Incidence, risk factors, and dose-volume relationship of radiation-induced rib fracture after carbon ion radiotherapy for lung cancer

Abstract

Background: The purpose of this study was to assess the incidence, risk factors, and dose-volume relationship of radiation-induced rib fracture (RIRF) after carbon ion radiotherapy for lung cancer.

Material and methods: Fifty-seven ribs of 18 patients with peripheral stage I non-small cell lung cancer treated with carbon ion radiotherapy were analyzed on rib fracture. The patients were treated at a total dose of 52.8 Gy [relative biologic effectiveness (RBE)] or 60.0 Gy (RBE) in 4 fractions and were followed at least six months. Patient characteristics and dosimetric parameters were analyzed for associations with RIRF.

Results: Eighteen patients and 57 ribs were included in this study. The median length of follow-up was 36.5 months. RIRF was observed in seven (39%) of the 18 patients, and in 11 (19%) of 57 ribs. Only one patient developed symptomatic fracture. The distance from the ribs to the tumor site was significantly shorter in fractured ribs than in non-fractured ribs (1.4 ± 0.3 cm vs. 2.5 ± 0.3 cm). Receiver operating characteristic curve analysis showed that as a cut-off value for discriminating RIRF had the largest area under the curve (AUC =0.78). Comparison of the two-year cumulative incidence of RIRF among two groups as determined by cut-off values, yielded the following result: 53% vs. 4% [, ≥ 38.2 Gy (RBE) or less]. Results from the two groups were significantly different (p < 0.05).

Conclusion: The crude incidence of RIRF after carbon ion radiotherapy was 39% but incidence of symptomatic fracture was low. The as cut-off values may be helpful for discriminating the risk of RIRF.

The use of stereotactic body radiotherapy (SBRT) with x-rays has rapidly emerged as a non-invasive cancer treatment modality due to improvements of radiation delivery techniques [1]. Several studies have shown that the treatment is effective and well tolerated in lung cancer patients [2–4]. However, radiation-induced rib fracture (RIRF) has been recognized as a late toxicity of this treatment [5–11]. SBRT with x-rays uses large doses per fraction to achieve good local tumor control, which might increase the risk of late toxicities. In fact, the incidence of RIRF after SBRT with x-rays for lung cancer has been addressed in numerous reports [6–11]. Carbon ion radiotherapy has been used for lung cancer patients since 1994 and has showed good local control [12,13]. Carbon ion beams deposit maximum dose sharply at the end of their range. This enables the delivery of a higher dose to the tumor site, while minimizing the dose delivered to surrounding normal tissue [14]. Carbon ion beams also have a larger mean linear energy transfer compared with photons and protons, resulting in a higher relative biological effectiveness (RBE) [15]. Hypofractionated carbon ion radiotherapy for early stage lung cancer has been utilized since 2010 at the Gunma University Heavy Ion Medical Center (GHMC). As mentioned above, carbon ion beams have a higher RBE on cells than photons and protons, which may increase the incidence of adverse events. However, few reports have addressed rib fracture after carbon ion radiotherapy [16,13]. The aim of this study was to assess the incidence, risk factors, and dose-volume relationship of RIRF after carbon ion radiotherapy for lung cancer.

Material and Methods

Patients

Since 2010, peripheral stage I non-small cell lung cancer patients at GHMC have been treated with carbon ion radiotherapy. All patients included in this study were treated based on a prospective protocol that was approved by the institutional review board of the university [UMIN: 000003797]. Participating patients underwent at least six months of follow-up. Patient and tumor characteristics are summarized in Table I. Median patient follow-up period was 36.5 months (range 7–53 months). Median patient age was 81.5 years (range 47–83 years). Patients included 12 men and six women. Numbers of T1a, T1b and T2a were five, eight and five, respectively. Cancer stages were stage IA in 13 patients and stage IB in five patients.

Table I. Patient characteristics.

Characteristic
Age  (median)	47 – 83 y.o. (81.5)
Sex
Male	12 patients
Female	6 patients
T stage
1a	5 patients
1b	8 patients
2a	5 patients
Rib-tumor distance
≧ 2.0 cm	4 patients
< 2.0 cm	14 patients

Footnote: Rib-tumor distance was defined as the minimum distance from the rib to the edge of gross tumor volume.

Carbon ion radiotherapy

Patients with stage IA and IB cancer were treated with a total dose of 52.8 Gy (RBE) and 60.0 Gy (RBE), respectively, in 4 fractions over the course of a week. Absorbed dose was 5 Gy in 1 fraction. In this study, the dose of carbon ion radiotherapy is expressed as Gy (RBE), and calculated by multiplying the carbon absorbed dose (Gy) by the RBE of 3. Carbon ion beams were generated using the heavy particle accelerator at GHMC. Energies of the beams were 290 MeV/u, 380 MeV/u, and 400 MeV/u. Beam energy was determined for each case based on the depth of tumor. Usually, non-opposing four beams are used to reduce the dose of chest wall and skin. Patients were immobilized using fixation cushions and thermoplastic shells of 3 mm thickness. After immobilization, respiratory-gated computed tomography (CT) images were taken for treatment planning. The gross tumor volume (GTV) was defined as lesions visible on CT lung window images. The clinical target volume (CTV) margin, including subclinical disease invasion, was added to the GTV, with 8 mm in all directions (excluding bony structures or the chest wall). The internal margin (IM) was determined by the demonstrable extent of tumor motion shown in four-dimensional (4D) CT images to compensate for tumor respiratory motion. The planning target volume (PTV) included the CTV, IM, and setup margin [17]. Dose distribution was normalized at isocenter which was same as weight point. Planning aim was to cover the PTV for at least 90% of the prescribed. Maximum dose allowed in PTV was below 107% of prescribed dose. When tumors were located near the ribs, priority was usually given to covering the PTV with the prescribed dose rather than avoiding ribs.

Follow-up and clinical evaluations

Data were prospectively collected according to the study protocol. CT was taken every three months during first one year after treatment and was performed every six months after that. Rib fractures were diagnosed by cortical discontinuity of rib bones on CT images (Supplementary Figure 1, available online at http://www.informahealthcare.com). RIRF was graded using Common Terminology Criteria for Adverse Events, Version 4.0.

Dose-volume evaluations

All rib bones that received more than 20 Gy (RBE) were contoured on the treatment planning CT, and dose-volume histograms (DVH) were generated [11]. There was no patients who were excluded from the study because they did not receive more than 20 Gy (RBE) to the ribs. Ribs were automatically contoured using MIM Maestro, Version 5.6 (MIM Software Inc. Cleveland, OH, USA). Rib contours were manually adjusted by a single observer (TA) after automated contouring. The maximum point dose to the ribs (D_max), volumes receiving more than 20 Gy (RBE) (V20), 30 Gy (RBE) (V30), 40 Gy (RBE) (V40), 50 Gy (RBE) (V50), and 60 Gy (RBE) (V60), and the minimum dose in the most irradiated tissue volume (0.5, 1, 2, 3, 4, and 5 cm³) were calculated based on the DVH.

Statistical analysis

The patients were divided into two groups by their characteristics and cumulative incidence of RIRF was estimated using the Kaplan-Meier method, and was compared between two groups with the log-rank test. Mean DVH parameters in the two groups were compared with the Student’s t-test. Receiver operating characteristic (ROC) curves were calculated to assess optimal cut-off values for DVH parameters and their ability to discriminate RIRF. The curve was defined as the plot of the true-positive rate (sensitivity) versus the false-positive rate (specificity). Areas under the curve (AUCs) were calculated to compare each cut-off value. p < 0.05 was considered statistically significant. All statistical analyses were performed using IBM SPSS Statistics for Windows, Version 21.0 (SPSS Inc., Armonk, NY, USA).

Results

RIRF was observed in seven (39%) of the 18 patients, and in 11 (19%) of 57 ribs. The median period for development of rib fracture was 14 months (range 5–43 months) from the initiation of the treatment. Six patients developed Grade 1 rib fractures, and only one patient developed a Grade 2 rib fracture, which required non-steroidal anti-inflammatory drug. RIRF was observed more frequently in patients who received dose of 60.0 Gy (RBE) than in patients who received 52.8 Gy (RBE) [73% vs. 41% (cumulative incidence), p < 0.05]. Significant differences were not observed in other clinical factors, such as gender or age (Table II). Rib-to-tumor distances were significantly shorter in fractured ribs than in non-fractured ribs (1.4 ± 0.3 cm vs. 2.5 ± 0.3 cm, p <0.05). Analysis was conducted on 57 ribs that received more than 20 Gy (RBE). Dosimetric parameters relating to RIRF are described in Table III. The D_max was 53.7 ± 1.8 Gy (RBE) in fractured ribs, as compared to 41.7 ± 2.0 Gy (RBE) in non-fractured ribs, indicating a significant difference (p < 0.05). The V30–V60 were significantly higher in fractured ribs than non-fractured ribs, while V20 was not significantly higher in fractured ribs. The ROC analysis showed maximum AUC with (AUC =0.78). In addition, and V40 showed large AUCs (AUC =0.76 and 0.76, respectively). Comparison of the two-year cumulative incidence of RIRF among two groups, as determined by cut-off values from ROC analysis, yielded the following results: 53% vs. 4% [, ≥38.2 Gy (RBE) or less]. The incidence of RIRF in the two groups were significantly different (p <0.05).

Table II. Comparison of the probability of radiation-induced rib fracture (RIRF) for each patient characteristics.

Characteristic		number of cases	2 - year cumulative incidence of RIRF (%)	p value
Age	≧ 75 y.o.	n = 8	45	0.67
	< 75 y.o.	n = 10	55
	Sex	Male	n = 12	33	0.11
Female		n = 6	75
Prescribed dose		52.8 Gy(RBE)	n = 13	41	0.03
	60.0 Gy(RBE)	n = 5	73

Footnote: RIRF, radiation induced rib fracture; Gy (RBE), the dose of carbon ion radiotherapy calculated by multiplying the carbon absorbed dose (Gy) with the relative biological effectiveness.

Table III. Comparison of dosimetric parameters between fractured ribs and non-fractured ribs.

Parameter	Fractured ribs	Non-fractured ribs	p value
V60 (cm^3)	1.0 ± 0.5	0.4 ± 0.2	<0.001
V50 (cm^3)	1.9 ± 0.5	0.7 ± 0.2	0.03
V40 (cm^3)	3.1 ± 0.6	1.6 ± 0.3	0.04
V30 (cm^3)	4.9 ± 0.6	2.6 ± 0.9	0.04
V20 (cm^3)	6.2 ± 0.6	4.3 ± 0.4	0.13
Dmax (Gy(RBE))	53.7 ± 1.8	41.7 ± 2.0	<0.001

Footnote: Vx, volume of rib receiving > x Gy (RBE); D_max, maximum point dose of the ribs; , the minimum dose in the most irradiated rib volumes of X - cm³.

Table IV. Optimal cut-off value determined by ROC analysis and calculated area under the curve.

	Opitmal cut-off value	AUC	Incidence of RIRF above/ below the cut-off value
	18.9 Gy (RBE)	0.68	56/10%
	31.3 Gy (RBE)	0.75	49/8%
	38.2 Gy (RBE)	0.78	53/4%
	43.6 Gy (RBE)	0.76	52/7%
Dmax	49.7 Gy (RBE)	0.74	54/8%
V60	0.1 cm^3	0.67	68/21%
V40	0.9 cm^3	0.76	52/7%
V30	4.7 cm^3	0.74	59/16%

Footnote: ROC, receiver operating characteristic curve; AUC, area under the curve.

Discussion

The crude incidence of RIRF in this study was 39%. Studies of RIRF after SBRT using photon beams reported an incidence ranging from 21% to 42% at a dose of 45–70 Gy in 3–10 fractions [6–10]. These results are summarized in Supplementary Table I (available online at http://www.informahealthcare.com). Although the incidence of RIRF in this study was relatively high, it did not deviate from other reports. Asai et al. reported a rib-to-tumor distance of 2.0 cm to be a significant cut-off value for classifying the risk of RIRF, and, in this study, 78% of patients (14 of 18 patients) had less than 2.0 cm of rib-to-tumor distance which was calculated as smallest distance in each patient. One of the causes of the high incidence of RIRF was the high rate of patients with small rib-to-tumor distances.

Although many reports have analyzed the correlation between patient characteristics and the incidence of RIRF, the results are controversial. Two studies reported that women were at significantly higher risk for RIRF [6,9], while two other studies reported that patient characteristics were not significant risk factors [10,18]. In this study, patient characteristics, including age and gender, were not significantly correlated with RIRF. Asai et al. hypothesized that the incidence of RIRF was influenced more by dose-volume factors than by patient characteristics because of the higher biologically effective dose (BED) of hypofractionated SBRT [10]. The incidence of RIRF after carbon ion radiotherapy may also be affected more by dose-volume factors than by patient characteristics because of the high BED of hypofractionated carbon ion radiotherapy.

As for RIRF after SBRT, Taremi et al. reported that a D_0.5cm³ of 60 Gy was an optimal cut-off value and Asai et al. reported that a D_max of 42.4 Gy in 4 fractions was an optimal cut-off value, and was associated with a 50% risk of RIRF after SBRT [8,10]. The ROC analysis showed to have the largest AUC, with and V40 also showing large AUCs. The cumulative fracture rate was approximately 50% when was above the cut-off value (Figure 1). These results indicate that cut-off value for discriminating the risk of RIRF after carbon ion radiotherapy was similar to other reports with photon therapy [8,10]. In addition, a D_max of 49.7 Gy (RBE) was shown to be a significant cut-off value for discriminating the risk of RIRF. Kanemoto et al. reported a D_max of 150 Gy (RBE) was a significant cut-off value for estimating the risk of RIRF after proton beam therapy for hepatocellular carcinoma (HCC) [18]. A D_max of 150 Gy (RBE) in 10 fractions is equal to 53.8 Gy (RBE) as the BED in 4 fractions, as calculated using the linear quadratic equation [19], assuming an α/β ratio of 3 Gy. Although, ROC analysis does not tell an absolute risk of RIRF at a specific dose, the results of these studies were quite similar. The dose-volume relationship of RIRF after carbon ion radiotherapy seems to be consistent with that after SBRT using photon and proton beams.

Figure 1. Thick line showed cumulative incidence of radiation-induced rib fracture per patients. Among 18 patients, seven patients (39%) developed rib fracture. Dashed line and thin line showed comparison of the cumulative incidence between two groups [, ≥38.2 Gy (RBE) or less] divided by optimal cut-off value derived from ROC analysis (p <0.05).

Seven of 18 patients developed RIRF in this study but only one patient complained of chest wall pain. Miyamoto et al. reported that nine (11%) of 79 patients complained of costal bone pain and tenderness after carbon ion radiotherapy for lung cancer [17]. Kanemoto et al. reported that 11 (17%) of 67 patients developed rib fracture after proton therapy for HCC, and four patients (6%) were symptomatic. Asai et al. also reported that 28 (24%) of 116 patients developed rib fractures and 12 patients (10%) were symptomatic. In this study, crude incidence of rib fracture was 39%, but the crude incidence of symptomatic fractures was 5%. This result justifies the therapeutic strategy of prioritizing PTV coverage with the prescribed dose, rather than decreasing the dose administered to the ribs when tumors are located near the ribs.

In conclusion, the crude incidence of RIRF after carbon ion radiotherapy for lung cancer was 38.9%, but the crude incidence of symptomatic fracture was 5%. The may be useful for reducing the risk of RIRF.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.