If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Stereotactic body radiotherapy (SBRT) has been increasingly utilized for medically inoperable early stage non–small-cell lung cancer. However, a lower biological equivalent dose (BED) is often used for central tumors given toxicity concerns, potentially leading to decreased local control (LC). We compared survival, LC, and toxicity outcomes for SBRT patients with centrally versus peripherally located tumors.
We included patients with primary cT1-2N0M0 non–small-cell lung cancer treated with SBRT at our institution from September 2007 to August 2013 with follow-up through August 2014. Central tumor location was defined as within 2 cm of the proximal bronchial tree, heart, great vessels, trachea, or other mediastinal structures. Kaplan–Meier analysis and multivariable Cox regression modeling were used for overall survival (OS) and LC, and the χ2 test and multivariable logistic regression modeling were used for toxicity.
We included 251 patients (111 central, 140 peripheral) with median follow-up of 31.2 months. Patients with central tumors were more likely to be older (mean 75.8 versus 73.5 years; p = 0.04), have larger tumors (mean 2.5 cm versus 1.9 cm; p < 0.001), and be treated with a lower BED (mean 120.2 Gy versus 143.5 Gy; p < 0.001). Multivariable analysis revealed that tumor location was not associated with worse OS, LC, or toxicity. Patients with central tumors were less likely to have acute grade greater than or equal to three toxicity than those with peripheral tumors (odds ratio: 0.24; p = 0.02).
Central tumor location did not predict for inferior OS, LC, or toxicity following SBRT when a lower mean BED was utilized.
Stereotactic body radiotherapy (SBRT) has been increasingly utilized in the management of medically inoperable early stage non–small-cell lung cancer (NSCLC). For peripheral tumors, the Radiation Therapy Oncology Group (RTOG) 0236 phase II trial demonstrated a 3-year primary tumor control of 97.6% and 3-year lobar control of 90.6% when a dose of 54 Gy was delivered in three fractions.
However, there is considerable concern that treatment of centrally located tumors could lead to increased toxicity. A phase II trial at the University of Indiana noted increased grade 3–5 toxicity for patients with central tumors (within 2 cm of the proximal bronchial tree) compared with peripheral tumors (27.3% versus 10.4%; p = 0.09) using SBRT prescribed to at least 54 Gy in three fractions (corrected for tissue heterogeneity).
This led to the exclusion of patients with central tumors from RTOG 0236, as well as the formulation of a separate dose escalation study for central tumors using a lower biological equivalent dose (BED; RTOG 0813).
Experience in treating central tumors with SBRT suggests that regimens using more than three fractions may be reasonably well tolerated.
However, decreasing the BED with a more conservative dose-fractionation regimen could be concerning for decreased local control (LC) and possibly overall survival (OS). Two large multiinstitutional studies have suggested that a BED of at least 100–105 Gy may be necessary to achieve optimal LC outcomes,
The aim of this study was to examine a large, single-institution experience with SBRT for both peripherally and centrally located NSCLC. Our goal was to assess whether or not central tumor location would predict for worse OS, LC, or acute and late toxicity in an era of more conservative dose-fractionation regimens for centrally located tumors.
PATIENTS AND METHODS
We prospectively maintained an institutional database of patients treated with SBRT for primary NSCLC from September 2007 to August 2013. From this database, we selected all patients with AJCC 7th edition stage cT1-2N0M0 disease, who had at least one follow-up appointment with medical oncology, radiation oncology, or pulmonology. Central tumor location was defined as within 2 cm of the proximal bronchial tree (RTOG definition
Patients were immobilized in a customized full-length vacuum cushion and underwent a four-dimensional computed tomography (CT) scan during free breathing. Abdominal compression was used only for select cases in which tumor excursion exceeded 1 cm. An internal target volume was contoured to include the entire respiratory excursion of the tumor using the Advantage Workstation (GE Healthcare, Waukesha, WI). An isovolumetric 7 mm expansion was added to the internal target volume create a planning tumor volume (PTV). The heart, lung, esophagus, proximal tracheobronchial tree, spinal cord, and brachial plexus were contoured.
Treatment plans were generated using Eclipse (Varian Medical Systems, Palo Alto, CA), with tissue heterogeneity corrections based on the anisotropic analytical algorithm. Priority was given to PTV coverage, at the expense of normal tissue exposure. All plans were normalized such that 95% of the PTV was covered by 100% of the prescription dose and 99% of the PTV was covered by at least 90% of the prescription dose, with an expected maximum heterogeneity of 111–143% within the tumor (corresponding to 70–90% of the maximum dose at the edge of the PTV). A higher priority was given to obtaining full coverage of the PTV to the prescription dose than to remaining within RTOG guidelines for organs at risk. Patients were treated initially with multiple nonopposed, noncoplanar beams, or more recently with a dynamic conformal arc technique (described in detail by Ross et al.
). Intensity-modulated radiation therapy using static fields or volumetric modulated arc therapy was reserved for the minority of cases when forward planning resulted in an inferior plan.
All patients were treated in three to five fractions on nonconsecutive days, completing therapy within 15 calendar days. Cone-beam CT image guidance was used for all patients.
Follow-up included a history and physical examination approximately 4 weeks after completing SBRT. Patients were then evaluated clinically and with a noncontrast chest CT every 3 to 4 months for the first year, then every 3–6 months thereafter. Local failure was defined as a recurrence within the treated lobe as determined by biopsy or clinical judgment of the treating physician. OS was recorded based on most recent evidence of vital status in the medical record or by death date, based on obituary or in-hospital death. LC and toxicities were followed as of last radiographic or clinical follow-up.
Treatment-related toxicity was scored with the National Institute of Health Common Terminology Criteria for Adverse Events (CTCAE), version 4.0. An acute toxicity was defined as a treatment-related side effect occurring within 90 days of the first fraction; a late toxicity was one that occurred after this time.
The BED was calculated using the linear quadratic equation, assuming α/β = 10. For univariable analysis, the χ2 test was used for categorical variables, and the student's t test was used for continuous variables. The Kaplan–Meier method was used to estimate OS and LC. Subgroups were compared with the log-rank test. Additional univariable and multivariable analyses for time-to-event analyses were performed using Cox proportional hazards modeling. Acute toxicity and late toxicity were evaluated with the χ2 test and multivariable logistic regression modeling. Multivariable analyses adjusted for potential covariates, including patient factors like age, sex, and performance status; clinical factors like biopsy versus nonbiopsy diagnostic method, tumor histology, tumor size, and T-stage; and treatment-related factors like BED, total number of targets, maximum lung point dose, mean lung dose, volume of lung receiving greater than or equal to 5 Gy (V5), volume of lung receiving greater than or equal to 10 Gy (V10), and volume of lung receiving greater than or equal to 20 Gy (V20). All analyses were performed with SPSS version 19 (IBM, Armonk, NY).
This study was granted approval from the institutional review board at our institution.
There were 251 patients with 272 tumors included in this analysis, among whom 111 patients (44.2%) received SBRT for centrally located tumors. Median follow-up was 31.2 months for OS and 21.4 months for all other outcomes. A total of 181 patients (72.1%) had biopsy-proven NSCLC, and 82 patients (32.7%) had invasive mediastinal staging. Patients with central tumors were more likely to be older (mean 75.8 versus 73.5 years; p = 0.04), have larger tumors (mean 2.5 cm versus 1.9 cm; p < 0.001), undergo invasive mediastinal staging (46.8% versus 21.4%; p < 0.001), and be treated with a lower BED (mean 120.2 Gy versus 143.5 Gy; p < 0.001) compared with those with peripheral tumors. The BED used for peripheral tumors was 151.2 Gy (54 Gy in three fractions) in 80.0% of patients, whereas BED for central tumors was more variable (151.2 Gy in 36.9% of patients, 112.5 Gy [50 Gy in four fractions] in 30.6% of patients, and 100 Gy [50 Gy in five fractions]) in 17.1% of patients. There were no other significant differences in patient, clinicopathologic, or treatment-related factors between patients with central versus peripheral tumors (Table 1).
TABLE 1Patient, Clinicopathologic, and Treatment-Related Characteristics for Patients Treated for Central versus Peripheral Tumors (n = 251)
Central (n = 111)
Peripheral (n = 140)
Age (mean ± SE, yr)
75.8 ± 0.8
73.5 ± 0.8
Sex (women vs. men)
ECOG PS (0–1 vs. ≥2)
Diagnostic method (biopsy vs. no biopsy)
Invasive mediastinal staging (yes vs. no)
Histology (adenocarcinoma vs. other)
Tumor size (mean ± SE, cm)
2.5 ± 0.1
1.9 ± 0.1
T-stage (T2 vs. T1)
BED (mean ± SE, Gy)
120.2 ± 2.5
143.5 ± 1.5
Number of targets (1 vs. ≥2)
Maximum lung point dose (mean ± SE, Gy)
70.3 ± 4.6
61.2 ± 0.7
Mean lung dose (mean, Gy)
4.2 ± 0.2
4.8 ± 0.5
Lung V5 (mean, %)
24.2 ± 3.2
22.2 ± 1.1
Lung V10 (mean, %)
12.4 ± 0.6
12.1 ± 0.6
Lung V20 (mean, %)
5.4 ± 0.3
5.3 ± 0.3
ECOG PS, Eastern Cooperative Oncology Group performance status; V5, volume of lung receiving ≥5 Gy; V10, volume of lung receiving ≥10 Gy; V20, volume of lung receiving ≥20 Gy.
Overall, most acute and late grade 2 and grade greater than or equal to three toxicities manifested as dyspnea or pneumonitis for both central and peripheral tumors, in addition to relatively frequent grade 2 chest wall pain (Table 2). Although there were no significant differences in acute grade greater than or equal to two, late grade greater than or equal to two, and late grade greater than or equal to three toxicity by tumor location on univariable and multivariable analysis, acute grade greater than or equal to three toxicity was significantly less frequent in patients with central tumors compared with peripheral tumors (4.5% versus 12.9%; p = 0.03; adjusted OR: 0.24, 95% confidence interval [CI]: 0.07–0.82; p = 0.02; Table 3).
TABLE 2Toxicity Grade and Type of Grade 2 and Grade 3–5 Toxicity for Patients Treated for Central versus Peripheral Tumors (n = 251)
Kaplan–Meier analysis revealed that tumor location was not associated with OS (71.6% central versus 71.0% peripheral at 2 years, median 34.8 months central [95% CI: 27.1–42.5 months] versus 36.1 months peripheral [95% CI: 29.1–43.0]; p = 0.30; Fig. 1), LC (87.1% central versus 88.6% peripheral at 2 years; p = 1.00; Fig. 2), regional nodal control (88.6% central versus 90.4% peripheral at 2 years; p = 0.95; Supplementary Figure 1, Supplemental Digital Content 1, http://links.lww.com/JTO/A797), or distant control (91.7% central versus 88.6% peripheral at 2 years; Supplementary Figure 2, Supplemental Digital Content 2, http://links.lww.com/JTO/A798). Among the 21 local recurrences within the treated lobe for which we could verify the location of the recurrence relative to the target volume, 12 occurred within the PTV. After controlling for potential covariates like age and tumor size, Cox proportional hazards modeling also showed no significant association between tumor location and OS (adjusted hazard ratio: 0.84, 95% CI: 0.55–1.28; p = 0.41) or LC (adjusted hazard ratio: 1.09, 95% CI: 0.45–2.66; p = 0.85; Table 3).
To the best of our knowledge, this study represents the largest and most comprehensive analysis in North America comparing OS, LC, and toxicity outcomes of central versus peripheral NSCLC treated with SBRT. Our practice has been to treat central tumors with a more conservative dose-fractionation scheme than that used for peripheral tumors, which is reflected in the lower mean BED of patients with central tumors. After this strategy, we found that our patients with central tumors had noninferior OS, LC, and toxicity compared with those with peripheral tumors.
Our early experience treating centrally located lung tumors (both primary NSCLC and metastatic lesions) was reported by Rowe et al.
. A total of 47 patients with 51 central lesions were evaluated, with a median follow-up of 11.3 months (range: 4.8–40.8 months). Toxicity grade greater than or equal to three occurred in five patients (10.6%) and local recurrences occurred in two patients (4.3%). Both patients who had local recurrences were treated with a median BED of 76 Gy, compared with 112.5 Gy for the patients without local recurrence (p = 0.04). The 2-year lobar LC was 100% for those receiving BED greater than or equal to 100 Gy compared with 80% for those receiving BED less than 100 Gy (p = 0.02). Based at least in part on this initial experience, the majority of patients in the more recent era were treated with a higher BED. In this updated analysis, there was no difference in LC by BED level, likely reflecting the fact that few patients in either cohort were treated with BED less than 100 Gy.
One likely explanation for our encouraging results for patients with central tumors is related to the increased utilization of more conservative dose-fractionation regimens for these patients. Among our cohort of patients with centrally located tumors, the majority received 50 Gy in four or five fractions (BED 112.5 or 100 Gy, respectively). Importantly, those with central lesions experienced lower and similar rates of acute and late grade greater than or equal to three toxicity rates, respectively, compared with those with peripheral disease. These results contrast with the high grade greater than or equal to three toxicity rates observed in the early experience of the University of Indiana phase II trial treating central tumors to at least 54 Gy in three fractions.
Furthermore, our findings are consistent with those of a recent systematic review of 20 publications including outcomes for 563 central lung tumors, the majority of which were treated with SBRT in four to eight fractions. Central versus peripheral tumor location did not impact OS, with LC greater than or equal to 85% (when BED greater than or equal to 100 Gy) and grade greater than or equal to three toxicities less than 9% found for patients with central tumors overall.
Even with the delivery of a relatively low BED in five fractions, caution must still be utilized when treating central tumors. A tumor control probability model from the Elekta Collaborative Lung Research Group has suggested greater LC with a BED of 151.2 Gy (54 Gy in three fractions) compared with 100 Gy (50 Gy in five fractions).
Despite the risk of decreased LC and increased toxicities for central tumors compared with peripheral tumors, there continues to be considerable interest in treating centrally located NSCLC with SBRT. In an analysis of 117 survey responses from practicing radiation oncologists, Daly et al.
found that 58% of practitioners would recommend SBRT in the absence of a clinical protocol and 23% would offer SBRT on a clinical protocol for a patient with a cT1aN0M0 central tumor with a negative mediastinal workup. Among those who would recommend SBRT, 65% would recommend 50–55 Gy in five fractions, 18% would recommend 48–50 Gy in four fractions, 7% would recommend 54–60 Gy in three fractions, and 9% would recommend 60 Gy in 8–10 fractions.
In addition, alternative hypofractionation schemes involving 10 or more fractions continue to be explored for centrally located lung tumors. Chang et al.
attempted 50 Gy in four fractions for 100 patients who fulfilled the criteria for central tumor location, but if normal tissue constraints were exceeded and optimal coverage could not be reached, 70 Gy in 10 fractions (BED 119.0 Gy) was utilized in 18% of patients. LC outcomes for the subgroup receiving 70 Gy in 10 fractions were not reported; however, Soliman et al.
utilized 48–60 Gy in 12–15 fractions (BED 67.2–84.0 Gy), achieving an actuarial 2-year LC rate of 76.2%. However, only 13 of 124 tumors (10.5%) were centrally located. Further study will be needed to assess the effectiveness of similar dose-fractionation regimens for larger populations of patients with central lung tumors.
Limitations of this study include those inherent to retrospective analyses, even those performed with prospectively collected data. A certain degree of selection bias cannot be excluded, especially because patients with peripherally located tumors were less likely to undergo invasive mediastinal staging and may have been more likely to harbor occult node-positive disease. Second, it is challenging to separate fully the effect of tumor location from BED in this study, given the heterogeneity of dose-fractionation regimens delivered, especially among patients with centrally located tumors. We attempted to account for this limitation by utilizing multivariable modeling adjusting for BED, as well as sensitivity analyses with BED stratification. Third, our a priori definition of central tumor location is broader than that used by RTOG 0236, to include tumors at high risk of toxicity due to their close proximity to critical structures outside the region immediately surrounding the proximal bronchial tree. This definition allows for the inclusion of patients with significant dose to mediastinal structures, so that the analysis of dose is more valid and complete. We recognize that this definition of central tumor location may include tumors that some radiation oncologists might consider peripheral. Finally, longer follow-up will be needed to assess for late toxicities and local recurrences.
In conclusion, we found that patients with centrally located NSCLC who received SBRT appeared to have noninferior toxicity, likely due to more conservative dose-fractionation schemes. Despite older age, larger tumor size, and lower BED, patients with central tumors had similar OS and LC as those with peripheral tumors. Although we await the results of RTOG 0813, an ongoing phase I/II trial evaluating the maximal tolerated dose and primary tumor control rate for patients with central tumors using various five-fraction SBRT regimens, our data indicates that 50 Gy in four or five fractions for central tumors is safe and effective without appearing to compromise OS or LC.
This research was presented in part as an oral presentation at the 96th Annual Meeting of the American Radium Society (ARS, April 26–29, 2014) and as an oral presentation at the 56th Annual Meeting of the American Society for Radiation Oncology (ASTRO, September 14–17, 2014).
Disclosure: Dr. Decker receives research funding from Merck & Co., Inc. and holds stock in Bristol-Myers Squibb. There are no relevant conflicts of interest for this manuscript.