Clinical Oncology 28 (2016) 28e35 Contents lists available at ScienceDirect
Clinical Oncology journal homepage: www.clinicaloncologyonline.net
Original Article
Predictors of Chest Wall Toxicity after Lung Stereotactic Ablative Radiotherapyq I. Thibault *, A. Chiang *, D. Erler *, L. Yeung y, I. Poon *, A. Kim *, B. Keller *, F. Lochray *, S. Jain z, H. Soliman *, P. Cheung * * Sunnybrook
Odette Cancer Centre, Department of Radiation Oncology, University of Toronto, Toronto, Ontario, Canada Department of Paediatrics and Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada z Centre for Cancer Research and Cell Biology, Queen’s University Belfast, Belfast, UK y
Received 22 February 2015; received in revised form 15 May 2015; accepted 10 June 2015
Abstract Aims: To determine the incidence and predictive factors of rib fracture and chest wall pain after lung stereotactic ablative radiotherapy (SABR). Materials and methods: Patients were treated with lung SABR of 48e60 Gy in four to five fractions. The treatment plan and follow-up computed tomography scans of 289 tumours in 239 patients were reviewed. Doseevolume histogram (DVH) metrics and clinical factors were evaluated as potential predictors of chest wall toxicity. Results: The median follow-up was 21.0 months (range 6.2e52.1). Seventeen per cent (50/289) developed a rib fracture, 44% (22/50) were symptomatic; the median time to fracture was 16.4 months. On univariate analysis, female gender, osteoporosis, tumours adjacent (within 5 mm) to the chest wall and all of the chest wall DVH metrics predicted for rib fracture, but only tumour location adjacent to the chest wall remained significant on the multivariate model (P < 0.01). The 2 year fracture-free probability for those adjacent to the chest wall was 65.6%. Among those tumours adjacent to the chest wall, only osteoporosis (P ¼ 0.02) predicted for fracture, whereas none of the chest wall DVH metrics were predictive. Eight per cent (24/289) experienced chest wall pain without fracture. Conclusions: None of the chest wall DVH metrics independently predicted for SABR-induced rib fracture when tumour location is taken into account. Patients with tumours adjacent (within 5 mm) to the chest wall are at greater risk of rib fracture after lung SABR, and among these, an additional risk was observed in osteoporotic patients. Ó 2015 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
Key words: Chest wall pain; lung cancer; lung metastases; rib fracture; stereotactic body radiotherapy
Introduction Stereotactic body radiotherapy, also called stereotactic ablative radiotherapy (SABR), has emerged over the past decade as an important therapeutic modality in the management of early-stage non-small cell lung cancer (NSCLC). It is now considered the preferred modality in medically inoperable stage I lung cancer, and some also advocate its q This project was an oral presentation at the 56th ASTRO meeting in San Francisco, California, 2014.
Author for correspondence: P. Cheung, Department of Radiation Oncology, Sunnybrook Odette Cancer Centre, University of Toronto, 2075 Bayview Avenue, Toronto, Ontario, M4N 3M5, Canada. Tel: þ1-416-4806000; Fax: þ1-416-480-6002. E-mail address: [email protected] (P. Cheung).
use in selected operable patients [1e4]. In elderly patients with stage I NSCLC, SABR is associated with the lowest risk of mortality within the first 6 months after treatment, as shown by the large population-based study of Shirvani et al. [2], which compared the effectiveness of five different therapeutic options. In addition, SABR is increasingly used in the setting of pulmonary oligometastases. When delivering SABR, organs at risk are routinely considered to include spinal cord, brachial plexus, heart, oesophagus, trachea, proximal bronchial tree and normal lung. More recently, the chest wall has been identified as a relevant organ at risk, with risk of chest wall toxicities ranging from 2 to 45% post-SABR [5e9]. Radiation-induced chest wall toxicities can be classified into two distinct entities e rib fractures and chest wall pain
http://dx.doi.org/10.1016/j.clon.2015.06.009 0936-6555/Ó 2015 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
I. Thibault et al. / Clinical Oncology 28 (2016) 28e35
without fracture. Although chest wall toxicities are reported in the literature, the distinction between chest wall pain from rib fracture versus chest wall pain without fracture is not always highlighted. With regards to SABR treatment planning, there are currently no standard doseevolume constraints established for the chest wall and no definite consensus about whether tumour coverage should be compromised in order to spare the chest wall. This is a question of particular interest, especially when treating tumours immediately adjacent to the chest wall. To add to the complexity, different dose-fractionation regimens are used for lung SABR, potentially affecting the doseevolume histogram (DVH) metrics of the chest wall and potentially the incidence of chest wall toxicities. At the Sunnybrook Odette Cancer Centre, we deliver 48e60 Gy in four fractions for peripheral lung tumours and we do not compromise the internal target volume (ITV) or planning target volume (PTV) coverage to limit the dose to the chest wall. The aim of the current study was to review our experience and to evaluate the risk of rib fracture, symptomatic or asymptomatic, the risk of chest wall pain without fracture and to identify any potential dosimetric and clinical predictive factors, particularly for tumours adjacent to the chest wall.
Materials and Methods This study was approved by the research ethics board committee of the Sunnybrook Health Sciences Centre. Patients were eligible if they received lung SABR between 2008 and 2011 at the Sunnybrook Odette Cancer Centre and had a minimum follow-up of 6 months, with computed tomography. Consecutive patients were identified from a prospective lung SABR database. In total, 289 lung tumours in 239 patients were eligible. Both primary and metastatic pulmonary tumours were included in the study. The lung SABR technique used at our institution has been previously described [10,11]. Peripheral primary NSCLC received 48e52 Gy/four fractions, peripheral metastases 52e60 Gy/four fractions and central tumours 50 Gy/five fractions. The immobilisation system consisted of the blue bag from BodyFix with an abdominal compression plate and patients were positioned supine with arms up. Threedimensional conformal radiotherapy (81/289 tumours) and intensity-modulated radiotherapy (208/289 tumours) were used for treatment planning. The aim was to deliver 100% of the prescribed dose to >99% of the ITV and 95% of the prescribed dose to >99% of the PTV. The PTV consisted of the ITV plus a 5 mm margin. ITV and PTV coverage were never compromised to spare the chest wall. For the purpose of this study, we categorised chest wall toxicities into two categories: SABR-induced rib fracture and chest wall pain without fracture. The main objective was to determine the incidence of chest wall toxicities and to evaluate specific clinical and dosimetric predictive factors for the risk of rib fracture, and chest wall pain without fracture. The secondary objective was to evaluate predictive factors for symptomatic and asymptomatic rib fracture.
29
In our cohort, 30/239 (13%) patients were treated for multiple synchronous lesions (71 tumours in 30 patients) and most (43/71 tumours, 60%) of these synchronous lesions were located in different lobes. Given that most were in different lobes and because chest wall toxicities tend to occur in close proximity to the treated site, we based our primary end point on the number of treated tumours rather than on the number of patients. For example, if a patient was treated for three separate tumour nodules, the risk of rib fracture was evaluated for each one. In the uncommon situation where two PTVs were located in the same lobe and close to each other, and at a similar distance from the chest wall, the occurrence of rib fracture was attributed to both tumours, because distinction of the risk was not possible. Dosimetric parameters obtained and evaluated for the chest wall were Dmax, D1cc, D2cc, D5cc, V30, V40 and V50. Dose calculation was computed with heterogeneity correction using convolution/superposition algorithm. Dose parameters were reported in biologically effective dose, using a/b ¼ 3 for the chest wall. The chest wall was contoured retrospectively on each treatment plan for consistency and similarity in contouring. The chest wall was delineated on the average scan, using a 2.0 cm circle tool [6], with inclusion of the ipsilateral chest wall soft tissues and ribs, and exclusion of vertebra, sternum, lung, and skin, cranio-caudally 3.0 cm above and below the PTV. Because the risk of rib fracture was calculated per treated tumour, dosimetric values were evaluated per treated site as well. Distinct chest wall contours were carried out for each treated site. In the scenario of having 1 cm distance between two PTVs, a single chest wall contour was delineated. Clinical factors evaluated for their predictive value were age, gender, osteoporosis, diabetes, and tumour location (adjacent to the chest wall or not). These clinical data were prospectively collected, recorded in our institutional electronic patient chart system and retrospectively reviewed. Patients had computed tomography and follow-up every 4 months for the first 3 years after completion of lung SABR, and every 6 months thereafter. The incidence of chest wall pain limiting daily living activities or requiring medication, related either to a symptomatic rib fracture or to chest wall pain without fracture, was also collected retrospectively from the electronic patient record system. The occurrence of a rib fracture was determined retrospectively based on imaging, after review of the initial planning computed tomography, all follow-up computed tomography and their respective radiology report. When there was difficulty in distinguishing between post-SABR changes confined within the rib and true rib fracture, imaging was reviewed with a radiologist. Statistical Analysis Descriptive statistics were used to list the relevant patient, tumour and treatment characteristics. KaplaneMeier methodology was used to compute non-parametric estimates of time to rib fracture. The time-dependent outcome of rib fracture was evaluated using univariate analysis by the parametric Weibull
30
I. Thibault et al. / Clinical Oncology 28 (2016) 28e35
variable hazard model. Continuous predictor variables included: the total prescription dose, chest wall Dmax, D1cc, D2cc, D5cc, V30, V40, V50 and patient age. The following dichotomous predictor variables were evaluated using univariate analysis by Log-rank test: gender, osteoporosis, diabetes and tumour location (adjacent to the chest wall or not). Tumours with PTV overlying the chest wall (i.e. ITV 5 mm from the chest wall) were categorised as adjacent to the chest wall. Statistically significant variables (P < 0.05) from univariate analysis were used in multivariate regression. The above methods were repeated in a subset analysis of tumours adjacent to the chest wall. Logistic regression was used to evaluate the same predictors to compare occurrence of symptomatic and asymptomatic rib fractures. For the outcomes of (a) chest wall pain without rib fracture and (b) chest wall pain (with and without facture), logistic regression was also used to evaluate the aforementioned predictor variables.
Results The baseline characteristics of the lung tumours, prescribed dose of SABR and chest wall DVH metrics are presented in Table 1. Of the 289 tumours, 210 (73%) were primary NSCLC and 79 (27%) were pulmonary metastases (from lung, n ¼ 12; colorectal, n ¼ 43; renal cell, n ¼ 9; and other cancer origin, n ¼ 15). In total, 164/289 (57%) were located in the upper lobes, 20/289 (7%) in the right middle
lobe and 105/289 (36%) in the lower lobes. The median prescribed total dose/number of fractions was 48 Gy/four fractions. The median ITV V100, PTV V100 and PTV V95 were 99.9% (range 71.8e100.0%), 73.3% (range 35.8e100.0%) and 99.5% (range 75.3e100.0%), respectively. The median 95% conformity index was 1.45 (range 0.97e2.60). The median absolute maximum point dose (Dmax) to the chest wall was 50.5 Gy (range 17.9e76.9) and the median biologically effective dose Dmax was 260.9 Gy3 (range 44.7e569.1 Gy3). After SABR, the median time to last imaging and clinical follow-up was 21.0 months (range 6.2e52.1 months). We observed 50/289 (17%) rib fractures, with 22/50 (44%) being symptomatic and 28/50 (56%) asymptomatic. The 1 and 2 year fracture-free probabilities were 97.4 and 77.6%, respectively. The median time to fracture among these 50 cases was 16.4 months (range 4.7e44.0 months). On univariate analysis, female gender (P ¼ 0.024), osteoporosis (P < 0.001), tumours adjacent to the chest wall (P < 0.001) and every one of the chest wall DVH metrics (P 0.001) predicted for rib fracture. On multivariate analysis, only tumour location adjacent to the chest wall remained a significant predictor (hazard ratio, 2.9; 95% confidence interval (CI) 1.7e5.0; P < 0.001). The 1 and 2 year fracture-free probabilities were 96.0 and 65.6% for tumours adjacent to the chest wall compared with 99.1 and 94.0% for those not adjacent to the chest wall, respectively (Figure 1; P < 0.001). In total, 158/289 (55%) tumours were adjacent to the chest wall and the crude risk of developing a rib fracture after SABR in tumours adjacent to the chest wall was 28%
Table 1 Baseline characteristics for lung stereotactic ablative radiotherapy (SABR) in 289 tumours grouped between those with and without a SABRinduced rib fracture Characteristics
All tumours (%)
Rib fracture (%)
No rib fracture (%)
n ¼ 289
n ¼ 50
n ¼ 239
74 (47e91) 35 (70%) 16 (32%) 7 (14%)
73 (35e92) 121 (51%) 27 (11%) 39 (16%)
33 (66%) 17 (34%) 2.0 (0.5e5.8) 45 (90%)
177 (74%) 62 (26%) 2.0 (0.3e5.8) 113 (47%)
2 (4%) 24 (48%) 24 (48%)
16 (7%) 127 (53%) 96 (40%)
Patient characteristics Age, median (range), in years 73 (35e92) Female gender 156 (54%) Osteoporosis 43 (15%) Diabetes 46 (16%) Treated tumour Primary lung cancer 210 (73%) Pulmonary metastasis 79 (27%) Maximal diameter, median (range),cm 2.0 (0.3e5.8) Adjacent to the chest wall 158 (55%) SABR regimen 50 Gy/5 fractions 18 (6%) 48 Gy/4 fractions 151 (52%) 52e60 Gy/4 fractions 120 (42%) Chest wall doseevolume histogram metrics, median/mean values Dmax, absolute dose in Gy (BED, Gy3) 50.5 (260.9)/48.1 (244.0) D1cc, absolute dose in Gy (BED, Gy3) 48.2 (238.8)/44.7 (216.9) D2cc, absolute dose in Gy (BED, Gy3) 47.5 (231.4)/43.6 (208.3) D5cc, absolute dose in Gy (BED, Gy3) 45.2 (213.2)/41.4 (191.3) V30 (in cc) 30.8/35.7 V40 (in cc) 11.7/16.3 V50 (in cc) < 0.1/3.7
52.7 (279.6)/53.2 50.3 (255.1)/50.7 49.3 (250.4)/49.8 47.8 (235.7)/47.6 42.9/47.3 18.7/23.1 1.3/4.6
(289.3) (265.1) (256.9) (237.2)
50.2 (254.5)/47.0 47.4 (229.9)/43.4 46.3 (221.6)/42.3 43.4 (199.2)/40.1 26.8/33.3 8.9/14.8 0.0/3.5
(234.5) (206.9) (198.1) (181.7)
I. Thibault et al. / Clinical Oncology 28 (2016) 28e35
Fig 1. Incidence of rib fracture after lung stereotactic ablative radiotherapy in 158 tumours adjacent to the chest wall versus 131 tumours not adjacent to the chest wall (P < 0.001).
31
(45/158 tumours). Of these 45 rib fracture events, 23 (23/90, 26%) occurred after a lung SABR total dose of 48e50 Gy and 22 (22/68, 32%) after 52e60 Gy. On the univariate model, among this subgroup of 158 tumours adjacent to the chest wall at high risk for rib fracture, only osteoporosis (P ¼ 0.017; Figure 2) was a statistically significant predictor of increased risk for fracture. None of the chest wall DVH metrics nor the total prescription dose were predictive, and we observed a trend for female gender (P ¼ 0.078). In addition, the rate of fracture was not significantly different for tumours adjacent to the anterior, lateral or posterior ipsilateral chest wall. Figure 3 illustrates a case of rib fracture and the corresponding dosimetry. We did not observe a significant difference between the incidence of symptomatic and asymptomatic rib fractures (P ¼ 0.078). However, patients aged 65 years were four times more likely to develop an asymptomatic fracture (odds ratio, 4.4; 95% CI 1.3e15.1; P ¼ 0.019). In total, 24/289 (8%) experienced chest wall pain without fracture and the crude rate was also 8% (13/158) if tumour was adjacent to the chest wall. We also explored the predictive significance of age, gender, osteoporosis, diabetes, tumour location (adjacent to the chest wall or not), SABR total dose and the chest wall DVH parameters, but none of those factors predicted for chest wall pain without fracture on the univariate model. Overall, 46/289 (16%) were symptomatic of chest wall pain, with 22 symptomatic rib fractures and 24 having chest wall pain without fracture. On univariate analysis, tumours adjacent to the chest wall (P ¼ 0.009), higher chest wall V30 (P ¼ 0.027), chest wall D1cc (P ¼ 0.056), chest wall D2cc (P ¼ 0.045) and chest wall D5cc (P ¼ 0.041) predicted for a greater risk of chest wall pain, but none of these factors was significant on the multivariate model.
Discussion Fig 2. Incidence of rib fracture after stereotactic ablative radiotherapy in 158 tumours adjacent to the chest wall for osteoporotic versus non-osteoporotic patients.
We showed that pulmonary tumours adjacent to the chest wall are at greater risk of rib fracture after lung SABR and none of the chest wall DVH metrics independently predicted for SABR-induced rib fracture when tumour
Fig 3. Stereotactic ablative radiotherapy (SABR)-induced rib fracture. Primary non-small cell cancer located adjacent to the chest wall, delineation of the chest wall contour, SABR treatment plan using 48 Gy/four fractions and rib fracture observed 15.3 months after lung SABR.
32
Table 2 Summary of the published literature on chest wall toxicities following lung stereotactic ablative radiotherapy (SABR) Reference
Year
Tumour type
Total dose (Gy)/no. fractions
No. patients (no. lesions)
Organ at risk for dose evolume histogram data
% of CWP
% of severe CWP
No. rib fractures (%)
Symptomatic rib fractures (%)
Median time in months (range) to rib fracture
to CWP
Predictors
Principal findings
Fracture: smallvolume/highdose region; D2cc e
50% risk of rib fracture if D2cc ¼ 3 16.6 Gy (49.8 Gy)
2009
P
45/3
33
Ribs
e
e
13 (7/33 patients, 21%)
e
15 (8e38)
e
[17]
2009
P
54e60/3
42
e
26%
e
15 (9/42 patients, 21%)
13/15 (87%)
17
e
[5]
2012
P
54e60/3
46 (49)
Ribs
46%
e
41 (17/46 patients, 37%)
21 (7e40)
e
[13]
2012
P or M
48/4
116
Ribs
e
e
46 (28/116 patients, 24%)
14/17 patients (82%) 12/28 patients (43%)
22 (9e42)
e
[14]
2013
P
48e70/4e10
177z
e
21%
e
41 (41/177 patients, 23%)
14/41 (34%)
33 (24e94)
e
[15]
2013
P or M*
36e60/3e4
118 (126)
CW
25%jj
e
48 (48/126, 38%)
(18e61%)
17 (4e52)
e
Fracture: female gender, lateral location, D8cc
Chest wall pain [18] 2011
P or My
50/4
265
CW
25%
9%
e
e
e
6 (0e11)
[19] [20]
2012 2012
P P or M
48e60/3e10 60/3
102 (106) 45 (48)
CW CW
19%jj 22%
0.9%jj
e e
e e
e e
8 (6e10) 8.8 (3.7e12.0)
CWP: BMI, DB, CW V30 CWP: mEUD, BMI CWP: CW V30 and V60
[6]
2012
P
40e60/3e5
126
CW
43%
15%
8 (5/126 patients, 4%)
3/5 patients (60%)
27 (21e32)
9 (1e36)
CWP: CW V30
[21]
2012
P or M
50e54/3e5
140 (146)
CW
15%jj,{
1.4%
e
e
e
12.6 (4e36)
CWP: BMI, CW V30, V35, V40
60
CW
33%
28%
5 (5/60 patients, 8%)
5/5 (100%)
7 (0.6e32)
Chest wall pain and rib fracture [16] 2010 P or My 21e60/3e5
Fracture: rib D0.5cc, age, female gender Fracture: rib etumour distance 16 mm developed fracture. Female gender and lateral location are risk factors for rib fracture. 18% risk of rib fracture if V30 30 cc mEUD better predicted CWP Restricting V30 to
Clinical Oncology journal homepage: www.clinicaloncologyonline.net
Original Article
Predictors of Chest Wall Toxicity after Lung Stereotactic Ablative Radiotherapyq I. Thibault *, A. Chiang *, D. Erler *, L. Yeung y, I. Poon *, A. Kim *, B. Keller *, F. Lochray *, S. Jain z, H. Soliman *, P. Cheung * * Sunnybrook
Odette Cancer Centre, Department of Radiation Oncology, University of Toronto, Toronto, Ontario, Canada Department of Paediatrics and Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada z Centre for Cancer Research and Cell Biology, Queen’s University Belfast, Belfast, UK y
Received 22 February 2015; received in revised form 15 May 2015; accepted 10 June 2015
Abstract Aims: To determine the incidence and predictive factors of rib fracture and chest wall pain after lung stereotactic ablative radiotherapy (SABR). Materials and methods: Patients were treated with lung SABR of 48e60 Gy in four to five fractions. The treatment plan and follow-up computed tomography scans of 289 tumours in 239 patients were reviewed. Doseevolume histogram (DVH) metrics and clinical factors were evaluated as potential predictors of chest wall toxicity. Results: The median follow-up was 21.0 months (range 6.2e52.1). Seventeen per cent (50/289) developed a rib fracture, 44% (22/50) were symptomatic; the median time to fracture was 16.4 months. On univariate analysis, female gender, osteoporosis, tumours adjacent (within 5 mm) to the chest wall and all of the chest wall DVH metrics predicted for rib fracture, but only tumour location adjacent to the chest wall remained significant on the multivariate model (P < 0.01). The 2 year fracture-free probability for those adjacent to the chest wall was 65.6%. Among those tumours adjacent to the chest wall, only osteoporosis (P ¼ 0.02) predicted for fracture, whereas none of the chest wall DVH metrics were predictive. Eight per cent (24/289) experienced chest wall pain without fracture. Conclusions: None of the chest wall DVH metrics independently predicted for SABR-induced rib fracture when tumour location is taken into account. Patients with tumours adjacent (within 5 mm) to the chest wall are at greater risk of rib fracture after lung SABR, and among these, an additional risk was observed in osteoporotic patients. Ó 2015 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
Key words: Chest wall pain; lung cancer; lung metastases; rib fracture; stereotactic body radiotherapy
Introduction Stereotactic body radiotherapy, also called stereotactic ablative radiotherapy (SABR), has emerged over the past decade as an important therapeutic modality in the management of early-stage non-small cell lung cancer (NSCLC). It is now considered the preferred modality in medically inoperable stage I lung cancer, and some also advocate its q This project was an oral presentation at the 56th ASTRO meeting in San Francisco, California, 2014.
Author for correspondence: P. Cheung, Department of Radiation Oncology, Sunnybrook Odette Cancer Centre, University of Toronto, 2075 Bayview Avenue, Toronto, Ontario, M4N 3M5, Canada. Tel: þ1-416-4806000; Fax: þ1-416-480-6002. E-mail address: [email protected] (P. Cheung).
use in selected operable patients [1e4]. In elderly patients with stage I NSCLC, SABR is associated with the lowest risk of mortality within the first 6 months after treatment, as shown by the large population-based study of Shirvani et al. [2], which compared the effectiveness of five different therapeutic options. In addition, SABR is increasingly used in the setting of pulmonary oligometastases. When delivering SABR, organs at risk are routinely considered to include spinal cord, brachial plexus, heart, oesophagus, trachea, proximal bronchial tree and normal lung. More recently, the chest wall has been identified as a relevant organ at risk, with risk of chest wall toxicities ranging from 2 to 45% post-SABR [5e9]. Radiation-induced chest wall toxicities can be classified into two distinct entities e rib fractures and chest wall pain
http://dx.doi.org/10.1016/j.clon.2015.06.009 0936-6555/Ó 2015 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
I. Thibault et al. / Clinical Oncology 28 (2016) 28e35
without fracture. Although chest wall toxicities are reported in the literature, the distinction between chest wall pain from rib fracture versus chest wall pain without fracture is not always highlighted. With regards to SABR treatment planning, there are currently no standard doseevolume constraints established for the chest wall and no definite consensus about whether tumour coverage should be compromised in order to spare the chest wall. This is a question of particular interest, especially when treating tumours immediately adjacent to the chest wall. To add to the complexity, different dose-fractionation regimens are used for lung SABR, potentially affecting the doseevolume histogram (DVH) metrics of the chest wall and potentially the incidence of chest wall toxicities. At the Sunnybrook Odette Cancer Centre, we deliver 48e60 Gy in four fractions for peripheral lung tumours and we do not compromise the internal target volume (ITV) or planning target volume (PTV) coverage to limit the dose to the chest wall. The aim of the current study was to review our experience and to evaluate the risk of rib fracture, symptomatic or asymptomatic, the risk of chest wall pain without fracture and to identify any potential dosimetric and clinical predictive factors, particularly for tumours adjacent to the chest wall.
Materials and Methods This study was approved by the research ethics board committee of the Sunnybrook Health Sciences Centre. Patients were eligible if they received lung SABR between 2008 and 2011 at the Sunnybrook Odette Cancer Centre and had a minimum follow-up of 6 months, with computed tomography. Consecutive patients were identified from a prospective lung SABR database. In total, 289 lung tumours in 239 patients were eligible. Both primary and metastatic pulmonary tumours were included in the study. The lung SABR technique used at our institution has been previously described [10,11]. Peripheral primary NSCLC received 48e52 Gy/four fractions, peripheral metastases 52e60 Gy/four fractions and central tumours 50 Gy/five fractions. The immobilisation system consisted of the blue bag from BodyFix with an abdominal compression plate and patients were positioned supine with arms up. Threedimensional conformal radiotherapy (81/289 tumours) and intensity-modulated radiotherapy (208/289 tumours) were used for treatment planning. The aim was to deliver 100% of the prescribed dose to >99% of the ITV and 95% of the prescribed dose to >99% of the PTV. The PTV consisted of the ITV plus a 5 mm margin. ITV and PTV coverage were never compromised to spare the chest wall. For the purpose of this study, we categorised chest wall toxicities into two categories: SABR-induced rib fracture and chest wall pain without fracture. The main objective was to determine the incidence of chest wall toxicities and to evaluate specific clinical and dosimetric predictive factors for the risk of rib fracture, and chest wall pain without fracture. The secondary objective was to evaluate predictive factors for symptomatic and asymptomatic rib fracture.
29
In our cohort, 30/239 (13%) patients were treated for multiple synchronous lesions (71 tumours in 30 patients) and most (43/71 tumours, 60%) of these synchronous lesions were located in different lobes. Given that most were in different lobes and because chest wall toxicities tend to occur in close proximity to the treated site, we based our primary end point on the number of treated tumours rather than on the number of patients. For example, if a patient was treated for three separate tumour nodules, the risk of rib fracture was evaluated for each one. In the uncommon situation where two PTVs were located in the same lobe and close to each other, and at a similar distance from the chest wall, the occurrence of rib fracture was attributed to both tumours, because distinction of the risk was not possible. Dosimetric parameters obtained and evaluated for the chest wall were Dmax, D1cc, D2cc, D5cc, V30, V40 and V50. Dose calculation was computed with heterogeneity correction using convolution/superposition algorithm. Dose parameters were reported in biologically effective dose, using a/b ¼ 3 for the chest wall. The chest wall was contoured retrospectively on each treatment plan for consistency and similarity in contouring. The chest wall was delineated on the average scan, using a 2.0 cm circle tool [6], with inclusion of the ipsilateral chest wall soft tissues and ribs, and exclusion of vertebra, sternum, lung, and skin, cranio-caudally 3.0 cm above and below the PTV. Because the risk of rib fracture was calculated per treated tumour, dosimetric values were evaluated per treated site as well. Distinct chest wall contours were carried out for each treated site. In the scenario of having 1 cm distance between two PTVs, a single chest wall contour was delineated. Clinical factors evaluated for their predictive value were age, gender, osteoporosis, diabetes, and tumour location (adjacent to the chest wall or not). These clinical data were prospectively collected, recorded in our institutional electronic patient chart system and retrospectively reviewed. Patients had computed tomography and follow-up every 4 months for the first 3 years after completion of lung SABR, and every 6 months thereafter. The incidence of chest wall pain limiting daily living activities or requiring medication, related either to a symptomatic rib fracture or to chest wall pain without fracture, was also collected retrospectively from the electronic patient record system. The occurrence of a rib fracture was determined retrospectively based on imaging, after review of the initial planning computed tomography, all follow-up computed tomography and their respective radiology report. When there was difficulty in distinguishing between post-SABR changes confined within the rib and true rib fracture, imaging was reviewed with a radiologist. Statistical Analysis Descriptive statistics were used to list the relevant patient, tumour and treatment characteristics. KaplaneMeier methodology was used to compute non-parametric estimates of time to rib fracture. The time-dependent outcome of rib fracture was evaluated using univariate analysis by the parametric Weibull
30
I. Thibault et al. / Clinical Oncology 28 (2016) 28e35
variable hazard model. Continuous predictor variables included: the total prescription dose, chest wall Dmax, D1cc, D2cc, D5cc, V30, V40, V50 and patient age. The following dichotomous predictor variables were evaluated using univariate analysis by Log-rank test: gender, osteoporosis, diabetes and tumour location (adjacent to the chest wall or not). Tumours with PTV overlying the chest wall (i.e. ITV 5 mm from the chest wall) were categorised as adjacent to the chest wall. Statistically significant variables (P < 0.05) from univariate analysis were used in multivariate regression. The above methods were repeated in a subset analysis of tumours adjacent to the chest wall. Logistic regression was used to evaluate the same predictors to compare occurrence of symptomatic and asymptomatic rib fractures. For the outcomes of (a) chest wall pain without rib fracture and (b) chest wall pain (with and without facture), logistic regression was also used to evaluate the aforementioned predictor variables.
Results The baseline characteristics of the lung tumours, prescribed dose of SABR and chest wall DVH metrics are presented in Table 1. Of the 289 tumours, 210 (73%) were primary NSCLC and 79 (27%) were pulmonary metastases (from lung, n ¼ 12; colorectal, n ¼ 43; renal cell, n ¼ 9; and other cancer origin, n ¼ 15). In total, 164/289 (57%) were located in the upper lobes, 20/289 (7%) in the right middle
lobe and 105/289 (36%) in the lower lobes. The median prescribed total dose/number of fractions was 48 Gy/four fractions. The median ITV V100, PTV V100 and PTV V95 were 99.9% (range 71.8e100.0%), 73.3% (range 35.8e100.0%) and 99.5% (range 75.3e100.0%), respectively. The median 95% conformity index was 1.45 (range 0.97e2.60). The median absolute maximum point dose (Dmax) to the chest wall was 50.5 Gy (range 17.9e76.9) and the median biologically effective dose Dmax was 260.9 Gy3 (range 44.7e569.1 Gy3). After SABR, the median time to last imaging and clinical follow-up was 21.0 months (range 6.2e52.1 months). We observed 50/289 (17%) rib fractures, with 22/50 (44%) being symptomatic and 28/50 (56%) asymptomatic. The 1 and 2 year fracture-free probabilities were 97.4 and 77.6%, respectively. The median time to fracture among these 50 cases was 16.4 months (range 4.7e44.0 months). On univariate analysis, female gender (P ¼ 0.024), osteoporosis (P < 0.001), tumours adjacent to the chest wall (P < 0.001) and every one of the chest wall DVH metrics (P 0.001) predicted for rib fracture. On multivariate analysis, only tumour location adjacent to the chest wall remained a significant predictor (hazard ratio, 2.9; 95% confidence interval (CI) 1.7e5.0; P < 0.001). The 1 and 2 year fracture-free probabilities were 96.0 and 65.6% for tumours adjacent to the chest wall compared with 99.1 and 94.0% for those not adjacent to the chest wall, respectively (Figure 1; P < 0.001). In total, 158/289 (55%) tumours were adjacent to the chest wall and the crude risk of developing a rib fracture after SABR in tumours adjacent to the chest wall was 28%
Table 1 Baseline characteristics for lung stereotactic ablative radiotherapy (SABR) in 289 tumours grouped between those with and without a SABRinduced rib fracture Characteristics
All tumours (%)
Rib fracture (%)
No rib fracture (%)
n ¼ 289
n ¼ 50
n ¼ 239
74 (47e91) 35 (70%) 16 (32%) 7 (14%)
73 (35e92) 121 (51%) 27 (11%) 39 (16%)
33 (66%) 17 (34%) 2.0 (0.5e5.8) 45 (90%)
177 (74%) 62 (26%) 2.0 (0.3e5.8) 113 (47%)
2 (4%) 24 (48%) 24 (48%)
16 (7%) 127 (53%) 96 (40%)
Patient characteristics Age, median (range), in years 73 (35e92) Female gender 156 (54%) Osteoporosis 43 (15%) Diabetes 46 (16%) Treated tumour Primary lung cancer 210 (73%) Pulmonary metastasis 79 (27%) Maximal diameter, median (range),cm 2.0 (0.3e5.8) Adjacent to the chest wall 158 (55%) SABR regimen 50 Gy/5 fractions 18 (6%) 48 Gy/4 fractions 151 (52%) 52e60 Gy/4 fractions 120 (42%) Chest wall doseevolume histogram metrics, median/mean values Dmax, absolute dose in Gy (BED, Gy3) 50.5 (260.9)/48.1 (244.0) D1cc, absolute dose in Gy (BED, Gy3) 48.2 (238.8)/44.7 (216.9) D2cc, absolute dose in Gy (BED, Gy3) 47.5 (231.4)/43.6 (208.3) D5cc, absolute dose in Gy (BED, Gy3) 45.2 (213.2)/41.4 (191.3) V30 (in cc) 30.8/35.7 V40 (in cc) 11.7/16.3 V50 (in cc) < 0.1/3.7
52.7 (279.6)/53.2 50.3 (255.1)/50.7 49.3 (250.4)/49.8 47.8 (235.7)/47.6 42.9/47.3 18.7/23.1 1.3/4.6
(289.3) (265.1) (256.9) (237.2)
50.2 (254.5)/47.0 47.4 (229.9)/43.4 46.3 (221.6)/42.3 43.4 (199.2)/40.1 26.8/33.3 8.9/14.8 0.0/3.5
(234.5) (206.9) (198.1) (181.7)
I. Thibault et al. / Clinical Oncology 28 (2016) 28e35
Fig 1. Incidence of rib fracture after lung stereotactic ablative radiotherapy in 158 tumours adjacent to the chest wall versus 131 tumours not adjacent to the chest wall (P < 0.001).
31
(45/158 tumours). Of these 45 rib fracture events, 23 (23/90, 26%) occurred after a lung SABR total dose of 48e50 Gy and 22 (22/68, 32%) after 52e60 Gy. On the univariate model, among this subgroup of 158 tumours adjacent to the chest wall at high risk for rib fracture, only osteoporosis (P ¼ 0.017; Figure 2) was a statistically significant predictor of increased risk for fracture. None of the chest wall DVH metrics nor the total prescription dose were predictive, and we observed a trend for female gender (P ¼ 0.078). In addition, the rate of fracture was not significantly different for tumours adjacent to the anterior, lateral or posterior ipsilateral chest wall. Figure 3 illustrates a case of rib fracture and the corresponding dosimetry. We did not observe a significant difference between the incidence of symptomatic and asymptomatic rib fractures (P ¼ 0.078). However, patients aged 65 years were four times more likely to develop an asymptomatic fracture (odds ratio, 4.4; 95% CI 1.3e15.1; P ¼ 0.019). In total, 24/289 (8%) experienced chest wall pain without fracture and the crude rate was also 8% (13/158) if tumour was adjacent to the chest wall. We also explored the predictive significance of age, gender, osteoporosis, diabetes, tumour location (adjacent to the chest wall or not), SABR total dose and the chest wall DVH parameters, but none of those factors predicted for chest wall pain without fracture on the univariate model. Overall, 46/289 (16%) were symptomatic of chest wall pain, with 22 symptomatic rib fractures and 24 having chest wall pain without fracture. On univariate analysis, tumours adjacent to the chest wall (P ¼ 0.009), higher chest wall V30 (P ¼ 0.027), chest wall D1cc (P ¼ 0.056), chest wall D2cc (P ¼ 0.045) and chest wall D5cc (P ¼ 0.041) predicted for a greater risk of chest wall pain, but none of these factors was significant on the multivariate model.
Discussion Fig 2. Incidence of rib fracture after stereotactic ablative radiotherapy in 158 tumours adjacent to the chest wall for osteoporotic versus non-osteoporotic patients.
We showed that pulmonary tumours adjacent to the chest wall are at greater risk of rib fracture after lung SABR and none of the chest wall DVH metrics independently predicted for SABR-induced rib fracture when tumour
Fig 3. Stereotactic ablative radiotherapy (SABR)-induced rib fracture. Primary non-small cell cancer located adjacent to the chest wall, delineation of the chest wall contour, SABR treatment plan using 48 Gy/four fractions and rib fracture observed 15.3 months after lung SABR.
32
Table 2 Summary of the published literature on chest wall toxicities following lung stereotactic ablative radiotherapy (SABR) Reference
Year
Tumour type
Total dose (Gy)/no. fractions
No. patients (no. lesions)
Organ at risk for dose evolume histogram data
% of CWP
% of severe CWP
No. rib fractures (%)
Symptomatic rib fractures (%)
Median time in months (range) to rib fracture
to CWP
Predictors
Principal findings
Fracture: smallvolume/highdose region; D2cc e
50% risk of rib fracture if D2cc ¼ 3 16.6 Gy (49.8 Gy)
2009
P
45/3
33
Ribs
e
e
13 (7/33 patients, 21%)
e
15 (8e38)
e
[17]
2009
P
54e60/3
42
e
26%
e
15 (9/42 patients, 21%)
13/15 (87%)
17
e
[5]
2012
P
54e60/3
46 (49)
Ribs
46%
e
41 (17/46 patients, 37%)
21 (7e40)
e
[13]
2012
P or M
48/4
116
Ribs
e
e
46 (28/116 patients, 24%)
14/17 patients (82%) 12/28 patients (43%)
22 (9e42)
e
[14]
2013
P
48e70/4e10
177z
e
21%
e
41 (41/177 patients, 23%)
14/41 (34%)
33 (24e94)
e
[15]
2013
P or M*
36e60/3e4
118 (126)
CW
25%jj
e
48 (48/126, 38%)
(18e61%)
17 (4e52)
e
Fracture: female gender, lateral location, D8cc
Chest wall pain [18] 2011
P or My
50/4
265
CW
25%
9%
e
e
e
6 (0e11)
[19] [20]
2012 2012
P P or M
48e60/3e10 60/3
102 (106) 45 (48)
CW CW
19%jj 22%
0.9%jj
e e
e e
e e
8 (6e10) 8.8 (3.7e12.0)
CWP: BMI, DB, CW V30 CWP: mEUD, BMI CWP: CW V30 and V60
[6]
2012
P
40e60/3e5
126
CW
43%
15%
8 (5/126 patients, 4%)
3/5 patients (60%)
27 (21e32)
9 (1e36)
CWP: CW V30
[21]
2012
P or M
50e54/3e5
140 (146)
CW
15%jj,{
1.4%
e
e
e
12.6 (4e36)
CWP: BMI, CW V30, V35, V40
60
CW
33%
28%
5 (5/60 patients, 8%)
5/5 (100%)
7 (0.6e32)
Chest wall pain and rib fracture [16] 2010 P or My 21e60/3e5
Fracture: rib D0.5cc, age, female gender Fracture: rib etumour distance 16 mm developed fracture. Female gender and lateral location are risk factors for rib fracture. 18% risk of rib fracture if V30 30 cc mEUD better predicted CWP Restricting V30 to
