Can bronchoscopically implanted anchored electromagnetic transponders be used to monitor tumor position and lung inflation during deep inspiration breath-hold lung radiotherapy?

Abstract

Purpose

To evaluate the efficacy of using bronchoscopically implanted anchored electromagnetic transponders (EMTs) as surrogates for 1) tumor position and 2) repeatability of lung inflation during deep-inspiration breath-hold (DIBH) lung radiotherapy.

Methods

41 patients treated with either hypofractionated (HF) or conventional (CF) lung radiotherapy on an IRB approved prospective protocol using coached DIBH were evaluated for this study. Three anchored EMTs were bronchoscopically implanted into small airways near or within the tumor. DIBH treatment was gated by tracking the EMT positions. Breath-hold cone-beam-CTs (CBCTs) were acquired prior to every HF treatment or weekly for CF patients. Retrospectively, rigid registrations between each CBCT and the breath-hold planning CT were performed to match to 1) spine 2) EMTs and 3) tumor. Absolute differences in registration between EMTs and spine were analyzed to determine surrogacy of EMTs for lung inflation. Differences in registration between EMTs and tumor were analyzed to determine surrogacy of EMTs for tumor position. The stability of the EMTs was evaluated by analyzing the difference between inter-EMT displacements recorded at treatment from that of the plan for the CF patients, as well as the geometric residual (GR) recorded at the time of treatment.

Results

219 CBCTs were analyzed. The average differences between EMT centroid and spine registration among all CBCTs were 0.45±0.42cm, 0.29±0.28cm, and 0.18±0.15cm in superior-inferior (SI), anterior-posterior (AP) and lateral directions, respectively. Only 59% of CBCTs had differences in registration <0.5cm for EMT centroid compared to spine, indicating that lung inflation is not reproducible from simulation to treatment. The average differences between EMT centroid and tumor registration among all CBCTs were 0.13±0.13cm, 0.14±0.13cm and 0.12±0.12cm in SI, AP and lateral directions, respectively. 95% of CBCTs resulted in <0.5cm change between EMT centroid and tumor registration, indicating that EMT positions correspond well with tumor position during treatments. Six out of the 7 recorded CF patients had average differences in inter-EMT displacements to be ≤0.26cm and average GR ≤0.22cm, indicating that the EMTs are stable throughout treatment.

Conclusions

Bronchoscopically implanted anchored EMTs are good surrogates for tumor position and are reliable for maintaining tumor position when tracked during DIBH treatment, as long as the tumor size and shape are stable. Large differences in registration between EMTs and spine for many treatments suggest that lung inflation achieved at simulation is often not reproduced.

I. INTRODUCTION

Localizing lung tumors during lung radiotherapy is challenging and necessary due to the fact that lung tumors can move up to 30 mm or more ^1,2 and motion can vary from day to day ^3,4. Deep inspiration breath-hold (DIBH) can improve lung radiation treatment by minimizing tumor motion and inflating the lungs, thus reducing the volume of healthy lung tissue receiving high dose ^5–11. In some patients, DIBH can move the tumor away from the heart or spine, thus reducing dose to these organs^7,8,12,13. Expanding healthy lung volume is beneficial since the dosimetric indicators that have been found to correlate with radiation pneumonitis, such as the percent volume of the healthy lung receiving ≥20Gy ^14–17 and the mean lung dose^14,16–19 depend inversely on the total lung volume. DIBH reduces tumor motion and so allows for the reduction or omission of an internal target volume (ITV) based on a free-breathing 4D CT scan, thus reducing target margins, which can result in decreasing dose to healthy lung without compromising tumor control. However, it is hard to validate that the gross tumor volume (GTV) remains within the treated volume for the duration of the DIBH treatment and that the lung anatomy during treatment is consistent with what was seen during simulation.

External surrogates, such as a reflective block on the chest, or internal surrogates, such as the diaphragm, have been used to track tumor motion, but external patient anatomy and internal diaphragm position do not correlate well with internal tumor position and motion ^20–24. Radio-opaque fiducials have been used as internal surrogates to help localize targets and track the tumor during the course of treatment, but they are less commonly used in the lung due to limitations in the ability to safely place the markers and keep them from migrating²⁵. A real-time tracking and motion management platform (Calypso^® , Varian Medical Systems, Palo Alto, CA) has been shown to be successful for real-time tracking of inter- and intra-fraction motion for prostate cancer patients using electromagnetic transponders (EMTs)^26–29. More recently, the Calypso^® platform has incorporated a new, FDA cleared anchored EMT (Varian Medical Systems, Palo Alto, CA) designed to track lung tumor motion in real-time for intra- and inter fraction motion^30–36. The anchored EMTs are bronchoscopically implanted into small airways near or within the tumors of lung cancer patients and can be used for tracking with the Calypso^® system interfaced with the medical linear accelerator (LINAC) to gate the beam to only deliver radiation when the EMT locations are within the user’s pre-defined tolerance limits.

Several clinical studies have investigated the clinical feasibility of using anchored EMTs for tumor tracking with the Calypso system. These studies found that implantation and tracking/gating are feasible and that EMTs do not migrate and remain in stable positions throughout treatment ^{31–33,35–39}.

No studies to date have looked at whether EMTs are good indicators of the tumor position during a course of DIBH lung radiotherapy. The lung and tumor anatomic geometry is very dynamic; tumors may grow or shrink, patients may hold their breath in different ways and other conditions, such as atelectasis, may change the lung geometry, so tracking tumor position during DIBH treatment is important. Willmann et al. found that the EMTs are good surrogates for predicting tumor motion in the superior-inferior (SI) direction, but the study was based on a single 4D CT scan acquired during simulation and did not investigate longitudinal data⁴⁰. Additionally, no studies have investigated whether the EMTs can be a good indicator of the repeatability of breath-hold achieved during lung DIBH treatments. Han-Oh et al. found that the inter-breath-hold variation, especially in the superior-inferior (SI) direction is significant for pancreatic SBRT treatments by looking at both GTV and fiducial locations from multiple acquired breath-hold CBCTs during treatment, however this study did not look into breath-hold repeatability for lung patients⁴¹. The aims of this study are to evaluate if EMTs are good surrogates for 1) tumor position throughout the course of treatment and 2) repeatability of lung inflation, and hence anatomic geometry, from simulation to treatment for DIBH lung radiotherapy.

II. METHODS

II.A. Patient Selection and Characteristics

45 patients were enrolled on an IRB approved prospective protocol (clinicaltrials.gov: NCT02111681) to treat inoperable thoracic malignancies using coached DIBH treatment guided by bronchoscopically implanted anchored EMTs tracked by the Calypso^® real-time tracking and motion management platform (Varian Medical Systems, Palo Alto, CA). Anchored EMTs were provided by Varian Medical Systems. Patients who were able to hold their breath for at least 5 breath-holds for more than 20 seconds each and who were planned to undergo lung radiotherapy and were willing to undergo implantation were eligible for the study. The 5 breath-hold inclusion criterion was a minimal clinical screening requirement for patient eligibility for the protocol; it was expected that typical treatments would require more than 5 breath-holds. Exclusion criteria included patients with implants in the chest region that contain metal, patients with active implanted devices, such as pacemakers, defibrillators and drug infusion pumps, patients who are unable to tolerate anesthesia or flexible bronchoscopy, and pregnant patients. After determination of breath-hold eligibility, the patient was evaluated by the pulmonologists as to whether they would be suitable for implantation of the EMTs. Only after meeting all criteria were they enrolled in the protocol. Patients were treated with either hypofractionation (HF) or Conventional Fractionation (CF). HF was defined as treatments with 8 fractions or less. Table 1 shows the patient characteristics for the patients enrolled in the protocol.

Table 1:

Patient characteristics for the 45 patients enrolled on IRB approved prospective protocol for patients getting treated with Hypofractionation (HF) or Conventional Fractionation (CF).

		HF	CF
Sex	Male	15	4
	Female	22	4
Age (yr)	Mean	67	67
	Standard Deviation	13	10
PTV volume (cc)	Mean	67	610
	Standard Deviation	98	450
Stage	I	13	1
	II	2	2
	III	2	5
	IV	20	0
Tumor location	RUL	5	2
	RML	1	0
	RLL	12	2
	LUL	7	3
	LLL	10	1
	Hilum	2	0
Prescription	Median Total Dose (cGy)	5000	6375
	Median # of fractions	4	30

II.B. EMT Implantation

Three anchored EMTs were placed bronchoscopically under general anesthesia into small airways in or near the tumor. EMTs were implanted using either the superDimension^® electromagnetic navigational bronchoscopy (ENB; Medtronic, Minneapolis, MN) or the Ion^® shape-sensing robotic-assisted bronchoscopy (ssRAB; Intuitive, Sunnyvale, CA). These guided bronchoscopy platforms allow for minimally invasive airway navigation to access the periphery of the lungs with higher accuracy than conventional bronchoscopy alone ^42–44. Both ENB and ssRAB allow transbronchial forceps biopsies, fine needle aspirations and fiducial marker placements with a reduced risk of pneumothorax as compared to transthoracic approaches^42,44. The EMTs were implanted under 2D fluoroscopic guidance and a verification radiograph was obtained to confirm appropriate placement.

II.C. CT Simulation and Treatment Planning

After a minimum of 3 days from the EMT implantation, the patient underwent a CT simulation for radiation treatment planning. The Calypso® array is positioned a few cm above the patient and cannot pick up the EMT signal if it is more than 21 cm from the EMT. To ensure that the array would pick up the signal at treatment, a lateral scout view with the patient supine was acquired at simulation. If the distance measured on this view from the most posterior EMT to the most anterior part of the patient was larger than 17 cm, the patient was immobilized, simulated and treated prone with Calypso®. The immobilization for each patient consisted of an alpha cradle mounted in a device that indexes to the couch, to ensure external positional reproducibility at each treatment. Each patient received a Deep-Inspiration CT (DI-CT) acquired in a single breath-hold for DIBH simulation; a 4DCT and a free-breathing CT were obtained per departmental practice as back up to allow the patient to be replanned for a free-breathing treatment (without Calypso^®) if they were unable to perform breath-hold at treatment. The DI-CT had slice thickness of 1.25mm to ensure sufficient EMT resolution. The 4DCT and DI-CT were acquired using the Real-time Position Management (RPM) system (Varian Medical Systems, Palo Alto, CA).

Target volumes were contoured by the physician and organs at risk were contoured by either the physician or dosimetrist on the DI-CT. The dosimetrist contoured the EMTs according to vendor instructions. The EMT centroid coordinates were obtained using an automatic tool within the Eclipse Treatment Planning System (TPS) (Varian Medical Systems, Palo Alto, CA). A treatment plan with the property “Use gating” was made per our clinical practice for the prescribed fractionation for each patient in the TPS. The isocenter and EMT centroid coordinates based on the EMT contours from the DI-CT, along with the gating tolerances were input into the Calypso^® system prior to treatment and are referred to here as the Calypso^® plan.

II.D. Treatment

All patients were treated on a Truebeam LINAC (Varian Medical Systems, Palo Alto, CA). The Calypso system locates each of the implanted EMTs with respect to the isocenter. The accuracy of the system is checked daily by independently setting a phantom containing EMTs to the isocenter using lasers. After patient setup, and with the radiolucent antenna array positioned just above the patient, the Calypso^® system reports, in real-time, the displacement of the mean EMT position from the planned EMT mean position. If the EMTs were perfect surrogates, this would display the target setup error.

The following took place in the treatment room at each treatment. The patient was initially set up for alignment, rotation and roll using external tattoos and room lasers. Then, the therapists coached the patient into DIBH ¹⁰. During the breath hold, the therapists adjusted the couch position using the in-room Calypso^® console to bring the target into the planned alignment with the radiation isocenter. The couch is adjusted to align the real-time mean EMT position with the planned position. This was followed by one or two more verification breath-holds. Because Calypso^® tracking was active, the therapists were able to make sure the patient understood what to do and was prepared for the treatment procedure.

After leaving the treatment room, the therapists acquired the following images. Per protocol, two orthogonal radiographs were acquired during DIBH to verify the EMTs were present and in their correct relationship with each other. The radiographs were not used for set-up. A 40 second spotlight CBCT scan, requiring a gantry rotation of 200° without couch centering, was acquired during DIBH for every fraction for the HF patients and weekly for the CF patients. This allowed the Calypso^® system to monitor the EMT positions during CBCT acquisition. If the displacement from planned EMT positions exceeded 2 mm during the CBCT acquisition, or if the patient restarted breathing, the scan was manually interrupted. CBCT acquisition was then completed during a second or third breath-hold. Due to the novelty of using the anchored EMTs in the lung, the CBCT was done to check that the geometric relationship of the EMTs and tumor centroid was unchanged, thus validating the Calypso^® plan with imaging. The CBCT was reviewed by a radiation oncologist to verify that the match between the GTV in the CBCT and the planning CT was acceptable according to departmental image-guidance criteria. After setup and pre-treatment imaging, the patient was treated in DIBH, with gating from the Calypso^® system based on the verified positions of the EMTs.

II.C. Evaluation

A retrospective analysis was performed to determine the surrogacy of the EMTs for 1) tumor position and 2) repeatability of lung anatomy throughout the treatment. Rigid registrations were manually performed with in-house software between each Deep-Inspiration CBCT (DI-CBCT) acquired and the planning DI-CT to match to spine, EMTs and tumor. First, the DI-CT and DI-CBCT spines were registered; this registration determined the zero position for that day. For the EMT registrations, a registration was performed to calculate the shifts from the zero position (with respect to the spine) to match the DI-CBCT and DI-CT to each of the 3 EMTs separately; then the EMT centroid (EMT_cent) was defined as the centroid of the triangle formed by the 3 individual EMT centroids. More specifically, the EMT_cent registration coordinates are calculated by averaging the x, y and z coordinates from each separate EMT registration. For tumor registration, the visually apparent soft tissue mass that was defined as the GTV on the planning scan was used for matching. Differences in image shifts for the best registration between the EMT_cent and the spine in the x (left-right), y (anterior-posterior) and z (superior-inferior) were analyzed to determine the surrogacy of EMTs for repeatability of lung anatomy. Differences in lung anatomy between breath holds are largely due to differences in lung inflation. The length of the CBCT scans is limited to 15cm, so the entire lung was not captured in the scan, limiting the ability to calculate the total lung volume to compare with the planning scan. The differences in registration between EMT_cent and the tumor in the x, y and z directions were analyzed to determine the surrogacy of the EMTs for tumor position. All differences analyzed were in absolute value.

In addition to the quantitative registration-based analysis, stability of the EMTs was also evaluated by analyzing the difference between inter-EMT displacements recorded by the Calypso^® system and the inter-EMT displacements from the Calypso^® plan. This was of particular interest for the CF patients, since their treatment courses cover a longer period of time during which the GTV and surrounding lung may change^45,46. At each fraction, the Calypso^® system records the inter-EMT distances between the three pairs of EMTs (AB, BC, and CA), as well as the Geometrical Residual (GR) which is defined as the root mean squared error of the discrepancy in inter-EMT positions between the planned and Calypso^® detected geometries. The GR defines how well the EMT geometry entered as part of the Calypso^® plan matches with what is detected by the Calypso^® system. A small value of GR indicates that the geometry of the EMTs detected matches well with the planned geometry.

III. RESULTS

Of the 45 patients enrolled, 4 patients were excluded from the quantitative analysis of this study due to body habitus, tumor shape change or a limited field of view in the CBCTs. 41 patients were simulated and treated supine and 4 patients were simulated and treated in the prone position. For 3 of the CF patients, only the first 2 or 3 CBCTs were used because the tumor evidenced some shrinkage during treatment. Since for these cases the tumor boundaries could not be determined based solely on the CBCT without other diagnostic scans, the tumor registration was not performed for these later CBCTs. For one patient only 2 of the 3 EMTs were used in their Calypso^® plan because one was too far from isocenter to be used by Calypso^®. A total of 41 patients were used in the analysis, and a total of 219 CBCTs were registered and analyzed. The average number of CBCTs per patient was 5.4.

III.a. Registration between EMTs and Spine

Table 2 shows the results of the absolute difference between EMT_cent and spine registrations averaged across all 219 CBCTs along with the corresponding standard deviations. It also shows the absolute difference in tumor and spine registrations averaged across all 219 CBCTs along with the corresponding standard deviations, which yielded very similar results. The largest difference in registration was seen in the SI direction. The average difference in registration between EMT_cent and spine was >0.2cm in both the SI and AP directions, indicating that the sub-volume of the lung, represented by the EMTs, can change position relative to the spine from simulation to treatments and between treatments. Figure 1 is the histogram of the difference in registration between EMT_cent and spine for all CBCTs, visually showing a large spread in values. Only 59% of all evaluable 219 CBCTs had a registration difference between EMT_cent and spine of < 0.5cm in any direction, which again suggests lung inflation changes from simulation to treatment.

Table 2:

Mean and standard deviations for absolute differences in registration between EMT_cent with spine, tumor with spine, and EMT_cent with tumor in superior-inferior (SI), anterior-posterior (AP) and lateral directions.

	| EMTcent – Spine | [cm]	| Tumor – Spine | [cm]	| EMTcent – Tumor | [cm]
SI	0.45±0.42	0.47±0.46	0.13±0.13
AP	0.29±0.28	0.29±0.26	0.14±0.13
Lateral	0.18±0.15	0.20±0.17	0.12±0.12

Figure 1:

Difference in EMT_cent from spine registration for all directions. Difference is in absolute value.

III.c. Registration between EMTs and Tumor

Table 2 shows results of the absolute difference in registration between EMT_cent and tumor averaged across all 219 CBCTs along with the corresponding standard deviations. Average registration differences were <0.2cm for all directions. Figure 2 shows a histogram of the difference in registration between EMT_cent and tumor for all CBCTs, which gives a visual representation of the small spread of registration differences. 95% of all evaluable 219 CBCTs resulted in an absolute registration difference of EMT_cent and tumor of <0.5cm in any direction. These results indicate that EMTs are a reliable surrogate for tumor position throughout the course of treatment. Of the 11 CBCTs with ≥0.5cm change between EMT_cent and tumor registrations, 8 were due to unusually large differences in lung inflation between simulation and treatment, which resulted in visually apparent large lung anatomy changes, and 3 were due to tumor shape change and possible deformation of the tumor from simulation to treatment. Two of the 3 CBCTs where tumor shape changed markedly were from the same patient. Both patients had HF treatment. Figure 3 shows an example patient who had a large difference in lung inflation from simulation to treatment. Figure 4 shows an example patient whose tumor shape changed markedly.

Figure 2:

Difference in EMT centroid and tumor registration for all directions. Differences are in absolute value.

Figure 3:

Axial, coronal and sagittal images from an example patient with large change in lung inflation from simulation to treatment. The DIBH planning CT is shown in red and the DIBH on-board CBCT is shown in cyan. White and gray levels indicate a match. The red arrow indicates area of extreme difference of lung inflation. The images are registered to match to spine in this example.

Figure 4:

Axial, coronal and sagittal images from an example patient whose tumor shape changed markedly from simulation to treatment. The DIBH planning CT is shown in red and the DIBH on-board CBCT is shown in cyan. White and gray levels indicate a match. The red arrow indicates where the tumor is. The images are registered to match to spine in this example. The yellow contour represents the GTV from the DIBH planning CT

Table 3 shows the percentage of CBCTs resulting in absolute registration differences less than 0.5cm, 0.4cm, 0.3cm and 0.2cm for both EMT_cent compared to spine and EMT_cent compared to tumor. For differences in EMT_cent compared to tumor, the registration differences do become larger as evaluation criteria becomes lower. 86%, 75% and 53% of CBCTs resulted in registration differences <0.4 cm, <0.3cm and <0.2cm, respectively.

Table 3:

Percentage of all CBCTs resulting in absolute registration differences less than various amounts

	<0.5cm	<0.4cm	<0.3cm	<0.2cm
| EMTcent – Spine |	59%	47%	35%	16%
| EMTcent – Tumor |	95%	86%	75%	53%

III.d. Inter-EMT Displacements and Geometric Residual for Conventionally Fractionated Patients.

Because the CF treatment courses are 15-35 fractions and on-treatment GTV changes are known to occur^45–48, visual evaluation of their EMT stability may be more problematic. In this study, 3 CF patients had CBCTs after 2 or 3 weeks in which the GTV was not able to be registered accurately due to tumor shrinkage, indicated by * in Table 4. Therefore, the 8 CF patients were evaluated for EMT stability by recording the inter-EMT displacement differences (AB, BC, and CA) and the GR reported by Calypso^® for every fraction. The difference in inter-EMT displacement was calculated by taking the absolute value of the difference between the displacement reported by the Calypso^® system at the time of treatment and that of the same EMT pair from the Calypso^® plan. Table 4 shows the inter-EMT displacement differences averaged among AB, BC, and CA across all treatment days, as well as the average GR reported values for each patient. Patient 7 only had two EMTs used for tracking, so the value recorded corresponds to the displacement difference for EMTs AB only averaged across all treatment days. The Calypso^® records for Patient 8 had been inadvertently deleted, but the images in the TPS were available for analysis.

Table 4:

Inter-beacon displacement values and geometric residual (GR) for the 8 conventionally fractionated patients.

	1*	2	3*	4	5	6	7	8*
Mean ± STD inter-EMT displacement difference (cm)	0.52±0.21	0.26±0.20	0.12±0.10	0.08±0.06	0.08±0.05	0.14±0.13	0.08±0.070	N/A
Mean ± STD GR (cm)	0.33±0.04	0.22±0.07	0.08±0.05	0.09±0.01	0.09±0.01	0.13±0.06	0.04±0.04	N/A

The * indicates the CF patients whose CBCTs after 2 or 3 weeks weren’t registered due to tumor shape change. Patient 1 and 8, the first 3 CBCTs were used in the registration analysis and for patient 3, only the first 2 CBCTs were used in the registration analysis.

IV. DISCUSSION

To our knowledge, this is the first and largest study to quantitatively evaluate whether bronchoscopically implanted anchored EMTs are good surrogates for tumor position and repeatability of lung anatomy for gated DIBH lung radiotherapy. Several studies investigated the feasibility of implantation of the EMTs into the lung and the feasibility of tracking the EMTs with the Calypso^® system for both free-breathing and DIBH lung treatments^{31–36,38,39}. These studies found that the EMTs can be used for monitoring and tracking during treatment and that the EMTs are stable and do not migrate. One of these clinical studies evaluated 4 Stereotactic Body Radiation Therapy (SBRT) patients treated in DIBH and 3 patients treated in free-breathing and found that the EMT positions in the lungs remained stable during treatment by using on-treatment verification CBCTs to visualize whether the EMTs had good alignment with other structures.³² The other clinical studies were done using patients treated with free-breathing only, with one study using breath-hold CT images acquired before treatment and periodically throughout to assess if the EMTs were stable³⁹. One study used a verification CBCT acquired before treatment and compared the shifts determined from the CBCT and planning CT registration to the shifts reported by Calypso^® during the CBCT acquisition and found the differences to be ≤1.5mm³⁷. The same study also used the GR reported by the Calypso^® system at the time of set-up as an indication of EMT stability throughout treatment³⁸. McDonald et. al. evaluated the long term stability of the EMTs by registering follow-up CTs with the planning CT and found the long term stability was comparable to other commercially available lung fiducial markers³⁶. Schmitt et al. found that intra-and interfractional variations in EMT distances varied up to 3.2mm for patients treated free-breathing³³. No studies, however, looked at whether the EMTs are good surrogates for tumor position for DIBH lung radiotherapy and no studies evaluated if the EMTs are good surrogates for lung anatomy.

Nevertheless, it is important to know whether the EMTs that are being tracked are good indicators for the tumor position, which is imperative to ensure that the intended PTV has been treated. Additionally, DIBH treatments are relied upon to increase lung volume, which helps reduce the mean lung dose and the volume percentage of the healthy lung receiving 20Gy (V20Gy), both of which are commonly used dose predictors for radiation pneumonitis ^9–11. Differences in lung anatomy are largely due to differences in lung inflation, and if the lung inflation achieved at simulation is not reproducible at each fraction, lung doses may be higher than what the radiation treatment plan reflects and clinically unacceptable. It is important to be able to track the reproducibility of lung anatomy from simulation to treatment in order to ensure confidence that the lung constraints were met during treatment.

In our study, the DI-planning CT was registered to the on-board DI-CBCT acquired during treatment. Registration differences between spine and EMT_cent were used as indication of difference in lung anatomy from simulation to treatment. Since the spine is stable and well positioned by the standard clinical immobilization and the EMTs do not migrate in the lung ^36,38,39 a difference in the registration between the two would indicate that lung anatomy has changed relative to the spine between the two image sets, likely due to differences in lung volume or lung shape. Based on the results in Table 2 and Figure 1, the EMTs are not good surrogates for repeatability of lung inflation. The average absolute difference in registration was 0.45±0.42cm and 0.29±0.28cm in the SI and AP directions, respectively. These large average values and large spread in standard deviation are a clear indication that EMT positions are not reliable indication that repeatability of lung inflation has been achieved. Only 59% of evaluable CBCTs resulted in registration differences between the EMT centroid and spine of <0.5cm in any direction. Figure 3 shows an example case where the lung inflation from simulation was substantially different from treatment. Depth of inhalation and lung inflation is not a static condition, but subject to patients’ daily physiological variations such as overall wellbeing, daily level of comfort, pain level, etc., thus, it is highly dependent on patients’ cooperation. Other factors that can affect lung volumes are inflammatory changes, pleural effusions, tumor changes, etc. Much of the time lung volumes are reproducible, but not all the time, as was observed in this study. It is worth noting that for the patient shown in Figure 3, the CBCT shown is from the patient’s last fraction, and the other CBCTs from other treatment days did not show a similarly large change in lung inflation from the planning CT.

Repeatability of lung anatomy may be less clinically important for HF treatments since the PTV volumes are typically small and dosimetric constraints easily met. For CF treatments, however, lung metrics from the plan may be closer to departmental limits and the delivered metrics may exceed limits if the patient does not achieve as large a lung inflation at treatment as at simulation. It is difficult to determine whether the patient achieved as large a lung inflation at treatment; because standard CBCTs have limited Field-of-View (FOV), it is not feasible to directly calculate the lung volume at the time of treatment. One option could be to acquire an extended length CBCT or full-lung orthogonal kV images to help determine the lung size achieved during treatment. Similar investigation is suggested for HF patients whose lung metrics are close to department limits. In this study, 82% of the patient cohort was treated with HF. Another potential avenue to investigate is whether DIBH using spirometry gives valuable information about lung volume. It is worthy to point out that the overall pulmonary toxicities for the patients in this study were low. This is because most patients were treated with HF and the risk of radiation pneumonitis with SBRT is 10-15% ^49,50. There is not enough data in this study to make conclusions about the clinical impact or lack thereof in terms of pulmonary toxicity based on the level of lung inflation.

Registration differences between EMT_cent and the tumor were used as indication of whether the EMTs are a reliable surrogate for tumor position. Since registration was performed between the planning CT and CBCT taken during treatment, changes in tumor size or shape make this problematic in some patients. This is more common for CF patients whose treatments last weeks; we were able to register tumors over the full course for 5/8 CF patients. Studies have found that lung tumors continuously shrink throughout conventional treatment^45–48. Kupelian et al. found that non-small cell lung cancers typically shrink 1.2% per day⁴⁵. For 3 of the CF patients in this study (indicated with * in Table 4), only the first 2 or 3 CBCTs were used because the tumor shrunk throughout treatment. Based on results in Table 2 and Figure 2, the EMTs are a reliable surrogate for tumor position, as long as the tumor size and shape are stable. The average absolute differences in registration between EMT_cent and tumor were <0.2cm in all directions with small standard deviation spreads, indicating that the position of the EMTs and tumor were well aligned throughout treatments. In the evaluable patients, excluding the CBCTs in which the tumor size and shape changed during the irradiation period, 95% of CBCTs resulted in registration differences between EMT_cent and tumor of <0.5cm in any directions. This provides confidence that when tracking the EMT positions with the Calypso^® system, the tumor position can be assumed to be tracking similarly. During treatment, the treatment beam is automatically gated on or off depending on whether the EMT coordinates are within a specified tolerance gate relative to the Calypso^® plan. For this study, which is based on the analysis of the CBCT images, we extrapolate that the EMT-target relationship seen on the CBCTs hold for that day’s treatment, with the knowledge that the EMTs were within the tolerance gates during the treatment. If the tumor shape or size changes markedly, it may be beneficial to re-simulate the patient for an adaptive replanning and to re-establish the EMT-GTV correlation for subsequent treatments^47,48.

Based on results shown in Table 4, six out of the 7 recorded CF patients had average differences in inter-EMT displacements to be ≤0.26cm and average GR ≤0.22cm, indicating that the EMTs are relatively stable and do not migrate throughout treatment, which is consistent with previous literature. For Patient 1, the absolute difference between inter-EMT displacement values from the Calypso^® plan and those recorded from the Calypso^® system at the start of the first fraction for AB, BC and CA were 0.32cm, 0.23cm and 0.61cm, respectively, and the standard deviation in the values reported by the Calypso^® system throughout the patient’s 35 fraction treatment was ≤0.1cm for AB, BC and CA. This indicates that there may have either been small contouring inaccuracies of the EMTs in the treatment plan or a slight migration of the EMTs between simulation and the start of treatment that affected the displacement and GR measurements reported in Table 4 for Patient 1.

In our methods, the spotlight CBCT acquired during breath-hold takes 40 seconds, which is typically longer than a single breath-hold. The CBCT would often need to be broken up into 2 or 3 acquisitions. It is possible that this may introduce CBCT artifacts. This is also true for DIBH CBCTs in general, regardless of the method used to monitor breath-hold. Other artifacts in the CBCT may also contribute to limitations in defining the GTV for the tumor match.

In our study, the kV radiographs and CBCT images acquired at treatment were used to verify the EMT positions and that the Calypso plan was still valid. They were not used to shift the patient for setup. The CBCTs also had their usual role in our standard departmental clinical procedure wherein a radiation oncologist reviews the match between the GTVs visualized in the CBCT and in the planning CT and approves the setup if the match is satisfactory. Even though our study concludes that the EMTs are good surrogates for tumor position, it is helpful to evaluate 2D or 3D anatomical images prior to treatment. kV radiographic imaging and CBCT imaging can provide verification and confidence of the Calypso plan. In addition, the DI-CBCT provides other valuable anatomical information, such as whether the patient is straight or excessively rotated, severe weight changes, if there is additional fluid in the lungs, or if the tumor changed shape. Calypso^® assures the target position at each treatment based on our findings that the EMTs are good surrogates for tumor position.

V. CONCLUSIONS

Bronchoscopically implanted anchored EMTs are good surrogates for tumor position and are reliable for monitoring and maintaining tumor position during DIBH treatment when tracking with Calypso^®, as long as the tumor size and shape are stable. Larger differences in registration between EMT centroid and spine for many treatments suggest that the EMTs may not be good surrogates for confirming that the level of lung inflation achieved at simulation is reproduced at treatment.