Rethink radiotherapy – BIGART 2017

Introduction

The present issue of Acta Oncologica contains studies presented at the 15th Acta Oncologica Symposium, held in Aarhus, Denmark on June 13–16, 2017. The symposium was dedicated to aspects of Biology-Guided Adaptive Radiotherapy (BiGART) and attracted more than 200 physicians, physicists, radiobiologists and other scientists with an active interest in this specific area.

Acta Oncologica has sponsored scientific symposia since the late 80s. The aim of these meetings has been to focus on oncological issues of emerging interest, preferably with a multidisciplinary and multi-professional approach. The current series of Acta Oncologica meetings with focus on radiotherapy (RT) have encompassed stereotactic body RT (SBRT) [1], image-guided RT [2], particle RT [3] and BiGART [4–6], all reflecting development in RT during the last decennium within the fields of treatment planning [7–12], (biological) imaging [13–33], adoptive RT, [34–44] particle therapy [45–50], clinical outcome [51–63], quality assurance [64,65], and other important factors [66–72].

The aim of BiGART is treatment adaptation in time and space based on biological and anatomical features, maximizing the therapeutic ratio for each individual patient. Key topics for the BiGART2017 conference included:

• Biology of tumors and normal tissue to guide patient selection, target volumes and dose prescription in photon and particle therapy.

• Functional imaging of tumors and normal tissues with functional imaging techniques based on MRI and PET, and the use of such images for dose painting and normal tissue avoidance in photon and particle therapy.

• Treatment planning and delivery challenges in adaptation of photon and particle therapy based on changes in tumor and normal tissue biology, anatomy and/or function.

• Clinical outcome of biology-guided and adaptive RT and particle therapy.

Particle therapy and ENLIGHT

The meeting was held back-to-back with the annual meeting of ENLIGHT, The European Network for LIGht ion Hadron Therapy. Thus, there was more emphasis on particle therapy-related issues than in previous BiGART meetings. The meetings provided a good opportunity for interaction between scientists and clinicians within the rapidly emerging field of ion beam therapy. Particle therapy needs to be well aligned with other radiation techniques as well as with general developments in cancer research and patient care. The combined strengths of European initiatives including BiGART, ENLIGHT, ESTRO and the European Particle Therapy Network, is required to promote clinical and research collaboration between the rapidly increasing number of European particle therapy centers and to ensure that particle therapy becomes integrated into overall radiation oncology community.

With the establishment of more and more particle (proton) centers [45,73–79], it has increasingly become apparent that the classical RBE estimate of 1.1 for protons can be a dubious number [80–83]. The original data and the consequent compilation of these into a 1.1 global value have become a dogma, which deserves to be scrutinized. Not only is there a wide spectrum of RBE values, but as the presentations at the BIGART 2017 indicate, this variation is also related to different early and late morbidity endpoints and not least to the fractionation pattern [81]. The use of a RBE of 1.1 is far too simple and there is a crucial need to derive more exact values, which can be used in future dose planning.

The BIGART 2017 symposium clearly identified this problem and pointed to the need for an international collaboration with the aim to clarify in detail, variations in RBE according to tissue types and dose per fraction. Such activity is planned within the framework of the European Particle Network, and should urgently be implemented. This demands a multicenter collaboration with ample access to animal experiments and experimental particle beams – and this may prove to be more difficult than optimistically expected.

Not only is the RBE under challenge [81,84] but so is the homogeneous expression of this factor throughout the entire spread out Bragg peak. The reasons for this being the increase in LET at the back of the peak, which is coherent with a decrease of the proton dose [85–87]. The balance between a higher RBE value (due to increased LET) counter acted by the decrease in proton dose makes the real biological dose at the backside uncertain [87]. There are both clinical and experimental evidence for more pronounced late effects than anticipated [73,84,87], and the particle therapy world must clearly acknowledge this risk and counter act it until the problem is both qualitatively and quantitatively identified and included in the biological dose algorithms for treatment planning.

On top of this, comes the issue of whether proton irradiation, in fact, have a different quantitative cellular effect [88], together with an increasing awareness of the importance of radiogenomics for the probability of developing various early and late morbidities [79,89–91]. This should be explored in detail and taken into consideration, in relation to the individual patient’s risk of morbidity. Since many centers plan to give the treatment according to the model-based approach [92–95], it should be acknowledged that such risk models cannot be fully implemented without taking the aforementioned RBE, LET and genomic variations into consideration.

Despite the promising aspects, particle therapy and especially proton treatment, is still an issue where an understanding and clarification of the underlying radiobiology in especially normal tissues is far from fully described or understood. Before the magnitude of this is integrated into the coming treatment plans, we are probably unable to get the full benefit of these advanced treatment options.

Like in conventional photon-based RT, there is also a continuous development of delivery techniques for treatment with proton and ion beams. Spot-scanning delivery techniques are now mature for widespread clinical implementation, although challenges still remain, in particular, related to its use at tumor sites influenced by intra-fractional motion. At BiGART 2017, a number of interesting scientific and technological developments to further improve and exploit particle therapy delivery were addressed, such as laser-based accelerators [96] as well as so-called grid therapy, with protons [97].

The relatively high building and equipment costs are major challenges for the further adoption of particle therapy. Current facilities are based on cyclotrons or synchrotrons, requiring large building space. However, there is considerable interest in developing more compact facilities. One interesting avenue of research explores the use of laser-accelerated particles/protons [98–101], which is reviewed in this issue by Karsch et al. [96]. This principle is based on the ionizing effects of high-intensity laser pulses, causing ions to be accelerated out of the laser target foil. As they discuss in their review, important steps forward have been taken documenting the potential of this technology, in particular for protons, however, much work still remains to reach clinical maturity, including obtaining a stable beam of sufficient energy.

Grid therapy, or spatially fractionated RT, is currently attracting increasing interest although it dates back more than 100 years, to the work of the German radiologist Köhler [102] who also explored this principle clinically using the low-energy X-rays available at that time. The grid therapy principle implies irradiating the tumor by an array or grid of narrow beam-lets (pencil beams), usually from one beam direction. Grid therapy has been explored initially with X-rays/photons [103–106] as well as with synchrotron radiation [107], and later also with particle therapy beams [108]. It has been shown that tumor control can be achieved despite the rather inhomogeneous target dose distributions and that normal tissue effects can be reduced (exploiting the normal tissue volume effects). Henry et al. performed Monte Carlo simulations of n interlacing cross-firing grids of narrow (1–3 mm) proton beams, and achieved a much more homogenous dose throughout the target volume while keeping the grid therapy patterns in the normal tissues [97]. For tumor sites with favorable locations, this idea seems to have a considerable potential.

In treatment planning for particle therapy, the issue of range uncertainty is a major concern, and methods to decrease this uncertainty are sought. Increasing the accuracy of stopping power determination in tissue can be obtained by the use of dual-energy CT [109], and several presentations at BIGART 2017 addressed this. While range uncertainty remains an important issue, the problem in daily practice may not be so large as it is highly dependent on beam angles. A higher number of beams results in higher robustness concerning range uncertainties in treatment plans.

Given that there is still a significant range uncertainty, this needs to be included in the planning process, and this is most commonly done through robust optimization, which often also includes robustness toward geometrical uncertainties [110–114]. While robust optimization is mostly directed toward range uncertainty and geometrical optimization, the patients’ changing anatomy during a course of treatment is an unsolved problem in particle RT. Better robustness can be obtained through a variety of techniques including primarily simpler field arrangements and robust optimization algorithms. Most noteworthy was the introduction of robustness toward anatomical changes through model prediction of the changing anatomy between treatment fractions. The changing anatomy between fractions is potentially very detrimental to dose coverage and is presently not addressed.

Emerging technologies and treatment principles

The BIGART 2017 contributions on image guidance and motion management mainly focused on particle therapy. Compared to photon RT, scanning beam proton therapy is much more impacted by organ motion and anatomical changes and has a much larger discrepancy between needed and commonly available in-room imaging. While proton pencil beam scanning dose distributions may look excellent in the treatment planning system, there will often be larger differences than for photon RT between the planned and actually delivered doses and thoracic and abdominal intensity-modulated proton therapy (IMPT) has still not matured into routine clinical practice. As pointed out at the conference, this calls for a comprehensive framework comprising the joined forces of robust treatment planning with robustness evaluation, in-room motion monitoring with dose evaluation and possibly adaptive replanning if the conditions for the plan robustness are not fulfilled at treatment.

If treatment evaluation by a full dose reconstruction is not possible, simpler geometrical predictors for the dose degradation with anatomy changes may be used [115,116]. The monitoring should ideally be performed throughout the treatment fraction for observation of intra-fraction baseline drifts [117,118] or – in the case of breath-hold treatments – variations in the breath-hold level [119–121].

Tools to decide on individualized motion management techniques, such as treatment in breath-hold or in free breathing with or without gating, repainting strategy, must be implemented. Robust planning strategies that aim at sufficient target dose coverage for a range of likely patient positions, internal patient anatomies and proton ranges in the patient inherently result in more doses to the normal tissue. The level of proton plan robustness will thus impact the balance between photon and proton treatments in a model-based patient selection approach. A meaningful model-based patient selection, therefore, requires a level of proton plan robustness that ensures comparable confidence in the target dose coverage for both photons and protons in the comprehensive picture that includes motion management, in-room imaging and plan adaptation. Just like the model-based selection will change over time as more clinical data become available for improved morbidity modeling, it will also change as we gain a better understanding of the relationship between the planned doses used for decision making and the actually delivered doses.

While in-room imaging by on-board cone-beam CT has been standard in photon RT for a decade, CBCT has barely entered the proton treatment room. On the other hand, a massive growth to catch up with photon image guidance is expected for proton therapy over the next few years. Furthermore, exciting modality-specific technologies are under development for in vivo range verification in particle therapy by detection of the ultrasound signal or nuclear emissions that are generated in the patient by the ion beam. Another emerging imaging technique presented at BIGART 2017 is proton radiography for image-guidance in small animal experiments [122], and a few contributions explored the RBE uncertainties of proton therapy in the context of 4D or motion-including modeling [123,124].

Models which predict morbidity after RT are currently most often based on risk factors related to classical patient characteristics and to treatment-related parameters. Typical examples of factors which may influence the development of side effects are age, comorbidities, chemotherapy and dose parameters. Identification of dosimetric risk factors can be used to build dose-response models [125]. These models can be used to set clinically relevant priorities during treatment planning [126–129]. Furthermore, identification of genetic factors which can improve the prediction of radiosensitivity is a research field receiving increasing attention. The hope is to further facilitate individualization of RT according to the risk of side effects. Identification of genomic predictors has shown to be a complicated field with contradictory findings [130], many underpowered studies and lack of validation studies and convincing associations have required large amounts of data [89,131,132]. During the BIGART 2017 meeting, a novel approach to identification of genomic predictors was discussed by J. Deasy, New York. The idea is to apply machine learning to make a prediction in samples with a larger number of predictors than events [133]. The methodology includes the building of predictive models which are validated in independent patient samples. Further support of hypotheses may be established through identification of biological processes related to the genomic findings.

Assessment of the burden of side effects after RT is a complicated matter. The burden of symptoms and the related impact on quality of life [134] is associated with both extend of bother as well as development and persistence over time. Actuarial statistics has been recommended for evaluation of radiation induced toxicities [135–137], although this approach is limited by not taking into account the development of a given symptom after its first occurrence. Actuarial statistics is appropriate for irreversible symptoms but is less suited for evaluation of symptoms which fluctuate or heal over time. K. Kirchheiner, Vienna, demonstrated a novel statistical method to identify patients with late, persistent, substantial and treatment-related symptoms (LAPERS). This was done by taking into account baseline status as well as a progression over time as well as the duration of symptoms. The hypothesis is that LAPERS may identify those patients who are experiencing the most pronounced burden of persisting symptoms.

With the introduction of radiomics, use of images in RT is truly undergoing ‘rethinking’. The advanced analysis of images by quantitative feature extraction is still in early days, and validation studies to ensure the stability of features over for instance different scanners and different patient groups are still necessary and ongoing. However, at BIGART 2017, a range of radiomics applications were presented for several clinical sites, and for several imaging modalities (CT, CBCT, MR and PET) used throughout the course of RT. Prediction of recurrent disease was reported based on radiomics analysis for pretreatment CT versus PET images for head and neck cancer patients [138], adding to previous studies on CT radiomics markers for both head and neck and lung cancers [139,140]. Cone-beam CT images were used for the longitudinal investigation to detect changes in features during RT in lung cancer patients, for potential decision support in treatment adaptation [141,142]. Overall, the results of radiomics analysis so far are immature and preliminary, and it remains to be seen whether this search for biological information in images will prove to be successful.

Image-based adaptation to anatomical changes during the course of RT has gradually become a clinical reality for many tumor sites [143–157]. Practical guidelines on how to facilitate the clinical implementation of adaptive RT were presented with examples from rectal and cervix cancer. In locally advanced non-small cell lung cancer (NSCLC), a study showed how adaptive RT can result in dosimetric gain and treatment outcome prediction [158]. Tumor shrinkage adaptive RT allowed significant organ at risk (OAR) dose reduction without compromising initial CTV coverage. The improved outcome was observed for patients with higher tumor shrinkage gradient. A study showed reduced acute and late toxicity rates in bladder cancer. A brain MRI study showed that ultra-early ADC footprint can predict response to RT, and thus facilitate individualization [159]. Finally, a study of adaptation based on off-line CBCT dose reconstruction showed only marginal parotid sparing, when compared to daily CBCT image review by the radiation therapists.

Current indications and treatment principles for RT

The BIGART 2017 Symposium ended with a challenging look at the current indications and treatment principles for RT of some major tumor sites. That these are under continuous development is clearly demonstrated in the role of RT of breast cancer. This is by far the most prominent indication for (adjuvant) RT with more than half a million patients treated in Europe each year, and with a 5-year prevalence of more than 2 million women. Therefore, both the benefit of the treatment and the risk of side effects must be very clear. With an increase in the number of patients diagnosed after the screening, small tumors are becoming more frequent and at the same time, changed population demographics results in an increasing number of elderly women [160,161]. Together with a more intensive use of adjuvant systemic therapy, this has brought us into a situation where the benefit and risk profiles are both relatively small, but still, due to the large incidence and prevalence, the burden of treatment is relatively large.

The indication for RT is primarily to avoid local recurrence. This is successfully achieved, but maybe because the true incidence of recurrence is on a rapid decline [162,163], which for some low-risk patient groups, make the indications for RT at all questionable [163,164]. On the other hand, there is still a prominent enhanced risk in the very young patients [161], which cannot be explained by differences in intrinsic subtypes or presence of BRCA mutations [165], but still is an issue with an unsolved etiology. In more advanced cases, we still see an overall survival benefit of a magnitude of 2% (if patients have their internal mammary nodes includes in the radiation field) [166], but the price for such activity may be an enhanced risk of cardiac morbidity and radiation induced lung cancer [167–169], not least if the patient is a current smoker [164].

The status of breast irradiation is influenced by many factors, which affect both overall outcome and morbidity, and the mutual pros and cons may call for revised visions on the importance of breast RT. In general, it seems the current risk of morbidity is larger than anticipated and the overall benefits smaller (or overlapping with that produced with systemic therapy). Therefore, a more individual consideration of the risks and benefits must be applied to secure an optimal individual balance [79,89,164,165,170–174], and we must probably initiate trials to identify the indications for breast RT. In fact, such trials are already emerging.

Since this is happening on an ever changing epidemiological and demographic background, and where the number of events are relatively small (but in a large patient population), this can only be revealed in international collaborative studies. Fortunately, this activity has already started, not least based on the Skagen initiative, which has resulted in several collaborative protocols [175].

Head and neck cancer is the classical tumor site for studying radiobiological based treatment optimization. Squamous cell carcinomas do typically exhibit the characteristics with a clear dose-response relationship, the influence of volume, dose per fraction, overall treatment time and hypoxia [79,176,177]. In addition, individual prognostic factors are also clearly identifiable in this patient cohort [178,179]. In recent years, the scenario has changed dramatically with the almost epidemic increase in the number of human papillomavirus (HPV) positive oropharyngeal cancers [180]. This has two implications, one being the fact that HPV positive tumors are found to be much more radiosensitive [181] and consequently, these patients have a superior prognostic outcome which is further supported by reduced risk from smoking and comorbidity [180,182]. In turn, this has resulted in various attempts to reduce the treatment burden in HPV positive patients, either by descalation of the radiation dose or in small tumors to exchange RT with trans-oral surgery. Several studies are underway or under discussion to deal with this issue, but no conclusive results are available. Until this has happened, dose reduction in this patient cohort should only take place within the framework of controlled clinical trials. The other implication is linked to the treatment of the ‘traditional’ head and neck cancer patients, which are HPV negative and typically have a tobacco-related etiology with a consequential enhanced risk of late morbidity and comorbidity [182–187]. Such patients, which demand the most challenging treatment, unfortunately also are the ones who have major difficulties in tolerating it. Therefore, they may be candidates for more refined radiation techniques with particle beam therapy, based on risk selecting strategies and/or the recurrence risk would be challenged based on individual risk parameters, focusing on especially hypoxia, tumor stem cells, immunological parameters or other individual molecular genomic parameters [79,178,188,189].

To comfort this plentiful scenario of treatment options, we are moving toward multicenter matrix protocols addressing both modified therapy with e.g. hypoxic radiosensitizers and predictive selection based on genetic profiles, radiogenomics or biological imaging. A challenge persists in identifying the risk of failures relative to intratumoral variations in tumor resistance (e.g. hypoxic imaging) [190–192]. Currently, we are still awaiting results which may identify to which extent we can apply those escalation studies based on dose painting or if will need to include the entire cross tumor volume [191,193]. Most interestingly are we again back to placing our main focus on hypoxia, both the indication through biological imaging or through identification of predictive parameters [189,190,194–196] and especially through exploration of therapeutic intervention [185]. This takes e.g. place with the large international multicenter clinical trial (EORTC 1219/DAHANCA 29), which not only evaluates the importance of the hypoxic modifying sensitizer Nimorazole [185,189,197,198] but also investigates whether such potential benefit can be predicted by a hypoxic 15-gene probe [188,189]. The outcome of this study, together with similar other trials investigating Nimorazole [197,198], is likely to yield ultimate information on the role of hypoxic modification in head and neck cancer – and thereby answer a more than 100-years old question in clinical radiobiology [199].

An overview of dose-volume-and fractionation effects in lung cancer was given by D. De Ruysscher, Maastricht. By SBRT, extreme high biological doses are delivered over a short time to small volumes. From a radiobiological point of view, it is not the most appropriate approach, but numerous studies have shown that SBRT of peripheral NSCLC is efficient and safe [200–202]. Studies with very long follow-up confirm these findings [203]. SBRT is now the preferred radical therapy of patients with severe chronic obstructive lung disease (who are in high risk for severe complication and death with surgery). With the current knowledge, there is no indication that SBRT of small volumes of lung result in deterioration of lung function [204], but recent studies indicate that large doses to the heart may have detrimental effects and increase the patients risk for suffering non-cancer death [205]. Advisory Committee on Radiation Oncology Practice (ACROP) guidelines providing advice on patient selection, equipment used and methodology of SBRT planning and delivery, quality assurance and patient follow-up have recently been published [206].

SBRT of central tumors (within 2 cm from the bronchial tree) is related to a risk of fatal bleeding. Three of 71 patients in RTOG0813 had a grade 5 toxicity but in an analysis of ultra-central cases where PTV-overlap with the main bronchi and trachea this risk increased dramatically [207,208]. There is no current consensus on how to treat central tumors, but there is a general belief that they should be treated with softer fractionation if not conventional RT [209].

Several randomized studies have aimed to test SBRT against surgical resection but were closed because of poor recruitment. A pooled analysis of two of the studies revealed a superior survival in the patients treated with SBRT compared to surgery, but the pooled analysis is not sufficiently powered to give a firm conclusion [210]. A Scandinavian study compared SBRT with conventional fractionated RT [211,212]. It revealed no difference in survival and control of disease, but there was less toxicity in the SBRT treated patients. In stage III NSCLC, RT with total doses of 60–66 Gy concurrent with chemotherapy is considered the standard of care [213]. Disease control and survival were expected to improve with dose escalation, but RTOG 0617 revealed inferior survival with the higher doses which has never been understood [214]. Deaths because of excess toxicity are an obvious possibility and the toxicity should be further explored [215–219]. A Danish study randomized patients with stage III NSCLC to 60 versus 66 Gy, both arms with concurrent oral vinorelbine. In this study, there was no difference between the study arms [220]. Boost to FDG PET/CT-positive subvolumes of advanced stage NSCLC is currently being tested [221].

Modern therapy of rectal cancer is truly multidisciplinary, involving surgeons, medical oncologists, radiologists, pathologists together with the radiation oncologist. Vincenzo Valentini gave a status of the radiation oncologist’s view on how radiation therapy should be implemented in the multimodality management of rectal cancer as it is described in the European Registration of Cancer Care consensus document [222].

The Swedish Rectal Cancer Trial was the first to show that preoperative RT (pre-op RT) improved the outcome in patients with rectal cancer [223]. A benefit in terms of improved local control, cancer-specific survival and overall survival appeared in all Dukes stages. Since then a number of randomized studies have confirmed that pre-op RT improves local control, but the Swedish study using short course 5 Gy ×5 without chemotherapy and with immediate surgery is still the only study to demonstrate a survival benefit [224–228]. Studies using long-term chemoradiation with delayed surgery showed improved local control and 8–30% complete pathological responses without demonstrable viable cancer in the resected specimen [229,230]. A recent randomized Swedish three-arm study found no difference in local control between short course (5 Gy ×5) with and without delay before surgery or long course pre-op RT [231–233]. The relatively high percentage of patients with the complete pathologic response has led to the hypothesis that surgery may be modified or omitted in some patients obtaining complete clinical remission after long course pre-op radio-chemotherapy. This is currently being tested in clinical trials. It is a general belief that pre-op RT is not needed for low-risk cancers if state-of-the-art total mesorectal excision is used and long course radio-chemotherapy is preferred in the more high-risk patients because of the high chance of local response compared to the short course regimen. The short course is often reserved for patients who are not amendable for pre-op chemotherapy. Pre-op RT comes to a prize with increased loss of sphincter control, but it seems less pronounced than after postoperative RT [225,234]. Due to limited evidence, the recommendations for the selection of patients for pre-op RT – for short or long course of pre-op RT and for adjuvant chemotherapy – are weak and there are large national differences in the practice of pre-op RT [235].

In conclusion, the BiGART2017 meeting demonstrated a positive and strong development within the field of radiation oncology toward integrating clinical radiobiology, advanced imaging, state-of-the-art RT, and the most recent advances in ion beam therapy. This is a necessary development in order to secure that cancer patients will continue to be offered the most optimal therapy with photons or ion beam, in an integrated and evidence-based framework.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by a grant from Acta Oncological Foundation.