Background
Standard therapy for operable clinical stage I non-small cell lung cancer (NSCLC) is lobectomy, a radical surgery requiring complete removal of the involved lobe plus ipsilateral hilar and mediastinal lymph node dissection.[
1] Tumor recurrence is infrequent following lobectomy and limited to the regional lymph nodes or distant sites. However, despite recent improvements,[
2] lobectomy remains a major operation associated with early mortality,[
3] a decline in pulmonary function[
1] and multiple postoperative morbidities.[
4] Recently, sublobar resection with adequate margins (> 1 cm) has been advocated for marginally operable patients with small peripheral lesions.[
5] Such treatment in appropriately selected patients provides excellent local control without the early mortality and significant decline in lung function associated with lobectomy.
Treatment options for patients with clinical stage I NSCLC who are not surgical candidates are limited. Inferior outcomes with conventionally fractionated radiation approaches have been largely attributed to poor local tumor control.[
6] secondary to historically necessary prolonged treatment courses, which diminish the effectiveness of the therapy.[
7,
8] The development of the stereotactic body frame with abdominal compression to dampen respiratory lung motion has allowed for the treatment of small mobile peripheral lesions with comparatively tight margins (1 cm) on the gross tumor.[
9] Recently completed trials suggest that extremely high biologically effective doses may be delivered safely and rapidly to small peripheral lung tumors with this enhanced accuracy. [
10‐
12] As anticipated, such treatment has resulted in improved early local control rates.[
13]
We began treating small peripheral lung tumors in mid-2004 using the CyberKnife
® frameless robotic radiosurgery system (Accuray Incorporated, Sunnyvale, CA) with Synchrony
® respiratory motion tracking.[
14,
15] The accuracy and flexibility of the system allowed us to deliver dose distributions capable of eradicating both the gross tumor and the microscopic disease radiating from it.[
16,
17] The goal was similar to that of sublobar resection, i.e., to eliminate the tumor with 1 cm or greater margins, and thus the approach was designated radical radiosurgery.[
14] We report preliminary outcomes from 20 consecutive inoperable patients with small, peripheral, clinical stage I NSCLC treated using this novel treatment approach.
Discussion
Stage I NSCLC is curable.[
22] Peripheral lung tumors are more likely to be cured with radiosurgery than central cancers because there is less untreated lymphatic spread[
23] and a more favorable therapeutic window.[
11] However, consistently curing these patients without surgery will require adequate gross tumor doses with finely tailored dose gradients capable of eradicating known relatively radiation-sensitive microscopic tumor extensions, while adequately preserving lung function.
In late 2004, we initiated a radical CyberKnife protocol for medically inoperable patients with small, peripheral, stage I NSCLC. Ultimately, we treated a select group of patients with relatively good performance status and small tumor volumes because we were concerned about CT-guided fiducial placement and treatment-related pulmonary toxicity. Mandatory minimum gross (42 Gy)[
24] and microscopic tumor doses (30 Gy) [
25‐
27] were derived from historical clinical data. Continuous tracking of respiratory tumor motion and highly accurate beam alignment throughout treatment with the CyberKnife allowed us to deliver radical dose distributions with tighter margins on the GTV than historically feasible (5 mm).[
14] Numerous pencil beams were used to produce dose gradients that conform closely to the shape of the target, resulting in theoretically adequate microscopic disease doses extending an ample 1 cm or more from the tumor.[
28] Twenty patients have been treated in 30 months. With a median follow-up of 25 months for surviving patients, the 2-year Kaplan-Meier local control rate was 100%, the 2-year Kaplan-Meier overall survival rate was 87%, and there have been no severe (grade IV) treatment-related complications or early mortalities. Furthermore, despite the comprehensive nature of the treatment, the decrement in lung function remained at acceptable levels. Thus, we conclude that radical radiosurgery with real-time tumor motion tracking using the CyberKnife is a safe and effective treatment option for small peripheral Stage I NSCLC.
Despite promising preliminary results, critical issues concerning the evaluation of treatment efficacy and selection of patients merit additional consideration. Radiosurgery delivered to small peripheral tumors with margins adequate to treat radial microscopic extension (> 1 cm) will damage peritumoral lung tissue. The acute lung injury will often result in asymptomatic focal lung parenchyma fibrosis corresponding with the high-dose radiation volume. [
29‐
32] In the present study all tumors initially responded to treatment with a decrease in volume on CT. However, by the six-month mandatory evaluation, all patients had developed CT evidence of peritumoral radiation fibrosis. At 18 months, two thirds of the tumors were obscured by such fibrosis, making CT tumor assessment difficult. Preliminary analysis of PET imaging suggests that it too is unreliable following radiosurgery.[
33] Although this trial did not require PET/CT imaging, it was routinely completed. Moderate PET activity was often observed following treatment, but could uniformly be attributed to radiation fibrosis rather than tumor progression on serial imaging. Until a dependable noninvasive means to identify early local recurrence in the presence of fibrosis is developed, inoperable patients with fibrosis will require close follow-up which may include biopsy.[
31]
In late 2003, Timmerman et al. published preliminary results of the Indiana University inoperable stage I NSCLC SBRT dose escalation trial.[
12] The primary finding of this study was that 60 Gy in 3 fractions could be safely delivered to inoperable stage I NSCLC patients in less than 2 weeks if relatively tight gross tumor margins (1 cm) were used. However, prior to proceeding with phase II studies, maturing data suggested that critical central thoracic structures tolerated high-dose hypofractionated radiation poorly. Accordingly, the Radiation Therapy Oncology Group (RTOG) protocol 0236 was limited to small (< 5 cm) tumors that lay outside the central bronchial tree, a large area that extends 2 cm from the major airways.[
11] We chose to deliver a range of radical doses (42–60 Gy) to smaller tumors (< 4 cm) with tighter margins (5 mm) than RTOG 0236. Therefore, we felt confident selecting a more inclusive definition of peripheral tumor as those tumors located a sufficient distance from sensitive critical central thoracic structures so that radical radiation doses could be delivered to the PTV while adhering to maximum point dose limits (Table
1). To date, with sufficient follow-up to detect late radiation toxicity, clinically apparent radiation damage to critical central structures has not been observed despite our delivery of a mean dose of 53 Gy to the PTV.
Peripheral thoracic structures such as the lung parenchyma and the chest wall did sustain clinically apparent damage as anticipated. Despite the radical treatment intent, the mean percentage of the total lung volume receiving a minimum of 15 Gy was only 7.3% (range, 2.4% to 11.3%). Nonetheless, acute Grade III radiation pneumonitis occurred in a single patient treated with concurrent gefitinib, an alleged potentiator of radiation-induced lung fibrosis.[
34] Although it is tempting to further limit the percentage of lung receiving greater than 15 Gy in an attempt to decrease the risk of radiation pneumonitis, this should only be considered if the radical treatment intent is maintained.
All evaluated patients were heavy smokers, 80% of whom had stopped smoking in the distant past (> 3 years). Therefore, although the baseline pulmonary function was poor, PFTs just prior to and following the treatment were largely unaffected by recent smoking. The treatment did not adversely affect FEV1 or TLC; however, it did cause significant early reductions in the mean percent predicted DLCO. Predictably, the magnitude of this decline was small, as it was in a recently reported segmentectomy series.[
35] Regrettably, 2 patients with severe COPD, who required continuous supplemental oxygen prior to treatment, died 9 and 18 months after radiosurgery secondary to progressive pulmonary dysfunction. It is unknown whether their deaths were premature and attributable to treatment. Nonetheless, it is possible that a population of inoperable stage I NSCLC patients exists, possibly those whom are oxygen dependent, who may not benefit from radical treatment. Such patients may benefit from a more conservative radiosurgery approach with lower doses [
36,
37] or tighter margins.[
38]
Thoracic surgery uniformly results in permanent chest wall scarring and acute pain, which may persist. Nonetheless, it remains the standard treatment for stage I NSCLC. In contrast, carefully designed early stage lung cancer radiosurgery may result in only trivial radiation skin reactions and transient, mild to moderate chest wall pain. The use of hundreds of lightly weighted beams in this study rather than a few heavily weighted ones has prevented the infrequent but potentially severe skin injuries reported in prior thoracic radiosurgery series conducted using other radiosurgical instruments.[
12,
39] However, mild-to-moderate chest wall pain, typically lasting several weeks, was seen following treatment in the majority of patients with lesions within 5 mm of the pleura. Limiting the dose to the chest wall would likely diminish the severity of this complication; however, this may have led to potentially life threatening local failures and therefore is not recommended at this time given the acceptable and transient nature of the toxicity observed to date.
The current study required CT-guided fiducial implantation prior to treatment. This procedure frequently results in pneumothorax in high-risk patients, often necessitating tube thoracostomy and a short hospital stay.[
37] Fortunately, alternative fiducial placement approaches and a fiducial-free peripheral lung tumor tracking system are now available. [
40‐
42] Ongoing research will determine appropriate patient selection for these new approaches and their efficacy. In the meantime, fiducial tracking and CT-guided fiducial placement will remain our standard approach for small, peripheral lung tumors. The risk of pneumothorax is justified by the optimal intrafraction targeting verification made possible by properly placed fiducials. We have found that frequent intrafraction targeting verification and continuous tumor tracking with the Synchrony system allows the delivery of adequate dose to the gross tumor and its microscopic extension while keeping the volume of healthy lung exposed to radiation at safe levels. Although there are other ways of dealing with the problem of treating tumors that move with respiration, our experience has made us confident of the safety and accuracy of this approach.
Competing interests
BC is an Accuray clinical consultant. EA is paid by Accuray to give lectures.
Authors' contributions
BC drafted the manuscript, participated in treatment planning, data collection and data analysis. SV participated in data collection, data analysis and manuscript revision. KE participated in data collection, data analysis and manuscript revision. SC prepared the manuscript for submission, participated in data collection, data analysis and manuscript revision. SS created tables and figures and participated in data analysis and manuscript revision. XY participated in treatment planning, data collection and data analysis. YZ performed the biostatistical analysis and created figures. DS participated in data analysis and manuscript revision. CR participated in data analysis and manuscript revision. IS participated in data collection, data analysis and manuscript revision. GE participated in data collection, data analysis and manuscript revision. SY participated in data analysis and manuscript revision. CJ participated in treatment planning, data analysis and manuscript revision. PB participated in treatment planning, data collection, data analysis and manuscript revision. EA participated in treatment planning, data collection, data analysis and manuscript revision