Squamous cell carcinoma of the head and neck (SCCHN) is one of the leading cancers in India. It is estimated that in the year 2001 the total number of new cases will be 137,000.1 Cancer prevalence shows a geographical pattern attributable to lifestyle. Differences in lifestyle factors in determining regional variations of cancer prevalence may be crucial for a country like India with a wide ethnic and lifestyle diversity. It has been speculated that both ethnic background and lifestyle differences may contribute to the differences in predisposing factors to cancer development.1
A crucial molecular change measured in various types of cancers is genome-wide alteration in simple repeat sequences, which was observed initially in hereditary nonpolyposis colon cancer (HNPCC) as well as in sporadic colon cancer2–4 and subsequently in many other tumors.5 This type of instability, also known as microsatellite instability (MIN), can be detected as a change in the length of microsatellite sequences in tumor DNA compared with normal DNA from the same individual. Such alterations reflect a defect in replication or repair known as the replication error (RER) phenotype, which persists throughout the lifetime of the tumor.6, 7
In HNPCC, MIN is apparently due to inherited and somatic mutations in any of the five mismatch repair (MMR) genes such as hMSH2,8, 9hMLH1,10, 11hPMS1 and hPMS212 and hMSH6/GTBP,13 located on human chromosomes 2p22-p21, 3p21.3, 2q31-q33, 7p22 and 2p16, respectively. Somatic mutations in MMR genes were also identified in sporadic colorectal primary tumors and cell lines demonstrating the RER+ phenotype.14, 15 Colorectal cell lines with defective MMR genes exhibited at least a 10- to 100-fold increased rate of spontaneous mutation within expressed genes.16, 17 This observation provides a mechanism linking the RER+ phenotype to the accelerated development of cancer in HNPCC kindred. The link between DNA repair defects with a specific pathway of tumor progression has been made from the observations that some human colon cancer cell lines and primary tumors with high frequency of MIN harbor mutations within small repeated sequences in the TGF-β type II receptor gene,18 the insulin-like growth factor II receptor gene19 and the Bax gene.20
The MIN detected in different tumors varies considerably. For example, 55% to 86% of the HNPCC tumors and other tumors arising in HNPCC kindred exhibited MIN, but only 15% of the sporadic colorectal tumors were observed to have such instability.5 Similarly, the incidence of MIN in other sporadic tumors varies from as high as 67% in pancreatic tumors to as low as 3% in breast, brain, bladder and liver cancers.5 The available data suggest that not all sporadic cancers showing the RER+ phenotype can actually be accounted for by somatic mutations in known MMR genes.21–24 It has therefore been suggested that MIN, occurring in tumors other than HNPCC, is attributable to defects in genes distinct from those responsible for colorectal cancer.21
Incidences of RER+ type MIN reported for SCCHN in British and American patients are 28% and 30%, respectively.25, 26 In a recent French study, an even lower incidence of MIN (12.5%) has been reported.27 In this report we show that the frequency of RER+ type MIN in SCCHN tumors is much higher in the Indian patient population than that reported for the other populations. We also report that several features of the MIN in these tumors are distinctly different from that observed in colorectal tumors.
MATERIAL AND METHODS
Fifty freshly operated primary SCCHN tumors and 5 samples of severely dysplastic tissue, along with their corresponding normal tissues or peripheral blood leukocytes (PBLs), were collected from the patients prior to treatment. Samples were immediately frozen and stored at −80°C. All the tumors were histopathologically diagnosed as squamous cell carcinoma and graded and staged according to the UIC TNM classification. The detailed history of the 50 patients' primary SCCHN tumors is presented in Table I.
Table I. History of Patients with SCCHN Tumors Used in the Study1
Among the 50 patients, 35 were male and 15 female. The mean age of patients was 49.5 years. Patients were grouped into three categories, as follows: 1) oral cavity, 2) larynx and 3) orofacial, on the basis of affected primary sites. Thirty-four tumors were found in the oral cavity (4 alveolus, 14 buccal mucosa, 7 tongue, 5 tonsil, 2 palate and 1 each hypopharynx and lip), 6 tumors were from the larynx and 10 tumors were categorized as orofacial tumors (5 maxilla, 3 mandible and 2 nasal cavity). Severe dysplastic changes were observed in five tissues. Histopathologically the tumors were classified as stage I (3 tumors), stage II (16 tumors), stage III (16 tumors) and stage 4 (10 tumors). Among these 50 tumors, 22 were lymph node positive, and 28 were diagnosed as free from any lymph node involvement. For five patients (D6, 377, 333, 4717 and D35), no history was available except that they were from the head and neck region.
Microdissection and DNA isolation
The normal cells present as contaminants in the primary tumor tissues were removed by a microdissection procedure.28 More than 50 to 60 serial tumor sections (10 to 20 μm) were dissected and placed on glass slides using cryostat (model CM 1800, Leica, Heidelberg, Germany). The representative 5 μm tumor sections from different regions of the tumor (beginning, middle and end) were stained with hematoxylin and eosin for diagnosis as well as for marking the tumor-rich regions. The normal tissues present in the tumor sections were then removed by microdissection. The adjacent normal tissues of the primary tumor were similarly dissected for the presence of tumor cell infiltration. For those cases in which the “normal” tissue contained any tumor cell infiltration, PBLs of the corresponding patient were taken as normal. The microdissected tumor samples containing <60% tumor cells were not taken for further analysis. High molecular weight DNA from the tissues was extracted by proteinase K digestion followed by phenol-chloroform extraction.29
Polymerase chain reaction
Microsatellite loci (Table II) were amplified by PCR in a 25 μl reaction volume containing 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 10% DMSO, 0.2 mM of each dNTPs, 4 pmol of each primer, an optimal concentration of MgCl2 (1.5 to 3 mM) determined separately for each primer set, and 50 to 100 ng of template genomic DNA. Reaction mixtures were heated to 95°C for 3 min, cooled to annealing temperature (50° to 65°C) to add 0.5 U Taq DNA Polymerase (GIBCO BRL, Gaithersburg, MD) and cycled for 30 times in GeneAmp 9600 (Perkin-Elmer, Oak Brook, IL) PCR machine. Each cycle was 10 sec at 95°C, 5 sec at the appropriate annealing temperature (50° to 60°C) and 10 sec at 72°C. Final extension was carried out at 72°C for 7 min. Samples without DNA for each primer pair were included as negative controls. Prior to PCR, the forward primer in the reaction mixture was end-labeled with [γ32P]dATP (specific activity 3,000 Ci/mM, Amersham, Buckinghamshire, UK) using T4 polynucleotide kinase (GIBCO BRL). Two microliters of the amplified DNA was electrophoresed on a denaturing 6% polyacrylamide gel at 45 to 60 W for 2 to 4 hours. After electrophoresis the gels were exposed to X-ray films (KODAK X-Omat) for autoradiography.
Table II. Frequency of Instability of Microsatellite Markers in SCCHN Tumors1
Type of repeat
Percentage of tumors showing instability
Numbers in parentheses in the fourth column indicate tumors showing instability out of total tumors analyzed.
Alteration in BAT26 (mononucleotide repeat) was examined only in those tumors that showed instability in two or more markers.
Definition of microsatellite instability and loss of heterozygosity
The status of MIN was scored as positive if one or both alleles in the tumor exhibited size variation due to either expansion or contraction of the repeat sequences in comparison with paired normal tissue from the same individual. Loss of heterozygosity (LOH) was determined by densitometric scanning (Shimadzu, Singapore, Singapore CS-900) of the autoradiographs. A given locus was scored as having LOH if the relative band intensity of one allele was at least 50% decreased in the tumor samples compared with normal samples. The value was calculated as the ratio of the band intensities of the larger to the smaller bands in the tumor DNA samples divided by the same ratio in the normal DNA samples. An LOH index of >1.5 (loss of the smaller allele) or <0.67 (loss of the larger allele) corresponds to at least a 50% reduction in relative band intensities.
Frequency of MIN in SCCHN tumors
Fifty SCCHN tumors and five severely dysplastic tissues were examined for genomic instability at 18 microsatellite loci (1 mono-, 9 di-, 5 tri- and 3 tetranucleotide repeats) spread over eight different chromosomes (Table II). Three of these markers (HD, DM and DRPLA) are associated with different neurological disorders and show germline expansion. Since chromosomes 3 and 9 show frequent LOH in SCCHN tumors, among 18 markers selected, 4 were from chromosome 3 and five from chromosome 9. Also, five mononucleotide repeat sequences from different growth-related and repair genes such as BAX, IGFIIR, TGFβRII, hMSH3 and hMSH6 were analyzed18–20, 30 only in those tumors that exhibited instability in two or more of the loci listed above (Table II). Each microsatellite locus was amplified in paired normal and tumor DNA samples. Fifty-one of 55 tumors were analyzed for at least 10 markers, and the frequency of instability for each marker was determined (Table II). Differences in DNA band patterns between tumor and normal cell pairs were observed to be 1) at least in one locus in the case of 60% tumors (33/55) and 2) in two or more loci in the case of 45% tumors (25/55).
Relatively recently it has been proposed that in certain tumors larger repeats are more commonly altered than smaller repeats.31 This prompted us to determine MIN levels separately for di-, tri- and tetranucleotide repeats for each tumor that exhibited instability in more than two markers (Table III). Among 25 tumors examined, 21 exhibited instability for both di- and trinucleotide repeats, but only 9 tumors were observed to be unstable for tetranucleotide repeats. Thus, previously reported “elevated microsatellite alterations at selected tetranucleotide” (EMAST) repeats for non-HNPCC tumors including head and neck cancers31 have not been observed in the present study using a limited number (three) of tetranucleotide repeats. We therefore based our classification of the 25 tumors showing instability at two or more loci on overall MIN observed and compared with the data obtained in studies on SCCHN tumors in other populations.
Table III. Microsatellite Instability (MIN) in SCCHN Tumors Showing Alterations in ≥2 Markers
% of MIN for dinucleotide repeat
% of MIN for trinucleotide repeat
% of MIN for tetranucleotide repeat
Overall % of MIN
1Values in parentheses indicate number of markers observed to have instability out of total markers analyzed for that tumor.
Eight tumors (15% of the total 55 tumors) show instability at >40% of the markers and 17 tumors (30% of 55) show instability at 10% to 40% of the markers tested (Table III). Among 17 tumors in the low MIN group (10% to 40% instability), four (#3970, #0710, #0751 and #1295) exhibited instability for tri- and/or tetranucleotide markers only. The remaining 30 tumors exhibited MIN for <10% of the markers tested; these may be considered as microsatellite stable tumors (data not shown). Interestingly, seven of eight tumors that exhibited instability at >40% of the markers tested were observed to have MIN in the D3S1611 marker, located within hMLH1 gene. Five of these tumors exhibited LOH at D3S1611 (data not shown), suggesting that these tumors probably have a defective hMLH1 gene.
Types of MIN observed in SCCHN tumors
Six different types of alterations in the microsatellite sequences were observed in the present study (Fig. 1). Those markers that were heterozygous in normal tissues were observed to have three different types of alterations in tumors: 1) alteration in length of one allele (Fig. 1a, tumor 0693), 2) alteration in length of both alleles (Fig. 1a, tumor 4332) and 3) loss of one allele with increase in band intensity for the other allele (Fig. 1a, tumor 710). Similarly, those markers that were homozygous in normal tissues also exhibited three different types of alterations in tumors: 1) generation of a new allele in addition to the existing one (Fig. 1b, tumor 4332), 2) replacement of the existing allele with two new alleles (Fig. 1b, tumor 710) and 3) in a few cases replacement of the existing allele with one new allele (Fig. 1b, tumor 1452).
LOH was observed in 38 different genotypes. Among these 38 genotypes, in four cases one allele was observed to have altered length while the other allele was lost (data not shown). None of the SCCHN tumors analyzed were observed to have the type of MIN commonly found in colorectal tumors (i.e., new alleles observed in addition to those detected in normal cells). The SCCHN tumors with MIN were always observed to have no more than two alleles—either or both having an altered size compared with the marker in normal tissue, implying that these tumor tissues are composed of a homogeneous cell population with respect to the marker tested. The possibility that the bi-allelic alterations observed in our study might be due to sample mixup has been ruled out by genotyping the paired normal and tumor samples with four different RFLP markers (data not shown), which clearly distinguished between the paired samples. Relative frequencies of mono- and bi-allelic alterations as well as LOH observed in these tumors are shown in Figure 2. The results show that most of the instabilities are mono-allelic in nature. However, one significant observation is that the bi-allelic and LOH types of alterations were observed in equal frequencies. For 5% of the genotypes, the pattern was ambiguous and the result could not be ascertained with confidence.
Lack of instability of mononucleotide repeat in SCCHN tumors
A striking observation in the present study is the absence of any noticeable instability at mononucleotide markers in SCCHN tumors. The mononucleotide marker BAT 26, which amplifies a repeat of 26 deoxyadenosine, located in the fifth intron of the hMSH2 gene, is quasi-monomorphic in the general population but shows considerable instability in tumors and is considered to be the best indicator of MIN.32 However, this marker was completely monomorphic among all tumors tested (24/25) except for one (#4904), in which only one nucleotide difference was detected by sequencing (data not shown). Colorectal tumors with MIN have been reported to show high frequency of instability in mononucleotide runs in coding regions of genes known to be involved in growth regulation.18–20, 30 We therefore studied instability in the mononucleotide runs of BAX, IGFIIR, TGFβR2, hHSH3 and hMSH6 genes in SCCHN tumors (Fig. 3 and data not shown). However, none of the tumors tested exhibited instability in these mononucleotide runs. The observed lack of instability in the tested mononucleotide repeats due to experimental error was tested by inclusion of control in the PCR and gel electrophoresis. Thus, instability in hMSH6 locus was observed in case of both RER+ colorectal cell lines HCT116 (Fig. 3b) and DLD1 (data not shown) used as controls.
Comparison of MIN in different clinicopathological stages of SCCHN tumors
The frequency of MIN in 50 SCCHN tumors was compared using different clinicopathological parameters (including age, site, histological differentiation, TNM staging and positive node at pathology), and no correlation was observed (Table IV). Similar observations have also been made for SCCHN tumors from other ethnic populations.25 Tumors from all the sites examined, except the tongue, exhibited MIN. We had only one sample from the lip, which did not show instability. Tumors from TNM stages I and II exhibited a high incidence of MIN (Table IV), indicating that this phenomenon is common in the earliest stage of the disease. This argument is further strengthened from the observation that two severely dysplastic tissues (#0751 and #0710) also exhibited the MIN phenotype. Examination of the age of diagnosis of SCCHN tumors reveals that patients with early age of onset exhibit a high incidence of MIN in their tumors (Table IV).
Table IV. Comparison of Microsatellite Instability and Clinicopathological Parameters in SCCHN1
NS, not significant. See footnotes of Table I for other abbreviations.
The result of this investigation indicates that a subset of SCCHN tumors from Indian patients shows microsatellite instability in various short tandem repeat sequences. Various investigators used different criteria for defining a tumor as microsatellite unstable.33, 34 In some studies, a tumor with instability in as few as one locus has been classified as a microsatellite unstable tumor,33 while others used more stringent criteria such as instability in 40% of the loci tested.34 In the present study 45% of the tumors studied show instability at more than two loci, and 15% of the tumors exhibited instability at 40% of the markers tested. The incidence of microsatellite instability reported for SCCHN tumors in the American patient population is 29% (at one or more loci) and that of British patient populations is 28% (at two or more loci).25 Also, in the British study only 3 of 56 tumors exhibited instability at more than 40% of the markers tested.25 Another study reported only 7% SCCHN with MIN; all of them exhibited instability at 40% or more loci.35 In a French study only 1 tumor exhibited high MIN type, and 6 tumors were classified as low MIN type out of 56 tumors analyzed.27 Thus SCCHN tumors from Indian patients show much higher MIN frequency than patients from other populations.
In contrast to HNPCC and sporadic gastric and endometrial tumors, which show instability mostly at the smaller (mono- and dinucleotide) repeats, elevated levels of instability in tri- and tetranucleotide repeats were reported for other sporadic tumors.31 It has been recommended that this different type of MIN be termed elevated microsatellite alterations at selected tetranucleotide (EMAST) repeats to achieve uniformity in the classification of tumors exhibiting MIN.31 Using three tetranucleotide (AAAC, TCTA and TAAA) repeat markers, we observed no EMAST-like activity; on the contrary, the instability of tetranucleotide markers was lower (9/25) compared with the observed instability (21/25) of di- and trinucleotide repeat markers. However, the number of tetranucleotide repeats used in this study was limited and did not include (AAAG)n, which is reported to be particularly susceptible to a higher level of instability.36
In the present study, MIN observed in SCCHN tumors has two distinctive features. First, complete absence of instability at mononucleotide repeat sequences was observed. This observation suggests that genes that are defective in these tumors may be different from that of RER+ colorectal tumors. Second, the appearance of one or two new alleles with concomitant loss of one or both constitutional alleles was noted. This observation is in sharp contrast to the type of MIN commonly found in colorectal tumors, which would include new alleles in addition to the constitutive allele(s) present in the normal tissue. However, careful analysis of published data from various tumors including SCCHN reveals that similar results were also obtained by some workers.27, 28, 33 One plausible explanation for this observation might be that a transient form of MMR deficiency occurred in the early stage of tumor development and subsequently defective cells were selected that exhibited no further instability. Alternatively, it has been argued that such an observation could be made due to clonal expansion of an already altered cell from heterogeneous normal tissue.37
We are grateful to Dr. S. Gupta (Director, Cancer Centre and Welfare Home, Calcutta) for providing some of the tumor samples and to Dr. K. Chaudhuri (Indian Institute of Chemical Biology, Calcutta) for help with statistical analysis. We thank all the members of the Human Genetics Department for their kind cooperation and encouragement during the study. S.C. is grateful to the Council of Scientific and Industrial Research, New Delhi, India for a predoctoral fellowship.