• microsatellite instability;
  • hMLH1;
  • hMSH2;
  • squamous-cell carcinoma;
  • head-and-neck cancer


  1. Top of page
  2. Abstract
  6. Acknowledgements

The most prevalent risk factors in the development of head-and-neck squamous-cell carcinoma (HNSCC) are excessive tobacco and alcohol consumption. In young patients with HNSCC, these risk factors are often absent. Our purpose was to investigate the risk factors, microsatellite instability (MSI) changes and status of the mismatch repair genes hMLH1 and hMSH2 in a cohort of young patients with HNSCC. Fifty-seven HNSCC tumors were examined for the presence of MSI at 16 microsatellite sites using PCR. In the young patient group (24 cases, ≤44 years old), 100% of tumors had MSI at 1 site at least and 88% had MSI at 2 or more loci. In older patients (33 cases, ≥45 years), MSI at 1 or more sites was found in 61% of tumors (young vs. old, p = 0.0003) and instability at 2 or more sites was found in 36% of tumors (young vs. old, p = 0.0001). The involvement of the mismatch repair genes was investigated by examining promoter methylation, exon mutation and gene expression of hMLH1 and hMSH2. All results were negative, indicating that inactivation of these 2 genes does not play a role in the development of MSI in tumors from this patient group. Furthermore, the young patient group had a significantly lower incidence of smoking (46% young, 88% old; p = 0.001) and alcohol consumption (33% young, 67% old; p = 0.0169), emphasizing the probable importance of other environmental and/or genetic factors in the development of their disease. © 2001 Wiley-Liss, Inc.

Head-and-neck squamous-cell carcinoma (HNSCC) is a relatively common neoplasm, accounting for 2% to 7% of all tumors in the Western Hemisphere.1, 2 The most prevalent risk factors in this cancer are excessive tobacco and alcohol consumption.3 However, in young patients with HNSCC, these risk factors are not always present, implicating other factors in the etiology of their disease. Microsatellite sequences are 2 to 5 nucleotide repeats that can be found scattered throughout the eukaryotic genome. They have been widely used as genetic markers due to their high degree of polymorphism. Microsatellite instability (MSI) is characterized by expansion or contraction in the length of short tandem repeats and has been associated with various cancer types and inheritable diseases. MSI was initially described in hereditary non-polyposis colorectal cancer (HNPCC) as well as in sporadic colorectal tumors.4 In HNPCC, a replication error phenotype (RER+) has been associated with mutations in the hMSH2, hMLH1, hPMS1 and hPMS2 genes, which are involved in mismatch repair (MMR) mechanisms.4–7 The RER+ phenotype has also been described in other tumor types, such as gastric,8–13 lung,14, 15 ovary,16, 17 pancreatic,18 endometrial19, 20 and HNSCC.21–27 MSI in head-and-neck carcinoma has been found in tumors at a frequency of 7% to 30%.21, 25, 28

Head-and-neck tumors generally occur in older men (>50 years) with a history of smoking and/or alcohol abuse. As indicated above, in young patients, excessive smoking and drinking are not always associated with the development of HNSCC. MSI has been detected at frequencies as high as 52% in young patients with gastric cancer11 and 58% in colorectal cancer patients under the age of 35.6 In this study we compared the frequency of MSI in tumors from young and older patients with HNSCC and screened MSI+ tumors for mutation, promoter methylation and deregulated expression of the MMR genes hMLH1 and hMSH2. In this report, we show that tumors from young patients (<45 years) with HNSCC have a much higher frequency of MSI than those from their older counterparts. Furthermore, the genetic process leading to MSI in these HNSCC tumors is unlike that observed in HNPCC and does not involve the MMR genes hMLH1 and hMSH2.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Sample collection and DNA extraction

Paired normal and tumor specimens from 57 patients at the University Health Network were obtained. They were treated primarily by surgery between 1995 and 1999. Patient age was determined at the time of first surgical resection. Clinical data as well as information pertaining to lifestyle (smoking history, alcohol use) were obtained from the clinical chart and via personal communication. Statistical analysis was done using Fisher's exact test. Among the 57 patients, 24 were ≤44 years of age (young patient group), while 33 were ≥45 years (older patient group). DNA was extracted from paired tumor and nearby normal tissue of each patient. Both paraffin sections and frozen tissues were used from patients undergoing surgical resection for HNSCC. DNA from paraffin-embedded tissue was extracted and purified using a QIAamp column (Qiagen, Valencia, CA). DNA from frozen tissue was extracted using traditional phenol-chloroform methodology.29

PCR amplification of microsatellite loci

Sixteen microsatellite markers were used to analyze the frequency of MSI in tumors from patients with HNSCC. The chromosomal loci, primer sequences and annealing temperatures of these markers are listed in Table I. Primer sequences for these microsatellite loci were obtained from the Genome Database. Primers were chosen based on their index of heterozygosity and flanked microsatellite sites that showed instability in HNSCC or other tumor types, as described in the literature.4, 28, 30, 31 In a study examining the frequency of MSI in HNSCC, no instability was observed using the 2 mononucleotide markers BAT25 and BAT26 in 56 patients, including tumors classified as MSI-high (>30%) using dinucleotide markers.32 Therefore, despite the recommendations of the International Consensus Conference for use of these markers in colon carcinoma,33 we did not feel they would be informative for analysis of MSI in head-and-neck carcinoma.

Table I. Polymorphic Loci and Primer Sequences
MarkersLociPrimer sequences (5′–3′)Annealing temperature (°C)

PCR was carried out via standard protocols but scaled down to a 10 μl reaction volume containing 2 to 5 ng of patient DNA to facilitate the use of limited paraffin-embedded material. Each reaction contained 10 mM (NH4)SO4; 10 mM KCl; 1.5 to 3.0 mM MgCl2; 10 mM Tris-HCl (pH 8.3); 0.01% gelatin; 0.1% Triton X-100; 200 nM each primer; 1.3 μCi [α-33P]-dATP; 8 μM each of dATP, dTTP, dCTP and dGTP; and 0.1 U Tsg DNA polymerase (Bio Basic, Toronto, Canada). [α-33P]-dATP was incorporated into PCR products during PCR. DNA was initially denatured at 94.5°C for 2 min, then subjected to 35 cycles of denaturation at 94°C for 50 sec, annealing at the appropriate temperature (46° to 60°C) for 1 min, strand elongation at 72°C for 1 min and final elongation at 72°C for 5 min. PCR products were diluted 1:1 with loading buffer (95% formamide, 20 mM EDTA, 0.025% xylene cyanol and 0.025% bromophenol blue). Diluted samples (6 μl) were heat-denatured and electrophoresed through a 5% polyacrylamide gel containing 8 M urea for 2 to 3 hr at 45°C. Tumor DNA and normal control DNA from each patient were run side by side on the gel so that the stutter bands of microsatellite DNA could be objectively compared. Gels were dried and exposed to X-ray film at room temperature overnight.

Assessment of MSI

Samples were scored as positive for MSI if there were additional bands that were not observed in the corresponding normal sample or if there was a band shift in DNA from the tumor tissue that contrasted with the corresponding bands of DNA derived from normal tissue of the same patient. A sample was considered MSI+ when shifted bands were detected at 2 or more microsatellite sites.

PCR-SSCP analysis of hMLH1 and hMSH2

Twelve pairs of PCR primers, described in previous studies,34, 35 were used to amplify 12 exons: 6 for hMLH1 (exons 1, 10, 13, 15 16 and 19) and 6 for hMSH2 (exons 5, 7, 10, 12, 13 and 15). SSCP and sequencing analyses have shown that these exons are mutational “hot spots” in other tumor types.34, 35 PCR was performed with [α-33P]-dATP (NEN, Boston, MA). PCR conditions consisted of an initial denaturation step of 2 min at 94°C, followed by 36 cycles of 50 sec at 94°C, 1 min at 50° to 60°C and 1 min at 72°C. The PCR product was diluted with an equal volume of loading buffer containing 95% formamide, 20 mM EDTA, 0.05% bromophenol blue and 0.05% xylene cyanol. The mixture was then denatured at 80°C for 5 min and cooled in ice water for 5 min, and 3 μl of each sample were loaded on a non-denaturing 5% to 8% acrylamide:bis-acrylamide (37.5:1) gel with 5% glycerol. Electrophoresis was performed using a Bio-Rad (Hercules, CA) vertical gel in 0.6× Tris borate (pH 8.3)-EDTA buffer at 10 W for 16 hr at room temperature.

Sequence analysis

DNA fragments that displayed a modified electrophoretic pattern were selected for sequencing. The PCR product was first amplified using exactly the same primers used for SSCP analysis. Then, a nested PCR was carried out on the originally amplified PCR product, to ensure specificity. Nested primers were 5′-ACT GGT TGT ATC TCA AGC AT-3′ (upstream) and 5′-TGT CAG AAG TGA AAA GGA TC-3′ (downstream). The PCR product obtained from a 50 μl PCR was sufficient for analysis. We used PCR products that clearly showed a specific band with a high yield. For PCR products with background bands, we separated all bands by gel electrophoresis and purified the DNA using the Qiaquick Column (Qiagen). Both strands were sequenced for all samples, to exclude artifactual changes. Sequencing was done using the ABI 377 and Big Dye terminator kits (Applied Biosystems Foster City, CA) at the DNA-sequencing facility of the Center for Applied Genomics, Hospital for Sick Children (Toronto, Canada).

Promoter methylation assay for hMLH1 and hMSH2

Methylation of the hMLH1 and hMSH2 promoter regions was examined using previously described protocols, with some modification.36, 37 Genomic DNA (100 ng) was digested with either no enzyme, HpaII or MspI; PCR was carried out using 2 primer pairs spanning hMLH1 nucleotides 881 to 1470. This region contains 4 CCGG sequences that can be digested by HpaII and MspI. Two reactions were done for hMLH1 using the primers hMLH1-881 and hMLH1-1219 or hMLH1-881 and hMLH1-1470. Primer sequences were as follows: hMLH1-881, 5′-CGC TCG TAG TAT TCG TGC-3′; hMLH1-1219, 5′-CAG CCA ATA GGA GCA GAG A-3′; and hMLH1-1470, 5′-TCA GTG CCT CGT GCT CAC-3′. The primer sequences used for hMSH2 analysis were as follows: hMSH2-379, 5′-GTT TCC TTC TGA TGT TAC TCC-3′; hMSH2-468, 5′-CCT GGG TGG GGT GTA TGC-3′. The resulting amplicon has 107 bp and contains 2 CCGG sequences. A portion of the dystrophin gene (MD) was amplified as an internal control using the primers 5′-TCC CAG ATC TGA CTC CTG TAG-3′ and 5′-ACA GTC CTC TAC TTC TTC CCA C CA-3′.

Expression analysis of hMLH1 and hMSH2

Immunohistochemical analysis for the expression of hMLH1 and hMSH2 was performed on paraffin sections from tumor and normal material of 12 HNSCC patients. Human skin tissue was used as a normal control. Sections (3 μm) were cut, dewaxed in toluene and rehydrated through graded alcohol to water. Endogenous peroxidase activity was blocked using 3% hydrogen peroxide. After antigen retrieval using standard techniques, slides were incubated overnight with primary antibody for hMLH1 (clone 168-728 at 1/50 dilution; Pharmingen, San Diego, CA) or hMSH2 (clone G219-1129 at 1/200 dilution, Pharmingen) at room temperature. After washing in PBS, secondary incubations were carried out with biotin anti-mouse IgG followed by streptavidin-horseradish peroxidase for 30 min each. Immunoreactivity was revealed by incubation in 3-amino-9-ethylcarbazol, visualized as a red stain. Slides were counterstained with hematoxylin and coverslipped with crystal mount.


  1. Top of page
  2. Abstract
  6. Acknowledgements

MSI and clinicopathological features

Clinical information from all 57 HNSCC patients is presented in chronological order, to make it easier to group and compare young and older patients (Table II). The list contains an age range from 20 to 87 years; 38 men and 19 women; 40 smokers and 17 non-smokers; 30 alcohol drinkers and 27 non-drinkers; 24 tumors from tongue, 13 from larynx and 20 from other head-and-neck sites. We did not consider the presence of 1 microsatellite alteration to be diagnostic of MSI; thus, our clinical correlations were calculated on the basis of MSI being demonstrated using 2 or more markers. MSI was defined by the presence of an extra band(s) and/or a band shift in tumor DNA that was not present in the corresponding normal DNA (Fig. 1). Using these criteria, 33 patients had MSI+ tumors (57.9%) and 24 had MSI tumors (42.1%) (Fig. 2, Tables II, III).

Table II. Summary of MSI and Risk Factors of HNSCC Patients
Young patients (<45 years)Older patients (>44 years)
Sex/age (years)MSI1Smoking statusAlcohol consumption2Tumor stageTumor siteSex/age (years)MSI1Smoking statusAlcohol consumption2Tumor stageTumor site
  • 1

    Number of MSI sites showing instability.

  • 2

    +, 1 or 2 drinks/day; ++, 2 to 5 drinks/day; +++, >5 drinks/day.

M/251T2N0TongueM/48020 packs/yearT1N0Larynx
F/262>40 packs/yearT3N0TongueM/492++T1N1Hypopharynx
M/286T3N2ATongueM/50027 packs/yearT2N0Larynx, SSC
F/297T3N0Oral-tongueM/53240 packs/year+T1N2ANeck node, adenocarcinoma
M/54325 packs/year++T3N0Tongue SSC
M/32115 packs/year+++T2N0TongueM/548+++Floor of mouth
F/3310T4N20Oral-tongueM/54030 packs/year++T4N2Floor of mouth
F/368T2N0Tongue SSCM/57320 packs/year++T3N0Tongue SSC
M/364T2N2Oral-tongueF/570++Larynx, SSC
M/361020 packs/yearT1N0Floor of mouthM/59120 packs/year+++T2N0Floor of mouth
M/38620 packs/year+T4N1Floor of mouthM/60280 packs/year+++T3N0Tongue SSC
M/389++T4N0LarynxF/60122.5 packs/yearT3N0Tongue SSC
F/383T3N0Floor of mouthM/61140 packs/yearT3N0Floor of mouth
M/38625 packs/year+++T2N1Oral SSCM/61122 packs/year+++T1N0Larynx, glottic
F/3812.5 packs/yearT3N1Tongue SSCF/65130–40 packs/year+T4N1Supraglottic larynx, SSC
F/38410 packs/yearT2N0TongueM/660+++T2N2OHypopharynx
F/396T2N0TongueF/66050 packs/year+++T2N0Buccal
F/395T2N0TongueM/661>40 packs/year+++T4N0Oral cavity
F/675>50 packs/yearT1N0Floor of mouth
M/403++T2N0Oral-tongueF/67545 packs/year+T2N1Oral-tongue
F/411420 packs/year++T2N0LarynxM/68064 packs/year+T2N0Larynx, poor SSC
M/42320 packs/year+++T2N1Oral-tongueM/69050 packs/yearT3N0Larynx, SSC
M/433<5 packs/yearT1N1Nasopharynx
M/44420 packs/year++T1N0LarynxM/70050 packs/year+++T2N0Tongue
M/70020 packs/yearT4N0Larynx, SSC
M/71155 packs/year+++T4N1Oral SSC, FOM
F/72330 packs/year+T2N2Floor of mouth
M/72050 packs/year++T3N0Larynx, SSC subglottic
M/732T4N0Upper palate, SSC-oral
F/73450 packs/yearT4N0Tongue
M/740>50 packs/year++T4N1Tongue/FOM
M/760>50 packs/year+T3N0Larynx, SSC
thumbnail image

Figure 1. MSI detected in HNSCC. Case numbers are shown above and locus symbols below. N, normal DNA; T, tumor DNA. Novel alleles in the tumor sample are indicated by arrows.

Download figure to PowerPoint

thumbnail image

Figure 2. Frequency of MSI detected in tumors from 57 patients with HNSCC. The y axis shows frequency of MSI detected using 16 microsatellite markers. The x axis shows ages of patients, ranging from 20 to 87years.

Download figure to PowerPoint

Table III. Correlation of MSI With Clinicopathological Features
ParameterNumber of cases (%)MSI+ (%)1MSI (%)p
  • 1

    A sample was considered MSI+ when detected at 2 or more sites.

Age (years)
 <4424 (42.1)21 (63.6)3 (12.5)
 >4533 (57.9)12 (36.4)21 (87.5)0.0001
 Male38 (66.7)19 (57.6)19 (79.2)
 Female19 (33.3)14 (42.4)5 (20.8)0.1935
 +40 (70.2)19 (57.6)21 (87.5)
 −17 (29.8)14 (42.4)3 (12.5)0.0195
 +30 (52.6)15 (45.5)15 (62.5)
 −27 (47.4)18 (54.5)9 (37.5)0.2838
Tumor site
 Tongue24 (64.9)17 (85.0)7 (41.2)
 Larynx13 (35.1)3 (15.0)10 (58.8)0.0075
 T 1/231 (56.4)20 (62.5)11 (47.8)
 T 3/424 (43.6)12 (37.5)12 (52.2)0.4088
 N038 (69.1)21 (63.6)17 (73.9)
 N 1/217 (30.9)11 (34.4)6 (26.1)0.5668

MSI changes were compared to clinical variables such as age, gender, smoking, alcohol consumption, tumor site and TNM. Statistical analysis was carried out using Fisher's exact test (Table III). Among the 57 HNSCC patients, 24 were younger than 45 years and all had at least 1 MSI change using the 16 microsatellite markers tested. Two or more MSI changes were detected in 88% of young patient tumors (21/24). In older patients, this number was 36% (12/33). A comparison of the frequency of MSI in tumors from young vs. older patients is highly significant (p = 0.0001) (Fig. 2, Table III). Of patients with MSI+ tumors at 2 or more sites, 74% were women and 50% men; 48% were smokers and 82% non-smokers (p = 0.0195); 50% consumed alcohol daily, whereas 66.7% did not drink; and 71% had tumors of the tongue, while 23% had laryngeal tumors (p = 0.0075) (Table III).

Our data indicate that the frequency of MSI is significantly different in tumors from young vs. older patients whether the age defined as “young” is 40 (p = 0.0051), 45 (p = 0.0001) or 50 (p = 0.0011) years. Women under the age of 40 have a higher incidence of disease than men (p = 0.0393). Smoking and daily alcohol consumption were observed less frequently in young patients under the age of 40 (p = 0.0002 and p = 0.0016, respectively). In addition, the incidence of tongue tumors was higher than that of laryngeal tumors for patients under the age of 40 (ratio 13:1, p = 0.0111; Table IV).

Table IV. Comparision of Different Age Groups With MSI and Other Features
ParameterNumber of cases (%)<39 years (%)>40 years (%)p<44 years (%)>45 years (%)p<49 years (%)>50 years (%)p
  • 1

    A sample was considered MSI+ when detected at 2 or more sites.

 +33 (57.9)16 (84.2)17 (44.7)21 (87.5)12 (36.4)22 (81.5)11 (36.7)
 −24 (42.1)3 (15.8)21 (55.3)0.00513 (12.5)21 (63.6)0.00015 (18.5)19 (63.3)0.0011
 Male38 (66.7)9 (47.4)29 (76.3)13 (54.2)25 (75.8)16 (59.3)22 (73.3)
 Female19 (33.3)10 (52.6)9 (23.7)0.039311 (45.8)8 (24.2)0.099311 (40.7)8 (26.7)0.2785
 +40 (70.2)7 (36.8)33 (86.8)11 (45.8)29 (87.9)13 (48.1)27 (90.0)
 −17 (29.8)12 (63.2)5 (13.2)0.000213 (54.2)4 (12.1)0.00114 (51.9)3 (10.0)0.0011
Alcohol drinking
 +30 (52.6)4 (21.1)26 (68.4)8 (33.3)22 (66.7)9 (33.3)21 (70.0)
 −27 (47.4)15 (78.9)12 (31.6)0.001616 (66.7)11 (33.3)0.016918 (66.7)9 (30.0)0.0081
Tumor site
 Tongue24 (64.9)13 (92.9)11 (47.8)15 (83.3)9 (47.4)16 (80.0)8 (47.1)
 Larynx13 (35.1)1 (7.1)12 (52.2)0.01113 (16.7)10 (52.6)0.03824 (20.0)9 (52.9)0.047
 T 1/231 (56.4)11 (57.9)20 (55.6)16 (66.7)15 (48.4)19 (70.4)12 (42.9)
 T 3/424 (43.6)8 (42.1)16 (44.4)18 (33.3)16 (51.6)0.56678 (29.6)16 (57.1)0.0577
 N038 (69.1)13 (68.4)25 (69.4)16 (66.7)22 (71.0)18 (66.7)20 (71.4)
 N 1/217 (30.9)6 (31.6)11 (30.6)18 (33.3)9 (29.0)0.77519 (33.3)8 (28.6)0.7753

Promoter methylation for hMLH1 and hMLH2

As methylation-sensitive restriction enzymes, such as HpaII and MspI, cannot digest sequences that are protected by methylation, a PCR product can be obtained only when the original target DNA is methylated. Tumor DNA from 21 patients (15 MSI+ and 6 MSI) was examined for the presence of methylation in the promoter regions of hMLH1 and hMSH2. Among the 21 patients tested for hMLH1 promoter methylation, 20 were methylation-negative and 1 was not informative. For hMSH2, 19 were methylation-negative and 2 were not informative. The dystrophin gene fragment was clearly amplified, indicating that the template DNA for this study was not degraded (Fig. 3). These data indicate that there is no detectable methylation in the promoter regions of hMLH1 and hMSH2 in the tumor tested.

thumbnail image

Figure 3. Analysis of methylation status of the hMLH1 promoter. PCR analysis of the hMLH1 and hMSH2 promoter region before and after digestion with methylation-sensitive enzymes. U, undigested; H, digested with HpaII; M, digested with MspI; N, normal male genomic DNA; G, normal male genomic DNA as PCR condition control; R, no human DNA as reagent control; P, MD primer used to test the quality of template DNA. T1 to T6 are tumor samples from 6 patients with HNSCC.

Download figure to PowerPoint

Mutation analysis by SSCP and sequencing

Sequence analysis of hMLH1 exons 1, 10, 13, 15, 16 and 19 and adjacent intronic nucleotides showed that tumor and adjacent normal tissue in both MSI+ and MSI patients (n =24) can have an A-to-G point mutation in the –17 position of the splicing adapter site of exon 15. As this germline mutation is non-coding and does not appear to affect the splicing of hMLH1, its relevance is unknown (data not shown). SSCP analysis of hMSH2 exons did not show any differences between tumor and normal samples; therefore, mutation analysis by sequencing was not performed for hMSH2.

Protein expression of hMLH1 and hMSH2

Immunohistochemical analysis was used to determine whether deregulated expression of hMLH1 and hMSH2 was correlated with MSI. Tumor and normal tissue from 6 young patients (3 with high MSI and 3 with low MSI, none had no MSI) and 6 older patients (3 with high MSI and 3 who were MSI) was analyzed. There was no observable difference in hMLH1 and hMSH2 expression among all 12 patient samples. Tumor cells (moderately differentiated SCC with keratinization) showed strong nuclear staining with hMLH1 and hMSH2 antibodies, similar to normal skin; this was accentuated in the basal layer. The normal adjacent lymphoid tissue and minor salivary gland tissue present in tongue and base of tongue also had strong nuclear staining, as did the associated lymphocytes and fibroblasts surrounding the nests of invasive SCC (i.e., stromal response to tumor). In contrast, skeletal muscle, fibrovascular tissue and adipose tissue were essentially non-staining. These results verify our data (see above) that the high frequency of MSI in tumors from young patients with HNSCC is not a consequence of deregulation of hMLH1 and hMSH2 (Fig. 4).

thumbnail image

Figure 4. Immunohistochemical analysis of hMLH1 and hMSH2 expression. (a) hMLH1 expression in normal human skin. (b) hMLH1 expression in a tumor with high MSI. (c) hMLH1 expression in an MSI tumor. (d) hMSH2 expression in normal human skin. (e) hMSH2 expression in a tumor with high MSI. (f) hMSH2 expression in an MSI tumor. Magnification 250×.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  6. Acknowledgements

Mutation studies using microsatellite analysis reveal 2 types of alteration: MSI and loss of heterozygosity. MSI indicates genomic instability, the significance of which is poorly understood. MSI also characterizes the mutator phenotype and is considered to be the consequence of exonic mutation or promoter methylation of MMR genes, rendering them dysfunctional. MSI was first observed in colon cancer,4 and in HNPCC families. It is caused by mutations in the human homologues of the bacterial genes MutS and MutL (hMSH2, hMLH1,4PMS1 and PMS238).

HNSCC is the sixth most common solid tumor worldwide, accounting for about 2% to 7% of all neoplasms in the Western world.1, 2 In HNSCC, MSI is reported to be in the range of 7% to 30%.21, 23, 36 Our results show a higher frequency of MSI (58%) in tumors from our total cohort of 57 patients. If the 24 patients under the age of 45 are removed from the data set, the frequency of MSI becomes 36%, a figure close to that found in the literature for patients with HNSCC (Table II).21, 23, 36 The high number of young patients in our cohort who showed a high frequency of MSI (88%, at 2 or more sites) obviously increases the overall percentage of instability observed. This high frequency of MSI has not previously been reported for patients with head-and-neck carcinoma and is a function of having enough young patients with this disease to stratify young and older patients into 2 groups. In our study, of the 16 MSI markers used, 13 are dinucleotides and 3 are tetranucleotides. Our data did not show a higher frequency of instability at tetranucleotide microsatellite sites (17.5%, 10/57 patients) compared to dinucleotide sites (24.6%, 14/57 patients). Data using 13 dinucleotide microsatellite markers showed that 100% of HNSCC patients under 44 years old had MSI+ tumors in at least 1 microsatellite locus (24/24) and 88% had instability at 2 or more loci (21/24). These results are the same as those using all 16 microsatellite markers including the 3 tetranucleotide markers. For the older patient cohort (≥45 years), 60% (20/33) had instability at 1 site and 30% (10/33) at 2 or more sites. When using all 16 primers, 36% (12/33) of patients showed MSI in these tumors at 2 or more sites. Thus, no statistically significant difference in the frequency of MSI was found whether 13 dinucleotide markers or a mixture of 16 dinucleotide and tetranucleotide markers were used.

Interestingly, in our entire patient group, MSI was detected at a higher frequency in tongue tumors (17/24, 71%) than in laryngeal tumors (3/13, 23%, p = 0.0075; Table III). Friedlander et al.39 found that tongue cancer occurs at a high frequency in young individuals, which is consistent with the observation of tongue tumors in 62.3% (15/24) of our young patients.

In previously published studies, most of the data on young patients were gathered from an age range under 40 or 50 years. We did our analyses using the ages 40, 45 and 50 as our cut-offs, to differentiate between young and older patient groups. The frequency of MSI and risk-factor analysis results were not markedly different between the 3 groups, probably because of the high proportion of very young patients in our data set. Unless specified, we used the age of 44 to divide the patients into young and older groups for statistical analysis.

In previous studies, MSI has been found at a high frequency in tumors from young patients with gastric and colorectal cancers.4–7, 37, 40 In young patients with gastric cancer, MSI was found to be as high as 50% (<55 years) and 52% (<35 years).10, 11 In colorectal carcinoma, MSI was detected in 23% (<40 years) and 32% (<30 years) of patient tumors.31, 41 MSI+ tumors were also found in 35% of patients with breast cancer under the age of 35 years.42 MSI was not observed in any of the tumors from 32 patients under the age of 45 years with renal cell carcinoma, suggesting that not all tumor types show instability at microsatellite loci.43 In our study, 2 or more MSI changes were detected at a frequency of as high as 88% in tumors from young patients with HNSCC. This number increased to 100% when instability at only 1 microsatellite locus was considered positive for MSI (Table II). In colorectal carcinoma, the MSI phenotype is linked to a better survival rate;44 this was not the case in our data set, where the 5-year survival rate was similar for both young and older patients.45 Whether or not younger HNSCC patients have a better prognosis remains unclear, with conflicting results from different institutions.39, 45, 46 A previous study at this institution of 185 patients under the age of 40 compared to a group over the age of 40 matched for site, gender and date of presentation has been published.45 Despite the differences in MSI described here, which may suggest a genetic propensity to cancer development in the younger patient group, there was no difference in clinical outcome in the younger group. The 5-year cause-specific survival in both groups was not statistically different (72% vs. 68%, p = 0.91). Although one might expect that the young patient group with MSI+ tumors might be as prone as the older patient group to the development of second primaries, this was not the case, with only 8% of the young patients developing second primary cancers compared to 18% of the older patients (p = 0.005).45

HNSCC typically occurs in older men during the fifth through eighth decades of life. These patients generally have a high incidence of tobacco and alcohol abuse. HNSCC rarely occurs in patients under 40 years of age, and traditional risk factors may or may not be present, implicating other, perhaps genetic, factors in the etiology of this disease. Our data support the concept that it is not necessarily abuse of smoking or drinking that is linked to the risk of developing HNSCC in our young patient population. Eighty-eight percent (29/33) of HNSCC patients over the age of 45 had a history of tobacco abuse, whereas only 46% (11/24) of young patients smoked (p = 0.001). There was a significant association between MSI and smoking history, with instability being more frequent in tumors from non-smokers compared to smokers (p = 0.0002). Data from Field et al.24 confirm these latter findings, though their patient numbers were small (3/4 patients) and age was not disclosed. Furthermore, only 33% of young patients consumed alcohol daily, whereas 67% of older patients did (p = 0.0169). Our analysis also indicates that young patients presented more often with T1 or T2 and N0 tumors than their older counterparts. MSI was mainly detected in TNM stage I and II carcinomas, indicating that genomic instability may occur in the early stages of tumor initiation/progression.

Tumors from HNPCC patients have been found to harbor frequent mutations within simple microsatellite repeat sequences, suggesting numerous replication errors during tumor development.4, 47 Exon mutations and/or promoter methylation of MMR genes has been reported in many MSI+ tumors from HNPCC,5, 6 ovary,16, 48 endometrium20, 49 and gastric37, 40 cancers. In previous studies of head-and-neck cancer, inactivation of MMR genes by exon mutation or promoter methylation has been rare. However, tumors from young patients with a high frequency of MSI have not been exclusively studied. While our analysis of the mutation “hot spots” of hMLH1 showed the presence of an A-to-G mutation in the splice adapter region of exon 15, we could not correlate this with the presence of MSI. Furthermore, this germline mutation did not appear to affect the splicing of hMLH1 as it occurs only 2 bases away from the left side of the splicing region branch site, thus rendering its significance uncertain. Promoter analysis of hMLH1 and hMSH2 did not detect CpG island methylation, indicating that this mechanism of gene silencing was also not involved in cases with MSI+ tumors. Finally, we analyzed gene expression of hMLH1 and hMSH2 in paraffin-embedded tissue sections using immunohistochemistry. Expression of both proteins was detected in all patient groups, in both normal and tumor material. Combining the results of mutation screening, promoter methylation analysis and expression studies, we conclude that the 2 genes most frequently mutated in other MSI+ tumors are not associated with MSI in HNSCC.

In conclusion, tumors from young patients with HNSCC show a high frequency of MSI. Furthermore, MSI is seen more frequently in young non-smokers with tumors of the tongue. MSI does not appear to be associated with dysfunction of the MMR genes hMLH1 and hMSH2, as shown by SSCP, sequencing, promoter methylation analysis and immunohistochemistry. The involvement of other genetic mechanisms is currently being explored.


  1. Top of page
  2. Abstract
  6. Acknowledgements

We thank Drs. J. Dimitroulakos and J.L. Hummel for their review of and comments on this report.


  1. Top of page
  2. Abstract
  6. Acknowledgements