Introduction

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common solid tumor worldwide, with an annual incidence of more than 500 000. Traditional treatment of this tumor type is based on combinations of surgery, radiotherapy and chemotherapy with still a rather poor success rate for locally advanced disease. As a result, enormous efforts are currently being put into the search for new molecular markers and associated molecular treatment strategies that might stratify patients and individualize treatment options. With these aims in mind, microsatellite alterations such as microsatellite instability (MSI) and loss of heterozygosity (LOH) gained a lot of interest during the last decade.

The interest in MSI as possible tumor marker was greatly influenced by the marked presence and prognostic value of this phenomenon in Hereditary Non Polyposis Colon Cancer (HNPCC).1 MSI is also typically associated with Turcot's and Muir–Torre syndrome.2 However, the frequency and clinical value of MSI in other solid tumors including HNSCC differs widely, largely due to an obvious variance in used methods and criteria.2

Concerning LOH, different techniques have been developed to assess this type of alterations. The further report on LOH in this work will only refer to LOH detected with microsatellite markers. Problems of methodological variance trouble the reported frequencies, in alignment with MSI assessment. The prognostic and predictive value of this phenomenon varies along different tumor types (bladder, prostate, brain cancer).3, 4, 5 Although the potential of LOH as a molecular prognostic marker in HNSCC has gained confidence over the years, literature on this field stays confusing.

The aim of this paper was to review the current knowledge of microsatellite alterations, both MSI and LOH, in HNSCC and to provide the reader with an assessment of their prognostic and predictive value.

What are MSI and LOH?

Microsatellites are stretches of DNA in which a short motif (usually 1–5 nucleotides long) is repeated 5–100 times. Microsatellite regions are at high risk for variations in the number of repeats caused by slippage of the DNA polymerase during DNA replication. In normal cells, these errors get repaired by the mismatch repair system (MMR) involving proteins such as MLH1, MLH3, MSH2, MSH3, MSH6, PMS1 and PMS2. In tumors, defects in the MMR system may be present, so that variations in microsatellite regions are not repaired correctly, leading to definitive somatic changes with gain or loss of repeat units. MSI can thus indicate the presence of a defect MMR system. Interestingly, MMR deficiency can lead to carcinogenesis as a result of mutations in key genes, defining a so-called mutator phenotype. According to Field, MSI could serve as an indirect marker for the somatic mutations caused by the mutator phenotype.6 This mutator phenotype is an important carcinogenetic mechanism in HNPCC, which is associated with inherited MMR mutations.1

Genetic and epigenetic inactivation of the MMR pathway is however not a commonly event in solid tumor types other than HNPCC. Therefore, MSI in these other tumor types, if observed, may be due to factors other than defects in MMR genes such as malfunctioning of factors downstream of MMR components or in other systems that proof-read DNA for replication errors.7 Although not clearly established, environmental factors such as oxygen radicals, lipid adducts, smoking or diet also could play a role in generating MSI. These exposures can act alone or in concert with defective DNA repair pathways.8

LOH, in contrast, marks a suppressor phenotype that is characterized by wide variety in chromosomal numbers (aneuploidy) and extensive loss of genetic material. Even if nowadays most LOH investigations use comparitive genomic hybridization or single-nucleotide polymorphisms (SNPs) techniques, the presence of LOH can also be detected with the use of microsatellite markers. Accordingly, LOH can be recognized by loss of a genomic fragment, SNP or microsatellite allele in a tumor when compared with normal tissue. LOH is a common mechanism of inactivation of tumor suppressor genes (TSG) located in the neighborhood of the allelic marker that is being detected. TSG loss is normally confirmed by investigating gene deletion or mutation, silencing by DNA methylation or concurrent loss of protein expression.9

Prevalence of MSI and LOH in HSNCC

MSI

In HNSCC, the reported frequencies of MSI vary widely but in general the incidence is low leaving any significance unclear. Table 1 that shows the published studies till now on MSI in HNSCC indicates that the incidence varies from 321 to 88%24 of MSI-H, largely depending on the number of patients, the loci and markers used, the patient age and detection methods.

Table 1 MSI in HNSCC

In general, MSI in HNSCC seems to be a late event associated with tumor progression. Most studies show that invasive carcinomas manifest more MSI than precursor lesions, suggesting progressive accumulation of MSI during tumor development12, 28 (Table 1). Furthermore, precursor lesions that show MSI are more prone to progress to HNSCC.19 Arguments against this vision arise from studies showing high MSI incidences in pre-malignant lesions and in young patients.20, 24

Most studies report on a lack of MMR gene defects in HNSCC compared with HNPCC.14, 21, 24, 29 This could either indicate a low incidence of MSI or a higher proportion of non-MMR MSI cases. In HNSCC, this could be caused by common carcinogens for this patient group, such as smoking. However, it could also mean that ‘true’ MSI in HNSCC does not exist – at least not in the way it does in HNPCC.

LOH

Numerous different techniques have been used to assess LOH alterations in HNSCC. Perhaps this is one reason why, although the potential of this molecular prognostic marker has gained confidence over the years, the literature in this field stays confusing. Table 2 lists the published studies on LOH detected with microsatellite markers in HNSCC.

Table 2 LOH in HNSCC

Most suggest that the frequency of LOH is higher than of MSI, indicating that the suppressor phenotype could be more prevalent.11, 12, 36 LOH and the spectrum of chromosomal loss progressively increases at each histopathological step from benign hyperplasia to dysplasia to carcinoma in situ to invasive cancer.19, 33, 53 Some primary and recurrent HNSCC originate from the same precursor lesion that contains genetically related features to the tumors, including LOH.54

Overall, studies in HNSCC have shown that deletions at chromosome arms 3p, 4p, 8p and 9p represent early genetic changes and that loss at 18q, 17p and 11qter is associated with neoplastic progression.33, 34, 36, 45, 48, 55 Dysplastic regions in the head and neck region that show more LOH seem to be more at risk for neoplastic evolution.17, 18, 19, 33

There are several reports that LOH is associated with more advanced stage and more aggressive HNSCC tumors,13, 34, 36, 37, 56 and specifically with nodal involvement.32, 44 Other authors describe very specific correlations such as a link between LOH at 17p and mitotic index,39 a connection of LOH at MLH1 with lower grade HNSCC, and LOH at CDKN2A with higher grade,9 or a preference of LOH in tumors originating from the pharynx.56 A correlation between LOH and other patient-related factors like age and race, was described in some studies.34, 51

Methodological issues

The first critical point in MSI and LOH detection is microsatellite marker selection. The use of the Bethesda MSI reference set8 that was developed for HNPCC versus presumably more HNSCC-selective markers is hotly debated, as is the use of mononucleotide versus polynucleotide markers.17, 57, 58, 59 However, most researchers currently accept that for MSI assessment, quasimonomorphic mononucleotide markers are suitable. For the detection of LOH with microsatellite markers, locus selection is dependent on the specific chromosomal region and possible associated TSG that the researchers reckon to be relevant.

Second cause for the observed variance in MSI and LOH reports lies in the use of different detection methods. The modern automatic fragment analysis procedures offer high throughput analysis and a more precise and quantitative assessment of MSI and LOH in comparison to the previously used gel electrophoresis methods.60, 61, 62, 63, 64 Although other points of discussion are the cutoff values to be used for LOH ratios (the ratio between the peak height ratios of both alleles of normal and paired tumor sample),9, 39, 63, 65 there is fairly general agreement that appropriate cutoffs are >2 or <0.5. A last issue is on the different types of MSI. Most investigators discern MSI-High (MSI-H, MSI in ≥30–40% of the markers tested) and MSI-Low (MSI-L, MSI in <30–40% of the markers tested),8 although others call these Type I and II62 or Type B and A instability,66 respectively.

Prognostic and predictive value of MSI and LOH in HNSCC

MSI

Prognostic value

In HNSCC, the prognostic impact of MSI is not clear.2 Some studies showed that MSI positivity in pathologically negative surgical margins of HNSCC can independently predict local recurrence.67, 68 Some authors found no correlation between MSI in HNSCC and survival,6 whereas others describe a better prognosis for MSI-positive patients,69 as recently also published for colorectal cancer.1 A possible explanation for the lack of prognostic value is the lack of statistical power in a lot of studies owing to the overall low prevalence of MSI in HNSCC, next to the previously mentioned difference in methodology.

Predictive value

As MMR-proficient cells undergo a cytotoxic reaction after recognition of drug-induced adducts in DNA, MMR deficiency can theoretically impart resistance to cancer chemotherapy agents. Some mainly in vitro and animal work on different tumor types indeed suggested that MMR-deficient cells have a poorer response to cisplatin, carboplatin and methylating agents.7, 69, 70, 71 However, clinical studies in colorectal cancer, in which a relation between MSI and MMR deficiency is clearly established, have not shown a consistent predictive value of MSI.1, 72 For ionizing radiation (IR), a link with MMR defects is unlikely, as IR mainly causes DNA strand breaks and repair of this damage does not involve the MMR system. Therefore, it is not surprising that no predictive value of MSI towards chemotherapy or radiotherapy has been found in HNSCC, a malignancy in which the occurrence of MMR deficiency is uncertain.11

LOH

Prognostic value

In general, LOH at multiple loci is associated with a poor clinical course such as nodal invasion or high-grade disease.44, 55 Furthermore, several studies have identified a negative prognostic role for LOH,18, 37, 41, 73, 74 even in multivariate analyses.34, 35, 46 This negative prognostic value is mostly connected to allelic loss at 3p, 8p, 11q, 13q, 17p and 18q. On the other hand, LOH at 9p21 seems to be an early event in carcinogenesis (like LOH at 3p) that has not consistently been connected with a poor prognosis,34, 35, 75 but along with LOH at 7q31 has been related to tumor recurrence.76, 77 At the most common loci, the occurrence of LOH is suggested to be associated with loss of TSG: FHIT at 3p, p16INK4A at 9p, Rb at 13q and so on. For all of these TSG, a prognostic role has been described in HNSCC.78, 79, 80

Predictive value

One study showed that microsatellite alterations in a histopathologically negative surgical margin can predict local recurrence.68

LOH is a frequent mechanism of inactivation of TSG, which might be involved in resistance to chemotherapy. A study in HNSCC showed an association between LOH at 9p or 17p resistance to a chemotherapy regimen consisting of cisplatin and 5-fluorouracil, with an independent negative predictive role for LOH at the p53 locus on 17p. The authors concluded that p53 alterations could play a role in chemotherapy resistance in HNSCC.11

Conclusion

This review aimed to explore current knowledge about microsatellite alterations (MSI/LOH) in HNSCC. HNSCC typically evolves from normal epithelium through dysplasia, carcinoma in situ finally to the invasive carcinoma stage. During this tumorigenesis, cumulative genetic alterations including MSI and LOH occur.

In HNSCC, most recent research efforts have been put into the investigation of LOH at several chromosomal loci. These alterations, representing the suppressor phenotype, seem to be more common than MSI in HNSCC. Although both types of microsatellite alterations have been correlated with clinicopathological features of head and neck cancer, only LOH seems to have a clear prognostic value. The predictive value of both MSI and LOH towards surgery, radiotherapy and chemotherapy is debatable. Biggest challenges however remain in the methodological problems connected with these types of investigations.

We recently tried to determine the real MSI and LOH prevalence in HNSCC, using automatic fragment analysis as the preferred technique to assess MSI and LOH, several panels of microsatellite markers in an attempt to compare their sensitivity, and strict cutoff values for LOH detection. This study resulted in a very low (around 1%) percentage of MSI, suggesting that indeed the prevalence of MSI in HNSCC has been overestimated in literature, partly due to the use of non-optimal techniques (De Schutter et al, submitted).82 Based on this experience, a role for MSI as prognostic or predictive marker in this tumor type seems highly unlikely.

On the other hand, the detection of LOH with the use of microsatellite markers in HNSCC seems feasible and of clinical importance. As LOH at certain loci may be indicative for the loss of a TSG, therapeutic options would mainly be directed towards re-expression of the involved gene, which is the goal of several gene therapy trials. However, re-expression therapies are mainly experimental and still face a lot of difficulties.

More research has to be performed to establish clearly LOH detection as a translational application in the HNSCC field, aiming to predict response to treatments or outcome, and eventually to use as a therapeutic target.