Stereotactic Radiotherapy for Early Stage Non-Small Cell Lung Cancer

Current Standards and Ongoing Research

Eugenia Vlaskou Badra; Michael Baumgartl; Silvia Fabiano; Aurélien Jongen; Matthias Guckenberger


Transl Lung Cancer Res. 2021;10(4):1930-1949. 

In This Article

Research Perspective of SBRT for ES NSCLC

Clinically Oriented Research

SBRT as Neoadjuvant/Adjuvant Treatment to Surgery in Operable Patients. SBRT in NSCLC is most commonly used as a single modality treatment. While many studies aimed to improve R0 resection rates in locally advanced NSCLC,[101] such concepts are at a very early stage for ES NSCLC. The MISSILE-NSCLC trial was the first prospective phase II trial, which aimed to evaluate complete pathologic response (pCR) after neoadjuvant SBRT in ES NSCLC.[101] They reported a pCR rate of 60% at ten weeks after SBRT, a local control rate of 100% at two years, and unchanged QoL, while treatment-related toxicity was comparable to that of surgery alone.[102] The pCR rate appears low compared to local control rates after SBRT and has been critically discussed. However, it needs to be considered that definition of pCR shortly after high-dose is not well defined and pCR is well known to increase after follow-up longer than 2–3 months.

SBRT Instead of Surgery in Operable Patients. Promising outcomes in inoperable ES NSCLC have prompted attempts to implement SBRT in the management of medically operable, fit patients. To date, there are no published RCT comparing SBRT vs. lobectomy (VATS) in medically operable stage I NSCLC patients. Three prospective trials comparing SBRT with surgery (STARS, ROSEL and Z4099)[32,103] were terminated early due to poor accrual, while a feasibility study in the United Kingdom showed that a large RCT is not feasible owing to the same reason.[104] The pooled-analysis of the STARS and ROSEL trials however—while limited—reported promising results and an advantage of 15% in OS with SBRT,[60] while two prospective trials (JCOG0403 and RTOG 0618) reported high rates of tumor control and low treatment-related morbidity.[34,35] While retrospective data suggests likely equal or superior outcomes with surgery, more randomized trials comparing surgical approaches to SBRT in the medically operable are warranted. The ongoing POSITIVL,[105] VALOR[106] and STABLE-MATES[107] trials aim to evaluate SBRT vs. complete resection (not further specified), lobectomy, and sublobar resection respectively in medically operable patients within a randomized control trial.

SBRT Combined With Cytotoxic Chemotherapy. Although SBRT can achieve excellent local tumor control, overall survival is predominantly limited by regional and distant disease progression after SBRT for ES NSCLC. Robinson et al. reported 4-year local, regional, and distant control after SBRT of 93.6%, 78.1%, and 54%, respectively.[108] This forms the rationale for investigating the combination of systemic treatment with SBRT, aiming to improve OS.

To date, no randomized prospective studies have examined the addition of chemotherapy before or after SBRT for ES NSCLC.

A retrospective analysis of the National Cancer Database showed that only 3% of analyzed patients received adjuvant chemotherapy after SBRT between 2004 and 2014.[109] Those patients had a significantly worse OS as compared to patients receiving SBRT only (28.0 vs. 36.5 months, P=0.001). After propensity-score matching, this difference increased further (28.0 vs. 47.7 months, P<0.0001). For the subset of patients with tumors greater than 4 cm, no statistically significant difference in OS was found, even after propensity-score matching. Whether this surprising difference is the result of the different treatment protocols or of different patient and disease characteristics, which were not corrected in the propensity-score matching, remains unknown.

These results are in disagreement with the retrospective study of Chen et al., which showed an improved OS for patients receiving cisplatin-based adjuvant chemotherapy as opposed to those receiving SBRT alone (47 vs. 36 months, P=0.035).[110] It is however important to note that patients were not randomized and that those who did not receive chemotherapy were either considered too old (over 75 years of age) or had relevant comorbidities. It is therefore likely that the two populations were heterogeneous, which could explain the lower overall survival in the SBRT-only group.

Verma et al. also analyzed the National Cancer Database, focusing exclusively on tumors greater than 5 cm treated <10 SBRT fractions.[88] When comparing patients receiving chemotherapy (before or after SBRT) to those receiving SBRT solely, OS was significantly greater in the former group (30.6 vs. 23.4 months, P=0.027). The role of chemotherapy remained significant in multivariate analysis.[111] Those results suggest that adjuvant chemotherapy after SBRT for ES NSCLC may be beneficial, mainly in patients with larger tumors. Prospective data are needed to verify this hypothesis.

SBRT Combined With Immune Checkpoint Inhibition. In advanced-stage NSCLC, immunotherapy alone or in combination with chemotherapy has achieved significant and clinically relevant overall survival improvements in comparison with chemotherapy for patients with both squamous and non-squamous advanced NSCLC.[111,112] Despite this background from metastatic NSCL and a strong preclinical rational, there are currently no published studies combining SBRT with immunotherapy in ES NSCLC.

An abstract from Daly et al. was published in October 2019, reporting the results of a phase I study that included 15 patients receiving Atezolizumab, a PD-L1 inhibitor, and SBRT (50 Gy in four or five fractions). Patients received intravenous Atezolizumab every 21 days over six cycles, while SBRT was delivered concurrently at the beginning of the third cycle.[113] The dose-limiting toxicity was assessed for 12 patients and the combination was well tolerated, with no grade 4 or 5 events and only one grade 3 event requiring interruption of treatment.

Several randomized phase III clinical trials are currently ongoing. Amongst them, the PACIFIC-4 trial (NCT03833154) aims to recruit 706 stage I or II (with negative lymph nodes) patients by 2025.[114] This double-blind, multi-center trial will investigate the benefit in progression-free survival when adding monthly Durvalumab (PD-L1 inhibitor) versus placebo for 2 years following SBRT. Another phase III study (NCT04214262) was started this year to study the influence of Atezolizumab (another PD-L1 inhibitor) on OS, when administered before and after SBRT[115] (NCT04214262). Beyond the different drugs used, this clinical trial differs from the aforementioned one in some ways: it is not blinded and uses OS as a primary outcome. Results are expected in 2028 after the recruitment of 480 patients.

Until the publication of the results from those phase III trials, some insight will be gained by the many ongoing phase I and II clinical trials also investigating SBRT and immunotherapy for ES NSCLC. The largest of them, the ASTEROID trial (NCT03446547), is a randomized multicenter phase II trial, due to enroll 216 patients with T1–2N0M0 NSCLC receiving adjuvant Durvalumab after SBRT versus SBRT-only (3 or 4 fractions).[116] A second randomized phase II trial, led by a team from MD Anderson, will look into SBRT with or without concurrent and adjuvant Nivolumab.[117] The recruitment goal is set at 140 patients with stage IIA or less.

SBRT Combined With Targeted Therapies. Similar to immunotherapy, the established role of targeted therapies in advanced NSCLC has laid the groundwork for their evaluation in ES NSCLC. In 2014, Wang and his colleagues published the results of a prospective study on 14 patients with advanced NSCLC (stages IIIB or IV).[118] Those patients received 250 mg of Gefitinib (epidermal growth factor receptor inhibitor) daily, then concomitant SBRT in three fractions, and continued with Gefitinib for a year or until disease progression. Those patients showed good tolerance of the combined therapy, with few grade 3 toxicities and no grade 4 or higher adverse events. The median follow-up was 15.5 months and the median OS was 19.0 months. One-year local control was 83.9%. As of today, there are no publications studying the interaction of SBRT and targeted therapies in ES NSCLC, and no ongoing trials have been reported either.

Technology Oriented Research

Real-time Tumor Tracking. With the current technologies available, it is unambiguous to compensate for inter- and intrafractional motion of tumor and internal organs at risk. The most commonly practiced 4D motion compensation strategy uses the so-called internal target volume concept (ITV) with continuously irradiating the target during free breathing.[44] The most sophisticated concept for reducing (geometrical) safety margins is the so-called real-time target tracking.[119] The process of real-time tumor tracking can be divided into three components: continuous or repetitive assessment of target motion, prediction models to compensate for time delays and non-continuous target monitoring, as well as real-time dynamic motion compensation.

Regular assessment of target motion refers to accommodating respiratory motion by dynamically repositioning the radiation beam in order to follow the tumor location. Tracking of the tumor location can be achieved by radiographically tracking the tumor lesion itself or by tracking of a surrogate structure, using the following four methods:[120]

  • Radiographic imaging of the lesion itself: Well-defined, natively high-contrasted and conveniently located tumor lesions can potentially be detected in (kv-)imaging acquired during treatment.

  • Radiographic imaging of implanted fiducial markers: the implantation of metal fiducial markers allows for detection in kV imaging or fluoroscopy during treatment, and while a single marker enhances tumor detection, the implantation of multiple markers (three or more) and the measurement of distance between them accounts for tumor motion as well as marker migration.

  • Radiographic imaging of a surrogate structure: when continuous imaging of the tumor itself is not feasible, the correlation of the tumor position and an external respiration signal source such as anatomical structures or surface markers can be of use. If the relationship between the tumor position and the surrogate signal is stationary, measurement of the spatial relationship beforehand could be sufficient. However as respiratory physiology is complex, a constant correlation of displacement is not exclusively safe to assume.

  • Non-radiographic tracking of implanted signaling devices: Non-radiographic tumor tracking can be achieved by implanting signaling devices, that can be tracked remotely in three dimensions

Treatment delivery system latencies can have a disadvantageous impact during the treatment delivery using real-time tumor tracking systems. Prediction models might help to reduce the tumor localization error and improve gated treatment accuracy,[121] while adaptive filter algorithms can be used to adjust for nonstationary correlation of the empirical tumor motion.[120]

Real-time dynamic motion compensation can be achieved using MLC compensation, which adapts the leaves opening as the tumor moves. This adaptation is possible using real-time information using the Electronic Portal Imaging Device.[122]

Feasibility of MLC tracking has been shown on Varian, Elekta, and Siemens standard linear accelerators (linacs); however, it is not yet commercially available.[123–128] The first report of a lung cancer patient treated with implanted electromagnetic transponders and real-time adaptive radiotherapy using MLC tracking was published in 2014 in a non-commercial framework and on a standard linac.[127]

Additionally, markerless lung target tracking was performed on a modified programmable platform (HexaMotion, ScandiDos) with a Computerized Imaging Reference Systems (CIRS) phantom mimicking different breathing patterns on a standard linac[129] passing all required QA criteria.[119,130,131]

Non-ionizing imaging modalities become more and more important to be implemented in combination with real-time motion compensation on standard linear accelerators. Simulating characteristic tumor trajectories in a water tank using a 4D online ultrasound MLC tracking technique showed promising results to complement the current, commercially available MLC tracking techniques with a noninvasive approach.[132] Nevertheless, the main limitation of online ultrasound imaging in general is the speed-of-sound errors in soft tissue with different physical properties leading to a maximum distance error of several millimetres.[133] Additionally, MRI-linacs have become commercially available and integrated MR imaging allows for continuous tumor tracking during treatment delivery[134] (Figure 3).

Figure 3.

Overview of different tracking methodologies for the detection and compensation of tumor motion. The detection of tumor motion with ultrasound and the compensation of tumor motion with MLC and couch tracking was experimentally proven and is not yet clinically available.

Particle Therapy. The use of SBRT has grown by a factor of three over the past decade and growing numbers of patients with ES NSCLC are expected to be treated in the future.[135] However, traditional photon SBRT has some limitations. As outlined above, severe toxicity has been reported in patients with central tumor location. Using an SBRT technique that minimizes the dose to the OARs is desirable in order to reduce radiation-induced toxicity in the primary setting or in the setting of re-irradiation.[136,137] In this context, particle therapy (PT) with protons or carbon ions could potentially be advantageous. The unique depth-dose curve characteristics of charged particles compared to photons can be exploited to improve normal tissue sparing without compromising tumor control. In addition to the physical dosimetric advantage, carbon ions also have a biological advantage over photons due to the higher probability of inducing tumor DNA-damage associated with a high linear energy transfer.

Several dosimetric studies comparing PT and photon based SBRT in ES NSCLC have shown that PT can offer comparable or even better coverage than SBRT while reducing the dose to the lungs, heart, esophagus, and spinal cord.[138–143] However, it should be noted that the vast majority of the studies comparing dosimetry in ES disease have been performed using the passive scattering technique for PT. These benefits are likely to increase further with the use of pencil beam scanning (PBS) owing to the higher dose conformity, as demonstrated in dosimetric reports.[144–147]

Single-arm phase I/II trials and retrospective data for ES disease have shown that proton therapy results in lung toxicities no greater than grade 3, the ability of dose escalation, and 2-year OS rates of 74–97.8%.[148–151] For carbon ion radiotherapy, Japanese studies have reported OS rates at 3 and 5 years of 75% and 45–50%, respectively.[152–154] Although these studies are promising, they were performed using the passive scattering technique and conventionally fractionated or hypofractionated schemes that are no longer used in the ES setting. A recent retrospective study has investigated the safety and efficacy of PT using pencil beam scanning for ES NSCLC.[155] It has been observed that PBS-based PT is associated with PFS, LC, and OS rates at 2-year of 85.5%, 95.2%, and 90.7%, respectively, with mild acute and late toxicities.

Despite these encouraging results, the optimal clinical context for PT is still unclear. There are currently no clinical data demonstrating a clear benefit of hypofractionated PT over SBRT for ES NSCLC. A meta-analysis comparing the two modalities suggested that there is no statistically significant survival benefit from PT over SBRT after the inclusion of operability, but the 3-years LC favored PT.[156] It should be emphasized that the study reported no indication of the inferiority of PT compared to SBRT, although almost all PBT patients were treated with passive scattering technique and without image guidance.

To date, there is only one report on phase II randomized study comparing SBRT and stereotactic body proton therapy in ES NSCLC by MD Anderson.[157] The trial closed early due to low accrual attributable to the lack of volumetric image-guided radiotherapy (IGRT) and insurance coverage. Nonetheless, the authors concluded that both techniques have acceptable toxicity and lead to comparable results.

In light of these results, further comparisons between PT and SBRT in randomized studies that use advanced techniques are warranted to define the role of PT in ES NSCLC.

FLASH Radiotherapy. In the past decades, advances in high-precision radiotherapy treatment delivery and image guidance have led to significant improvements in the management of lung cancers. However, tumor motion during treatment remains clinically challenging to address. Recently, FLASH radiotherapy has emerged as a technique able to "freeze" intra-fraction motion as it involves the ultra-fast delivery of treatment at dose rates exceeding by several orders of magnitude those currently used in clinical practice. Moreover, many pre-clinical studies across different animal models have shown that FLASH radiotherapy has the potential to markedly improve normal tissue tolerance while maintaining tumor control level (the so-called FLASH effect).[158–160] In a pioneering study on the FLASH effect, Favaudon et al. investigated lung fibrogenesis in C57BL/6J mice after bilateral thorax exposure to pulsed, ultra-high dose rates (≥40 Gy/s) irradiation with 4.5 MeV electron beams given in a single dose.[161] Results showed that FLASH irradiation protects lungs from radiation-induced fibrosis at doses known to trigger the development of fibrosis in the totality of animals after conventional dose-rate irradiation (≤0.03 Gy/s). Cutaneous lesions were also reduced in severity, without modifying the anti-tumor efficiency compared to conventional irradiation.

To date, most studies investigating the FLASH effect have been performed using dedicated electron linear accelerators as a source of radiation, thus limiting its clinical viability in practice. It was recently shown that clinacs can be modified for delivery of FLASH radiotherapy with electrons, thus increasing the potential availability of FLASH irradiators and facilitating its clinical translation.[162,163] However, the poor penetration depth of 4.5–20 MeV electron beams limits FLASH radiotherapy to the treatment of superficial tumors only, or in the intra-operative radiation therapy (IORT) setting. Whilst US researchers are developing the PHASER platform that might represent the ideal approach to bring FLASH with high-energy X-ray beams into clinic,[164] to date, FLASH radiotherapy treatment of deep-seated tumors could potentially be performed only with proton beams. In fact, it has already been shown that modern proton therapy systems are potentially able to produce beams at very high intensities,[165] and it is now being investigated if FLASH dose rates can be achieved for clinical proton therapy treatments.[166]

Radiomics. Cross-sectional imaging is a pillar of modern diagnosis. With the steady improvement of imaging quality and the increase of imaging availability over the last decades, more and more valuable data is available for extraction and interpretation. Radiomics is a fast emerging research field, yielding to harness imaging features and provide additional quantitative information to build prediction models and/or characterize cancer phenotypes. Radiomics is currently evaluated for two purposes: pre-treatment risk assessment of ES NSCLC and post-SBRT assessment of radiation-induced fibrosis versus local tumor recurrence.

A study analyzing a longitudinal 18F-FDG-PET/CT dataset of 100 consecutive patients (ES NSCLC) reported that an unsupervised machine learning method based on 722 radiomics features showed promising outcome prediction compared to prediction models based on clinical characteristics only.[167] Such models would be highly desirable for selection of high-risk patients, which might benefit the most from treatment intensification.

Post-SBRT fibrotic changes are frequently difficult to distinguish from true recurrence after SBRT for ES NSCLC: longitudinal CT imaging improves the accuracy but might put the patient at an increased risk for further disease progression. Early studies have shown promising results of using radiomics for post-SBRT follow-up imaging. Mattonen et al. reported a study of 45 patients, 15 with local recurrence matched to 30 without, where radiomics was able to accurately predict local recurrence as early as 6 months after SBRT.[168]

Despite the promise of Radiomics, its prospective evaluation especially in a multi-institutional environment, with varying imaging hardware and protocols needs to be demonstrated.