A systematic review of outcomes following stereotactic ablative radiotherapy in the treatment of early-stage primary lung cancer

Abstract

Stereotactic ablative body radiotherapy (SABR) describes a radiotherapy (RT) technique where high doses of radiation are precisely delivered to an extracranial target within the body, using either a single fraction of RT or using multiple small numbers of fractions. SABR has now become the standard of care treatment for patients with early-stage non-small-cell lung cancer (NSCLC) for whom surgery is not appropriate. This systematic review considers the evidence supporting the use of SABR in early-stage NSCLC, reported toxicity rates, the use of SABR in centrally located NSCLC, the use of SABR as salvage therapy following surgery or RT, and future potential drug combinations with SABR.

INTRODUCTION

Stereotactic ablative body radiotherapy (SABR), also known as stereotactic body radiotherapy (SBRT), has been defined by the American Society of Radiation Oncology and the American Society of Radiology as an external beam radiation therapy method used to very precisely deliver a high dose of radiation to an extracranial target within the body, using either a single dose or a small number of fractions.¹ The UK National Radiotherapy Implementation Group report defines SBRT as the precise irradiation of an image-defined extracranial lesion, using a high total radiation dose delivered in a small number of fractions (hypofractionation).²

Even with the increasing number of patient outcome data being published on SABR in early-stage (ES) non-small-cell lung cancer (NSCLC), phase III evidence comparing the treatment against surgery and conventional radiotherapy (RT) is limited. The most compelling evidence for SABR in extracranial sites is available in early lung cancer where systematic reviews have shown 2-year survival rates and local control as high as 70% and 90%, respectively.³

Despite this, SABR is now a recognized standard of care in early peripherally located inoperable lung cancer. Use of SABR treatment can produce local control rates which are similar to surgery, with low reported toxicity and patient convenience due to the reduced number of visits required for treatment, when compared with conventionally fractionated RT (3–8 compared with 20–33 treatment visits). There is also an increasing evidence base/clinical experience for the use of SABR in other primary cancer sites such as the prostate, pancreas, liver (metastases and primary hepatocellular carcinoma) and spinal metastases, and for oligometastatic disease.^4–9

In order to deliver ablative doses of RT to tumours, it is necessary to accurately identify both the target and the surrounding organs at risk (OARs), in addition to accommodating inter- and intrafraction movement of both.¹⁰ The three key components required to achieve this involve patient immobilization, on-treatment image guidance and on-treatment patient motion management.¹¹ With the use of four-dimensional CT scanning for planning of the tumour motion, and image guidance with cone beam CT during treatment delivery, uncertainties/errors have been reduced significantly, thus increasing the confidence in delivering hypofractionated radiation precisely.¹² In addition, the ability to quantify individual patient's tumour motion can allow patient-specific margins.¹³ For tumours with observed significant motion, different methods have been developed to either limit motion (e.g. abdominal compression/breath-hold techniques) or to track the tumour with respiratory gating or real-time tumour localization using extra-anatomical or implanted fiducials.¹⁰

Further advances in planning such as the use of volumetric arc therapy or numerous non-coplanar beams allow increased conformity of dose to the tumour and hence better sparing of OARs. The introduction of flattening filter-free linear accelerators (linacs) with high dose rate delivery means that SABR treatments can be delivered as quickly as standard conventional RT.¹⁴ This is clearly beneficial and more comfortable for patients and means that there are less concerns regarding patient stability during radiation delivery.

In the UK, SABR doses for NSCLC aim to deliver between 54 and 60 Gy in three to eight fractions delivered on an alternate-day basis, depending on the proximity of the OARs.¹⁵ Where the tumour is close to the chest wall, patients are treated with five fractions to reduce the risk of late rib fractures and chest wall toxicity. Tumours close to the vertebral column, brachial plexus and major vessels are usually treated with eight fractions, with the three-fraction regimen mainly used for tumours surrounded by lung parenchyma only with no concerns of radiation toxicity to other OARs. Tumours that lie within 2 cm in the axial plane of the main airways are presently not routinely treated with SABR in the UK outside clinical trials due to increased toxicity seen in the dose escalation studies by Timmerman et al.¹⁶

Search strategy

PubMed, OVID and Web of Science were searched encompassing the search terms early stage, non-small cell lung cancer and stereotactic RT/SABR/SBRT, resulting in over 2400 articles. The search period was from January 2000 to April 2016. The search parameters were filtered using the terms trials, outcomes, surgery and English. The resulting articles were then selected by assessing the article for relevance, including patient cohorts undergoing SABR, and toxicity of SABR and presenting overall survival (OS) and local control estimates.

Treatment options for early-stage non-small-cell lung cancer

ES-NSCLC is defined according to the TNM seventh edition as T1 (<3 cm) and T2 (<7 cm) lung cancer, with metastatic disease in ipsilateral peribronchial and/or ipsilateral hilar lymph nodes and intrapulmonary lymph nodes, including direct extension (N1),¹⁷ however, this review focuses on N0 disease. When assessing a patient for optimal treatment of NSCLC, it is important to take into account patient fitness, comorbidities and wishes.¹⁸ In the UK, staging is completed with fluorine-18 fludeoxyglucose positron emission tomography CT (PET-CT) to exclude metastatic disease, in addition endoscopic or surgical assessment of equivocal mediastinal nodes is essential.¹⁹

The radical treatment options include surgical resection and radical RT, including SABR and conventional fractionated external beam RT (EBRT). Other ablative methods such as radiofrequency ablation are also used.^20,21 Surgical resection is the current treatment of choice for ES-NSCLC.²⁰ Surgery has the obvious advantage of pathological confirmation and further mediastinal staging; however, the increased morbidity and mortality associated with surgical intervention makes alternative treatments desirable. The use of minimally invasive video-assisted thoracoscopic surgery has the potential to reduce morbidity/mortality in patients at higher risk of surgical morbidity for both lobectomy and sublobar resections.²²

For those patients who are too frail for surgical intervention, non-surgical management is warranted because the survival of untreated Stage I NSCLC is poor and population-based analysis over time has shown a trend for improved survival with increased active treatment rates.^23,24 In patients who are advised or decide against surgery, the treatment options include non-invasive RT, invasive ablative therapies, systemic therapy or best-supportive/palliative care. There is a paucity of randomized evidence examining these treatments, and most publications are retrospective heterogeneous case series. Despite this, SABR has been adopted widely across the world. In the UK, the delivery of both conventional EBRT and SABR for ES lung cancer has recently been audited by McAleese et al.²⁵ They found high levels of conformance with quality standards, including respiratory compensation, and onset verification with cone beam image-guided RT in the UK.

Randomized evidence for SABR vs more conventionally fractionated RT regimes is emerging. The Phase II Stereotactic Precision And Conventional Radiotherapy Evaluation (SPACE) study (http://clinicaltrial.gov identifier NCT01920789) has been presented in abstract and randomized 102 patients to SABR (66 Gy in 3 fractions and 45 Gy at the periphery) or conventional RT (70 Gy in 35 fractions). No difference in local control was discovered.²⁶ They found that conventional treatment was associated with a higher risk of grade 1–2 toxicity (oesophagitis and pneumonitis) and with worse treatment compliance; in addition, the SABR group may have had more unfavourable prognosticating factors (T₂ tumours 47% vs 25%; male 45% vs 36%). The Trans Tasman Radiation Oncology Group (TROG) CHISEL Phase III trial comparing SABR to conventional RT is currently recruiting patients and is close to completing recruitment.²⁷

What is SABR and how is it delivered?

SABR, a synonym for SBRT, has been an “external beam radiation therapy method used to very precisely deliver a high dose of radiation to an extracranial target within the body, using either a single dose or a small number of fractions”.²⁸ Stereotaxis was initially defined when treating brain lesions, using an external three-dimensional frame directly fixed to the skull and allowing accurate image-guided surgery using Cartesian coordinates. This allowed development of stereotactic radiosurgery, or Gamma Knife^®, using multiple beamlets to ablate metastatic deposits within the brain. Radiosurgery is defined as ablative RT dose in a single fraction, whereas fractionated courses were initially defined as SBRT, and now SABR.

For lung SABR, a rigid frame is not practicable therefore for the initial development of SABR in Karolinska Institute, Sweden, a stereotactic body frame was developed.^29–31 With the advent of sophisticated image-guided RT, the use of stereotactic body frame for target localization has waned, although it is still used for immobilization and abdominal compression to restrict breathing motion. With the small margins used in SABR, target identification and delineation is crucial. To confidently plan lung SABR, planning margins should be individualized to each patient. The most common method to achieve this is a four-dimensional CT planning scan, where the tumour motion is assessed through multiple phases of breathing. A standard free-breathing three-dimensional CT alone can produce movement artefact, especially with small mobile lung cancers. Underberg et al³² and Bradley et al³³ have compared helical, average-intensity-projection and maximum-intensity-projection four-dimensional CT for use in lung SABR. The authors found maximum-intensity projection to be superior in defining full motion gross tumour volume (GTV). The motion GTV can be checked against the other breathing phases, and a composite GTV delineated. In SABR, there is no margin given for clinical target volume, therefore the composite GTV/motion GTV is equivalent to the internal target volume. A margin is then applied to allow for setup errors (usually 5 mm) to create the planning target volume (PTV).³⁴

Multiple treatment systems have been used to treat patients undergoing SABR. These include the robotic linac (CyberKnife^®, Accuray, Sunnyvale, CA), conventional gantry-based linac (Elekta^®, Stockholm, Sweden. Varian, Salt Lake City, UT), TomoTherapy^® (Accuray) and MRI-cobalt linac (ViewRay^™, Oakwood Village, OH).^35,36 The dose delivered is key to the high local control rates seen with SABR in lung cancer. Previous studies have demonstrated that a biologically equivalent dose (BED) for tissue with an alpha/beta ratio of 10 of >100 gives the best chance of local tumour control.^37,38 Koshy et al³⁹ examined the US National Cancer Database investigating whether even higher doses are more effective. Of 498 patients who underwent SABR for NSCLC between 2003 and 2006, the five most common doses were 60 Gy/3# (34%), 48 Gy/4# (16%), 54 Gy/3# (10%), 45 Gy/3# (10%) and 48 Gy/3# (4%) which represent calculated BEDs of 180, 105.6, 151.2, 112.5 and 124.8 Gy, respectively. They found a statistically significant improvement in OS in patients with T₂ tumours and a calculated BED of >150 Gy.

Lung SABR: reported outcomes and toxicity

There are an increasing number of studies reporting patient outcomes, dosimetry and toxicity following treatment of ES lung cancer with SABR. Table 1 illustrates the cross section of articles in the current literature. These articles report SABR outcomes including the details of 4570 patients in total. The group is heterogeneous and have had a variable number of different dose-fractionation treatments especially in the earlier articles. Gross analysis of the combined reported OS which was reported in 15 articles was 38.44 months, with an average follow-up of 29.4 months (although in three articles, the median survival was not reached at the time of analysis). The OS and local control rates are listed in Table 1. The OS rates in the SABR cohort should be viewed in the context that the vast majority of patients were medically inoperable.

Table 1.

Summary of the literature reporting stereotactic ablative body radiotherapy outcomes listed below for early-stage lung cancer reporting overall survival (OS) and local control (LC)

Survival reported	No. studies reporting	Average [mean (%), range (%)]
Median survival (months)	15	38.44 (F/u av 29.4) (27.3–57)
1 year OS	15	87 (78–100)
2 year OS	18	82.9 (48–96)
3 year OS	22	59.6 (32–95)
4–5 year OS	8	39.6 (17–83)
1 year LC	10	92.7 (64.7–100)
2 year LC	16	89.9 (77.4–98.5)
3 year LC	19	86.7 (40–97.6)
4–5 year LC	5	89.6 (83–95)

F/u av, follow-up median average.

Nagata et al⁴⁰ examined the outcomes of patients undergoing SABR for lung cancer in the Japanese Clinical Oncology Group (JCOG) 0403 study. They stratified 104 inoperable patients and 65 operable patients who underwent SABR for T1N0M0 pathologically proven NSCLC. The 3-year OS for inoperable patients was expected at 50% and the actuarial value found was 59.9% [95% confidence interval (CI) 49.6–68.8%]. As expected, the operable patient group was younger and had less comorbidity and a higher 3-year OS of 76.5% (95% CI 64–85.1%). As with other groups, the improvement in OS of patients with ES lung cancer compared with conventional fractionated RT puts forward the case for SABR as the first-line treatment for inoperable ES-NSCLC and as a potential alternative to surgery in this group (Table 2).

Table 2.

Outcomes following stereotactic ablative body radiotherapy (SABR) in the primary treatment of early-stage non-small-cell lung cancer

Author	Type of study	Technique/dose	No. Pts	Med follow-up (months)	OS	Local control	Med survival
Haseltine et al 201641	Retrospective analysis of central SABR	IMRT 60 Gy/5# 50 Gy/5# 48 Gy/4# 45 Gy/5#	108	22.7 (range 1.5–71.5)	63.9% at 2 years	77.4% at 2 years	Not reported
Shaverdian et al 201642	Retrospective analysis	IMRT 54 Gy/3# 50 Gy/4#	118	28.9	77% at 3 years	97% at 3 years	Not reported
Navarro-Martin et al 201643	Phase II trial	IMRT/VMAT 54 Gy/3#	42	42 (range 1.44–66)	92% at 1 year 75% at 2 years 66% at 3 years	92% at 3 years	57 months
Chiang et al 201644	Matched cohort comparison	SABR vs AH 50–52 Gy/4# 50 Gy/5#	192	32.5 (range 0.3–62.6)	72.4% at 3 years	89.3% at 3 years	Not reached, at 60 months
Chang et al 201545	Meta-analysis of STARs and ROSEL Phase III trials	Arc/IMRT 54 Gy/3# 50 Gy/4# Cent. or 60 Gy/5#	31	40.2 (IQR 23–40.7)	100% at 1 year (95% CI 100–100%) 95% at 3 years (95% CI 85–100%)	96% at 3 years	Not reached at 60 months
Murray et al 201546	Prospective analysis	Arc/IMRT 54 Gy/3# 55 Gy/5# 60 Gy/8#	273	19.63	78% at 1 year 54.9% at 2 years 38.6% at 3 years	95.7% at 3 years	27.3 months (95% CI 22.3–33.2)
Nagata et al 201540	Prospective analysis of inop and op patients	Arc/IMRT 48 Gy/4#	164 100 inop 64 op	47 (range 39–57)	Inop—at 3 years–59.9% (95% CI 49.6–68.8%) Op—at 3 years–76.5% (95% CI 64–85.1%)	Inop—at 3 years–87.3% Op—at 3 years–85.4%	Not reported
Gillespie et al 201547	Review of single centre published data	IMRT 48 GY/4# 50 GY/5#	320	23.75 (range 17–28)	64.25 2 years	95% at 2 years	Not reported
Chang et al 201248	Retrospective analysis	IMRT	130	26 (range 6–78)	93% 1 year 78.2% 2 years 65.3% 3 years	98.5% 2 years	60 55 inop Not reached for borderline
Abelson et al 201249	Retrospective review	Arc/IMRT 60 Gy/3# 54 Gy/3# 50 Gy/4# 40 Gy/4#	54	13.2 (range 3.2–60.5)	87% at 1 year 73% at 2 years	100% 1 year 87% 2 years	Not reached
Taremi et al 201250	Prospective analysis	IMRT 48 Gy/4# 54/3# 60 Gy/3#	108	19.1 (range 1–55.7)	84% at 1 year (95% CI 76–90%) 30% at 4 years (95% CI 15–46%)	92% at 1 year (95% CI 86–97%) 89% at 4 years (95% CI 81–965)	32 months
Senthi et al 201251	Retrospective analysis	IMRT/VMAT 54–60 Gy/3# 55–60 Gy/5# 60 Gy/8#	676	32.9 months (IQR 14.9–50.9)		4.9% (2 years LRR) (95% CI 2.7–7.1) 10.5% (5 years LRR) (95% CI 6·4–14·6)	40.7 months (95% CI 34.7–46.8)
Palma et al 201252	Prospective analysis	Arc/IMRT 60 Gy/3#–8# PB 54 Gy/3#, 55 Gy/5#, or 60 Gy/8# with AAA algorithm	176	21	79% at 1 year 47% at 3 years 25% at 5 years	89% at 3 years	32 months
Bongers et al 201153	Prospective analysis	IMRT 60 Gy/3# 60 Gy/5# 60 Gy/8#	500	33 (13–86)	53.1% at 3 years	90.4% at 3 years	–
Bral et al 201154	Phase II	IMRT 60 Gy/3# 60 Gy/4#	40	19.1 (range 5–33)	52% at 2 years (SE- 11%)	97% at 1 year (SE 5%) 84% at 2 years (SE 9%)	–
Andratschke et al 201155	Retrospective analysis	Arc/IMRT 30–45 Gy/3# 30–45 Gy/4–5#	92	21 (range 3–87)	38% at 3 years 17% at 5 years	89% at 1 year 83% at 3 years 83% at 5 years	29 months
Haasbeek et al 201156	Prospective analysis of central lesions with SBAR	IMRT	63	35	85.7% at 1 year 69% at 2 years 49.5% at 5 years	94.8% at 1 year 92.6% at 2 years 92.6% at 5 years	–
Haasbeek et al 201057	Prospective analysis (>75 years)’	IMRT 60 Gy/3# 60 Gy/5# 60 Gy/8#	193	12.6 (range 3–52)	85.7% at 1year 45.1% at 3years	89.3% at 3 years	32.5 months
Timmermann et al 201058	Phase II	IMRT 60 Gy/3# (54/3# after heterogeneity correction)	55	38.7 (range 4.8–49.9)	55.8% at 3 years (95% CI 41.6–67.9)	97.6% at 3 years	48.1 months (95% CI 29.6—not reached)
Hamamoto et al 201059	Prospective analysis primary vs metastatic lung lesions	IMRT 48 Gy/4#	52	14 (range 3–34)	96% 1 year and 2 years	91% at 1 year 88% at 2 years	–
Grills et al 201060	Retrospective comparison study	IMRT 48 Gy/4# 60 Gy/5#	58	30	72% at 30 months	96% at 30 months	–
Ricardi et al 201061	Phase II	IMRT 45 Gy/3#	62	28 (range 9–60.7)	69.2% at 2 years 57.1% at 3 years	92.7% at 2 years 87.8% at 3 years	–
Bradley et al 201033	Prospective analysis	IMRT Med 54 Gy/3# 45 Gy/5#	91	18	80% at 1 year 72% at 2 years	2 years 86%	38 months
Crabtree et al 201062	Retrospective analysis	SBRT 54 Gy/3# to 80 or 85% isodose	76	19	32% at 3 years	89% at 3 years	–
Baumann et al 200963	Phase II	IMRT 45 Gy/3#—67% isodose to PTV ∼66 Gy/3#	60	35 (range 4–47)	65% at 2 years	92% at 2 years	40.6 months
Fakiris et al 200964	Phase II	IMRT 60 (T1)-66 (T2) Gy/3# to 80% isodose to >95% PTV	70	50.2 (range 1.4–64.8)	42.7% at 3 years (95% CI 31.1–54.3%)	88.1% at 3 years	32.4 months
Lagerwaard et al 200865	Prospective analysis	IMRT 60 Gy/3# T1 (43%) 60 Gy/5# chest wall/T2 (45%) 60 Gy/8# Central (12%) to 80% PTV isodose	206	12 (range 3–55)	64% at 2 years	98% at 1 year 93% at 2 years	34 months
Chen et al 200836	Retrospective analysis	Body Gamma-Knife® 48–64 Gy in 3.6–8 Gy# daily. BED 71.8–115.2 GY	65	47 months (range 25–76)	92.3% at 1 year 80% at 2 years 57.3% at 3 years 35.1% at 5 years	–	39 months
Onishi et al 200737	Retrospective analysis	Arc/IMRT 1–14# 4.4–35 Gy/# 30–84 Gy Med BED 111 Gy (range 57.6–180 Gy)	257	38 (range 2–128)	56% at 3 years (95% CI 50.2–63.5%) 47.2% at 5 years (95% CI 38.7–53.5)	86% (3 years)	–
Koto et al 200766	Phase II	Arc/IMRT 45 Gy/3# 60 Gy/8#	31	32 (range 4–87)	71.7% at 3 years	T1—77.9% at 3 years T2—40% at 3 years	–
Hoyer et al 200667	Prospective analysis	IMRT 45 Gy/3# to PTV 67% isodose	40	28.8	48% at 2 years	85% at 2 years	-
Nyman et al 200668	Prospective analysis	IMRT 45 Gy/3# to 100% isodose at PTV	45	43 (range 24–74)	80% at 1 year 55% at 3 years 30% at 5 years		39 months
Nagata et al 200569	Phase I/II	Arc/IMRT 48 Gy/4# to isocentre	45	Ia—30 (range 6–71) Ib—22 (range 6–61)	93% at 1 year 72% at 2 years 83% at 3 years 83% at 5 years	100% at 1 year 95% at 5 years	Not reached
McGarry et al 200570	Phase I study—dose escalation	IMRT 24 Gy/3# to 80% isodose at PTV escalated to 72 Gy/3#	47	15.2	64% at 3 years	64.7% at 1 year	–
Zimmermann et al 200571	Prospective analysis	Arc/IMRT 24–37.5 Gy/3–5# 69% 37.5 Gy/3# to 60% isodose	30	18 (range 6–38)	80% at 1 year 75% at 2 years	100% at 1 year 83% at 2 years	–

AAA, Analytical Anisotropic Algorithm; AH, accelerated hypofractionation; Arc, Arc therapy; BED, biologically equivalent dose; CI, confidence interval; IMRT, intensity-modulated radiotherapy; inop, inoperable; IQR, interquartile range; LRR, local relapse rate; Med, median; op, operable; OS, overall survival; PB, pencil beam algorithm; Pts, patients; PTV, planning target volume; VMAT, volumetric modulated arc therapy; SE, Standard Error.

The toxicity of SABR appears to be well documented within the articles below (Table 3); 30 articles reported toxicity outcomes. Gross grade 5 toxicity appears to be rare, especially in peripheral tumours, with 11 patient deaths reported in these articles, and possibly attributable to SABR, 1 patient was thought to have underlying interstitial lung disease.⁴³ The final study patient reported as grade 5 toxicity had no data regarding the cause of death, and the patient died within 30 days of treatment.⁴⁶

Table 3.

Lung stereotactic ablative body radiotherapy toxicity

Study	Scale	Gd 1	Number	Gd 2	Number	Gd 3	Gd 4+
Navarro–Martin et al 201643	CTCAE V3	89% Gd 1 acute 100% chronic Gd 1 Pneumonitis (68%)	34 42 26	16% acute 2 pleuritis 1 dyspnoea 1 cough 1 dermatitis 1 pneumonitis 18% chronic Dyspnoea Pneumonitis Cough Dermatitis Anorexia	6 7	1 Gd 3 pneumonitis (5%) 1 chronic AF and dyspnoea	1 Gd 5 (developed late from Gd 3 toxicity −2.65)—Pt thought to have subclinical ILD
Heseltine et al 201641	Not reported					13 Gd 3+ events (12%) Significantly more Gd 3+ events in those pts with GTV <1 cm from PBT	4 potential Gd 5 events 2 pneumonia 2 pulmonary haemorrhage (both previous exposure to bevacizumab)
Shaverdian et al 201642	CTCAE v3	88% reported no ≥ Gd 2 toxicity		≥Gd 2 toxicity in 12% Pneumonitis (9%) Rib# (3%)	8 3	2 pneumonitis Gd 3	No Gd 3+ toxicity reported
Chang et al 201545	NCI-CTC V3					SABR 3 pts (10%) 2 cough/dyspnoea 3 Chest wall pain 1 fatigue and rib# Surgery 12 pts (44%)4 Gd 3 dyspnoea 2 Gd 3 pneumonia 4 Gd 3 chest pain	No Gd 4 or above in SABR arm Sx arm—1 Gd 5 death from surgical complications 12 Gd 3–4 Surgery 12 pts (44%) 1 Gd 4 dyspnoea
Murray et al 201546	NCI-CTC V4					Acute—1 Gd 3 dyspnoea 4 Gd 3 fatigue Delayed (>6 weeks, <12 weeks) 1 Gd 3 cough 10 Gd 3 dyspnoea 3 Gd 3 fatigue 1 Gd 4 fatigue Late (>12 weeks) 8 Gd 3 dyspnoea 1 Gd 3 pneumonitis	No grade 4 toxicity during tx 1 pt died within 30 days of SABR, but CoD unknown; had been asymptomatic (potential Gd 5) 1 patient died at 12 weeks from COPD
Nagata et al 201540	CTCAE V3					Inop—Gd 3 11 (10.6%) Dyspnoea 10 Hypoxia 8 Pneumonitis 8 Chest pain 2 Cough 1 Op—Gd 3 4 (6.2%) Dyspnoea 2 Hypoxia 1 Pneumonitis 1	No Gd 5 toxicity reported Gd 4 inop—2 pts 2 Dyspnoea 1 Hypoxia 1 pneumonitis No Gd 4 toxicity in operable pts.
Gillespie et al 201547	CTCAE V2			Over 3 studies Acute Gd 2 range 8.7–19%		Gd 3 toxicity range 0–2%	No Gd 4 or 5 toxicity reported
Palma et al 201252	CTCAE V3	Common Gd 1/2–55% Fatigue Dyspnoea/cough Chest wall pain Nausea/appetite	55 24 18 10			Acute—1 pt developed Gd 3 pneumonitis late (>6 weeks)—2 pts Gd 3 pneumonitis 2 pts rib# 1 pt haemoptysis	No Gd 4 or 5 toxicity reported
Chang et al 201248	CTCAE v3	Oesophagitis 1.5%	2	Chest wall pain (8.5%) Pneumonitis (9.2%)	12 15	1 Chest wall pain Gd 3 (0.8%) 8 Dermatitis Gd 2/3 (6.2%) 3 Pneumonitis Gd 3 (2.3%)	No reported Gd 4 or 5 toxicity
Taremi et al 201250	CTCAE V3	29% denied any acute toxicity 31% denied late toxicity		Late—rib# (14.8%) Mostly asymptomatic	16	Acute Gd 3–4 (3.7%) 1 Fatigue 2 Dyspnoea 1 Chest wall pain Late Gd 3 (>3 months)—6 pts 3 rib# 2 Dyspnoea 1 Pneumonia	No grade 4 or 5 toxicity reported
Bongers et al 201153	CTCAE v4			Chest wall pain Gd 1–2 (5.4%)	27	5 Chest wall pain Gd 3 (1%) 8 rib# (1.6%)	Not reported
Bral et al 201154	NCI-CTC V3/RTOG			Dyspnoea (12%)	2	Gd3—8 pts (20%) Pneumonitis 2 (12%) Cough 2 (12%) Stenosis 1 (10%)
Andratschke et al 201155	NCI-CTC V3	Radiological pneumonitis (34.8%)	32	Pneumonitis (13%)	12	Pneumonitis 2 (2.2%) Dyspnoea 7 (7.3%)	Gd 4 dyspnoea—4 (4.1%)
Haasbeek et al 201156	CTCAE v3	Gd 0 or 1 acute Gd 0 or 1 late	56 50	Gd 2 acute (10%) Grade 2 late (14%)	6 9	1 Gd 3 acute (2%) 4 Gd 3 late (6%)	No Gd 4 or 5 toxicity reported
Haasbeek et al 201057	RTOG acute toxicity RTOG/EORTC late toxicity	116 patients reported no acute toxicity		32% fatigue		Late—4 pts Gd 3–4 pneumonitis (2.1%) 3 pts rib# (1.6%) Chest wall pain 5 pts (2.6%) 3 pts non-malignant effusion	No grade 5 toxicity reported
Timmerman et al 201058	NCI-CTC V3					7 pts reported Gd 3 (12.7%) 2 reduced FEV1 2 Hypoxia 2 Pneumonitis 3 PFTS altered 2 pts reported Gd 4 (3.6%) 1 hypocalcaemia 1 Pulmonary	No Gd 5 toxicity reported
Grills et al 201060	NCI-CTC V3	Chest wall pain—9 (9.8%)		9% pneumonitis 6% rib# 4% chronic dyspnoea	Rib# 3 (3.3%)	2% pneumonitis Dyspnoea Myositis
Ricardi et al 201061	RTOG acute and late	Fatigue 15% Skin erythema 10% Dyspnoea and cough 10% Chest pain 7%		3 pts had chronic pain (4.8%)		2 patients had Gd 3 radiation pneumonitis (3.2%)
Bradley et al 201033	CTCAE V3			3 pts Gd 2 pneumonitis 4 pts rib pain or #		1 pt developed brachial plexopathy
Baumann et al 200963	NCI-CTC V2					16 (28%)	1 Gd 4 dyspnoea No Gd 5
Fakiris et al 200964	NCI-CTC V2					Gd 3 pneumonia 1 Pleural effusion 2 Pts 2 Gd 4 apnoea	Gd 5 (possible) Pneumonia 3 Haemoptysis 1 Resp failure 1
Lagerwaard et al 200865	–	51 % No toxicity Fatigue 31% Chest wall pain 12% Nausea 9% Dyspnoea 6% Cough 6%				Gd 3–6 pts (3%)—pneumonitis 4 rib fractures 3 chronic pain
Chen et al 200836	NCI-CTC/RTOG	28% had no reported toxicity 7 pts Gd 1 bone marrow toxicity		3.1% Gd 2 toxicity Bone marrow		No Gd 3 toxicity
Onishi et al 200737	NCI-CTC v2	Cough/dyspnoea Gd 1	28 (10.9%)	Gd 2 pneumonitis—14 pts (5.4%)		Gd 3–6 pts pneumonitis requiring O2 3 oesophagitis 3 pts skin reaction (1.2%) 4 pts rib# (1.6%)
Koto et al 200766	NCI-CTC v3	24 pts Gd 1 pneumonitis		Gd 2 pneumonitis 3 pts		Gd 3 pneumonitis in 1 pt 1 pt lobar collapse
Hoyer et al 200667	WHO PS and Tox	8 Gd 1–2 dyspnoea 6 Gd 1–2 nausea 14 Gd 1–2 pain				3 Gd 3 dyspnoea 2 Gd 3 pain 1 Gd 4 dyspnoea	1 death due to pneumonia felt unrelated to treatment
Nyman et al 200668	RTOG acute/late	9 pts skin reaction (20%) oesophagitis in 4 pts (9%) 4 pts had transient chest pain 4 pts had pneumonia/infections 3 pts had cough	51% did not have any toxicity			2 pts had rib# 3 pts atelectasis	No pneumonitis reported
Nagata et al 200569	NCI-CTC v2	10 pts Gd 1 cough or malaise (22%)		2 pts Gd pneumonitis		No Gd 3 toxicity reported
McGarry et al 200570	NCI-CTC	4 Gd 2 pts—bronchitis, pericardial effusion, pneumonitis, distant pneumonia				6 Gd 3 pts—2 pneumonitis, tracheal necrosis, hypoxia, dermatitis, pericardial effusion	1 pt Gd 4 pneumonitis 72 Gy/3#
Zimmermann et al 200571	CTC-RTOG RTOG late	Gd 1–2—fatigue 20 % 1 pt pain, fever, pneumonia. 8 pts Gd 1, 5 pts Gd 2 pneumonitis 1 nausea and dermatitis Gd 1				1 Gd 3 pneumonitis (3%)

AF, atrial fibrillation; CoD, cause of death; COPD, Chronic Obstructive Pulmonary Disease; CTC, common toxicity criteria; FEV1, forced expiratory volume in 1 second; Gd, grade; ILD, interstitial lung disease; inop, inoperable; NCI, national cancer institute; op, operable; PBT, proximal bronchial tree; PFTs, pulmonary function tests; Pt, patient; resp, respiratory; rib#, rib fracture; RTOG, Radiation Therapy Oncology Group; Sx, surgical; tx, treatment; WHO, World Health Organisation.

Grade 3 and 4 toxicity reports range between 2.7% and 27% of patients. Pneumonitis, dyspnoea, chest pain and pneumonia being the most commonly reported toxicity. These were usually self-limiting, but one patient who died had a grade 4 pneumonitis initially.⁴³

Grade 1 and 2 toxicities were very common and reported in up to 100% of cases, especially grades 1–2 fatigue. Many articles have only reported grade 3 and above toxicity due to the self-limiting nature of grades 1–2 toxicities.

SABR vs surgery in NSCLC

Chang et al⁴⁵ recently published a pooled analysis of two randomized trials in the USA and Netherlands (ROSEL and STARS) comparing SABR with lobectomy. Although small numbers, with 31 patients randomized to SABR and 27 patients to surgery, the results and conclusions drawn raised interest on publication. The authors advocate SABR for ES lung cancer even in the operable group of patients. They report an estimated 3-year OS of 95% (95% CI 85–100) in the SABR group compared with 79% (95% CI 64–97%) in the surgical arm, with a hazard ratio of 0.14 (95% CI 0.017–1.19), statistically significant log rank p = 0.037. There was no significant difference in recurrence-free survival. This report is the first of randomized data between the two treatment options, but interpretation of the data is severely limited by failure to complete recruitment in either study and by small patient numbers. Furthermore, the apparent outlier of surgical morbidity and perioperative mortality of 1 patient in a small cohort of 27 may be misleading. The surgical arm also had a significant proportion of three (11%) patients who did not undergo a lobectomy, confounding the results further.

The other studies listed in Table 4 are attempts at matched pair retrospective analyses. The larger studies have shown improved OS in the surgery groups; however, these are usually confounded by older patients with more comorbidities in the SABR arms due to inoperability.^72–76

Table 4.

Studies comparing outcomes between surgery and stereotactic ablative body radiotherapy (SABR)

Study	Study type	Patients	Follow-up (median)	Notes	Outcome
Eba et al 201672	Propensity score analysis of two separate trials	21 patients SABR 21 patients lobectomy		Older SABR group Median age 79 vs 62 years	HR 9 (95% CI 1.14–71.04) favouring lobectomy HR 1.19 (CI 0.38–3.73) in younger group (<75)
Chang et al 201545	Pooled analysis of two randomized controlled trials (STARS and ROSEL)	31 SABR 27 surgery (19 open vs 5 VATS)	SABR—40.2 months IQR 23–47.3 Surgery—35.4 months IQR 18.9–40.7	1 patient node positive 1 patient surgery aborted due to progression	Pooled estimated 1- and 3-year OS was 100% (95% CI 100–100) and 95% (95% CI 85–100) in the SABR group, and 88% (95% CI 77–100) and 79% (95% CI 64–97) in the surgical group Log-rank p = 0·037; HR 0·14 (95% CI 0·017–1·190)
Hamaji et al 201573	Retrospective analysis of VATS lobectomy vs SABR	VATS 413 SABR 104	48 months	Propensity score matching	3-, 5- and 10-year OS 80.1%, 68.5% and 61.65 with VATS vs 52.7%, 37.3% and 20.7% with SABR
Zheng et al 201474	Meta-analysis	40 SABR studies, 4850 patients with Stage 1 disease 23 Surgical 7071 patients	SABR 28 months Surgery 37 months		Mean 1-, 3- and 5-year OS with SBRT were 83.4%, 56.6% and 41.2% compared with 92.5%, 77.9% and 66.1% with lobectomy and 93.2%, 80.7% and 71.7% with limited lung resections
Shirvani et al 201475	SEER database review >66 years	9093 pts reviewed Lobectomy 7215 (79.4%) SLR 1496 (16.5%) SABR 389 (4.2%)	Unadjusted 90 days mortality and 3 years survival were analysed	Unadjusted 90-day mortality was the highest for lobectomy (4.0%) followed by sublobar resection (3.7%, p = 0.79) and SABR (1.3%, p = 0.008). At 3 years, unadjusted mortality was the lowest for lobectomy (25.0%), followed by sublobar resection (35.3%, p < 0.001) and SABR (45.1%, p < 0.001)	Compared with lobectomy, sublobar resection was associated with worse OS (HR 1.32; 95% CI 1.20–1.44; p < 0.001) and worse lung-cancer-specific survival (HR 1.50; 95% CI 1.29–1.75; p < 0.001)
Zhang et al 201476	Matched pair meta-analysis	864 matched patients from 6 studies 432 SABR 432 surgery	1- and 3-year OS OR 1.31 (0.90–1.91), and 1.82 (1.38–2.40)		3-year OS superior in surgery vs SABR. But CSS, DFS, LC and DC was not significant
Verstegen et al 201377	Retrospective matched analysis	64 SABR 64 VATS lobectomy	SABR—30 months VATS—16 months	Unsuspected nodal disease in 12 (18.8%)	Locoregional recurrence 1 and 3 years 96.8% and 93.3% vs 86.9% and 82.6%, respectively, p = 0.04 No significant difference in OS
Solda et al 20133	Systematic review and historical cohort comparison	3641 pts form 45 reports 2038 stage Ia pts from IASLC database			2-year OS 70% in SABR group (range 35–96%, 95% CI 67–72%) 67% in surgery (95% CI 66–70%)
Grills et al 201060	Prospective comparison SBRT vs wedge resection	58 SABR 69 Wedge resection	Potential F/U 2.5 years	N+ve patients excluded. 21% SABR vs 71% wedge underwent mediastinal LN sampling	Improved LR in SBRT 4% vs 20% for wedge (p = 0.07) at 30 months. OS superior with wedge 87% vs 72% SBRT; p = 0.01. CSS 94% wedge vs 93% SBRT; p = 0.53. No difference in regional recurrence, locoregional recurrence, distant mets, or freedom from failure
Crabtree et al 201062	Retrospective analysis	462 surgery 76 SBRT	Sx—31 months SBRT—19 months	80% bx proven in SBRT	OS 5 years 55% in Sx group Sx OS 3 years—68% SBRT OS 3 years—32% CSS—82% similar LC (3 years)—94% sx vs 89% SBRT| DFS (3 years)—77% vs 86%

Bx, Biopsy; CI, confidence interval; CSS, cause-specific survival; DC; disease control; DFS, Disease-free survival; HR, hazards ratio; IASLC, International association for the Study of Lung Cancer; IQR, interquartile range; LC, Local control; LR, local relapse; mets, metastases; N+ve, node positive; OR, odds ratio; OS, overall survival; SLR, sublobar resection; Sx, surgical group; VATS, video-assisted thoracoscopic surgery.

van den Berg et al⁷⁸ report outcomes of 340 patients treated with surgery (n = 143) or SABR (n = 197) for ES lung cancer. They found no difference in OS, local recurrence or distant recurrence. However, they did find that locoregional recurrence was significantly more frequent with SABR (adjusted sub-hazard ratio 2.51; 95% CI: 1.10–5.70; p = 0.028). Verstegen et al⁷⁷ matched two groups of patients with ES lung cancer using propensity scores based on patient demographics such as cTNM stage, age, gender, Charlson comorbidity score, lung function and performance score. They retrospectively matched 64 patients who received SABR to the same number of patients who underwent a video-assisted thoracoscopic surgery lobectomy. They found superior local regional control, but no difference in OS. The difference between the two publications may be the definitions of local and locoregional control, especially regional lymph node recurrence. van den Berg et al defined any mediastinal lymph node recurrence as locoregional failure, whereas Verstegen et al only defined ipsilateral lymph node recurrence as locoregional.

Grills et al⁶⁰ have an earlier report of outcomes between lung SABR or wedge resection. In the non-randomized group comparing SABR and wedge resection at 30 months, they found OS was improved in the surgical group (87% vs 72%, p = 0.01), however, cause-specific survival was equal. SABR was superior for local (5% vs 24%, p = 0.05), locoregional (5% vs 29%, p = 0.03) and regional relapse (0% vs 18%, p = 0.07).

The data supporting the use of SABR in ES lung cancer appear to be strengthening. Nagata et al⁴⁰ reported their outcomes of SABR in operable patients, which were improved compared with inoperable patients.

Further robust, appropriately powered trials are required to accurately compare the two treatment groups. In the UK, the SABRTooth Trial has recently opened.⁷⁹ This is a feasibility study to randomize “high-risk” operable patients with ES-NSCLC to surgery or SABR, depending on their perioperative risk. If recruitment is successful, it will extend into a Phase II study. In addition, the “Stablemates” trial (chief investigator Robert Timmerman) in the USA is also directly randomizing high-risk patients to SABR or sublobar resection,⁸⁰ and the Veterans Affairs lung cancer surgery or stereotactic RT study (chief investigator D Moghankaki) is due to open in the very near future.

SABR for central lesions in lung cancer

Central ES lung cancer tumours are defined by their proximity, usually within 2 cm, to central organs at risk, such as the proximal bronchial tree, oesophagus, great vessels, spinal cord or heart/pericardium. Earlier studies identified the increased risk of toxicity in treating central tumours, including 6 of 70 patients deemed to have suffered grade 5 toxicity due to SABR and a further early case report of proximal airway necrosis and haemoptysis.^16,81 The fact that these patients are usually medically inoperable may increase the risk of complications following central SABR.

Subsequently, the “no-fly-zone” planning organ-at-risk volume of 2 cm around the proximal airways has been adopted in SABR treatments for lung cancer. Patients with central tumours were excluded from the Radiation Therapy Oncology Group 0236 trial⁵⁸ which demonstrated the safety of 54 Gy/3# in peripheral ES-NSCLC.

However, the use of a risk-adaptive SABR technique to control the total BED to OARs has been demonstrated to be relatively safe.^56,82

Further studies have reported the use of SABR to treat tumours in ultracentral positions, defined as the GTV in direct contact with the proximal airway.^56,83 Chaudhuri et al⁸³ compared 68 patients with ES lung cancer suitable for SABR (50 Gy in 4 or 5#), 34 patients with peripheral tumours and 34 with central tumours, and found no significant difference in OS, primary tumour failure or local failure between the two groups (median OS 38.1 months, 2-year OS 73.8%). They reviewed six patients with “ultracentral” NSCLC tumours and found that the 2-year OS was 80.0% for patients with ultracentral NSCLC compared with 63.2% for the remaining patients with central NSCLC and 86.6% for patients with peripheral NSCLC (p = 0.62). They report that one patient had grade 4 pneumonitis with a central tumour and two patients had grade 3 chest wall pain with peripheral tumours. No grade 5 toxicity was reported, and they found no statistically significant correlation between toxicity and tumour site.

Tekatli et al^84,85 report the outcomes of 80 patients with tumour PTV within 2 cm of the proximal bronchial tree and 47 patients with more ultracentral tumours, defined as the PTV in contact with the trachea or main bronchi. The patients were treated with 60 Gy/8# (BED for tissue with an alpha/beta ratio of 10 = 105 Gy). Of the 80 patients with central tumours, a total of six (7.5%) patients were considered to have possible (n = 3) or likely (n = 3) treatment-related death. They made a contemporaneous comparison with a cohort of patients with peripheral tumours treated with SABR and found no significant difference in OS between the two groups. Median OS was 38 months (95% CI: 26–50) for central and 44 months (95 CI%: 38–51) for peripheral tumours.

The 47 ultracentral patients were treated with SABR (70 Gy/12#) and reported to have a median OS of 15.9 months and a 3-year OS of 20.1%. Of more concern was the expected increase in toxicity, Grade 3 or higher toxicity was recorded in 38% of patients, with 21% scored as having a “possible” (n = 2) or “likely” (n = 8) treatment-related death between 5.2 and 18.2 months after treatment. Fatal pulmonary haemorrhage was observed in 15% of patients.⁸⁵

The promising reported outcomes and toxicity profile, as well as the prognostic sequelae of not adequately treating ES central lung cancer would suggest that SABR will be offered to these patients with increasing frequency.⁸⁶ However, those patients with tumours involving the trachea or main bronchus will be at a much higher risk of treatment toxicity, including death. Bezjak et al⁸⁷ have recently reported the primary study end point analysis of NRG Oncology/Radiation Therapy Oncology Group 0813 SBRT for central NSCLC dose escalation trial. They report a dose-limiting toxicity rate of 7.2% with the maximum dose of 12 Gy/# over 5 #. However, they report four deaths, one at 10.5 Gy/#, two at 11.5 Gy/# and one at 12 Gy/#. Efficacy data are awaited.

In Europe, the European Organisation for Research and Treatment of Cancer, LungTech study is currently evaluating 60 Gy in eight fractions for central tumours in a prospective Phase 2 study with robust RT quality assurance and robust patient follow-up data.⁸⁸ This should hopefully provide high-quality data on the efficacy and safety of this regimen for central tumours.

SABR as salvage following surgery or radiotherapy

Recurrence rates of lung cancer following radical surgical treatment have been reported between 30% and 75%, increasing with the more advance stages; those patients with the pathological Stage I disease reported 27–38% recurrence, and those patients with Stage IIIa disease having the highest risk.^89,90 The majority of these are distant recurrences, and usually occur within 2 years of surgery. Post-surgical local recurrence can be managed with surgical resection, EBRT or more recently SABR has been used.⁹¹ However, this cohort of patients, who are usually of poor performance status, are at a significant increased risk of complications due to intra-thoracic re-operation.

The detection on local recurrence post-SABR is also challenging. Fibrotic change is common and can be progressive for a number of months following treatment, in addition to causing local anatomical changes around the radiation field.⁹² Current UK guidelines recommend regular post-SABR CT assessment of the thorax,¹⁵ and high risk CT features and PET-CT are being investigated to improve the diagnosis of post-SABR local recurrence.^93,94

Trovo et al⁹⁵ report 17 patients with “in-field” central recurrence of NSCLC using SABR 30 Gy/5–6#, following previous radical RT (50–60 Gy). They reviewed 17 patients, with a local control rate of 86% at 1 year due to 2 patients having local failures. However, Kaplan Meier OS estimates at 1 and 2 years were 59% and 29%, respectively. They also report one fatal occurrence of pneumonitis and one patient who died of fatal haemoptysis 2 months following SABR; in addition, they report 4 patients who developed grade 3 pneumonitis. The treatment of local recurrence following surgical resection is not straightforward. There are increased risks of mortality and toxicity associated with surgical resection as well as SABR.

Trakul et al⁹⁶ report their single institution experience of SABR reirradiation of in-field recurrent lung cancer, previously treated with conventional fractionated RT and SABR; 15 patients with 17 lung tumours compared with a standard group undergoing SABR with no previous RT. Local control at 1 year was 65.5% compared with 92.1% of the control group. 1-year actuarial progression-free survival and OS were 58.2% and 80%, respectively. They had no grade 4+ toxicities in either group; and grade 2+ pneumonitis was not seen in the reirradiated group but reported in 13 patients (11.6%) in the control group. Hearn et al⁹⁷ identified 22 of 436 patients with local recurrence following treatment with SABR for ES lung cancer. 10 of these were deemed suitable for re-treatment SABR, and 30% are disease free since treatment (follow-up range 11.7–43.5 months). Two patients developed distant disease, with four other patients with local relapse at a median 9.9 months.

As patients initially treated with SABR usually have significant comorbidity that precludes surgery, re-treatment SABR may be a viable treatment option in local recurrent disease.

Lung cancer SABR and adjuvant treatments

The use of SABR for ES lung cancer raises the question of benefit of additional treatments. SABR induces local ablation of tumour cells and surrounding normal stroma. Lung cancer is known to have high metastatic potential, and micrometastatic nodal disease is beyond the scope of PET-CT to detect.^98,99 This raises the concern of undetected metastatic deposits which may then affect outcome. Interestingly, there are additional hypotheses postulated around the tumour microenvironmental changes associated with SABR. These include the release of tumour-associated antigens, which can prime the host immune system, thereby potentially encouraging the abscopal effect on distant sites of disease.¹⁰⁰

Conventional treatment such as chemotherapy may be considered, certainly for those tumours with associated increased risk of nodal involvement. There are currently no reported randomized trials comparing SABR for ES lung cancer with or without adjuvant chemotherapy. Chen et al³⁶ report a small analysis of SABR for early lung cancer with a cohort of 65 patients, of which 17 patients then received adjuvant chemotherapy. Albeit a small sample size, they report a trend in improvement of OS of patients receiving adjuvant chemotherapy with three to four cycles of a cisplatin-containing regimen. Although not statistically significant, the 3- and 5- year survival rates for the patients who received SABR plus adjuvant chemotherapy were 80.5% and 46%, compared with 49.6% and 31.5% receiving SABR alone, respectively.

SUMMARY AND OUTLOOK

The use of stereotactic ablative RT for ES lung cancer is currently routine for medically inoperable patients and is well tolerated with high local control rates. There remains equipoise regarding the effectiveness of SABR compared with surgical resection in higher-risk patients and medically operable patients, which is hopefully being addressed by randomized clinical trials in this setting. For patients with ES-NSCLC suitable for surgery, there may be an argument for using SABR upfront and for reserving surgical resection for those patients who relapse locally. Finally, there is growing interest in the potential abscopal effect of SABR and the addition of immune-modulating systemic therapies in combination with SABR, and this may have the potential to eradicate potential micrometastatic deposits within central draining lymph nodes and beyond.