EXPRESSION PROFILING TO PREDICT OUTCOME AFTER COMBINED RADIOTHERAPY AND CHEMOTHERAPY IN HEAD AND NECK TUMORS  

Purpose: The objective is to improve the prediction of local tumor control after cisplatin-chemo-radiotherapy in squamous cell carcinomas of the oral cavity, larynx and pharynx using gene expression profiles of pre-treatment biopsies.

Background: In head and neck squamous cell carcinomas, radiotherapy and chemo-radiotherapy are the preferred treatment modalities for organ preservation but fail in 20-30% of cases. Surgery, concomitant chemo-radiotherapy, accelerated radiotherapy, or oxygenation modifiers with radiotherapy are employed in larger tumors , although the choice between these modalities is often arbitrary. Clinical parameters, even assessed multifactorially, have limited additional value for selection of the optimal treatment strategy and do not predict response accurately. Tumor response is hoofd-halswn to be multifactorial, and both biological and clinical factors need to be considered concurrently. No reliable biological predictive markers are yet available for aiding treatment choice. Detailed analysis of gene expression using DNA microarrays can reveal profiles reflecting intrinsic radiosensitivity and chemosensitivity, as well as to parameters related to tumor hypoxia, repopulation and malignancy characteristics. Such expression profiles have proven their value in predicting treatment response in other tumors and cell cultures, and are likely to ultimately help in choosing between different treatment options and thus individualizing treatment.

Plan of Investigation:

Fresh frozen tumor tissue has been collected from pretreatment biopsies and recurrences from patients with advanced oral cavity, laryngeal and pharyngeal carcinomas treated within a trial with high dose cisplatin-based chemo-radiotherapy (RADPLAT) or low daily dose cisplatin with concurrent radiotherapy. RNA from biopsies will be extracted and amplified. Fluorescent-label aRNA will be used for competitive microarray hybridization on oligo-arrays from the NKI Central Microarray Facility. The expression profiles of non-responders (with a local recurrence) will be compared to a matched group of responders (without a local recurrence). As of January 2004, 198 fresh frozen pretreatment tumor biopsies from eligible patients have been collected and stored in our tissue bank. 44 of these patients suffered a local/regional recurrence. We will use material from RADPLAT patients treated between 1998 until 2002, giving sufficient follow-up time for definitive outcome assessment by 2005, as the test series to define the predictive profile. We will then validate the profile on two further series. Firstly, material from patients treated with the same RADPLAT trial after 2002, and secondly, from a parallel series treated with low daily dose cisplatin concurrent with radiotherapy. For validation, we expect to have at least 50 patients in each of the two series. At the conclusion of the project, sufficient follow-up time will have accrued for the validation series.

Further, we will carry out limited validation on the predictive gene panel using QRT-PCR and immunohistochemistry on tissue microarrays. For the IHC we will use a panel of antibodies reported to be of predictive value, as well as a panel derived from the results of the gene expression profile.

Possible results and relevance

Identification of a gene expression profile predicting loco-regional failure after combined chemo-radiotherapy will have immediate practical benefits and more long-term benefits. The immediate benefit is that it will aid clinicians in treatment choice, particularly chosing between surgery and chemoradiation in advanced tumors. Long term, the expression profile will indicate genes associated with treatment resistance in human cancer, providing leads for future research aimed at a better understanding of radio-and chemo-resistance, and pointing to alternative intervention strategies such as EGFR inhibition.