Prediction of Outcome After Chemoradiotherapy for Head and Neck Cancer Using Functional Imaging and Tumor Biology

Overview

Despite uniform histopathological definition the response of locally advanced squamous cell carcinomas of the head and neck (HNSCC) to ionizing radiation differs greatly with locoregional recurrences burdening this patient population. The addition of concurrent chemotherapy and the use of altered fractionation schedules has significantly increased locoregional control and overall survival over the last decade however, this has come at the cost of increased acute and late toxicity, preventing further treatment intensification in all patients. If the investigators want to increase the therapeutic index of HNSCC, we need to be able to tailor the treatment more individually to each patient. The project aims at developing a prognostic model for head and neck cancer patients based on the combination of known clinical parameters with 1) genetic characteristics of the tumor and 2) parameters derived from diffusion weighted and dynamic contrast enhanced magnetic resonance imaging (MRI) obtained before and during treatment. The investigators plan a prospective trial where 120 patients with locally advanced head and neck cancer treated with chemoradiotherapy will be included. Prior to treatment biopsy material will be collected for genetic analysis and before and during treatment functional MRI with diffusion weighted and dynamic contrast enhanced imaging will be performed. All patients will be followed up multidisciplinary afterwards with follow-up of tumor status and toxicity.

1. BACKGROUND AND SETTING 1.1. Introduction Concurrent (chemo-) radiotherapy (CRT) is the current standard of care for patients with locally advanced head and neck squamous cell carcinoma (HNSCC). The proximity of important functional structures with the tumour makes treatment however highly complex. While locoregional control rates have improved over the last decade, treatment related toxicity can be severe with xerostomia and dysphagia gravely complicating the patient's quality of life. Furthermore, it becomes increasingly clear that while these tumours can be identical in location and basic histology, their response to treatment differs greatly. This implies that for a subgroup of patients, equal locoregional control rates could be achieved using a less intense and consequently less toxic treatment schedule. This in contrast to the group of patients who develop a locoregional recurrence during follow up, for whom current treatment is apparently insufficient. These patients might benefit from a more intense treatment schedule, i.e. a higher dose of radiation. These differences in sensitivity to treatment can be explained by variations in biological, molecular and genetic factors. Clinical parameters alone are insufficient for response prediction. Identifying the different molecular and genetic factors, would help us increase the accuracy of response prediction and based on these factors tailor the treatment more individually to each patient. Therefore we want to develop a prognostic and predictive model incorporating well-defined molecular and imaging parameters which show great promise in response prediction after ionizing radiation. 1.2. Genetic and molecular tumour characteristics The first parameter we want to investigate further is tumour hypoxia. Hypoxia has been identified as a factor that increases tumour aggressiveness and decreases radiosensitivity. In the past, several techniques have been applied to detect biologically relevant tumour hypoxia, but none of them are used today in routine clinical practice. Recently, a polymerase chain reaction (PCR) -based hypoxia classifier gene signature was published by Toustrup et al. This classifier allows us to study biologically relevant tumour hypoxia and consequently tumour aggressiveness and resistance to ionizing irradiation. This analysis will be performed on biopsies obtained prior to treatment. Part of this biopsy material will also be stored into our biobank. This will facilitate future research on promising molecular and genetics parameters (such as HPV correlated overexpression of p16 for example) on a well-defined and structured patient database. 1.3. Imaging parameters Aside from these important molecular and genetic tumour characteristics, several functional imaging parameters will also be included in our model. In contrast to anatomical imaging, functional imaging modalities provide us with a deeper insight in the tumour's underlying biological activity and microstructure. Diffusion weighted magnetic resonance imaging (DWI) can characterize tissues based on the random displacement of water protons. This displacement can be quantified using the apparent diffusion coefficient (ADC). Preliminary studies in HNSCC have demonstrated the prognostic and predictive potential of repetitive DWI early during and after treatment. Dynamic contrast enhanced magnetic resonance imaging(DCE-MRI) is a second technique, which shows great promise as an early indicator of response to ionizing radiation. Many malignant tumors manifest neovascularity or angiogenesis. These processes are however flawed and as a result these newly synthesized vessels manifest a high permeability, tortuosity and generally a poor functionality. This might result in poor oxygen supply to the tumour cells, which may compromise the effectiveness of radiation treatment of the tumour. Therefore the vascular properties of a tumour, assessed with DCE-MRI, could prove a prognostic indicator of its aggressiveness. 2. STUDY OBJECTIVES 2.1. Primary objectives The main objective of this study is to correlate clinical, molecular and radiological predictors with outcome, as defined by locoregional control and disease free survival. In this way we will develop a prognostic/predictive model. The predictive model will be instrumental in the individualization of treatment ensuring optimized treatment and avoiding under- and overtreatment. 2.2. Secondary objectives * We want to get new insights in the tumor biology and microstructure by correlating imaging and molecular/genetic markers in a well-defined patient population. * We want to validate the different imaging techniques. * We want to make a correlation between hypoxia and other predictive biomarkers in the future, by storing the tissue obtained in this study in our biobank. 3. STUDY DESIGN We propose to set up a study where we will prospectively include patients with locally advanced head and neck cancer who will be treated with concurrent chemoradiotherapy as decided after multidisciplinary consultation. An outline of the trial is presented in the figure below. All patients recognized eligible (non-metastatic locally advanced head and neck squamous cell carcinoma, treated with chemoradiotherapy, karnofsky performance status ≥ 70% and ≥ 18 years old) will be included after written informed consent. Staging with a laryngoscopy, CT of the neck, MRI of the neck and a PET-CT will be routinely performed. The CRT treatment consists of radiotherapy up to 72 Gy using an accelerated hyperfractionated schedule. Day 1 and day 22 of the treatment Cisplatin at 100mg/m² will be administered. Tumour biopsies will be obtained a few days prior to treatment. The hypoxia gene expression profile will be derived from the tumour material (15 genes on PCR that can be analyzed individually or as a group through one binary variable). The tissue used for RNA and DNA extraction needs to be flanked by H\&E staining confirming tumor presence. From the biopsy the two ends will be cut off and fixed in paraffin. This tissue will be stored in the biobank. The middle part of the tissue will be stored into RNA-later, and will be used in part to extract DNA and RNA. RNA will be used to synthesize cDNA to perform the qPCR analysis for the hypoxic classifier. Patients will also undergo a DWI MRI and DCE imaging before and 3 weeks into CRT. Lesions will be quantitatively assessed by manual delineation of regions of interest (ROI) over the tumour on the native DWI and DCE-MRI images. ADC values and (semi-) quantitative DCE parameters will be calculated respectively. All the above described data will be integrated into the prognostic model together with the available clinical data. 4. DEFINITION OF ENDPOINTS The main endpoint of this study is to validate the above described prognostic model. Using this prognostic model, we can predict response to treatment. This will help us to tailor the treatment more individually to each patient.