A joint physics and radiobiology DREAM team vision - Towards better response prediction models to advance radiotherapy.

Radiotherapy developed empirically through experience balancing tumour control and normal tissue toxicities. Early simple mathematical models formalized this practical knowledge and enabled effective cancer treatment to date. Remarkable advances in technology, computing, and experimental biology now create opportunities to incorporate this knowledge into enhanced computational models. The ESTRO DREAM (Dose Response, Experiment, Analysis, Modelling) workshop brought together experts across disciplines to pursue the vision of personalized radiotherapy for optimal outcomes through advanced modelling. The ultimate vision is leveraging quantitative models dynamically during therapy to ultimately achieve truly adaptive and biologically guided radiotherapy at the population as well as individual patient-based levels. This requires the generation of models that inform response-based adaptations, individually optimized delivery and enable biological monitoring to provide decision support to clinicians. The goal is expanding to models that can drive the realization of personalized therapy for optimal outcomes. This position paper provides their propositions that describe how innovations in biology, physics, mathematics, and data science including AI could inform models and improve predictions. It consolidates the DREAM team's consensus on scientific priorities and organizational requirements. Scientifically, it stresses the need for rigorous, multifaceted model development, comprehensive validation and clinical applicability and significance. Organizationally, it reinforces the prerequisites of interdisciplinary research and collaboration between physicians, medical physicists, radiobiologists, and computational scientists throughout model development. Solely by a shared understanding of clinical needs, biological mechanisms, and computational methods, more informed models can be created. Future research environment and support must facilitate this integrative method of operation across multiple disciplines.

Current ideas to reduce or salvage radiation damage to salivary glands.

Radiation-induced hyposalivation is still a major problem after radiotherapy for head and neck cancer. Current and promising new thoughts to reduce or salvage radiation damage to salivary gland tissue are explored. The main cause underlying radiation-induced hyposalivation is a lack of functional saliva-producing acinar cells resulting from radiation-induced stem cell sterilization. Current methods to prevent that damage are radiation techniques to reduce radiation-injury to salivary gland tissue, surgical techniques to relocate salivary glands to a region receiving a lower cumulative radiation dose, and techniques to make salivary gland cells more resistant to radiation injury. These preventive techniques cannot be applied in all cases, also reduce tumor sensitivity, or do not result in a sufficient amelioration of the dryness-related complaints. Therefore, alternative methods on techniques to salvage salivary glands that are damaged by radiation are explored with promising results, such as stem cell therapies and gene transfer techniques to allow the radiation-injured salivary gland tissue to secrete water.

Lung irradiation induces pulmonary vascular remodelling resembling pulmonary arterial hypertension.

BACKGROUND: Pulmonary arterial hypertension (PAH) is a commonly fatal pulmonary vascular disease that is often diagnosed late and is characterised by a progressive rise in pulmonary vascular resistance resulting from typical vascular remodelling. Recent data suggest that vascular damage plays an important role in the development of radiation-induced pulmonary toxicity. Therefore, the authors investigated whether irradiation of the lung also induces pulmonary hypertension. METHODS: Different sub-volumes of the rat lung were irradiated with protons known to induce different levels of pulmonary vascular damage. RESULTS: Early loss of endothelial cells and vascular oedema were observed in the irradiation field and in shielded parts of the lung, even before the onset of clinical symptoms. 8 weeks after irradiation, irradiated volume-dependent vascular remodelling was observed, correlating perfectly with pulmonary artery pressure, right ventricle hypertrophy and pulmonary dysfunction. CONCLUSIONS: The findings indicate that partial lung irradiation induces pulmonary vascular remodelling resulting from acute pulmonary endothelial cell loss and consequential pulmonary hypertension. Moreover, the close resemblance of the observed vascular remodelling with vascular lesions in PAH makes partial lung irradiation a promising new model for studying PAH.

Recurrent scoliosis one year after surgical correction.

A year after anterolateral spondylodesis for progressive scoliosis, the patient showed a flexion gait pattern with recurrent deformity, due to late infection. Surgical debridement resolved all symptoms. Whereas most postoperative infections occur after posterior spondylodesis and present with back pain and mild increase of infection parameters, late infection after anterolateral approach is rare. In this case the patient did not present with the classic symptoms.

Estimation of parameters of dose-volume models and their confidence limits.

Predictions of the normal-tissue complication probability (NTCP) for the ranking of treatment plans are based on fits of dose-volume models to clinical and/or experimental data. In the literature several different fit methods are used. In this work frequently used methods and techniques to fit NTCP models to dose response data for establishing dose-volume effects, are discussed. The techniques are tested for their usability with dose-volume data and NTCP models. Different methods to estimate the confidence intervals of the model parameters are part of this study. From a critical-volume (CV) model with biologically realistic parameters a primary dataset was generated, serving as the reference for this study and describable by the NTCP model. The CV model was fitted to this dataset. From the resulting parameters and the CV model, 1000 secondary datasets were generated by Monte Carlo simulation. All secondary datasets were fitted to obtain 1000 parameter sets of the CV model. Thus the 'real' spread in fit results due to statistical spreading in the data is obtained and has been compared with estimates of the confidence intervals obtained by different methods applied to the primary dataset. The confidence limits of the parameters of one dataset were estimated using the methods, employing the covariance matrix, the jackknife method and directly from the likelihood landscape. These results were compared with the spread of the parameters, obtained from the secondary parameter sets. For the estimation of confidence intervals on NTCP predictions, three methods were tested. Firstly, propagation of errors using the covariance matrix was used. Secondly, the meaning of the width of a bundle of curves that resulted from parameters that were within the one standard deviation region in the likelihood space was investigated. Thirdly, many parameter sets and their likelihood were used to create a likelihood-weighted probability distribution of the NTCP. It is concluded that for the type of dose response data used here, only a full likelihood analysis will produce reliable results. The often-used approximations, such as the usage of the covariance matrix, produce inconsistent confidence limits on both the parameter sets and the resulting NTCP values.

Techniques for precision irradiation of the lateral half of the rat cervical spinal cord using 150 MeV protons [corrected].

Techniques for high precision irradiation experiments with protons, to investigate the volume dependence of the tolerance dose of the rat cervical spinal cord are described. In the present study, 50% of the lateral cross section of the spinal cord was irradiated. The diameter of the cross section of this part of the rat spinal cord is at maximum 3.5 mm. Therefore, a dedicated procedure was developed to comply with the needs for a very high positioning accuracy and high spatial resolution dosimetry. By using 150 MeV protons a steep dose gradient (20-80% = 1 mm) in the centre of the spinal cord was achieved. This yields a good dose contrast between the left and right halves of the cord. A home-made digital x-ray imager with a pixel resolution of 0.18 mm/pixel was used for position verification of the spinal cord. A positioning accuracy of 0.09 mm was obtained by using information of multiple pixels. The average position stability during the irradiation was found to be 0.08 mm (1 SD) without significant systematic deviations. Profiles of the dose distribution were measured with a 2D dosimetry system consisting of a scintillating screen and a CCD camera. Dose volume histograms of the whole spinal cord as well as separately of the white and grey matters were calculated using MRI imaging of the cross section of the rat cervical spinal cord. From the irradiation of 20 animals a dose-response curve has been established. MRI showed radiation-induced damage at the high dose side of the spinal cord. Analysis of the preliminary dose-response data shows a significant dose-volume effect. With the described procedure and equipment it is possible to perform high precision irradiations on selected parts of the spinal cord.

Detector line spread functions determined analytically by transport of Compton recoil electrons.

To achieve the maximum benefit of conformal radiation therapy it is necessary to obtain accurate knowledge of radiation beam penumbras based on high-resolution relative dosimetry of beam profiles. For this purpose there is a need to perform high-resolution dosimetry with well-established routine dosimeters, such as ionization chambers or diodes. Profiles measured with these detectors must be corrected for the dosimeter's nonideal response, caused by finite dimensions and, in the case of an ionization chamber, the alteration of electron transport and a contribution of electrons recoiled in the chamber wall and the central electrode. For this purpose the line spread function (LSF) of the detector is needed. The experimental determination of LSFs is cumbersome and restricted to the specific detector and beam energy spectrum used. Therefore, a previously reported analytical model [Med. Phys. 27, 923-934 (2000)] has been extended to determine response profiles of routine dosimeters: shielded diodes and, in particular, ionization chambers, in primary dose slit beams. The model combines Compton scattering of incident photons, the transport of recoiled electrons by Fermi-Eyges small-angle multiple scattering theory, and functions to limit electron transport. It yields the traveling direction and the energy of electrons upon incidence on the detector surface. In the case of ionization chambers, geometrical considerations are then sufficient to calculate the relative amount of ionization in chamber air, i.e., the detector response, as a function of the detector location in the slit beam. In combination with the previously reported slit beam dose profiles, the LSF can then readily be derived by reconstruction techniques. Since the spectral contributions are preserved, the LSF of a dosimeter is defined for any beam for which the effective spectrum is known. The detector response profiles calculated in this study have been verified in a telescopic slit beam geometry, and were found to correspond to experimental profiles within 0.2 and 0.3 mm (full width at half-maximum) for a Wellhoefer IC15 chamber in a 6 and 25 MV-X x-ray beam, respectively. For a shielded diode these figures were found to be 0.2 and 0.1 mm, respectively. It is shown that a shielded diode in a primary beam needs only a small size-based correction of measured profiles. The effect of the LSF of an IC15 chamber on penumbra width has been determined for a set of model penumbras. The LSFs calculated by the application of the analytical model yield a broadening by 2 mm of a 3 mm wide penumbra (20%-80%). This is 0.5 mm (6 MV-X) to 1 mm (25 MV-X) smaller than found with the experimental LSFs. With a spatial correction based on the LSFs that were determined in this study, this broadening of up to 2 mm is eliminated, so that ionization chambers like the IC15 can be used for high-resolution relative dosimetry on a routine basis.

Collimator scatter and 2D dosimetry in small proton beams.

Monte Carlo simulations have been performed to determine the influence of collimator-scattered protons from a 150 MeV proton beam on the dose distribution behind a collimator. Slit-shaped collimators with apertures between 2 and 20 mm have been simulated. The Monte Carlo code GEANT 3.21 has been validated against one-dimensional dose measurements with a scintillating screen, observed by a CCD camera. In order to account for the effects of the spatial response of the CCD/scintillator system, the line-spread function was determined by comparison with measurements made with a diamond detector. The line-spread function of the CCD/scintillator system is described by a Gaussian distribution with a standard deviation of 0.22 mm. The Monte Carlo simulations show that protons that hit the collimator on the entrance face and leave it through the wall of the aperture make the largest scatter contribution. Scatter on air is the major contribution to the extent of the penumbra. From the energy spectra it is derived that protons with a relative biological effectiveness greater than 1 cause at most 1% more damage in tissue than what would be expected from the physical dose.

Performance of a fluorescent screen and CCD camera as a two-dimensional dosimetry system for dynamic treatment techniques.

A two-dimensionally position sensitive dosimetry system has been tested for different dosimetric applications in a radiation therapy facility with a scanning proton beam. The system consists of a scintillating (fluorescent) screen, mounted at the beam-exit side of a phantom and it is observed by a charge coupled device (CCD) camera. The observed light distribution at the screen is equivalent to the two-dimensional (2D)-dose distribution at the screen position. It has been found that the dosimetric properties of the system, measured in a scanning proton beam, are equal to those measured in a proton beam broadened by a scattering system. Measurements of the transversal dose distribution of a single pencil beam are consistent with dose measurements as well as with dose calculations in clinically relevant fields made with multiple pencil beams. Measurements of inhomogeneous dose distributions have shown to be of sufficient accuracy to be suitable for the verification of dose calculation algorithms. The good sensitivity and sub-mm spatial resolution of the system allows for the detection of deviations of a few percent in dose from the expected (intended or calculated) dose distribution. Its dosimetric properties and the immediate availability of the data make this device a useful tool in the quality control of scanning proton beams.

Slit x-ray beam primary dose profiles determined by analytical transport of Compton recoil electrons.

Accurate measurement of radiation beam penumbras is essential for conformal radiotherapy. For this purpose a detailed knowledge of the dosimeter's spatial response is required. However, experimental determination of detector spatial response is cumbersome and restricted to the specific detector type and beam spectrum used. A model has therefore been developed to calculate in slit beam geometry both dose profiles and detector response profiles. Summations over representative photon beam spectra yield profiles for polyenergetic beams. In the present study the model is described and resulting dose profiles verified. The model combines Compton scattering of incident photons, transport of resulting electrons by Fermi-Eyges small-angle multiple scattering theory, and functions to limit electron transport. This analytic model thus yields line spread kernels of primary dose in a water phantom. It is shown that the spatial response of an ideal point detector to a primary photon beam can be well described by the model; the calculations are verified by measurements with a diamond detector in a telescopic slit geometry in which all dose contributions except for the primary dose can be excluded. Effects of photon detector behavior, source size of the linear accelerator (linac) and detector size are studied. Measurements show that slit dose profiles calculated by means of the kernel are accurate within 0.1 mm of the full-width at half-maximum. For a theoretical point source and point detector combined with a 0.2 mm wide slit, the full-width half-maximum values of the slit beam dose profiles are calculated as 0.37 mm and 0.42 mm in a 6 MV and 25 MV x-ray beam, respectively. The present study shows that the model is adequate to calculate local dose effects that are dominated by approximately mono-directional, primary photon fluence. The analytic model further provides directional electron fluence information and is designed to be applied to various detectors and linac beam spectra.

Fast 2D phantom dosimetry for scanning proton beams.

A quality control system especially designed for dosimetry in scanning proton beams has been designed and tested. The system consists of a scintillating screen (Gd2O2S:Tb), mounted at the beam-exit side of a phantom, and observed by a low noise CCD camera with a long integration time. The purpose of the instrument is to make a fast and accurate two-dimensional image of the dose distribution at the screen position in the phantom. The linearity of the signal with the dose, the noise in the signal, the influence of the ionization density on the signal, and the influence of the field size on the signal have been investigated. The spatial resolution is 1.3 mm (1 s.d.), which is sufficiently smaller than typical penumbras in dose distributions. The measured yield depends linearly on the dose and agrees within 5% with the calculations. In the images a signal to noise ration (signal/1 s.d.) of 10(2) has been found, which is in the same order of magnitude as expected from the calculations. At locations in the dose distribution possessing a strong contribution of high ionization densities (i.e., in the Bragg peak), we found some quenching of the light output, which can be described well by existing models if the beam characteristics are known. For clinically used beam characteristics such as a Spread Out Bragg peak, there is at most 8% deviation from the NACP ionization chamber measurements. The conclusion is that this instrument is a useful tool for quick and reliable quality control of proton beams. The long integration-time capabilities of the system make it worthwhile to investigate its applicability in scanning proton beams and other dynamic treatment modalities.