|Home | About | Journals | Submit | Contact Us | Français|
Evidence supports stereotactic body radiotherapy (SBRT) as a curative treatment option for inoperable early stage non-small-cell lung cancer (NSCLC) resulting in high rates of tumour control and low risk of toxicity. However, promising results are mainly derived from SBRT of peripheral pulmonary lesions, whereas SBRT for the central tumours can lead to severe radiation sequelae owing to the spatial proximity to the serial organs at risk. Robust data on the tolerance of mediastinal structures to high-dose hypofractionated radiation are limited; furthermore, there are many open questions regarding the efficiency, safety and response assessment of SBRT in inoperable, centrally located early stage NSCLC, which are addressed in a prospective multicentre study [sponsored by the European Organization for Research and Treatment of Cancer (EORTC 22113-08113—LungTech)]. In this review, we summarize the current status regarding SBRT for centrally located early stage NSCLC that leads to the rationale of the LungTech trial. Outline and some essential features of the study with focus on a summary of current experiences in dose/fraction-toxicity coherences after SBRT to the mediastinal structures that lead to LungTech normal tissue constraints are provided.
Stereotactic body radiotherapy (SBRT) is a technique in which high doses of radiotherapy are very precisely delivered with steep dose gradients and a short overall treatment time (OTT). This achieves a very high biological dose.1 Evidence supports SBRT as a curative treatment option for inoperable early stage non-small-cell-lung cancer (NSCLC).2–5 High-precision, hypofractionated dose delivery enables a significant reduction in both target volume size and exposure of normal tissue (NT) to high doses, resulting in a decreased toxicity risk and high rates of local control.6–8 However, promising results are mainly derived from the SBRT of small peripheral pulmonary lesions with low risk of treatment-related toxicity, as these tumours are surrounded by parallel organs at risk (OARs), specifically lung tissue. In contrast, the SBRT for central tumours can lead to severe, potentially life-threatening radiation sequelae owing to the spatial proximity to serial OARs. As robust, prospective and multicentre data on the tolerance of mediastinal structures to high-dose hypofractionated radiation is limited, there is generally caution in implementing SBRT for central lung tumours. Current published data are rather inconsistent regarding the definition of “central tumour”, SBRT techniques, dose prescription and reporting, calculation algorithms, image guidance and evaluation of outcome. A recent systematic review has summarized the current limited evidence on this topic, mainly from single-centre experiences,9 but there are still many open questions regarding patient selection, efficiency, safety and response assessment, which hamper the use of SBRT for central lung tumours in routine practice.10 A prospective study sponsored by the European Organization for Research and Treatment of Cancer (EORTC 22113-08113–LungTech) was launched in late 2014 and is expected to answer the question of the efficacy and toxicity of SBRT in inoperable and centrally located early stage NSCLC in a multicentre setting.
The incidence and mortality of lung cancer in the European Union are 52/100.000 and 47/100.000 per year, respectively.11 The NSCLC accounts for approximately 80%11 of all cases with a poor 5-year survival rate of 16%, mainly owing to patients being diagnosed at advanced stages. If diagnosed at an earlier stage, 5-year survival for NSCLC can be as high as up to 80%.12,13 The current standard of care for early stage tumours is anatomic surgical resection in medically fit patients, consisting of lobectomy or pneumonectomy accompanied by a systematic mediastinal lymph node sampling or lymphadenectomy.13–15 Associated 5-year survival rates are commonly accepted to be 60–80% for Stage I and 40–50% for Stage II NSCLC.13 The efficacy of routine sublobar resections for tumours <2cm is currently being investigated worldwide. This technique spares lung tissue and preserves pulmonary function when compared with lobectomy. However, the results of a randomized controlled trial indicate that this is associated with inferior oncological outcome.16 In contrast to peripheral tumours, those located centrally often show compression and/or invasion in vessels, major bronchi or other critical mediastinal structures and, therefore, require more extensive surgical procedures associated with higher mortality and morbidity.17–19 Approximately 20% of all patients with NSCLC Stage I are medically inoperable because of poor general condition or coexisting morbidities such as chronic obstructive pulmonary disease and/or heart disease,20 and <50% of all patients with early stage NSCLC older than 75years undergo surgery.21 It is expected that as the global population ages and lung cancer screening of high-risk populations is implemented,22 the proportion of inoperable patients with lung cancer with comorbidities will increase.23–25 In Stage I patients with NSCLC who refuse surgery and do not receive other treatments such as radiotherapy, 5-year overall survival (OS) and cancer-specific survival (CSS) are low, 6% and 16%, respectively.26 The standard of care treatment option for medically inoperable patients with early stage NSCLC has been conventionally fractionated radiotherapy (CFRT), superior to best supportive care with CSS rates of about 30% after 5years27,28 and 5-year OS of 29–37% for T1 tumours.29 Evidence supports a clear dose effect for local control and survival in NSCLC,30,31 however, traditional CFRT-planning and delivery techniques are associated with low accuracy and broad safety margins resulting in increased toxicity rates,32 therefore, limiting the delivery of higher doses in inoperable patients with early stage NSCLC. Although advanced techniques and positron emission tomography (PET) staging allow modern three dimensional (3D)-CFRT the application of high biological doses, resulting in local control rates >85% in Stage I patients with NSCLC,33,34 the implementation of such 3D-CFRT regimes leads to longer OTT of approximately 3–6weeks,33,34 whereas SBRT usually can be performed in OTT of days rather than weeks.
In the past two decades, SBRT has been accepted as a curative treatment alternative for inoperable patients with small (<5cm) peripheral early stage NSCLC.2–5 Several prospective studies reported excellent local tumour control rates of up to approximately 90%,5,6,8,35,36 in the same range as those obtained with surgery.37,38 Low rates of toxicity have been observed, including elderly, medically, inoperable patients with severe comorbidities.6,8,35,36,39,40 Survival after SBRT is modest (approximately 50% at 3years) but has to be interpreted in the context of patients with multiple comorbidities. As expected, survival after SBRT is higher in fitter, operable patients who refused surgery.41 Furthermore, population-based studies indicated improved OS in medically inoperable patients following the introduction of SBRT.42 Based on these data, SBRT for patients with early stage NSCLC has been recommended as the standard of care for medically inoperable cases by the European Society For Medical Oncology, Lugano, Switzerland,43 the National Comprehensive Cancer Network, Fort Washington, PA14 and by the Deutsche Gesellschaft für Radioonkologie e.V., Berlin, Germany.20
Although histological or cytological confirmation of NSCLC is recommended before performing SBRT, biopsy is sometimes not feasible or associated with an unacceptably high risk because of the severe comorbidities of the patient. Thus, an inevitable percentage of patients are treated based on clinical suspicion and imaging criteria of malignancy only. However, in Western Europe, the likelihood of having NSCLC has been demonstrated to be well over 90% using a calculation based on clinical parameters, CT characteristics of malignancy and significant uptake of fluorine-18 fludeoxyglucose (18F-FDG) on PET scan.43 Moreover, when malignancy is highly likely based on these imaging criteria, SBRT without pathological confirmation was shown to lead to the same outcome as in pathology-proven NSCLC.44
With regard to dose/fractionation, there is vast heterogeneity in the SBRT regimes used routinely. Internationally, 3×18Gy is one of the most frequently used regimes, whereas in Germany for small peripheral lesions, lower doses of 3×13.5–15.0Gy are recommended20 and have been demonstrated to result in local control rates of >90%.45 Local tumour control has repeatedly been reported to show a dose-dependent increase with a minimum biologically effective dose (BED; α/β ratio, 10Gy) of 100Gy to the planning target volume (PTV) surrounding isodose resulting in local tumour control rates >90%,46,47 that even translates into improved OS.46,48 Interestingly, in a recent meta-analysis, Zhang et al49 demonstrated a plateau effect with regard to OS with doses of 83.2–146.0Gy BED and a detrimental effect on OS with doses >146Gy.
The major difference between peripherally and centrally located NSCLC is the spatial proximity to centrally located serial OARs, such as, main airways, large blood vessels, the heart, the oesophagus, the phrenic nerves or the brachial plexus, where hypofractionated high doses might lead to severe, potentially life-threatening consequences. Several articles have reported high rates of toxicity after SBRT of centrally located NSCLC. A prospective study by Timmerman et al50 from the Indiana University reported a 2-year freedom from severe toxicity of only 54% for central tumours compared with 83% for peripheral tumours, in 70 patients treated with 60–66Gy total in three fractions. Furthermore, SBRT may have contributed to the events leading to the death of six patients (four patients from bacterial pneumoniae; one patient from pericardial effusion; and one patient owing to massive haemoptysis), of which four of the six patients had central tumours.50 These deaths occurred after a median of 10.4months following SBRT (range, 1–20months). The 2-year incidence of toxicity grade ≥3 was 17% and 46% for peripherally and centrally tumours, respectively. The observation of severe toxicity after SBRT of centrally located NSCLC was subsequently confirmed by two studies: grade 5 toxicity was observed in 1 out of 17 patients after treatment with 60Gy in 4 fractions36 and in 1 out of 9 patients after irradiation with 48Gy in 4 fractions51 delivered on consecutive days. Moreover, a case report was published by the University of Pennsylvania, Philadelphia, PA, reporting on a fatal central airway necrosis more than 8months after 50Gy in five fractions was delivered to a central lung tumour.52 Conversely, investigators from the Free University in Amsterdam demonstrated that the use of “risk-adapted SBRT” using a more fractionated regime (60Gy in eight fractions) did not result in excess toxicity for centrally located early stage lung tumours (n=63), and clinical outcomes were comparable with those seen for peripheral tumours.53 However, none of the retrospective data included detailed dose–volume histogram/toxicity analyses. A PRISMA structured literature review published in 2013 identified 20 publications reporting outcomes for 563 central lung tumours treated with SBRT, including 315 patients with early stage NSCLC.9 The majority of these studies were retrospective and conducted at single institutions. Only four studies were prospective and reported on 68 patients with central tumours. Local control was 85% when the prescribed BED10 was >100Gy. Tumour location (central vs peripheral) did not impact on OS, and treatment-related mortality was 2.7% overall. Grade 3/4 toxicities following SBRT for central tumours were more common than for peripheral tumours, and occurred in <9% of patients. It should however be noted that there was heterogeneity in the common toxicity criteria used to define toxicities and treatment–response criteria. Furthermore, the follow-up of these studies was relatively short (median, 18months), and long-term toxicity data are needed given the high dose and hypofractionated nature of the treatments delivered. The BED10 delivered ranged between 60 and 180Gy. It is important to note that published studies vary widely in their reporting of dose/fractionation as well as dose specifications and calculations. Only 11 of the 20 reviewed studies used tissue heterogeneity correction, and the prescription isodoses ranged between 50% and 100%.
In summary, there is a clear need for a prospective, multicentre study using standardized RT techniques, treatment–response criteria and toxicity assessment. As the maximum tolerated doses and optimal fractionation schemes for SBRT near mediastinal structures are currently unknown, collective knowledge of the complex relationship between the radiation dose and the volume of irradiated tissue for each mediastinal structure is a pre-requisite to the safe future routine use of SBRT for central tumours. The lack of high-quality evidence to support SBRT in centrally located tumours is reflected in international guidelines that only recommend SBRT for peripheral and medically inoperable early stage NSCLC.43,54 We therefore advocate that treatment of central lesions should preferably be performed within the context of a well-conducted prospective trial.
The LungTech trial aims to evaluate the efficiency and toxicity of SBRT for patients with histologically or cytologically proven early stage, centrally located, inoperable NSCLC in a multicentre setting. The primary end point will be freedom from local progression at 3years as assessed by serial CT scans and confirmed by 18F-FDG-PET/CT. If still unclear, biopsies or 3 monthly repeated imaging will be performed. Using one-sided 5% Type I error and 80% power, 150 patients (included so far from 23 planned centres in Belgium, France, Germany, Poland, Switzerland and UK) should be sufficient to reject the hypothesis of freedom from the local progression rate at 3years of ≤80% under the assumption that SBRT is expected to achieve a freedom from local progression rate of 90%. The trial will also investigate acute and late toxicities as well as patterns of local and distant recurrence, including mediastinal failure, assessed by serial CT scans and confirmed by 18F-FDG-PET/CT and subsequent repeat imaging or biopsy, if necessary. Moreover, this study offers a unique opportunity to evaluate the role of 18F-FDG-PET/CT to monitor disease progression and toxicity, as translational research end points, including staging comparison between 3D- and four-dimensional (4D) 18F-FDG-PET/CT assessment, possible impact of 4D 18F-FDG-PET/CT on target volume contouring and external validation of Huang et al55 response criteria.
Recently, the Radiation Therapy Oncology Group (RTOG) has performed a dose-escalation Phase I/II study (RTOG 0813) in centrally located NSCLC.56 Derived from the North American standard, the study is based on a five-fraction regime, the results are awaited. Although the RTOG study includes only T1/T2 tumours, the LungTech study will also include a limited population of T3 tumours and therefore supplement the available data pool. In Europe, owing to the risk of toxicity with three- to five-fraction regimes as described above, it was decided to investigate a less hypofractionated “risk-adapted” approach (60Gy in eight fractions) as reported by the Free University in Amsterdam.53 To ensure the feasibility and safety of the planned treatment, only those centres that have used SBRT for a period of at least 12months and have treated a minimum of 20 patients will be eligible. The study will be conducted only within EORTC centres and will benefit from the EORTC and Radiation Oncology Group Quality Assurance infrastructure. A pooled data analysis of both studies maybe an interesting future option.
Some essential features of the protocol are outlined below.
In current literature, the definition of what constitutes a “central” tumour varies widely and further hampers the comparison between trials. Only 11 out of the 20 studies reviewed by Senthi et al9 used the same definition as RTOG 0813;56 Modh et al57 even included two different definitions of “central tumours” within the same publication of their study.
Haasbeek et al53 defined central tumours as located within a 1-cm zone from the mediastinal envelope. In RTOG 0236,58 the following definition is used: a tumour within 2cm of the proximal bronchial tree (carina, right and left main bronchi, and bronchial tree to the second bifurcation), whereas RTOG 0813 sticks to the definition of Timmerman et al:50 a tumour located within 2cm in all directions around the proximal bronchial tree and immediately adjacent to mediastinal or pericardial pleura with a PTV expected to touch or include the pleura. This definition has also been slightly modified in the LungTech trial: “Centrally located tumour defined as tumour within 2cm or touching the zone of the proximal bronchial tree or tumour that is immediately adjacent to the mediastinal or pericardial pleura, with a PTV expected to touch or include the pleura.”
Some tumours might be “too central” to be safely treated with SBRT when applying BED >100Gy. Even using the “risk-adapted” LungTech regime, there will be central tumours, for example, invading the proximal bronchial tree and/or hilar structures, which will not be treatable within the trial owing to a very high risk of severe toxicity (Figure 1). In current literature, a selection bias with regard to the inclusion of certain anatomical subgroups within the SBRT “danger zone” cannot be excluded. Indeed, the examples and description of patient cohorts, for example, in the Haasbeek's study53 show that “very central” tumours have not been included in the larger published series. However, a clear definition of what constitutes “too central tumours” is lacking. In the LungTech protocol, T3 tumours >7cm, all T4 tumours and tumours abutting the oesophagus or presenting with separate tumour nodule(s) in the same lobe are excluded to increase the safety of this treatment. Furthermore, all potential LungTech cases will undergo a central expert review of eligibility relating to tumour anatomical location and treatment planning before enrolment in the study treatment, thus confirming central tumour localization and excluding tumours that are “too central” according to the eligibility criteria and dose constraints (e.g. proximity to the central bronchi, Figure 1).
In terms of tumour dose, the LungTech protocol will investigate a medium hypofractionated approach, applying 8.0×7.5Gy to a total dose of 60Gy; this is equal to a BED of 105Gy (α/β=10). Referring to report 83 of the International Commission on Radiation Units and Measurements (ICRU), 95% of the PTV has to receive at least the nominal fraction dose, and 99% of the PTV receives a minimum of 90% of the nominal dose. The maximum dose within the PTV should not be <110%, nor should it exceed 130% of the prescribed dose. Gross tumour volume (GTV) delineation is based on 4D-CT in treatment position, mandatory 3D-PET/CT scan and supplementary clinical information, for example, results of bronchoscopy. To account for tumour motion during the breathing cycle, the protocol allows the individual internal target volume (ITV) and the average mean position approach. The PTV margin based on the ITV concept is primarily meant to take into account patient set-up uncertainties, thus requiring an isotropic ITV expansion of 3–5mm. For PTV generation, based on the average mean position of the tumour, the margin should take into account both the set-up error and breathing-induced motion and should not be <3mm. All patients treated in this trial will receive image-guided SBRT. Further details on the technical aspects of the radiotherapy planning including dose specification and radiation therapy quality assurance (RTQA) procedures will be reported elsewhere.
Defining NT constraints for thoracic OARs, in the context of SBRT, is a major challenge, as the maximum-tolerated doses and optimum fractionation for mediastinal structures are currently unknown. Another challenge is the assessment of treatment-related toxicity in patients with multiple comorbidities, which may be a confounding factor. Whilst toxicity for SBRT delivered to peripheral tumours is well documented, this is not the case for central tumours, and long-term data are necessary given that late effects can become apparent more than 1year post treatment.65 For this reason, patients will be prospectively assessed for at least 3years after treatment in the LungTech study.
Although the linear quadratic model (LQM) has been criticized for not being applicable to SBRT,76 data from Guckenberger et al77 suggested accurate modelling of local tumour control in fractionated SBRT for Stage I NSCLC with the traditional linear-quadratic formalism. Brown et al78 reported that there is compelling in vitro and in vivo NT evidence that the LQM provides a reasonable estimate of dose–response relationships including single high doses. Therefore, the NT constraints defined in LungTech have been derived from the available literature on SBRT coupled with LQM estimates related to CFRT experience (Tables 1 and and22).
While translating stated dose/fraction from the varying data to comparable equivalent dose in 2Gy fractions (EqD2), wide ranges of allegedly comparable equivalent doses become apparent (Tables 1 and and2).2). This is most likely because different α/β ratios have been used and/or other factors than LQM estimates might have contributed to dose calculation or recommendation. This obvious heterogeneity underlines the need for systemically collected data for defining maximum-tolerated doses and optimum fractionation for mediastinal structures, and thus robust NT constraints for thoracic OARs. The dose constraints and possible deviations chosen for the LungTech trial are presented in the EORTC 22113-8113RTQA guidelines (Table 3).
Clinical symptoms of acute radiation-induced lung injury develop within approximately 3–6months after treatment. A proportion of patients will subsequently develop radiation fibrosis (from 6months after SBRT). The assessment of tumour response after SBRT by CT-based criteria [response evaluation criteria in solid tumour (RECIST)] is therefore challenging, as such changes are typically seen within the high dose area. However, some changes, independently of the tumour size change, have been identified as more reliable indicators of local recurrence (e.g. opacity with a convex border, disappearance of air bronchograms55). In order to confirm a suspicion of local recurrence, it has been shown that 18F-FDG-PET/CT may play an important role in increasing the sensitivity of the diagnosis. In a systematic review of the literature, Huang et al55 proposed a set of criteria for the evaluation of local recurrence with CT and PET findings in patients with primary lung tumours or lung metastases treated with SBRT. A value of maximum standard value uptake (SUVmax)>5 was suggested79,80 for the differential diagnosis between local recurrence and post-treatment changes. However, the quantitative measurement criteria set by Huang et al,55 are derived from various single-centre studies with heterogeneous PET imaging protocols. Therefore, Huang et al55 recommend conducting all PET scans for a specific patient on the same machine and with a standardized scanner. This is obviously not possible in the context of a multicentre trial.
In the LungTech trial, 18F-FDG-PET/CT will be requested in case of equivocal findings or progressive soft-tissue abnormalities. The integration of quantitative measurement criteria would require the standardization of all the 18F-FDG-PET/CT scanners. Given that such standardization, e.g. the “European Association of Nuclear Medicine” (EANM) accreditation, is not used in all the participating centres, a modified Huang et al55 criterion is being applied with local progression defined as “focal 18F-FDG accumulation significantly above the mediastinal blood pool” (Figure 2). However, quantitative assessment (SUVmax) of the treated tumour and regional lymph nodes will also be measured and collected in order to prospectively evaluate the robustness of the criteria set by Huang et al55 in a subset of centres that have EANM accreditation. In difficult cases, reimaging or biopsy will be requested to confirm progression if clinically indicated.
An important secondary end point of the LungTech trial is the prospective evaluation of toxicity. As previously described, there is a paucity of prospective data regarding the relationship of dose fractionation, irradiated volume and toxicity of SBRT in centrally located tumours. One of the objectives of LungTech is to elaborate on such coherences and thus be able to provide robust recommendations on central NT constraints in patients treated with SBRT. In order to make recommendations that will be applicable to other SBRT dose/fractionation, pooling the data of LungTech and RTOG 081356 (five-fraction regime) is furthermore an interesting future option. Adverse events are assessed according to Common Terminology Criteria for Adverse Events (CTCAE) v. 4 at baseline, at the end of treatment, 6weeks after SBRT, 3 monthly for the first 3years, 6 monthly for up to 5years and yearly thereafter. Additionally, CT imaging is performed and pulmonary function tests recommended. Serious Adverse Event and Suspected Unexpected Serious Adverse Reactions reporting are conducted according to the Good Clinical Practice.81
Differences in delivery of care between institutions may induce variations affecting trial outcome. These variations may be multifactorial such as failure to adhere to protocol guidelines or differences in the equipment quality and/or its use. To prevent these biases, a high-technology quality assurance (QA) procedure has been developed. The RTQA procedure aims to ensure for each treatment plan an acceptable level of conformity to the protocol guidelines.82 The process includes the requirement to submit a benchmark case for review prior to entry of the first patient in the trial (Figure 3). During the recruitment phase, all treatment plans are sent for a prospective central review by a team of experts, who will decide whether the patient is allowed to be treated based on the treatment plan provided (Figure 4).83 Another aspect of RTQA is the verification of the correct use of RT techniques within the trial. End-to-end tests will be performed during site visits using a specific breathing phantom (carrying films and an ionization chamber). Owing to the moving nature of the target, the RTQA procedure also includes an evaluation of the PET/CT, CT and 4D-CT techniques to assess the impact of potential motion artefacts on target volume accuracy. Furthermore, the LungTech QA evaluation will also include the assessment of consistency of Hounsfield units and SUV values, which is of particular relevance given the use of PET/CT data in this study.
With technology advancement, the conduct of modern clinical research requires a major upgrade of clinical research organizations and the infrastructure for performing such sophisticated trials. Indeed, implementing this high-technology research brings new challenges to international multicentre clinical trials in terms of QA and standardization. For instance, imaging technologies such as PET scans have existed for decades and are widely available, their full use in clinical research, however, remains challenging. The quality and comparability of images collected within international multicentre clinical trials are not always optimal. Reliable clinical research involving imaging or radiotherapy can only be achieved within quality-assured, multicentre trials supported by robust methodology and operational infrastructures, allowing the processing, storage and analysis of imaging or treatment plan data by experienced researchers to be fully integrated with clinical and biological data (Figure 5). Maintaining quality-assured clinical trials infrastructure is a cornerstone of independent clinical research, which will guarantee consistent, long-term and reliable research to patients. To have the necessary academic clinical research infrastructure available, skilled, experienced staff must be retained.
In summary, the LungTech trial is expected to provide high-quality, prospective, multicentre data on the efficacy and toxicity of moderately hypofractionated SBRT for central early stage NSCLC. The data generated by this trial will inform future recommendations on technique, prescription, risk estimation and response assessment for the routine use of this promising new radiotherapy technology. This is of particular importance given the likelihood of an increase in the proportion of patients with lung cancer diagnosed at an earlier stage through screening programs.
The authors thank Sally Falk for language support.
This publication was supported by the EORTC Academic Fund, the Deutsche Krebshilfe e.V., the EORTC Charitable Trust and Cancer Research UK.