Presented in oral form at the 52nd Annual Meeting of the American Society for Therapeutic Radiology and Oncology; October 31 to November 4, 2010; San Diego, California.
We are grateful to Drs. Hiroyuki Ogino, Chikao Sugie, Rumi Murata, Takeshi Yanagi, Shinya Otsuka, Katsura Kosaki, and Shinya Takemoto for their contribution in collecting data and to Hiroshi Fukuma, Kazuhiro Komai, Dr. Naoki Hayashi, and Hisato Nakazawa for treating patients.
The most common regimen of stereotactic body radiotherapy (SBRT) for stage I nonsmall cell lung cancer in Japan is 48 grays (Gy) in 4 fractions over 4 days. Radiobiologically, however, higher doses are necessary to control larger tumors, and interfraction intervals should be >24 hours to take advantage of reoxygenation. In this study, the authors tested the following regimen: For tumors that measured <1.5 cm, 1.5 to 3.0 cm, and >3.0 cm in greatest dimension, radiation doses of 44 Gy, 48 Gy, and 52 Gy, respectively, were given in 4 fractions with interfraction intervals of ≥3 days.
Among 180 patients with histologically proven disease who entered the study, 120 were medically inoperable, and 60 were operable. The median patient age was 77 years (range, 29-92 years). SBRT was performed with 6-megavolt photons using 4 noncoplanar beams and 3 coplanar beams. Isocenter doses of 44 Gy, 48 Gy, and 52 Gy were received by 4 patients, 124 patients, and 52 patients, respectively.
The overall survival rate for all 180 patients was 69% at 3 years and 52% at 5 years. The 3-year survival rate was 74% for operable patients and 59% for medically inoperable patients (P = .080). The 3-year local control rate was 86% for tumors ≤3 cm (44/48 Gy) and 73% for tumors >3 cm (52 Gy; P = .050). Grade ≥2 radiation pneumonitis developed in 13% of patients (10% of the 44-Gy/48-Gy group and 21% of the 52-Gy group; P = .056). All other grade 2 toxicities developed in <4% of patients.
Stereotactic body radiotherapy (SBRT) has become an indispensable treatment modality for medically inoperable patients with stage I nonsmall cell lung cancer (NSCLC).1-5 Also, operable patients who choose SBRT based on their own wishes appear to be steadily increasing. Despite the increasing popularity of SBRT even among operable patients, many technical issues exist. One of the most important unresolved issues is the optimal fractionation schedule.
Various fractionation regimens have been used, including 45 or 60 grays (Gy) in 3 fractions over 1 week, 50 Gy in 5 fractions over 5 to 8 days, and 48 Gy in 4 daily fractions.5-9 In Japan, 48 Gy given in 4 daily fractions has been the most frequently used schedule.6 The Japan Clinical Oncology Group (JCOG) study 0403, which was conducted along with the current study, used 48 Gy in 4 fractions over 4 to 8 days.10 This dose was used for tumors of any size. However, a question arose regarding whether the same dose should be applied to all tumors regardless of size. The clonogenic cell number, hypoxic fraction, and quiescent cell fraction all increase as the tumor enlarges,11, 12 and these all lead to elevated radioresistance. Therefore, higher radiation doses are necessary to control larger tumors. Another issue is the overall treatment time (OTT) or interval between fractionation. In Japan, SBRT sessions have been delivered most often on consecutive days or within a week; in the JCOG 0403 study, 4 fractions of 12 Gy were delivered in 4 days in most cases.10 However, this schedule does not seem to fully use the benefit of reoxygenation in tumors. In most murine tumors, reoxygenation is not complete at 24 hours after irradiation, and reoxygenation proceeds further after 24 hours until the third day and thereafter.13-15 Therefore, we believed it would be better to adopt an interval of at least 72 hours, which we did in the current study.
In the current article, we report the results from a multi-institutional SBRT study for NSCLC with a protocol based on the above-mentioned radiobiologic background. SBRT was delivered in 4 fractions with an interfraction interval of at least 3 days. The total dose was 44 Gy, 48 Gy, or 52 Gy, depending upon the tumor size.
MATERIALS AND METHODS
Study Design and Eligibility Criteria
This was a prospective study involving 3 institutions of the Nagoya City University (NCU) group (NCU-0401). Approval was obtained from the institutional review boards at the respective hospitals. Eligibility criteria were as follows: 1) histologically confirmed NSCLC; 2) T1N0M0 or T2N0M0 disease according to the International Union Against Cancer 2002 system diagnosed by computed tomography (CT) scans of the chest and upper abdomen, bone scintigraphy, and brain magnetic resonance imaging; 3) greatest tumor dimension ≤5 cm; 4) a World Health Organization performance status (PS) ≤2 or a PS 3 when its cause was not a pulmonary disease; 5) no prior therapy for the NSCLC to be treated by SBRT; 6) no active concurrent malignancy; and 7) arterial oxygen pressure ≥60 mm Hg and a forced expiratory volume in 1 second ≥700 mL. When 18F-deoxyglucose-positron emission tomography (FDG-PET) or FDG-PET-CT studies were performed, bone scintigraphy was omitted. FDG-PET was not a mandatory examination, because 2 of the 3 institutions had no PET scanner; however, when metastases were suspected, a PET scan was obtained in all patients. No restrictions were imposed with regard to the tumor location if the dose limits for organ tolerance were met. Patients who did not meet these criteria, such as those with no histologic confirmation and those with larger tumors, also were treated on the same protocol, but they were not included in the current study. All patients consented to the treatment after being informed of the method and rationale of the study.
The primary endpoint of the study was the local control rate at 3 years. Secondary endpoints were overall survival, cause-specific survival, rates of regional lymph node and distant metastases, and incidence of acute and late toxicities. Local control rates of 90% for T1 tumors and 75% for T2 tumors were expected with this protocol; to estimate the local control rate for T1 tumors with a 95% confidence interval of ±7%, 115 patients were considered necessary. The total number of patients with T2 tumors was expected to be much lower than that for patients with T1 tumors; therefore, to estimate the local control rate for T2 tumors with an 80% confidence interval of ±10%, 50 patients were considered necessary. Patient accrual was stopped shortly after the numbers of both T1 and T2 patients exceeded these values.
Between May 2004 and November 2008, 185 patients were considered for entry, but 5 patients did not enter the study because the dose prescription criteria were not met. Thus, 180 eligible patients entered the study, and all 180 completed the planned treatment. Patients were categorized according to medical operability; they were deemed medically inoperable when they had poor pulmonary function (the ratio of forced expiratory volume in 1 second to forced vital capacity <60% and/or a percentage vital capacity <75%) or other debilitating conditions that preclude surgery. Advanced age alone was not a factor for inoperability. Patient and tumor characteristics are listed in Table 1. Patients ranged in age from 26 years to 89 years (median, 77 years). The greatest tumor dimension ranged from 12 mm to 50 mm (median, 27 mm). The tumor location was classified as either central or peripheral according to Radiation Therapy Oncology Group (RTOG) criteria.16
For patient immobilization, the BodyFIX system (Medical Intelligence, Schwabmuenchen, Germany) was used at 2 institutions (NCU and Nagoya Daini Red Cross Hospital [Nagoya, Japan]), whereas a custom-made thermoplastic cast (Hip-Fix; Med-Tec, Orange City, Iowa) was used at another institution (Nagoya Kyoritu Hospital, Nagoya, Japan) for the Novalis image-guided system (Varian Medical Systems, Palo Alto, Calif). Details of the 2 systems were described previously.17, 18 CT images for treatment planning were obtained with a 1.25-mm or 2.5-mm slice thickness under normal breathing and with breath holding during the expiratory and inspiratory phases. The clinical target volume (CTV) was defined as the visible gross tumor volume. The CTV on CT during the 3 phases was superimposed on 3-dimensional radiation treatment planning systems (Eclipse version 126.96.36.199; Varian Medical Systems; BRAINSCAN version 5.31; BrainLAB, Feldkirchen, Germany; or Pinnacle3; Philips Medical Systems, Madison, Wis) to represent the internal target volume (ITV). The planning target volume (PTV) margin for the ITV was 5 mm in the lateral and anteroposterior directions and 10 mm in the craniocaudal direction. All treatment planning was checked and approved by 1 of the study integrators (Y.S. or M.M.).
SBRT was delivered by a linear accelerator (CLINAC 23EX or 21EXS; Varian Medical Systems) or the Novalis image-guided system, all with 6-megavolt photons. Three coplanar and 4 noncoplanar static ports were used. SBRT treatment generally was performed twice weekly. Three or more fractions per week were not permitted. Consecutive treatments were given at 3-day or longer intervals, except for rare exceptions. Because of national holidays, patient schedule convenience, and machine availability, the actual overall treatment period ranged between 9 days and 21 days; in 92% of patients, it was 10 to 14 days. The median treatment period was 12 days.
Prescription Dose and Dose Constraints Regarding Normal Tissues
The prescribed dose was 44 Gy in 4 fractions for tumors with a maximum dimension <1.5 cm, 48 Gy in 4 fractions for tumors with a maximum dimension of 1.5 cm to 3.0 cm, and 52 Gy in 4 fractions for tumors with a maximum tumor dimension >3 cm. Pencil-beam convolution with Batho power law correction was used as the dose-calculation algorithm in all institutions. The prescribed dose represented the dose delivered to the isocenter. It was recommended to cover 95% of the PTV with at least 90% of the isocenter dose; and, in all patients, 95% of the PTV received at least 80% of the prescribed dose. Consequently, 95% of the ITV was covered with ≥94% of the prescribed dose in all but 1 patient. Only 1 of the beams was allowed to pass the spinal cord, so that the maximum dose to the cord was <18 Gy. Other dose constraints for normal tissues were: 1) volume of the lung receiving 20 Gy, ≤20%; 2) 40 Gy for <1 mL of the pulmonary artery and esophagus; and 3) 36 Gy for ≤10 mL of the stomach. Four patients received 44 Gy, 124 received 48 Gy, and 52 received 52 Gy.
The protocol for follow-up after SBRT was as follows. Chest and upper abdominal CT scans were obtained at 2-month intervals up to 6 months and every 2 to 4 months thereafter. FDG-PET or FDG-PET-CT studies were obtained whenever necessary. Local recurrence was suspected when enlargement of a consolidated fibrotic mass was detected on CT images without signs of inflammation and was diagnosed by high uptake on an FDG-PET study (standardized uptake value ≥5) and/or biopsy. Local recurrence was confirmed by biopsy in 2 patients. Pleuritis carcinomatosa unaccompanied by local recurrence was regarded as distant metastasis. Toxicity was evaluated using the Common Terminology Criteria for Adverse Events version 3. Grade 2 radiation pneumonitis was defined as symptomatic but not interfering with activities of daily living.
Survival and failure-free rates were calculated using the Kaplan-Meier method from the start of SBRT. To evaluate isolated lymph node (regional) failure, patients were censored when they developed local recurrence or pulmonary metastasis. To evaluate isolated distant metastasis, patients were censored when they developed local and/or regional recurrence. Local status was followed until death, and patients were not censored even when they developed regional or distant metastasis. The log-rank test was used to compare the survival and control rates between the subsets. The incidence of adverse events was compared using the Fisher exact test and the chi-square test. Statistical analysis was carried out using the computer programs StatView version 5.0 (SAS Institute, Cary, NC) and HALWIN (Gendaisuugakusha, Kyoto, Japan).
The median follow-up was 36 months for all patients and 42 months for the patients who remained alive. Sixty-five patients died, including 39 deaths from lung cancer and 26 deaths from intercurrent disease or suicide. Twenty-six patients developed local recurrence, 24 developed regional recurrence, and 32 developed distant metastasis. Overall and cause-specific survival curves and local, regional, and distant metastasis control curves for all patients are provided in Figure 1. For all patients, the overall survival rate was 69% at 3 years and 52% at 5 years. The local control rate was 83% at 3 years and thereafter. The regional and distant metastasis control rates at 3 years were 85% and 80%, respectively.
The local control curves according to tumor size and prescribed dose are provided in Figure 2. No patients who received 44 Gy developed local recurrence. At 3 years and thereafter, the local control rate was 86% for 124 patients who received 48 Gy and 73% (80% confidence interval, 63%-82%) for those who received 52 Gy. For 128 patients with T1 tumors (ie, ≤3 cm; 44 Gy and 48 Gy groups), the local control rate was 86% (95% confidence interval. 80%-93%). All local recurrences developed within 3 years after SBRT. Figure 3 illustrates the regional and distant metastasis control curves according to tumor classification. Both regional and distant metastasis control rates tended to be higher in T1 tumors than in T2 (≥3 cm) tumors.
Figure 4 provides the overall survival curves according to operability. At 3 years, the overall survival rate was 74% for 60 operable patients and 59% for 120 inoperable patients. The projected 5-year data were 70% and 44%, respectively. The 3-year overall survival rate was 80% for 45 operable patients with T1 tumors and 58% for 15 operable patients with T2 tumors.
Overall and cause-specific survival and local, regional, and distant metastasis control data according to potential prognostic factors are summarized in Table 2. Younger patients and women had more favorable local control rates. Women also had better overall survival rates. Patients with T1 tumors had better cause-specific survival rates than those with T2 tumors. Otherwise, no significant differences were observed between the groups.
Table 2. Three-Year Data According to Patient and Tumor Characteristics
No. of Patients
Abbreviations: SCC, squamous cell carcinoma.
Grade 2 or greater radiation pneumonitis was observed in 24 of 180 patients (13.3%); among these, only 2 episodes (1.1%) were grade 3. The incidence of grade ≥2 pneumonitis was 13 of 128 patients (10.2%) with T1 tumors and 11 of 52 patients (21%) with T2 tumors (P = .056 [Fisher exact test]; P = .049 [chi-square test]). Grade 2 or greater esophagitis, rib fracture, and dermatitis were observed in 3 of 180 patients (1.7%), 3 of 180 patients (1.7%), and 7 of 180 patients (3.9%), respectively. Grade 3 pleural effusion and grade 2 cardiac effusion were reported in 1 patient each (0.6%).
In SBRT for stage I NSCLC, various issues remain unresolved; however, in the current report, we focus our discussion on dose and fractionation issues. By using the same dose for T1 and T2 tumors, most previous investigators reported a poorer outcome for patients with T2 tumors. Differences in reported local control or cause-specific survival rates between T1 and T2 tumors range from 12% to 38%, and the average rate is approximately 20%.3, 19-22 In an exceptional report, Takeda et al23 observed excellent outcomes for both stage IA and IB tumors, but their median follow-up appeared to be much shorter than 3 years. Both radiobiologic24 and clinical data25, 26 suggest that higher doses are necessary for larger tumors. In 1 study in which a greater dose was prescribed for T2 tumors, no difference was observed in median survival or in 3-year cause-specific survival rates between stages IA and IB.8 In the current study, the 3-year local control rate was 86% for T1 tumors and 73% for T2 tumors (P = .050), and overall survival did not differ significantly. This observation may reflect the use of a higher dose for T2 tumors.
With respect to the OTT in SBRT, most previous investigators administered total doses within 1 week; however, in our opinion, such schedules do not take full advantage of the reoxygenation phenomenon. On the basis of laboratory investigations of reoxygenation in murine tumors,13-15 we adopted an interfraction interval of at least 3 days. Contrary to our claim, a recent multi-institutional retrospective study suggested that an OTT of <11 days was associated with a more favorable outcome.27 However, that study analyzed various fractionation schedules together, so it is questionable whether a shortened treatment time of <11 days really is better in SBRT for stage I NSCLC. The radiation effect is hampered if the OTT becomes >40 days because of the repopulation of tumor cells; however, OTTs <3 weeks do not appear to adversely affect outcomes.28, 29 In contrast, a retrospective study using 48 Gy in 4 fractions suggested a trend toward a poorer outcome when the OTT was <1 week.21 Our own results seem to compare favorably with those obtained using similar doses with a shorter OTT.10, 21 Thus, we advocate that fractions of SBRT should not be delivered every day.
The local control rates in our study may not be sufficient compared with those reported in some other recent reports.4, 7 In particular, the recent RTOG 0236 study obtained a 3-year local control rate of 97.6% using 54 Gy in 3 fractions delivered to the periphery of the PTV.4 However, caution must be taken when interpreting local control rates. In the RTOG study, which investigated medically inoperable patients, many patients died of other causes before they developed local recurrence. Therefore, the reported local control rate may have become artificially high because of these censored patients. At the completion of the current study, we revised the protocol to increase the doses.
In summary, we obtained reasonable local control rates using a higher dose for T2 tumors and relatively long interfraction intervals. The optimal fractionation schedule with dose escalation should be investigated further. Interfraction intervals ≥72 hours may be adequate from a radiobiologic standpoint.