2003;14:47– Waal I. Malignant transformation of oral leukoplakia: a
62. follow-up study of a hospital-based population of 166

1608 Treatment and Follow-Up of Oral Dysplasia HEAD & NECK—DOI 10.1002/hed December 2009
 patients with oral leukoplakia from the Netherlands. 30. Silverman S, Bhargava K, Mani NJ, Smith LW, Malao-
Oral Oncol 1998;34:270–275. walla AM. Malignant transformation and natural history
24. Lumerman H, Freedman P, Kerpel S. Oral epithelial of oral leukoplakia in 57,518 industrial workers of
dysplasia and the development of invasive squamous cell Gujarat India. Cancer 1976;38:1790–1795.
carcinoma. Oral Surg Oral Med Oral Pathol Radiol 31. Mincer HH, Coleman SA, Hopkins KP. Observations
Endod 1995;79:321–329. on the clinical characteristics of oral lesions showing
25. Hogewind WFC, van der Kwast WAM, van der Waal I. histologic epithelial dysplasia. Oral Surg 1972;33:389–
Oral leukoplakia, with emphasis on malignant transfor- 399.
mation. J Craniomaxillofac Surg 1989;17:128–133. 32. Gupta PC, Mehta FS, Daftary DK, et al. Incidence rates
26. Lind PO. Malignant transformation in oral leukoplakia. of oral cancer and natural history of oral precancerous
Scand J Dent Res 1987;95:449–455. lesions in a 10-year follow-up study of Indian villagers.
27. Silverman S, Gorsky M, Lozada F. Oral leukoplakia and Community Dent Oral Epidemiol 1980;8:287–333.
malignant transformation. A follow-up study of 257 33. Silverman S, Gorsky M, Lozada F. Oral leukoplakia
patients. Cancer 1984;53:563–568. and malignant transformation. A follow-up study of 257
28. Gupta PC, Mehta FS, Daftary DK, et al. Incidence rates patients. Cancer 1984;53:563–568.
of oral cancer and natural history of oral precancerous 34. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis
lesions in a 10-year follow-up study of Indian villagers. of observational studies in epidemiology: a proposal
Community Dent Oral Epidemiol 1980;8:287–333. for reporting. Meta-analysis of observational studies in
29. Banoczy J, Csiba A. Occurrence of epithelial dysplasia in epidemiology (MOOSE) group. JAMA 2000;283:2008–
oral dysplasia. Oral Surg 1976;42:766–773. 2012.

Treatment and Follow-Up of Oral Dysplasia HEAD & NECK—DOI 10.1002/hed December 2009 1609