Presented at the 45th Annual Meeting of the American Society of Therapeutic Radiation Oncology (ASTRO), Salt Lake City, Utah, October 20–23, 2003.
Stereotactic irradiation (STI) has been actively performed using various methods to achieve better local control of Stage I nonsmall cell lung carcinoma (NSCLC) in Japan. The authors retrospectively evaluated results from a Japanese multiinstitutional study.
Patients with Stage I NSCLC (n = 245; median age, 76 years; T1N0M0, n=155; T2N0M0, n=90) were treated with hypofractionated high-dose STI in 13 institutions. Stereotactic three-dimensional treatment was performed using noncoplanar dynamic arcs or multiple static ports. A total dose of 18–75 gray (Gy) at the isocenter was administered in 1–22 fractions. The median calculated biologic effective dose (BED) was 108 Gy (range, 57–180 Gy).
During follow-up (median, 24 months; range, 7–78 months), pulmonary complications of National Cancer Institute-Common Toxicity Criteria Grade > 2 were observed in only 6 patients (2.4%). Local progression occurred in 33 patients (14.5%), and the local recurrence rate was 8.1% for BED ≥ 100 Gy compared with 26.4% for < 100 Gy (P < 0.05). The 3-year overall survival rate of medically operable patients was 88.4% for BED ≥ 100 Gy compared with 69.4% for < 100 Gy (P < 0.05).
Nonsmall cell lung carcinoma (NSCLC) represents a leading cause of mortality worldwide. Lung carcinomas are being detected increasingly early, thanks to routine use of computed tomography (CT) scans. For patients with Stage I (T1N0M0 or T2N0M0) NSCLC, full lobar or greater surgical resection represents a treatment choice that promises local control rates ≥ 80% and overall survival rates > 50% after 5 years.1 However, surgical resection is often not feasible or involves excessive risk for some patients with lung carcinoma with tobacco-related illness, severe cardiovascular disease, or other medical conditions. A small proportion of patients who are eligible for surgery may refuse procedures for personal reasons. Radiotherapy can offer a therapeutic alternative for these patients, but outcomes for conventional radiotherapy are unsatisfactory, and are potentially amplified by selection bias, with local control rates of 40–70% and 5-year survival rates of only 5–30%.2–4 Doses of conventional radiotherapy to treat NSCLC have been suggested to be too low to achieve tumor control. However, providing a higher dose to the tumor without increasing adverse effect was previously impossible, due to technical uncertainties over focusing irradiation only on the tumor-bearing area of the lung.
With the increasing accuracy of localization for tumor-bearing areas using various imaging techniques, hypofractionated or single high-dose stereotactic irradiation (STI) has been actively investigated for Stage I NSCLC in Japan.5–8 STI can also substantially reduce overall treatment time from several weeks for a conventional radiotherapy schedule to a few days, offering important advantages to the patient. A landmark study by Uematsu et al.,5 one of the pioneers of STI for extracranial lesions, revealed excellent survival rates for medically operable patients, approximating those for full lobar surgical resection. Under the guidelines of the Japanese Society of Radiation Oncology study group,9 Stage I NSCLC has been treated using small-volume STI in numerous Japanese institutions since the late 1990s, with far fewer symptomatic adverse effects than conventional radiotherapy. Although optimal STI techniques and schedules for Stage I NSCLC remain unclear, the number of patients with Stage I NSCLC treated nationwide using small-volume, high-dose STI has accumulated rapidly. Although differences in techniques and schedules may vary widely, retrospective investigation of the results of STI for Stage I NSCLC from the many institutions that have used small-volume, high-dose irradiation in this short period should yield some meaningful data. The current study retrospectively evaluated Japanese multiinstitutional results for high-dose STI for Stage I NSCLC, and sought to answer the following questions: 1) What is the optimal dose to limit toxicity and still obtain local control? 2) Are the results from single-institution studies reproducible? 3) Are STI results comparable to those of surgery?
MATERIALS AND METHODS
All patients enrolled in the current study satisfied the following eligibility criteria: 1) identification of T1N0M0 or T2N0M0 primary lung carcinoma on chest and abdominal CT scans, bronchoscopy, bone scintigraphy, or brain magnetic resonance imaging scans; 2) histologic confirmation of NSCLC; 3) tumor diameter < 60 mm; 4) performance status ≤ 2 according to World Health Organization guidelines; and 5) inoperable tumor due to poor medical condition or refusal to undergo surgery.
No restrictions were utilized concerning the location of eligible tumors, irrespective of whether they were located adjacent to a major bronchus, blood vessel, chest wall, or the esophagus or spinal cord. However, the spinal cord was kept out of the high-dose area.
Patients were informed as to the concept, methodology, and rationale of this treatment. Written informed consent was obtained from all patients. The study was approved by the ethics committee of each institution and was performed in accordance with the 1983 revision of the Helsinki Declaration.
A summary of patient characteristics is provided in Table 1. From April 1995 to February 2003, 245 patients with primary NSCLC were treated with hypofractionated high-dose STI in 13 institutions. Of the 245 patients, 158 (65%) were considered to be medically inoperable, due predominantly to chronic pulmonary disease, advanced age, or other chronic illness. The remaining 87 patients (35%) were considered to be medically operable, but had refused surgery or had been advised to select STI by medical oncologists.
All patients were irradiated using stereotactic techniques. For the purposes of the current study, all stereotactic techniques fulfilled three requirements: 1) reproducibility of the isocenter ≤ 5 mm, as confirmed in every fraction; 2) slice thickness on CT scan ≤ 3 mm for three-dimensional (3D) treatment planning; and 3) irradiation with multiple noncoplanar static ports or dynamic arcs. Table 2 summarized various techniques and instruments introduced to achieve STI in 13 institutions. To fulfill the first requirement, a CT scan or two-directional portal graph was undertaken before every treatment regimen in 12 institutions, whereas real-time tumor tracking using a gold marker inserted around the tumor10 was performed in 1 institution. A CT scanner sharing a common couch with the linear accelerator was placed in an irradiation room in two institutions.11, 12
Table 2. Treatment Schemes
BED: biologic effective dose; Gy: gray.
6-MV X-ray, 12; 4-MV X-ray, 1
Measures for respiratory motion
Respiratory gating, 5; breath hold, 2; non, 6
Fixation of patients
Vacuum pillow, 5; body frame, 4; non, 4
Irradiation port shape
Regular, 4; conformal, 9
1–25 (multiple, 11; single, 2)
Multiple (6–20) static ports, 7; dynamic arc, 6
Single dose (at the isocenter)
Total dose (isocenter) of stereotactic irradiation
30–44 Gy/15–20 fractions in 27 patients; non, 218 patients
BED = nd(1+d/a/b) at the isocenter
57–180 Gy (median, 108 Gy)
Treatment planning with irregularly shaped beams using noncoplanar multiple (3–10) dynamic arcs or multiple static ports (6–20 ports) was established with the help of a 3D treatment-planning computer. Beam shaping was performed in some institutions using an integrated motorized multileaf collimator with 0.5–1-cm leaf width at the isocenter. Furthermore, various techniques using breathing control or gating methods and immobilization devices such as a vacuum cushion with or without a stereotactic body frame were utilized to reduce respiratory internal margins. Respiratory gating or breath-hold methods were used in seven institutions.
Planning CT scans were performed with 2 or 3-mm slice thickness and displayed using a window level of −700 Hounsfield units (HU) and a window width of 2000 HU. In some institutions, irradiation and planning CT scans were performed under breath-hold conditions. In other institutions, irradiation and planning CT scans were performed under free shallow breathing, with images taken using slow scanning (4 seconds per slice).
The clinical target volume (CTV) marginally exceeded the macroscopic target volume by 0–5 mm. The planning target volume (PTV) comprised the CTV, a 2–5-mm internal margin, and a 0–5-mm safety margin. An example of an STI dose distribution for Stage I lung tumors is shown in Figure 1. A high dose was concentrated on the tumor-bearing area while sparing surrounding normal lung tissues using STI.
Irradiation schedules also differed among institutions. The number of fractions ranged between 1 and 25, with single doses of 3–12 Gy. A total dose of 18–75 Gy at the isocenter in 1–25 fractions was administered with 6-MV X-rays within 20% heterogeneity in the PTV dose. Twenty-seven patients had received conventional irradiation doses of 30–44 Gy in 15–20 fractions before STI due to physician preferences. No chemotherapy regimens were administered before or during radiotherapy.
To compare the effects of various treatment protocols with different fraction sizes and total doses, a biologic effective dose (BED) was utilized in a linear-quadratic model.13 BED was defined as nd(1+d/α/β), with units of grays, where n is the fractionation number, d is the daily dose, and α/β is assumed to be 10 for tumors. BED was not corrected with values for tumor-doubling time or treatment term.
The median BED at the isocenter was 108 Gy (range, 57–180 Gy). BED was ≥ 100 Gy in 173 patients and < 100 Gy in 72 patients.
Dose constraints were set for the spinal cord only. The BED limitation for the spinal cord was 80 Gy (α/β was assumed to be 2 Gy for chronic spinal cord toxicity). This dose constraint for the spinal cord was achieved in all patients who satisfied all eligibility criteria.
The objectives of the current study were to retrospectively evaluate toxicity and the local control and survival rates according to BED. Follow-up examinations were performed by radiation oncologists for all patients. The first examination took place 4 weeks after treatment, and patients were subsequently seen every 1–3 months. Tumor response was evaluated using previously published National Cancer Institute (NCI) criteria.14 Chest CT scans (slice thickness, 2–5 mm) were usually obtained every 3 months for the first year, and repeated every 4-6 months thereafter. A complete response (CR) indicated that the tumor had completely disappeared or was replaced by fibrotic tissue. A partial response (PR) was defined as a ≥ 30% reduction in the maximum cross-sectional diameter. Distinguishing between residual tumor tissue and radiation fibrosis was difficult. Any suspicious residual confusing density after radiotherapy was considered to be evidence of PR, so the actual CR rate may be higher than presented in the current study. Local disease recurrence was considered to have occurred only when enlargement of the local tumor continued for > 6 months on follow-up CT scans. Findings on CT scans were interpreted by two radiation oncologists. Absence of local disease recurrence was defined as locally controlled disease.
Lung, esophagus, bone marrow, and skin were evaluated using Version 2 of the National Cancer Institute-Common Toxicity Criteria (NCI-CTC).
Local disease recurrence rates in the two groups were compared using the chi-square test. BED among patient groups at each pulmonary toxicity grade was compared using Kruskal–Wallis tests. Cumulative survival curves were calculated and drawn using Kaplan–Meier algorithms with the day of treatment as the starting point. Subgroups were compared using log-rank statistics. Values of P < 0.05 were considered to be statistically significant. Statistical calculations were conducted using Version 5.0 StatView software (SAS Institute Inc., Cary, NC).
All patients completed treatment with no particular complaints. The median period of follow-up was 24 months (range, 10–78 months). BED (α/β is assumed to be 2 Gy for chronic toxicity of the spinal cord) did not exceed 80 Gy in any of the patients.
Local Tumor Response
Of the 245 patients evaluated using CT scans, CR and PR were achieved in 57 (23.3%) and 151 (61.6%) patients, respectively. The overall response rate (CR+PR) was 84.8%. Overall response rates for tumors with BED ≥ 100 Gy (n = 173) and < 100 Gy (n = 72) were 84.5% and 83.3%, respectively. An example of a CR is shown in Figure 2.
Treatment toxicities are summarized in Table 3. Symptomatic radiation-induced pulmonary complications (NCI-CTC criteria Grade > 1) were observed in 17 patients (6.9%). No significant differences in BED were identified among patient groups at each pulmonary toxicity grade. Pulmonary fibrosis or emphysema before treatment was observed in 15 (88%) of the 17 patients with pulmonary complications > Grade 1. Pulmonary symptoms resolved in most patients with or without steroid therapy, but continuous oxygen supply was required in three patients who displayed poor respiratory function before irradiation. Chronic segmental bronchitis and wall thickening causing atelectasis on the peripheral lung was observed in one patient. Grade 3 esophagitis was temporarily observed in two patients with tumors adjacent to the esophagus. Grade 3 or 4 dermatitis was observed in two patients with tumors adjacent to the chest wall. No vascular, cardiac, or bone marrow complications had been encountered as of the last follow-up.
Table 3. Toxicity
Graded according to National Cancer Institute-Common Toxicity Criteria (Version.2.0).
Local disease recurrence occurred in 13.5% of all patients, with rates being significantly lower for BED ≥ 100 Gy (8.1%) compared with < 100 Gy (8.1% vs. 26.4%, P < 0.01). Patients with Stage IB disease displayed significantly higher rates of local disease recurrence compared with patients with Stage IA disease. However, no differences in the local disease recurrence rate were observed between patients with Stage IA disease and patients with Stage IB disease for BED ≥ 100 Gy. Rates of local disease recurrence were also significantly lower in the total group and Stage IA and Stage IB subgroups for BED ≥ 100 Gy compared with < 100 Gy. In particular, when BED was < 100 Gy, the local disease recurrence rate in patients with stage IB disease was 41.4% (12 of 29) compared with 16.3% (7 of 43) for patients with Stage IA disease. For BED ≥ 100 Gy, the local disease recurrence rate was 7.5% for BED ≥ 120 Gy (n = 80) and 9.8% for BED ≥ 140 Gy (n = 40). The local disease recurrence rates for adenocarcinoma and squamous cell carcinoma were 13.6% (15 of 110) and 13.8% (15 of 109), respectively.
The patterns of first disease recurrence are listed in Table 4. Some sites of disease recurrence overlapped, and isolated local, lymph node, and distant disease recurrences were observed in 8.6%, 3.3%, and 9.8% of patients, respectively. The local disease recurrence rate of patients with Stage IB was twice that of patients with Stage IA disease, whereas lymph node and distant disease recurrence rates were basically identical in the two subgroups.
Table 4. Patterns of First Disease Recurrences According to Stage and BED
Some of the disease recurrences overlapped each other.
Local disease recurrence
Regional lymph node recurrence
The overall 3 and 5-year survival rates were 56% and 47%, respectively. The cause-specific 3 and 5-year survival rates were both 78%. Overall survival rates differed significantly according to medical operability. For example, intercurrent deaths occurred in 19.1% of inoperable patients and in 3.4% of operable patients (Fig. 3). Overall survival rates according to BED in all patients revealed significant differences between the subgroups for BED < 100 Gy and ≥ 100 Gy (Fig. 4). Overall survival rates according to BED in operable patients revealed identical 3 and 5-year survival rates of 88% for BED ≥ 100 Gy (Fig. 5). Overall 5-year survival rates according to stage in operable patients irradiated with BED ≥ 100 Gy were 90% for patients with Stage IA disease and 84% for patients with Stage IB disease (Fig. 6).
Surgical resection remains the standard management for patients with Stage I NSCLC. The 5-year overall survival rates for patients undergoing resection range from 55% to 72% for Stage I NSCLC.15–17 Results for treating early-stage NSCLC using conventional radiotherapy are disappointing. Qiao et al.18 reviewed 18 studies on Stage I NSCLC treated using conventional radiotherapy alone, and reported that the 3-year overall and cause-specific survival rates were 34 ± 9% (mean ± standard error of the mean) and 39±10%, respectively. Although CR represents an important prognostic factor, particularly for tumors < 5 cm in diameter,19, 20 local disease recurrence is common after conventional radiotherapy for early-stage NSCLC.18, 21, 22 Several studies have shown the value of dose escalation in Stage I NSCLC.18–20 Although increased radiation dose to the tumor is essential, escalating the dose is difficult under conventional radiotherapy techniques, given the relatively large amount of normal lung tissue enclosed in the high-dose region, including internal and safety margins to accommodate respiratory movements and daily setup errors. The most common reactions caused by radiation dose escalation are pneumonitic changes, which can induce acute symptoms of fever and cough, leading to interstitial fibrosis and subsequent reduction in lung capacity. In patients with already compromised respiratory function, such reductions can prove fatal.
Because excessive dose escalation, which improves local control in patients with NSCLC,18, 23, 24 is so hard to obtain using conventional techniques, new approaches must be taken to improve outcomes. In 1995, Blomgren et al.25 introduced a new STI technique for extracranial radiotherapy that was analogous to cranial radiosurgery. The advantages of this radiotherapeutic technique include narrow X-ray beams, concentrated in such a manner as to provide intense irradiation to small lesions at high doses, and a small number of treatment fractions. The ability to concentrate radiotherapy on a small tumor while sparing surrounding tissues had already been made possible using STI. Results from treating small brain metastases are excellent, with local control rates of approximately 90%. Application of STI techniques to the treatment of small lung tumors is reasonable, as the ratio of high-dose radiation volume to normal tissue volume should be smaller than that for the brain. Moreover, the limited volume of radiation damage on the lung or adjacent structures is unlikely to result in the severity of symptoms possible with damage to cerebral tissues. The current data reveal that Grade 3 or 4 radiation pneumonitis was observed in few patients (4%). Acute esophagitis, dermatitis, and chronic bronchitis were also observed in relatively few patients for whom tumors bordered on these organs. No other life-threatening toxicities were encountered. However, the chronic effects of hypofractionated irradiation on major vessels, bronchus, esophagus, heart, and spinal cord remain unknown. Lethal pulmonary bleeding has been reported after a schedule of 24-Gy single-dose irradiation (BED = 81.6 Gy).26 Long and careful follow-up is therefore warranted.
Recently, STI for small lung tumors using a linear accelerator has gained acceptance as an effective treatment modality. Irradiation methods and local disease recurrence rates from several institutions in which STI has been performed for primary Stage I NSCLC are listed in Table 5.5–7, 27–29 Although BED analysis using the linear-quadratic model is not quite appropriate for radiotherapy with a large single dose or short treatment period,30 the model is useful to compare outcomes from a variety of treatment schedules using different single doses and number of fractions. Cheung et al.31 summarized several clinical studies. In their study, crude local recurrence rates with conventional radiotherapy were 36–70%, with BED of 59.6–76.4 Gy at an α/β ratio of 10. They recommended dose escalation to increase the local control rate. STI appears to represent an ideal modality for dose escalation. Local tumor recurrence rates of Stage I NSCLC after STI with a BED of 99–137 Gy were 0–6% for a median follow-up period of 19–60 months.5–7, 27, 28 The comparatively high local disease recurrence rate reported by Hof et al.29 may be attributable to lower BED. In the current study, the local control rate was 91.9% for BED ≥ 100 Gy. For BED < 100 Gy, the local control rate was poor, particularly in patients with Stage IB disease. Given our clinical results, additional dose escalation studies may be possible. However, patients receiving BED ≥ 120 Gy or ≥ 140 Gy did not display significantly better local control rates than patients receiving lower BED, even for patients with Stage IB disease. Satisfactory BED to achieve local control for Stage I NSCLC is approximately 100 Gy. Representative examples of dose regimens performed in the current study that provided approximate BEDs > 100 Gy were 48 Gy/4 fractions or 50 Gy/5 fractions. However, treatment outcomes for patients who received conventional irradiation before STI in our study were not significantly different from those of other patients. Although a longer follow-up is necessary to determine final control rates of tumors in our study, local control rates for STI may be equivalent to surgical results, as most local disease recurrences generally occur within 3 years after treatment.18
Table 5. Comparison of STI Methods and Local Control Rates for Stage I Nonsmall Cell Lung Carcinoma
In our study, the overall survival rates were excellent for limited patients considered operable before treatment and with BED ≥ 100 Gy. The 88% three-year overall survival rate in operable patients treated with BED ≥ 100 Gy was consistent with single institutional results (a 3-year overall survival rate of 88% in 29 medically operable patients) reported by Uematsu et al.5 The patients in that study (from the Medical Defense College, Saitama, Japan) were not included in the current multiinstitutional study. Survival rates after STI for BED ≥ 100 Gy may well match those after lobectomy for Stage I NSCLC. We believe that good treatment outcomes from STI depend on a high BED, a large single dose, a short treatment period, and delivery of a modest dose to a large lung volume. STI can reduce substantially overall treatment time from several weeks of conventional radiotherapy to a few days, offering important advantages to the patient.
In conclusion, hypofractionated high-dose STI with BED < 150 Gy represents a feasible and beneficial method for obtaining curative treatment of patients with Stage I NSCLC. Local control and survival rates were better for BED ≥ 100 Gy than for BED < 100 Gy for all treatment methods and schedules. Survival rates for STI in selected patients (medically operable and BED ≥ 100 Gy) were excellent and reproducible among institutions, irrespective of specific treatment methods, and were potentially equivalent to those of surgery. The current study was a retrospective review, and unknown selection biases for treated and analyzed patients may have been present. Moreover, treatment parameters were very heterogeneous. However, STI may become a standard radical treatment strategy for Stage I NSCLC, at least for compromised patients. More patients and longer follow-up, or a prospective Phase II study based on a single treatment schedule followed by a Phase III trial comparing surgical outcomes with those of STI, are necessary to determine standard treatments for Stage I NSCLC.