Vol. 119 No.

1 January 2015

MutSa and MutLa immunoexpression analysis in diagnostic

grading of oral epithelial dysplasia and squamous cell carcinoma
M. Jessri, BDSc, PhD Candidate,a,b A.J. Dalley, PhD, Research Officer,a and
C.S. Farah, BDSc, MDSc, (Oral Med Oral Path), PhD, FRACDS (Oral Med), Head, Oral Oncology Research
Programa,b

Objectives. This study explored the expression of DNA mismatch repair (MMR) proteins in a range of oral biopsies. We further
evaluated the significance of MMR protein expression combined with basic demographic data in differentiating grades of oral
epithelial dysplasia (OED) and oral squamous cell carcinoma (OSCC).
Study Design. Immunohistochemical expression of MutSa (hMLH1 and hPMS2) and MutLa (hMSH2 and hMSH6) were
compared in 98 formalin-fixed paraffin embedded oral biopsies: 21 normal, 24 mild-dysplasia (MD), 8 moderate-to-severe-
dysplasia (SD), and 45 OSCC.
Results. Expression of hMLH1, hPMS2, and hMSH2 was reduced in MD, SD, and OSCC compared with the normal. Reduced
hMSH2 immunoreactivity discriminated poorly differentiated OSCC from well-differentiated OSCC. The diagnostic model
correctly classified 71.4% of cases and revealed that hPMS2-negative biopsies were more likely to be cancerous (odds ratio
[OR], 0.11; 95% confidence interval [CI], 0.000-0.813; P ¼ .040).
Conclusion. The results suggested a diagnostic role for MMR proteins in OED and OSCC. (Oral Surg Oral Med Oral Pathol
Oral Radiol 2015;119:74-82)

Oral squamous cell carcinoma (OSCC) accounts for acquisition of DNA microsatellites (termed microsat-
more than 90% of oral cavity and oropharynx malig- ellite instability) is the hallmark of defective MMR
nancies.1 Oral malignant transformation is a multistep pathways and has been associated with numerous
process,2 which can initially manifest as oral potentially malignancies.7-9
malignant disorders (OPMD).3 The most important The nomenclature and structure of two MMR pro-
prognostic factor for malignant transformation is the teins (MutS and MutL) are conserved in prokaryotes
degree of epithelial dysplasia,4 with lesion site, patient and eukaryotes, but in contrast to bacterial Mut pro-
age, gender, and alcohol/smoking habits also consid- teins, eukaryotic Mut proteins are heterodimers of two
ered.5 The concept of field cancerization was first dis- related but distinct proteins.10 The present study
cussed in 1953 and is likely to have a molecular focused on one form of both the major Mut proteins
foundation in DNA repair mechanisms, which are MutSa and MutLa, in each case using antibodies
susceptible to both genetic and epigenetic compromise, against both of their constituent proteins. MutSa pro-
since surveillance of DNA integrity is a fundamental teins are a heterodimer of hMSH2 and hMSH6 and
defense against localized mutation, clonal selection, and discriminate between mismatched and perfectly paired
malignancy. DNA. Similarly, MutLa proteins are a heterodimer of
The human DNA mismatch repair (MMR) genes are hMLH1 and hPMS2 and are believed to repair post-
pivotal in preserving genomic integrity by repairing replication errors.11
base substitution mismatches and insertion/deletion Hereditary nonpolyposis colorectal cancer, otherwise
mismatches that evade the “proofreading” function- known as Lynch syndrome, is an autosomal dominant
ality of DNA polymerases. Defective MMR mecha- syndrome caused by mutations in the MutSa and
nisms generate novel genomic microsatellites, MutLa genes.12 A diagnosis of Lynch syndrome is
otherwise known as simple sequence repeats (SSRs) reached through a combination of clinical and tumor
or short tandem repeats (STRs), which can be passed biomarker assessments. Semiquantitative assessment of
clonally throughout cellular progeny.6 Cumulative immunohistochemical staining for MutSa and MutLa

This work was supported by a grant from the Australian Dental

Research Foundation, Australia.
a
The University of Queensland, UQ Centre for Clinical Research, Statement of Clinical Relevance
Herston, QLD 4029, Australia.
b
The University of Queensland, School of Dentistry, Brisbane, QLD
This study proposes a diagnostic model based on the
4000, Australia.
Received for publication Jan 6, 2014; returned for revision Jun 2, expression of mismatch repair proteins and patient
2014; accepted for publication Jun 15, 2014. gender, which can correctly classify 71.4% of
! 2015 Elsevier Inc. All rights reserved. normal, oral epithelial dysplasia, and oral squamous
2212-4403/$ - see front matter cell carcinoma cases.
http://dx.doi.org/10.1016/j.oooo.2014.06.017

74
OOOO ORIGINAL ARTICLE
Volume 119, Number 1 Jessri, Dalley and Farah 75

Table I. Characteristics of patient samples (n ¼ 98)

Dysplasia OSCC
Moderately/poorly
Normal mucosa Mild Moderate/severe Well differentiated differentiated
Gender
Male 8 13 6 19 9
Female 13 11 2 7 10
Age
<45 (range 18-45) 20 4 2 3 1
#45 (range 45-73) 1 20 6 23 18
Biopsy site
Tongue 4 8 1 3 1
Palate 2 1 1 11 2
Gingiva 9 5 2 1 4
Labial mucosa 0 6 1 7 4
Buccal mucosa 6 4 3 4 8
Smoking history
Positive 1 15 3 15 12
Negative 20 1 0 3 2
Unavailable 0 8 5 8 5

genes has been shown to be a useful asset in screening of the Oral Oncology Research Group, the University
for carriers of these gene mutations in patients with of Queensland. Samples had been collected between
Lynch syndrome.13 Similarly, expression of hMLH1 1947 and 2012 and classified histologically as 21
and hMSH2 is either lost or significantly reduced in negative for dysplastic changes (referred to here as
head and neck squamous cell carcinoma and lesions “normal”), 24 mild dysplasia (MD), 8 moderate-to-
with a high degree of dysplasia.14-20 Allelic imbalance severe dysplasia (SD), and 45 oral squamous cell
in hMLH1 has been suggested to be an etiologic factor carcinoma (OSCC) to form a cohort of 98 patients
in head and neck carcinogenesis,20 and promoter with a male-to-female ratio of 1:0.78 (Table I). A
methylation of this gene has been shown to be an early hematoxylin and eosin (H&E)estained section from
event in oral carcinogenesis.14 Previous studies in this the samples was assessed by an oral pathologist (CSF)
area have been limited to expression of hMLH1 and according to the World Health Organization (WHO)
hMSH2 in either OPMD or squamous cell carcinoma. classification system for oral epithelial dysplasia.21
In the present study, we hypothesized that decreased OSCC cases were further graded into well-differentiated
MMR protein expression is associated with oral and moderately or poorly differentiated groups accord-
epithelial dysplasia (OED) and OSCC. We explored ing to the WHO criteria.21
MMR protein expression with immunohistochemistry
(IHC) staining of hMLH1, hPMS2, hMSH2, and
hMSH6 across a range of normal, dysplastic, and Immunohistochemical staining
cancerous oral biopsies. Furthermore, we combined Immunohistochemical staining was performed as
MutSa and MutLa expression as a panel with minimal described by us previously.22 Briefly, 5-mm sections
patient demographic cofactors (age and gender) in a were dewaxed and dehydrated in xylene and graded
multinomial logistic regression model to evaluate sta- ethanol. Heat-induced epitope retrieval utilized Diva
tistical significance of MMR protein expression in decloaker solution (DV2004 MX Biocare Medical,
differentiating grades of OED and OSCC. Concord, CA) in a Decloaking Chamber (Model
DC2002 Biocare Medical, Concord, CA) at 125" C for
30 seconds and then returned to 90" C, and this was
MATERIALS AND METHODS
followed by a 5-minute wash in distilled water. Sec-
Ethical review
tions were incubated in endogenous peroxidase
This study was approved by the Ethics Committees of
blocker (Peroxidazed1 PX968 M Biocare Medical,
the University of Queensland (2007001478) and the
Concord, CA) and then in nonspecific background
Royal Brisbane Hospital (HREC/10/QRBW366).
stain-blocking reagent (Background Snipper, BS966
G, Biocare Medical, Concord, CA). Slides were
Patient samples incubated for 15 minutes with primary monoclonal
A total of 98 formalin-fixed paraffin-embedded spec- antihuman antibodies (Biocare Medical Inc., Concord,
imens were chosen randomly from the tissue archives CA) as follows: HMLH1 (G168-15), hPMS2 (A 16-4),
ORAL AND MAXILLOFACIAL PATHOLOGY OOOO
76 Jessri, Dalley and Farah January 2015

Fig. 1. Representative immunostaining of the positive control for (i) hMLH1; (ii) hPMS2; (iii) hMSH2; (iv) hMSH6.

hMSH2 (FE11), and hMSH6 (BC/44). Secondary on immunoexpression thresholds, per Barrow et al.,
antibody and revealing agent were obtained from where the genetic alterations in the MMR genes were
Biocare Medical (MACH1 Universal HRP polymer associated with their immunohistochemical expres-
kit, M1 U539 L10, Biocare Medical, Concord, CA). sion.13 Although immunoexpression of hMSH6 did not
Sections were revealed using Biocare Betazoid DAB meet the minimum requirements for carrier mutation
(BDB900 G, Biocare Medical, Concord, CA). Sections detection in Barrow’s study, an H-score of 5 or less was
were counterstained with CAT hematoxylin (CATH-M considered a carrier for hMSH6 mutations (sensitivity:
Biocare Medical, Concord, CA) and mounted with Leica 81.3%, specificity: 80.85%).13
CV mount (Leica Microsystems, Wetzlar, Germany).
Positive (colon) and negative staining controls were
conducted. Statistical analysis
Nonparametric analysis was performed with IBM
SPSS Statistics V.20 software (IBM Corporation,
H-score of positive immunoreactivity in stained Armonk, NY). H-score data normality was established
sections by using the Kolmogorov-Smirnov test. Immunoloc-
The immunoreactivity of hMLH1, hPMS2, hMSH2, alization H-score data for disease groups were
and hMSH6 was analyzed semiquantitatively. Epithe- compared with the normal, using two-tailed Mann-
lial cells with brown nuclear staining, regardless of Whitney U tests with Bonferroni correction for n ¼ 6
intensity, were considered positive. Both negative and comparisons (P < .008) unless stated otherwise.
positive cells were counted in 16 randomized, high- The relationship between MMR protein expression
power ($400) microscopic fields,23 using light micro- (H-score) and lesion severity was established by
scopy and an eyepiece grid (Standard 20, Olympus, Spearman’s productemoment correlation coefficient
Japan). Epithelial cells from the invasive front of OSCC after ordinal grading of lesions: normal, 0; mildly
cases were included in the pool of all epithelial cells dysplastic, 1; severely dysplastic, 2; OSCC, 3.
and also recorded separately for further analysis. In Multinomial logistic regression was used to identify
small samples where there were less than 16 fields per predictors of dysplasia or neoplasia in comparison
slide, all fields were evaluated. The percentage of with normal tissue, with H-score thresholds for the
positively stained cells was scored as described by mismatch repair panel13 and gender of the participants
Muller et al.24 A score of 0 indicated no cell immu- as independent variables, after adjusting for age.
nopositivity, 1 represented 1% to 10% positive cells, 2
represented 11% to 50% positive cells, 3 represented RESULTS
51% to 80% positive cells, and 4 represented more than Mismatch repair protein intracellular distribution
80% positive cells.24 Figure 2 shows typical immunostaining of four MMR
The immunoreactivity intensity was scored on a genes (hMLH1, hPMS2, hMSH2, and hMSH6) in
0-3 scale based on comparison of epithelial cell normal oral tissue, mild dysplasia, moderate-to-severe
staining intensity with positive control cells dysplasia, and oral squamous cell carcinoma (OSCC).
(Figure 1). A score of 3 indicated immunoreactivity Immunoreactivity for these proteins was localized pre-
equivalent to that of positive control, and a score of dominantly in cell nuclei, with a small number of
0 denoted no immunoreactivity. The H-score, which samples exhibiting cytoplasmic expression of MMR
ranged from 0 to 12, was the product of categorized proteins (hMLH1 in 2 OSCC and 1 moderate-severe
percentage positive score and the scaled stain in- dysplasia; hPMS2 in 2 OSCC, 3 mild dysplasia, 1
tensity score.13 normal; hMSH2 in 3 OSCC; hMSH6 in 12 mild
In order to perform multinomial logistic regression, dysplasia and 10 normal; all 4 proteins in 3 OSCC and
patients were further categorized into two groups based 1 moderate-severe dysplasia).
OOOO ORIGINAL ARTICLE
Volume 119, Number 1 Jessri, Dalley and Farah 77

Fig. 2. Representative photomicrographs. A, Normal samples for (ii) hMLH1, (iii) hPMS2 , (iv) hMSH2 and (v) hMSH6
(H-score ¼ 12 across the board. B, Mild dysplasia for (ii) hMLH1 (H-score ¼ 8), hPMS2 (H-score ¼ 9), hMSH2 (H-score ¼ 8),
and hMSH6 (H-score ¼ 9). C, Moderate dysplasia for (ii) hMLH1 (H-score ¼ 2), hPMS2 (H-score ¼ 4), hMSH2 (H-score ¼ 6),
and hMSH6 (H-score ¼ 9). D, Epithelium from well-differentiated oral squamous cell carcinoma (OSCC) samples for hMLH1
(H-score ¼ 6), hPMS2 (H-score ¼ 9), hMSH2 (H-score ¼ 4), and hMSH6 (H-score ¼ 6). E, Well-differentiated OSCC cell nests
and the underlying connective tissue for hMLH1 (H-score ¼ 6), hPMS2 (H-score ¼ 4), hMSH2 (H-score ¼ 6), and hMSH6
(H-score ¼ 9). F, Poorly differentiated OSCC for hMLH1 (H-score ¼ 0), hPMS2 (H-score ¼ 1), hMSH2 (H-score ¼ 0), and
hMSH6 (H-score ¼ 2). Relevant hematoxylin and eosin photomicrographs for each sample are shown in column i (Ai to Fi).
ORAL AND MAXILLOFACIAL PATHOLOGY OOOO
78 Jessri, Dalley and Farah January 2015

Fig. 3. Distribution of immunohistochemical expression scores for mismatch repair proteins in normal, mild dysplasia (MD),
moderate-to-severe dysplasia (SD), and oral squamous cell carcinoma (OSCC) samples. (*P < .0125)

Mismatch repair protein association with oral Table II. Correlation of mismatch repair gene expres-
disease severity sion with each other and patient’s graded diagnosis
Figure 3 provides a nonparametric representation of MutLa MutSa
H-score distribution grouped by lesion classification. hMLH1 hPMS2 hMSH2 hMSH6 Diagnosis
Relative to normal tissue, the overall trend was for all MutLa
lesion groups to exhibit reduced expression of hMLH1, hMLH1 d 0.403* 0.396* 0.131 %0.311*
hPMS2, and hMSH2, which was supported in most hPMS2 d 0.214y 0.191 %0.514*
cases (6/9) by Mann-Whitney U tests, although large MutSa
interquartile ranges (IQRs) of H-score impaired the hMSH2 d 0.457* %0.47
hMSH6 d 0.34
attainment of statistical significance for severe
dysplasia lesions stained for hMLH1 and hPMS2 and *Correlation is significant at P < .01 (two-tailed).
y
for OSCC lesions stained for hMSH2. Interestingly, the Correlation is significant at P < .05 (two-tailed).
H-scores that describe expression of hMSH6 showed
equivalent median values and very high IQRs for Correlation of mismatch repair protein expression
all lesion groups. OSCC cases were graded into well- with oral disease severity
differentiated and moderately or poorly differentiated The relationship between MMR protein expression and
groups according to the WHO criteria.21 Median disease severity was investigated by Spearman’s pro-
H-scores for hMSH2 expression were higher in well- ductemoment correlation coefficient of H-score data
differentiated (median score ¼ 7; IQR ¼ 3.7 to 9; versus an ordinal grading of lesion severity. Relatively
n ¼ 26) compared with poorly differentiated cases weak negative correlations with graded lesion severity
(median score ¼ 4; IQR ¼ 4 to 6; n ¼ 19), (U ¼ 159.500; were observed for hPMS2, hMSH2 and hMLH1, but not
z ¼ %2.051; P ¼ .040; Mann-Whitney U test without hMSH6 (Table II). The strongest relationship between
Bonferroni correction P < .05). MMR protein expression and disease severity was
OOOO ORIGINAL ARTICLE
Volume 119, Number 1 Jessri, Dalley and Farah 79

Table III. Predicted versus observed diagnosis according to multinomial logistic regression adjusted for age of the
participants
Predicted
Observed Normal Mild dysplasia Mod/severe dysplasia OSCC % Correct
Normal 20 0 0 1 95.2%
Mild dysplasia 0 9 1 14 37.5%
Moderate/Severe dysplasia 1 1 4 2 50.0%
OSCC 1 6 1 37 82.2%
Overall percentage 22.4% 16.3% 6.1% 55.1% 71.4%
OSCC, oral squamous cell carcinoma.

observed for hPMS2 (r ¼ %0.514; n ¼ 98; P < .0001), general formula by which the model was designed is as
the only other statistically significant correlation being follows:
that for hMLH1 (r ¼ %0.311; n ¼ 98; P < .002). ! " X# $
pðDxÞ
Ln ¼ bþ bxi $ xi
Coexpression correlations for mismatch repair pðNormalÞ
protein expression in which P ¼ probability, Dx ¼ diagnosis of sample
On the basis of what is known about the four MMR (mild dysplasia, moderate-to-severe dysplasia, and
proteins under investigation, one might expect coex- OSCC), b ¼ interface, b ¼ coefficient for each in-
pression of hMSH2 with hMSH6 (to comprise the dependent variable, x ¼ independent variable, and
MutSa heterodimer) and coexpression of hMLH1 with i ¼ score for which model is calculated.
hPMS2 (to comprise the MutLa heterodimer). Coex- Below are the three equations used for the data
pression correlation coefficients for each antigen are shown in Table IV.
also presented in Table II. All MMR proteins exhibited ! "
weak but statistically significant positive coexpression probabilityðmild dysplasiaÞ
Ln ¼ %0:255
correlations. As anticipated, the strongest coexpression probabilityðnormalÞ
correlations were between hMSH2 and hMSH6 þ 0:154ðageÞ % 3:215ðsex ¼ 1:0Þ
(r ¼ 0.457; n ¼ 98; P < .0001) and between hMLH1
% 0:1566ðhMLH1Score ¼ 1:00Þ
with hPMS2 (r ¼ 0.403; n ¼ 98; P < .0001).
% 0:4641ðhMSH2Score ¼ 1:00Þ
Construction of a diagnostic model of oral % 1:220ðhMSH6Score ¼ 1:00Þ
carcinogenesis based on DNA mismatch repair
% 0:2:915ðhPMS2Score ¼ 1:00Þ
protein immunoexpression
Multinomial logistic regression was performed to assess ! "
the impact of independent variables (gender, MMR probabilityðmoderate to severe dysplasiaÞ
Ln
pathway protein H-scores) on the likelihood and degree probabilityðnormalÞ
of dysplasia or neoplasia in participants, adjusted for ¼ %2:673 þ :172ðageÞ þ ð%5:144Þðsex ¼ 1:0Þ
age. Due to the small number of samples, site of the þ ð1:278ÞðhMLH1Score ¼ 1:00Þ
lesion and history of smoking were not included in this
þ ð%2:054ÞðhMSH2Score ¼ 1:00Þ
model. The model was able to differentiate between the
different degrees of dysplasia or neoplasia c2 [18, þ ð%1:484ÞðhMSH6Score ¼ 1:00Þ
n ¼ 98] ¼ 114.409; P < .0001. The model classified þ ð%2:502ÞðhPMS2Score ¼ 1:00Þ
71.4% of cases correctly (Table III) and explained be-
tween 68.9% (Cox and Snell R square) and 75.2% ! "
probabilityðOSCCÞ
(Nagelkerke R square) of variance in diagnosis, with Ln ¼ %2:104 þ 0:202ðageÞ
the best prediction power for the normal at 95.2% and probabilityðnormalÞ
the weakest prediction power for mild dysplasia at % 4:377 ðsex ¼ 1:0Þ
37.5%. The strongest association for MMR proteins % 1:746 ðhMLH1Score ¼ 1:00Þ
was between hPMS2 and ability of the model to
þ 1:058 ðhMSH2score ¼ 1:00Þ
differentiate between normal and OSCC (Table IV).
Here, the model identified cases with loss of hPMS2 as % 1:413 ðhMSH6Score ¼ 1:00Þ
being more likely to have OSCC (OR, 0.011; 95% CI, % 4:554 ðhPMS2Score ¼ 1:00Þ
0.000-0.813; P ¼ .040) compared with the normal. The
ORAL AND MAXILLOFACIAL PATHOLOGY OOOO
80 Jessri, Dalley and Farah January 2015

Table IV. Environmental and genetic predictors of diagnosis according to multinomial logistic regression model
Lesions compared to normal cohort OR (CI 95%)
Mild dysplasia Severe dysplasia OSCC
Demographic
Age 1.167 (1.045-1.303)* 1.188 (1.052-1.340)* 1.223 (1.088-1.375)*
Gender 0.040 (0.000-4.773) 0.006 (0.000-1.147)y 0.013 (0.000-1.697)
Expression level
hMLH1 0.209 (0.007-6.531) 3.589 (0.078-164.260) 0.175 (0.005-6.162)
hMSH2 0.629 (0.015-26.617) 0.128 (0.002-8.054) 2.879 (0.061-136.511)
hMSH6 0.295 (0.010-9.125) 0.227 (0.006-8.930) 0.243 (0.007-8.052)
hPMS2 0.054 (0.001-3.859) 0.082 (0.001-7.840) 0.011 (0.000-0.813)*
*P < 05.
y
P ¼ .056.

DISCUSSION comparative evaluation of DNA repair in epithelial

This study analyzed immunoreactive protein expression samples must encompass the entire epithelium. The
of the four subunits that comprise the two major design, implementation, and interpretation of our
MMR proteins MutSa and MutLa in an archive of 98 semiquantitative assessment of MutSa and MutLa
formalin-fixed, paraffin-embedded oral tissue biopsies drew upon the seminal work of Barrow et al., who
bearing a spectrum of disease, ranging from normal found immunohistochemical staining for MutSa and
through mild and moderate-to-severe dysplasia to oral MutLa to be an asset for identification of carriers of
squamous cell carcinoma (OSCC). Our principal aim mutated MMR genes in patients with Lynch syn-
was to demonstrate alterations in the expression of drome.13 Lynch syndrome is associated with inherited
MMR proteins in oral dysplasia and neoplasia, as pre- genetic mutations in MMR proteins and a higher risk
viously reported for other cancer types. This was ach- for acquiring several forms of malignancy, including
ieved by construction of a predictive model of OSCC colorectal tumors, breast cancer,26 lung cancer, adrenal
that was based on immunohistochemical staining of cortical neoplasm, and pancreatic acinar cell carci-
hMLH1, hPMS2, hMSH2, and hMSH6, combined noma.27 Only a limited number of studies have focused
with minimal patient demographic cofactors (age and on the association and pathologic features of MMR-
gender). A trend between oral disease severity and related tumors in the upper gastrointestinal tract, spe-
reduced expression of hMLH1, hPMS2, and hMSH2 cifically oral squamous cell carcinoma.
(but not hMSH6) was evident, and interestingly, The work of Barrow et al. is significant, as it used
reduced hMSH2 immunoreactivity distinguished poorly receiver operating characteristic curves to link MMR
differentiated OSCC from well-differentiated OSCC. In gene mutations to their histochemical immunoreactivity
addition, this study investigated coexpression correla- in Lynch syndrome. The present study utilized the
tions for the four MMR-associated proteins and docu- immunoexpression thresholds that identify MMR gene
mented the sporadic anomalies in their intracellular mutations in Lynch syndrome to categorize oral bi-
distribution. The results affirm the role of genetic opsies into two independent groups that became the
instability in oral carcinogenesis by demonstrating a nominal dependent variables, that is, age and gender, in
negative trend between MutSa and MutLa immuno- multinomial logistic regression analysis of immunore-
score data and diagnostic grading of dysplasia and activity H-scores.
OSCC severity. These findings suggest potential The diagnostic model of oral carcinogenesis pre-
application of MutSa and MutLa expression analysis to sented here was effective in distinguishing oral lesion
diagnostic grading of dysplasia and OSCC and support disease severity (c2 [18, n ¼ 98] ¼ 88.543; P < .0001),
the hypothesis that defective MMR pathways play a identifying 71.4% of cases correctly and explaining
functional role in oral cancer. At this stage, however, 68.9% to 75.2% of diagnostic variance. Subsequent
there is no suggestion that MutSa and MutLa are po- work is intended to validate this model in a larger in-
tential therapeutic targets for OSCC. dependent cohort of biopsies from patients with dysplasia
Under normal conditions, expression of MMR pro- or OSCC.
teins has been associated with a proliferative epithelial Defective MMR pathways are associated with
compartment (transient amplifying cells adjacent to several pathologic conditions.7-9 Phenotypic conse-
basement membrane with an active replication pro- quences of microsatellite accumulation in the coding
cess).25 The proliferative region within epithelial and noncoding regions of DNA may account for the
dysplasia or neoplasia, however, is more difficult to observation that genomic instability is an early event in
define. Therefore the scoring of these proteins as a carcinogenesis.6 The role of MMR proteins in oral
OOOO ORIGINAL ARTICLE
Volume 119, Number 1 Jessri, Dalley and Farah 81

carcinogenesis has not yet been studied exten- covariate model of MMR proteins improved the
sively,18,28 as the studies tend to be limited to the ability of the model in classifying moderate-to-severe
presence or absence of hMLH1 and hMSH2 genes in dysplasia samples by 13%. Hence we believe that
dysplastic or neoplastic oral lesions. The present study the decision to exclude hMSH6 from this panel should
is more inclusive, encompassing loss, reduction, and be postponed until replication of our findings is
overexpression of hMLH1, hMSH2, hMSH6, and undertaken.
hPMS2 in a range of normal, dysplastic, and neoplastic The present results affirm the role of genetic insta-
oral samples. bility in oral carcinogenesis by demonstrating a nega-
We provide coexpression correlations for the four tive trend between MutSa and MutLa immunoscore
MMR proteins under investigation. These data support data and diagnostic grading of oral epithelial dysplasia
known heterodynamic associations between hMSH2 and OSCC severity. These findings are derived from a
and hMSH6 within the MutSa complex and between small sample size with limited number of dysplastic
hMLH1 and hPMS2 within the MutLa complex. cases, particularly moderate and severe dysplasia. Further
Weaker correlations that cross the MutSa and MutLa validation in an independent cohort of samples and pa-
homologs are also of interest, since they suggest that tients is therefore recommended.
MutSa and MutLa expression levels become compro- This is a retrospective study performed on randomly
mised in oral lesions simultaneously. selected representative oral biopsies collected over an
Both MutLa subunits, hMLH1 and hPMS2, exhibi- extended period, and, inevitably, moderate or severely
ted reduced expression with increased disease severity. dysplastic cases are underrepresented. Through a
This is the first study of hPMS2 expression in a range mathematical model, we proposed a potential diag-
of oral tissues, whereas expression of hMLH1 in nostic role for MMR proteins as an immunohisto-
oral lesions has been the subject of prior publications. chemical panel; however, this model is only an adjunct
The finding of reduced immunoreactivity of hMLH1 to the diagnosis made by a trained pathologist and
in dysplastic lesions and OSCC complements previous should be considered within its innate limitations. In
reports of negative correlations between hMLH1 conclusion, our data support the hypothesis that
expression and disease severity for OPMD and defective MMR pathways play a potentially functional
OSCC.16-18 Promoter methylation and subsequent loss role in oral cancer. Although there is no suggestion that
of hMLH1 expression has previously been associated MutSa and MutLa are potential therapeutic targets for
with oral carcinogenesis and tumor progression.14,29 OSCC, application of MutSa and MutLa expression
Positive immunoreactivity for hMLH1 has great spec- analysis to diagnostic grading of oral epithelial
ificity and sensitivity for microsatellite instability dysplasia and OSCC has the potential to promote early
detection,30 and it is possible that previously reported detection and improved management of OPMD and
intratumor heterogeneity in microsatellite instability for oral cancer.
OSCC31 may correlate with similar hMLH1 heteroge-
neity. However, our study did not demonstrate a sig- This work was funded by the Australian Dental Research
nificant difference between hMLH1 expression at the Foundation. The authors acknowledge Michael McCullough
invasive tumor front compared with central or superfi- at the University of Melbourne for providing some tissue
cial areas, and only four OSCC cases revealed intra- samples, and Anthony Chan for his technical laboratory
tumor heterogeneity in hMLH1 expression. assistance.
For the MutSa subunits hMSH2 and hMSH6,
reduced expression with increased disease severity was
only observed for hMSH2. There were no noteworthy
trends between hMSH6 immunoexpression and lesion REFERENCES
grading. Favorable oral disease stratification was ach- 1. Neville BW, Day TA. Oral cancer and precancerous lesions. Ca-a
ieved with hMSH2, which distinguished mild and Cancer J Clinicians. 2002;52:195-215.
moderate-to-severe dysplasia from normal tissue and 2. Califano J, vanderRiet P, Westra W, et al. Genetic progression
model for head and neck cancer: implications for field cancer-
distinguished well-differentiated OSCC from poorly
ization. Cancer Res. 1996;56:2488-2492.
differentiated OSCC. Similar studies in colon tissues 3. Haya-Fernandez MC, Bagan JV, Murillo-Cortes J, Poveda-
have shown hMSH2 to be a good marker of MMR Roda R, Calabuig C. The prevalence of oral leukoplakia in 138
mutation carrier state.24,32 Although hMSH2 and patients with oral squamous cell carcinoma. Oral Dis. 2004;10:
hMSH6 contribute to the MutSa complex, a previous 346-348.
study in head and neck cancer indicated a low level 4. Napier SS, Speight PM. Natural history of potentially malignant
oral lesions and conditions: an overview of the literature. J Oral
(12%) of allelic imbalance at the hMSH2-hMSH6 loci. Pathol Med. 2008;37:1-10.
Although expression of hMSH6 did not correlate with 5. Dalley AJ, Pitty LP, Major AG, Abdulmajeed AA, Farah CS.
disease severity, inclusion of this protein in an associate Expression of ABCG2 and Bmi-1 in oral potentially malignant
ORAL AND MAXILLOFACIAL PATHOLOGY OOOO
82 Jessri, Dalley and Farah January 2015

lesions and oral squamous cell carcinoma. Cancer Med. 2014;3: Genetics of Head and Neck Tumours. 9th ed. Lyon, France: IARC
273-283. Press; 2005.
6. Tanic N, Tanic N, Milasin J, Vukadinovic M, Dimitrijevic B. 22. Dalley AJ, Abdulmajeed AA, Upton Z, Farah CS. Organotypic
Genomic instability and tumor-specific DNA alterations in oral culture of normal, dysplastic and squamous cell carcinoma-
leukoplakias. Eur J Oral Sci. 2009;117:231-237. derived oral cell lines reveals loss of spatial regulation of CD44
7. Jones S, Wang TL, Kurman RJ, et al. Low-grade serous carci- and p75 NTR in malignancy. J Oral Pathol Med. 2013;42:37-46.
nomas of the ovary contain very few point mutations. J Pathol. 23. Fernandes AM, De Souza VRDC, Springer CRA, et al. Tobacco
2012;226:413-420. and inflammation effects in immunoexpression of hMSH2 and
8. Ramírez-Ramírez MA, Sobrino-Cossío S, de la Mora-Levy JG, hMLH1 in epithelium of oral mucosa. Anticancer Res. 2007;27:
et al. Loss of expression of DNA mismatch repair proteins in 2433-2437.
aberrant crypt foci identified in vivo by magnifying colonoscopy 24. Müller W, Burgart LJ, Krause-Paulus R, et al. The reliability of
in patients with hereditary nonpolyposic and sporadic colon rectal immunohistochemistry as a prescreening method for the diagnosis
cancer. J Gastrointest Cancer. 2012;43:209-214. of hereditary nonpolyposis colorectal cancer (HNPCC)dresults
9. Buerki N, Gautier L, Kovac M, et al. Evidence for breast cancer as of an international collaborative study. Familial Cancer. 2001;1:
an integral part of lynch syndrome. Genes Chromosomes Cancer. 87-92.
2012;51:83-91. 25. Modrich P, Lahue R. Mismatch repair in replication fidelity, ge-
10. Kunkel TA, Erie DA. DNA mismatch repair. Annu Rev Biochem. netic recombination, and cancer biology. Annu Rev Biochem.
2005;74:681-710. 1996;65:101-133.
11. Hsieh P, Yamane K. DNA mismatch repair: molecular mecha- 26. D’Arcy C, Wen YH, Stadler ZK, Brogi E, Shia J. Synchronous
nism, cancer, and ageing. Mech Ageing Dev. 2008;129:391-407. breast cancers with different morphologic and molecular pheno-
12. Lynch HT, De la Chapelle A. Hereditary colorectal cancer. types occurring in lynch syndrome: what does the heterogeneity
N Engl J Med. 2003;348:919-932. imply? Am J Surg Pathol. 2011;35:1743-1748.
13. Barrow E, Jagger E, Brierley J, et al. Semiquantitative assessment 27. Karamurzin Y, Zeng Z, Stadler ZK, et al. Unusual DNA
of immunohistochemistry for mismatch repair proteins in Lynch mismatch repair-deficient tumors in Lynch syndrome: a report of
syndrome. Histopathology. 2010;56:331-344. new cases and review of the literature. Hum Pathol. 2012;43:
14. González-Ramírez I, Ramírez-Amador V, Irigoyen-Camacho ME, 1677-1687.
et al. HMLH1 promoter methylation is an early event in oral 28. Souza LR, Fonseca-Silva T, Pereira CS, et al. Immunohisto-
cancer. Oral Oncol. 2011;47:22-26. chemical analysis of p53, APE1, hMSH2 and ERCC1 proteins in
15. Czerninski R, Krichevsky S, Ashhab Y, Gazit D, Patel V, Ben- actinic cheilitis and lip squamous cell carcinoma. Histopathology.
Yehuda D. Promoter hypermethylation of mismatch repair genes, 2011;58:352-360.
hMLH1 and hMSH2 in oral squamous cell carcinoma. Oral Dis. 29. Liu K, Huang H, Mukunyadzi P, Suen JY, Hanna E, Fan CY.
2009;15:206-213. Promoter hypermethylation: an important epigenetic mechanism
16. Fernandes AM, Ramos-Jorge ML, Cardoso SV, Loyola AM, for hMLH1 gene inactivation in head and neck squamous cell
Mesquita RA, Aguiar MCF. Immunoexpression of hMSH2 and carcinoma. Otolaryngol Head Neck Surg. 2002;126:548-553.
hMLH1 in oral squamous cell carcinoma and its relationship to 30. Shia J, Ellis NA, Klimstra DS. The utility of immunohisto-
histologic grades of malignancy. J Oral Pathol Med. 2008;37: chemical detection of DNA mismatch repair gene proteins.
543-548. Virchows Arch. 2004;445:431-441.
17. Caldeira PC, Abreu MHNG, Batista AC, Do Carmo MAV. 31. Wang X, Fan M, Chen X, et al. Intratumor genomic heterogeneity
hMLH1 immunoexpression is related to the degree of epithelial correlates with histologic grade of advanced oral squamous cell
dysplasia in oral leukoplakia. J Oral Pathol Med. 2011;40: carcinoma. Oral Oncol. 2006;42:740-744.
153-159. 32. Wahlberg SS, Schmeits J, Thomas G, et al. Evaluation of mi-
18. Caldeira PC, Aguiar MCF, Mesquita RA, do Carmo MAV. Oral crosatellite instability and immunohistochemistry for the predic-
leukoplakias with different degrees of dysplasia: comparative tion of germ-line MSH2 and MLH1 mutations in hereditary
study of hMLH1, p53, and AgNOR. J Oral Pathol Med. 2011;40: nonpolyposis colon cancer families. Cancer Res. 2002;62:3485-
305-311. 3492.
19. Tawfik HM, El-Maqsoud NMRA, Hak BHAA, El-Sherbiny YM.
Head and neck squamous cell carcinoma: mismatch repair
immunohistochemistry and promoter hypermethylation of
hMLH1 gene. Am J Otolaryngol Head Neck Med Surg. 2011;32: Reprint requests:
528-536. Camile S Farah
20. Nunn J, Nagini S, Risk JM, et al. Allelic imbalance at the DNA Oral Oncology
mismatch repair loci, hMSH2, hMLH1, hPMS1, hPMS2 and PO Box 88, Royal Brisbane and Women’s Hospital
hMSH3, in squamous cell carcinoma of the head and neck. Oral Herston
Oncol. 2003;39:115-129. QLD 4029
21. Barnes L, Eveson JW, Reichart P, Sidransky D, eds. World Australia
Health Organization Classification of Tumours. Pathology and [email protected]