< Back to previous page

Project

The role of functional imaging and molecular profiling in image guided highly conformal radiotherapy for head and neck cancer: towards a patient tailored treatment.

Treatment of head and neck squamous cell carcinoma (HNSCC) is highly complex. Not only because of the variety of disease subsites, butalso because of the complicated anatomical relationship between the tumor and the normal structures. The importance of organ preservation normal">has shifted treatment of locally advanced HNSCC towards (chemo-) radiotherapy (CRT). The current treatment paradigm however is far from optimal. The implementation of altered fractionation schedules and concurrent chemotherapy
significantly improved locoregional control (LRC) and overall survival (OS) rates, but this also increased both acute and latetoxicity, with long-term xerostomia and dysphagia severely implicating the patients quality of life (QOL). The use of highly conformal radiotherapy techniques such as intensity modulated radiotherapy (IMRT) allows us to sculpt the dose more conformally around the target volume, thus reducing the dose to the surrounding organs at risk (OAR), while maintaining the dose to the target volume (TV). This high conformality however harbours an increased risk for marginal misses and requires adequate compensation for set-up uncertainties, appropriate selection and accurate delineation of the TV, proper dose prescription and extensive quality control. Therefore, an accurate evaluation of the clinical benefit of new techniques is very important. Aside from these technological innovations, there is also the clinical observation that there is a heterogeneity in the response of HNSCC to treatment. This  heterogeneity is rooted invariable biological, molecular and genetic characteristics of the tumor. Functional imaging techniques potentially allow us to visualize and quantify this heterogeneity. Diffusion weighted magnetic resonance imaging(DWI) is a functional MRI technique which provides us with a deeper insight into the tumors microstructure by depicting molecular diffusion, i.e. the Brownian motion of water protons in biologic tissues. This diffusion can be quantified as an apparent diffusion coefficient (ADC) and ithas shown great promise as a potential biomarker for treatment responsebefore and early during treatment in a variety of tumor sites. This knowledge can be very helpful in patient selection and treatment adaptation. However to date the exact value of DWI and its relation to other prognostic factors remains unclear.  

The main focus of thisresearch project was to explore new waysto improve the therapeutic index in HNSCC patients treated with (chemo-)radiotherapy. On the one hand we aimed to decrease toxicity through the optimal use of highly conformalIMRT. On the other hand we wanted to investigate the potential of DWI in individualizing treatment in HNSCC patients. 
OPTIMIZING HIGHLY CONFORMAL RADIOTHERAPY
IMRT has rapidly become the standard of care in the management of locally advanced HNSCC. To fully assess its current value we retrospectively investigated the effect of introducing IMRT on outcome and treatment related toxicity compared to parotid sparing 3D conformal radiotherapy (3DCRT). 
For this analysisa total of 245 patients with stage III and IV HNSCC treated with primary (chemo-)radiotherapy between January 2003 and December 2010, were included. After a median follow-up of 35 months in  the IMRT group and 68 months in the 3DCRT group no significant differences were found in 3 year LRC and OS rates between both groups. There was significantly less late xerostomia ≥ grade 2 in the IMRT group compared to the 3DCRT group (23% vs 68%, p<0.001). While our analysis confirmed that IMRTeffectively reduces the dose to the parotid glands, thus reducing late xerostomia without compromisingtumor control it failed to demonstrate a significant benefit on the
development of late dysphagia. Thus late dysphagia is rapidly becoming the prime dose-limiting toxicity in HNSCC patients. Several studies have identified that the dose delivered to the swallowing muscles is of paramount importance in the development of latedysphagia. Multivariate analysis has identified that the extent of lymph node irradiation is an important predictive factor for swallowing dysfunction after treatment. And while the necessity of elective lymph node irradiation in the eradication of subclinical disease has long been established in HNSCC, the dose suggested to achieve microscopic sterilization,
i.e. 50 Gy in 2 Gy fractions is largely based on empirical, nonrandomized data.We hypothesized that with more accurate tumor staging, moreprecise treatment planning and delivery and the improved outcome using both concurrent chemotherapy altered fractionation schedules lower doses  will sufficient to obtain microscopic sterilization. Especially considering our analysis of 368 patients with advanced HNSCC treatedwith primary (chemo-)radiotherapy between May 1987 and March 2008. In this analysis, only 2 patients (0.54%) developed an isolated recurrence in the electively treated nodal levels. Similar results were seen after treatment of HNSCC patients with IMRT, with the majority of locoregional recurrences arising within the originally defined tumor volume. Considering these observations, the technical possibilities provided by IMRT and the contribution of elective lymph  node irradiation to the dose delivered to the entire swallowingapparatus, a dose reduction might be effective in reducing late toxicity, and more specifically late swallowing disorders. This, of course, should not come at the cost of treatment outcome. Therefore a multicentre, randomized clinical trial was initiated with the aim to investigate whether a dose-reduction to the elective nodal sites and off-target regions of the swallowing apparatus would result in a reduction of both acute and late side effects without compromising tumor control. In this trial, 200 patients were randomly assigned to the standard or experimental arm. In the standard arm the elective nodal volumes (PTVelect)  were irradiated up to an equivalent dose of 50 Gy in 2 Gy fractions respectively. In the experimental arman equivalent dose of 40 Gy in 2 Gy fractions was prescribed to the PTVelect and the dose to the swallowing apparatus was kept as low as reasonably possible without compromising coverage of the therapeutic PTV (PTVther). Using IMRT we were able to significantly reduce the dose to the tothe PTVelect without compromising the dose to the PTVther. This resulted in a significant reduction of the dose to swallowing structures. During treatment this did not result in a significant difference in dysphagia. However, 3 months after radiotherapy there was already significantly less grade 3+ dysphagia in the experimental arm compared to the standard arm (2% vs 11%;  p=0.03). So far however, this did not translate into a reduction of PEG tube dependency 6 months after treatment.
Median follow-up was 6 months with excellent locoregional control numbers in both arms. So far no patients had a recorded recurrence within the de-escalated elective nodal volume. While these results are promising, the follow-up is currently quite short and more time is needed to assess the full value of the
dose escalation on both treatment outcome and treatment related toxicity. 
EXPLORING THE POTENTIAL OF DWI
In exploring the potential of DWI, we firstly wanted to assess the prognostic value of pretreatment apparent diffusion coefficient (ADC) and incorporate this into a multivariable prognostic model. For this study 175 patients withpathology proven HNSCC between August 2004 and May 2010 were prospectively included. In multivariable analysis ADChigh was a clear independent predictor of disease recurrence. The validated 3-year DFS rates forthe groups with worst, intermediate, and best prognosis were 44%, 67% and 69% respectively. ADChigh predicted LRC more strongly, while tumor and nodal volume seemed to be more prognostic of DM. This study demonstrated that a pretreatment ADC value derived from high b-values is an independent prognostic factor for HNSCC and can be incorporated into a multivariable prognostic model. However, the performance of the model is still suboptimal. And further optimization of the model is needed to increase its performance. This might include biological, molecular and genetic biomarkers. One of these potential biomarkers is HPV associated p16 expression, which has been identified as a promising prognostic factor in oropharyngeal SCC. To investigate the relationship between p16 and pretreatment ADC values we performed a subgroup analysis in patients with oropharyngeal SCC.  A total of 45 patients with oropharyngeal SCC were included. On all biopsy samples, p16 immunohistochemistry (IHC) was performed and prospectively scored. This analysis suggested that aside from p16 status, pretreatment ADC value might be a valuable prognostic factor in oropharyngeal SCC. Especially in the subgroup of p16 negative patients pretreatment ADC appeared to identify a subgroup of patients with abetter LRC. The ability to further stratify patients within the group of p16 negative patients might be very useful as these patients currentlylack favourable prognostic markers.  Secondly, the ability ofmolecular and functional imaging to depict and quantify spatial
information on cellular or microenvironmental phenotypes associated with radioresistance makes them an attractive tool in the conception of novel dose prescription strategies. DWI-parameters can be both quantified and visualized as an ADC map in which the calculated voxel intensity is related to the amount
of water proton diffusion in the voxel. Any form of necrosis or tumor lysis induced by radiation will lead to an increase in extracellular space water diffusion, with a corresponding increase in the tumors ADC value. This loss of cellularity generally precedes the macroscopic changes in tumor size making it
a potentially powerful tool for early treatment response assessment. A voxelwise evaluation of the ΔADC
based on registered pretreatment and early treatment ADC maps could visualize regions of relative radioresistance early during treatment. Currently treatment-induced anatomical changes and modality-specific distortions still inhibit the straightforward comparison of these ADC maps and its correlation to
an anatomical reference image. To counter these problems we developed a semi-automatic registration tool which could compensate for these distortions and allow us to visualize andlocalize regional  differences in ΔADC during treatment.For this analysis, twenty patients with stage IV HNSCC were included inthis study. A MRI including DWI was acquired prior to and after 2 weeksinto treatment. The primary tumor was delineated on the pertreatment T1-weighted image and markers were manually placed at anatomical landmarkson the different modalities at both time points. A registration method,consisting of a fully automatic rigid and nonrigid registration and twosemi-automatic thin-plate spline (TPS) warps was applied to the image sets. After each registration step the mean registration errors were calculated and ΔADC was compared between good and responders. Overall 1838 markers were placed on 99 slices in 20 patients. The TPS warps significantly reduced registration error (in mm, 6.3 ± 6.2 vs 3.2 ± 3.3  mm, p<0.001). Linear regression analysis further showed a significant correlation between the registration error and the number of placed markers per slice (t-value: 6; p<0.001). Prior to registration no significant difference in ΔADC was seen between good and poor responders. However, after the marker based registration the median ΔADC in the poor responders was significantly lower than in the good responders (7% vs. 21%;  p<0.001). Within each delineated tumor volume the ΔADC could be visualized on a voxel-wise basis.
In conclusion, we found that using this registration method we were able to significantly reduce the mean registration error. Furthermore the voxel-wise calculation of the ΔADC early during radiotherapy allows us to visualize regions of low ΔADC within the tumor. This information could be useful in an adaptive dose-escalation approach.

CONCLUSION

The use of IMRT in the treatment of HNSCC, enables us to generate steep dose gradients between the target volumes and the critical OAR, thus potentially decreasing toxicity and enabling the delivery of a higher radiation dose to the tumor (i.e. dose-escalation). However, with this increased conformality there
is also an increased risk of a marginal miss. Our analysis confirmed that IMRT does not compromise long-term tumor control, and effectively reduced the incidence and severity of late complications,  especially xerostomia.  In a randomized, multicenter setting we found that we can use IMRT to effectively reduce the dose to the elective lymph nodes without compromising target coverage and this way  effectivelyreduce the dose to the swallowing apparatus. This already resulted in asignificant decrease
of  severe dysphagia 3 months after the end of treatment, while excellent LRC and  DFS rates were seen. While these results are very promising, further follow-up is essential to accurately  assess the
definitive effect of such a dose de-escalation on both late toxicity and treatment outcome. 

As for patient selection, we demonstrated that pretreatment ADC derived from high-b-values is a significant prognostic factor in HNSCC and can be incorporated into a multivariable prognostic model. Furthermore, it appearsthat pretreatment ADC remains valid even in the prognostically unfavorable
group of p16 negative oropharyngeal HNSCC. However, these data needs to be confirmed and validated in a prospective setting. Also the integration of other imaging parameters, biological, molecular and genetic factors is needed to further increase the performance of the model. Finally the developed semi-automatic registration method allowed us to capture the heterogeneity of the tumors response early during treatment. Together with the possibilities of IMRT this knowledge gives us both the rationale and the means to further optimize our radiation treatment. </></></></></></></></></></></></></></></></></></></></></></></>
Date:1 Aug 2009 →  17 Dec 2013
Keywords:Head and neck cancer, Diffusion Weighted MRI, IGRT, Radiotherapy, Functional imaging
Disciplines:Medical imaging and therapy, Other paramedical sciences, Otorhinolaryngology, Speech, language and hearing sciences, Morphological sciences, Oncology
Project type:PhD project