The authors developed fluoroscopic real-time tumor-tracking radiation therapy (RTRT) by insertion of a gold marker using bronchofiberscopy to reduce uncertainties in organ motion and set-up error in external radiotherapy for moving tumors. The purpose of the current study was to evaluate RTRT's feasibility in lung carcinoma treatment.
The three-dimensional position of a 1.0-2.0 mm gold marker in or near the tumor was detected by two sets of fluoroscopies every 0.03 seconds. The treatment beam was gated to irradiate the tumor only when the position of the marker coincided with its planned position using the RTRT system. Bronchofiberscopic equipment for insertion of the marker into the lung tumor was developed and used for 20 lung tumors in 18 patients. Patients were given high dose hypofractionated focal irradiation (35–48 Gy in 4–8 fractions in 4–10 days) with a planning target volume margin of 5 mm for the tumor.
The markers were successfully inserted and maintained at the inserted position during and after the radiotherapy in 14 (88%) of 16 peripheral-type lung tumors and in none of four central-type lung tumors, indicating that this method of RTRT was not feasible for central-type lung tumors. Tracking of the marker was successfully performed in 1 of 2 tumors with a 1.0 mm marker and in all of 12 tumors with a 1.5–2.0 mm marker. On the whole, 13 (65%) of the 20 tumors were successfully treated with RTRT. Local tumor control was achieved and maintained for all 12 patients (13 tumors), who were treated with RTRT, with a median followup of 9 months (range, 5–15). Localized radiation pneumonitis was found radiographically at the lung volume that was irradiated with about 20 Gy, without symptoms in all but one patient.
High-dose, focused radiation therapy is attractive in the treatment of small lung cancers, particularly for elderly or medically compromised patients, since in these patients surgical resection of tumors is life-threatening and another highly curative treatment modality is needed. However, the treatment results of localized mega-voltage x-ray therapy have not always been satisfactory. Morita et al. showed that 57% of Stage I nonsmall cell lung carcinomas (NSCLCs) died of local tumor regrowth within five years after conventional radiotherapy (55–74 Gy, mean 64.7 Gy; 2 Gy per fraction and five fractions per week).1 Kaskowitz et al. reported that local primary tumor progression occurred in 22 out of 53 Stage I NSCLC patients treated with definitive radiotherapy alone, resulting in a 51% three-year actuarial freedom from local progression (39.9–79.2 Gy, median 63 Gy; data on dose per fraction, not available).2 Radiotherapy was applied for the elderly patients or medically compromised patients in both of these studies.
Failure to control a tumor with radiation therapy may be due to insufficient dose of radiation or inadequate coverage of the tumor with the treatment portal. Recently, imaging analysis showed that respiratory movement of the tumor causes considerable changes in the area and displacement of the tumor edge.3 In conventional radiotherapy, a large proportion of normal tissues may be unnecessarily irradiated, or part of the tumor may not be satisfactorily irradiated because of uncertainty regarding tumor location. It may be more appropriate to irradiate the tumor by gating the timing of irradiation to the movement of the tumor. Treatment planning should incorporate information regarding organ motion in the individual patient and set-up error in each treatment session.4 However, it has been difficult to decrease the safety margin of lung tumor because of insufficiency in visualizing movement of the tumor and in gating techniques.
We previously reported that a new radiotherapy system can gate the irradiation to the movement of the tumor itself by insertion of an internal marker in or near the tumor and use of a fluoroscopic tracking system.5 In the current study, we investigated the safety and early tumor response of real–time tumor–tracking radiation therapy (RTRT) for lung carcinoma by the aid of insertion of a gold marker using bronchofiberscopy.
PATIENTS AND METHODS
Real-time Tumor-tracking System
The fluoroscopic real-time tumor-tracking system (Mitsubishi Electronics Co. Ltd., Tokyo, Japan; Fig. 1) consists of four sets of diagnostic X-ray television systems, a moving object recognition system, and a patient couch composed of carbon fiber placed in the linear accelerator room. The details have been described elsewhere.6, 7 In brief, the diagnostic X-ray television system is composed of an X-ray tube embedded under the floor, an image intensifier mounted on the ceiling, and a high-voltage X-ray generator. All four sets of the system are adjusted such that the central axis of the diagnostic X-rays will cross at the isocenter of the linear accelerator. Two of the four X-ray television systems are selected to display the gold marker in the patient's body during radiotherapy. When the appropriate two sets are used, the gantry of the linear accelerator does not interfere with the fluoroscopic fields.
Coordinates of the tumor center and the gold marker are transferred from the three-dimensional radiotherapy planning system to the RTRT system through the network. The information is transformed using projective geometry and overlapped on the two X-ray television images displayed on the cathode ray tube monitor. If necessary, the location of the patient couch is adjusted such that the marker will cross over the planned coordinates.
The three-dimensional position of the marker is automatically determined during irradiation as follows. The markers are first found in the digital images by means of a template matching algorithm, which is performed using special hardware. A fluoroscopic image is digitized into 1024 × 1024 pixels with a pixel size of 0.09 × 0.09 mm2. The pattern of the gold marker in the X-ray image is already registered as a template image of 24 × 24 pixels. Image processors compare the digitized image and the template image of a metallic marker to detect the location of the marker, which is considered representative of the position of the tumor. A recognition score, derived from the regression co-efficient between the searched area and the template, is then calculated. The location x, y that gives the highest correlation is considered to be the marker position in the image. If the correlation score is below a certain value, a machine interlock is issued. The same procedure is applied to both images, rendering tumor marker coordinates A (XA, YA) and B (XB, YB). In the central processing unit, the tumor marker coordinates are corrected for image intensifier distortion, and converted to straight lines using fluoroscopic transformation matrices, which are stored in advance during the calibration procedure. The 3D coordinates of the tumor marker are calculated by intersecting these two lines. If the coordinates of the marker are within the limits of the predetermined permitted dislocation, the system allows the linear accelerator to irradiate the patient (Fig. 2). These calculations are performed within 0.03 seconds. The time delay from the matching of the coordinates to the start of irradiation is 0.09 seconds. The phantom experiment showed that the geometric accuracy of the tumor-tracking system is better than 1.5 mm for moving targets up to a speed of 40 mm/s. As long as the marker is situated within 50 mm from the center of the isocenter (i.e., the center of the tumor), the localization accuracy was better than 1.5 mm. The accuracy does not depend on the location or size of the target volume.
Insertion of the Gold Marker
Insertion of the gold marker (99.9% Au) was an essential part of the RTRT as an interface between virtual reality during the planning and actual reality in the treatment room. We developed a technique to insert the marker into or near lung tumors. The procedure consisted of five steps, as follows. 1) A bronchofiberscope (BF 1T-240, Olympus Co. Ltd., Tokyo, Japan) was placed at the target bronchus. The patient was under local anesthesia in a supine position. 2) A catheter made of polytetrafluoroethylene was inserted through the bronchofiberscope channel using video-guidance. The tip of the catheter was visualized by the fiberscope during the following procedure. 3) A spherical 1.0, 1.5, or 2.0 mm gold marker was inserted into the catheter from the back of the catheter. 4) A hard plastic wire was inserted from the back of the catheter, pushing the gold marker to the top of the catheter. 5) The gold marker was inserted into the tumor or wedged at one of the bronchial trees near the tumor by observing the top of the catheter on the video-image (Fig. 3). The whole procedure took about 10 minutes per patient.
Between May 1999 and October 2000, patients were entered into the current RTRT study for lung tumors 1) when they met the criteria of a pathologic diagnosis of NSCLC and were not suitable for surgery and chemotherapy because of old age, poor respiratory, cardiac, and renal function, or patient refusal, or 2) when patients with an expected survival of longer than one year were observed to have one or two metastatic lung tumors that could be symptomatic. All 18 patients who met the above criteria agreed to be treated with RTRT after giving informed consent: 11 males and 7 females, with a mean age of 68.5 years (range, 43–87; Table 1). Only 1 of the 18 patients (Patient 15 in Table 1) was previously treated with thoracic radiation therapy eight months before the RTRT. Among these 18 patients, there were a total of 20 lesions: 4 central-type and 16 peripheral-type. A central-type tumor was defined as a tumor located in the trachea, main bronchus, or lobar bronchus. A peripheral-type tumor was defined as a tumor located in the segmental bronchus or a more peripheral region.
Table 1. Characterics of Patients Who Enrolled in the Current Study
Diagnosis & Pathology
Tumor size (mm)
Marker size (mm)
Reason for no surgery
Recurrent: recurrent lung cancer; Primary: primary lung cancer; Metastatic: metastatic lung cancer; Sq: squamous cell carcinoma; Ad: adenocarcinoma; ACC: adenoid cystic carcinoma; Tr: trachea; LMB: left main bronchus; LUL: left upper lobe; RLL: right lower lobe; RUL: right upper lobe; LLL: left lower lobe; NM: not measurable by computed tomography.
Patients were given high dose hypofractionated focal irradiation with a planning target volume (PTV) margin of 5 mm for the tumor. Three-dimensional conformal irradiation using five to seven static beams was performed. Dose schedule was 35–40 Gy in four fractions in principle. No chemotherapy was given throughout the study period. Patient characteristics are shown in Table 1.
Two orthogonal chest radiographies at the planned respiratory phase were taken just after the insertion of the marker on the same day, a day after the insertion, and during treatment period to monitor the marker position. Large dislocation of the marker, such as the drop of the marker from the inserted position, can be easily detected by the chest radiography. Small migration of the marker requires estimation by computed tomography (CT) measurement, of which results for solid organs such as liver and prostate have been published elsewhere.8 The CT measurement for lung is more complex because of respiratory change of lung volume and is beyond the scope of this paper. In the current study, the migration of the marker for lung tumor was estimated using the distance between the marker and the tumor and anatomical bronchial structure on the followup CT scan taken one month after the treatment, which usually does not yet show radiation pneumonitis and still has residual tumor. Followup CT scans were transferred to the three-dimensional radiotherapy planning (3DRTP) system, and the contours of the tumor and the marker were determined again to measure the distance. Possible decrease in tumor size was not taken into account in the measurement.
Examinations with CT scan and/or bronchofiberscopy were performed to estimate the responsiveness of RTRT every month for the initial three months and approximately every three months thereafter. Tumor sizes were measured bidimensionally with CT scan for the response evaluation. Complete response (CR) was defined as the total disappearance of all clinically detectable disease for at least four weeks. Partial response (PR) was defined as a 50% or more decrease in tumor size for at least four weeks without the appearance of new lesions or progression of any lesion. Stable disease (SD) was defined as less than a 50% decrease in total tumor size or less than a 25% increase in the size of measurable lesions. Progressive disease (PD) was defined as a 25% or more increase in the size of measurable lesions or the appearance of new lesions. Radiation pneumonitis and fibrosis were evaluated by CT image according to the scoring system for late effects of normal tissues (LENT):9 Grade 1, radiologic abnormality; Grade 2, patchy dense abnormalities on radiograph; Grade 3, dense confluent radiographic changes limited to radiation field; Grade 4, dense fibrosis, severe scarring, and major retraction of normal lung. Radiographic differentiation between tumor recurrence and radiation pneumonitis/fibrosis was performed with the characteristics and changes of shadows in the observation by periodic CT scan.
Insertion of the Gold Marker
The gold marker was successfully inserted in 19 of 20 tumors (95%; Fig. 4), but in the remaining central-type tumor it could not be inserted because it was so hard (Patient 4). The patient in whom the marker could not be inserted was the second case from the start of the current study and was seen at time when the bronchofiberscopic procedure was not fully developed. The marker dropped from the inserted position within 24 hours in four tumors; two were central-type tumors (Patients 2 and 3) and two were peripheral-type tumors in the left superior segment (Patients 11 and 16). The marker dropped within a week in one patient with a central-type tumor (Patient 1). The marker in Patient 1 was in the tumor itself, which decreased in size during irradiation and probably could not hold the marker. On the whole, the markers were held at the same position throughout the treatment period in 14 patients (74%) and dropped from the inserted position in the remaining 5 patients (26%). The 5 patients in whom the markers dropped from the inserted position were among the first 10 patients from the start of the current study. The peripheral-type tumors (14 out of 16, 87%) were statistically better than the central-type tumors (0 out of 4) for insertion and successful treatment (P < 0.01). All six 2.0 mm gold markers, all seven 1.5 mm gold markers, and one of two 1.0 mm gold markers were visible on the fluoroscopy and used for real-time tracking. One 1.0 mm marker in a patient with a peripheral-type tumor (Patient 7) was not visualized well on fluoroscopy, which was therefore unusable for tracking the marker, indicating that the markers greater than 1.0 mm were appropriate for tumor tracking.
Put in order, real-time tumor-tracking radiotherapy was thus begun for fourteen (79%) lesions. Among these 14 lesions, a marker dropped during RTRT from 1 lesion (Patient 1). Overall, 13 (65%) out of the initial 20 eligible lesions were treated with RTRT throughout the whole treatment period. Dose fractionation for each patient is shown in Table 2. The remaining seven lesions were treated by stereotactic radiation therapy without tumor-tracking with a larger PTV margin of 10 mm for lateral and antero-posterior directions and 15 mm for cranio-caudal direction to account for set-up error and internal error due to respiratory motion after the drop of the marker.
Table 2. Fourteen Tumors from 12 Patients Treated with RTRT
The marker was inserted into the tumor tissue in eight lesions and near the tumor in the other five lesions (median, 35.4 mm; range 16.0 mm to 45.3 mm, from the isocenter; Table 2). For those seeds that remained in place, the distance from isocenter was within the limits determined as needed in the phantom studies (i.e., within 50 mm from the center of the isocenter). No change in position was apparent in these 13 lesions in the orthogonal chest radiographies and followup CT scans (median, 0.6 mm; range, 0.0 mm to 2.9 mm change from the planned position).
Tumor response was evaluated by CT image and/or bronchofiberscopy in 11 of 14 lesions, except in 3 lesions for which 2 cavitary lesions (Patient 18) or radiation pneumonitis due to previous radiotherapy (Patient 15) made the estimation difficult. The response rate was 100% (11 out of 11; CR: 18%, PR: 82%). Neither SD nor PD were observed. Response rate was 100% both in NSCLC primary lung tumors (6 out of 6; CR: 17%, PR: 83%) and in metastatic or recurrent lung tumors (5 out of 5; CR: 20%, PR: 80%). There was no apparent difference in response rate between those with a marker in the tumor (8 out of 8, 100%) and those with a marker near the tumor (3 out of 3, 100%). In five tumors treated with stereotactic radiation therapy without tumor-tracking, no CRs, one PR, three SDs, and no PDs, were observed; in the remaining patient followup data were not obtained. The response rate was 25% (1 out of 4). Overall response rate in patients enrolled in the current study was 80% (12 out of 15).
Radiation-induced pneumonitis was found in 11 of 12 patients treated with RTRT on chest x-ray and/or CT scan. All shadows of pneumonitis were localized at or near the PTV. The mean appearance time from the irradiation was 2.8 months (range, 1–6). All patients but one were asymptomatic despite the radiographic appearance of pneumonitis. One patient (Patient 13) with recurrent primary lung adenocarcinoma, having low pre-irradiation vital capacity (1.8 liters) after surgery, required the administration of oral steroid for one month because of cough and fatigue associated with the pneumonitis on CT images (LENT score Grade 3) at three months after irradiation. She became asymptomatic one month after the initiation of steroid intake and was alive without relapse of tumor at 15 months after irradiation. Her vital capacity was 1.7 liters at last followup. In five tumors treated with stereotactic radiation therapy without tumor-tracking, radiation-induced pneumonitis was found in three out of four patients, and in the remaining patient followup data were not obtained. Overall, radiation-induced pneumonitis was found in 14 of 16 patients in the current study.
Among 12 patients treated with RTRT, one patient (Patient 3) died of acute bacterial pneumonia at 10 months after treatment without evidence of local recurrence or radiation pneumonitis, and another patient (Patient 12) died of lymphangitis carcinomatosa of the lung at 15 months after treatment without evidence of local recurrence. The remaining 10 patients were still alive without apparent local recurrence in median observation period of 9 months (range, 5–15).
Currently, there are two major methods to reduce the uncertainty of lung tumor location caused by respiratory motion. One includes active breathing control methods, in which patients hold their breath at a certain respiratory level during the irradiation.10–12 However, patients with poor pulmonary function such as those in the current study cannot hold their breath even for a few seconds. The other method is the respiration-gated intermittent irradiation system, by which the movements of the skin surface or other physiologic parameters are monitored.13–17 It has been used in particle therapy for over 10 years with encouraging results.18 However, it has been suggested that respiration depth is not always consistent with skin movement.4 To date, inter-fractional and intra-fractional baseline shift of the tumor location has not been fully investigated.
Optimally, the actual shape of the tumor can be seen completely in real-time during irradiation. Computed tomography on the linear accelerator's gantry has the potential to visualize the shape of the tumor, but at present the image cannot be generated in real-time during irradiation. Fiducial markers in or near the tumor are expected to represent the movement of the tumor in RTRT to some extent. The RTRT system has been shown to reduce translational set-up error and internal error with an accuracy of ± 1.5 mm for moving objects in a solid phantom.7 We expect that this technique will considerably reduce the uncertainty of moving tumors in radiotherapy. However, it is still not certain whether the marker in or near the tumor is sufficient to represent the actual movement of the tumor. In fact, precise estimation of the marker movement is difficult because of the slice thickness of the CT scan and respiratory change of lung volume. Systematic and random error of the distance between the marker and the tumor center may depend on the proximity of the marker to the tumor, the location or size of the target lesion, and the size of the marker. Conventional CT scan at every treatment can be useful but not perfect because respiratory change of lung volume would change the distance between the tumor center and the marker. Multi-detector CT scans located in the treatment rooms will help reduce the influence of respiratory change, but the cost makes this an impractical solution for most treatment centers. We are now improving the system by using three markers for one tumor to reduce the uncertainty about marker migration.
To our knowledge, the current study is the first report on the clinical feasibility of RTRT for patients with lung carcinoma. In the current series, the insertion of a gold marker using bronchofiberscopy was safely performed without apparent complications. However, for central-type tumors, it was difficult to keep the gold markers in place throughout the treatment period, probably due to the fragility of the tumor tissue. Conversely, for peripheral-type tumors, it was possible to insert the marker into the tumor or wedge the marker into one of the narrow bronchial trees. To push the marker through the flexible catheter was difficult when the fiberscope needed to be angled more than 90 degrees. Tumors at the upper lobe and superior segment presented difficulties for inserting and holding the marker. Stenosis or deformation of the bronchus after surgery was also an obstacle for insertion of the marker. Encouragingly, the marker was not observed to migrate from the initially inserted position in peripheral lesions in followup radiographies and CT. We suppose that the gold marker was not displaced due to fibrotic adhesion following inflammation in response to the insertion of the gold marker. The skill of the operator is an important factor for successful implantation, considering that unsuccessful implantation occurred in the initial cases in the current study, when the operator had less experience. A new insertion technique for central lesions is under consideration.
Initial response to RTRT was excellent. No progression of the disease was observed, although the followup period was short. Because our response criteria were based entirely on changes in tumor size as judged on CT images or bronchofiberscopy, the response rate may have been underestimated as a result of the formation of local lung fibrosis within several months of treatment. As expected, radiation-induced pneumonitis occurred at the treated area in almost all cases on CT images. Several authors have reported that precise set-up using CT scan in the linear accelerator room can achieve a high local control rate for small lung tumors without gating or tracking during radiotherapy.16, 19 Comparative studies between radiation therapies with and without real-time tumor-tracking are required to confirm this.
In conclusion, a 1.5–2.0 mm gold marker was safely inserted in or near peripheral-type lung tumors through bronchofiberscopy and used efficiently in the fluoroscopic RTRT system. Larger studies with a longer followup are needed not only to investigate tumor response and radiation-induced pneumonitis but also to determine the relapse, survival, and long-term side effects relative to the clinical benefits of this system.