Quantitative assessment of lung density changes after 3-D radiotherapy for breast cancer

Abstract

Our aim was to reduce the rates of clinical and radiological pneumonitis in local-regional radiotherapy (RT) for breast cancer compared to a previous treatment series by applying a pre-planned lung dose volume constraint. 3-D dose planning was performed in 66 women with the aim of not exceeding an ipsilateral V20 of 30%. The patients were followed for short-term signs/symptoms of post-RT pneumonitis and radiological changes on chest CT 4 months after RT. Radiological abnormalities were scored with a CT-adapted modification of a semi-quantitative classification system originally proposed by Arriagada which accounts for severity and affected lung regions. The abnormal subvolumes were contoured and the mean doses were calculated. Three cases of mild and one case of moderate symptomatic pneumonitis were diagnosed. The mean V20 was higher in symptomatic than in unaffected patients, 29% vs. 24% (p =0.04). Mild/moderate radiological changes were frequently observed on CT in regions with average doses >30 Gy. According to multivariate modeling, a trend for association was found between the studied dosimetric factors V13, V20, V30 and mean lung dose, and moderate-severe changes on CT but not with patient specific covariates, e.g. chemotherapy or tamoxifen exposure. 3-D planned local-regional RT with a preplanned lung dose volume constraint of V20 <30% resulted in few cases of moderate symptomatic pneumonitis. Mild/moderate radiological changes were still detectable on CT in subvolumes receiving doses >30 Gy. Long-term follow-up for evaluation of possible late morbidity is warranted.

It is well known that adjuvant local-regional radiotherapy (RT) in addition to systemic therapy reduces the risk of recurrence and death in women with high-risk breast cancer (BC) compared to systemic therapy alone [1–3]. The treatment has, however, also been associated with both pulmonary [4] and cardiac side effects [5]. Our [6], [7] and other research groups, e.g. at the Netherlands Cancer Institute [8–10] and at Duke University in USA [11], have, during the last decade, extensively studied RT-related lung side effects in patients with BC with respect to clinical [11], [12], radiological [13–15], and physiological sequelae [16]. The risk and severity of acute and chronic RT-induced lung morbidity are most importantly influenced by total dose, dose per fraction, and volume of incidentally irradiated lung. In our previous work we found no case of moderate clinical radiation pneumonitis in patients who received a dose ≥20 Gy (V20) to less than 30% of the ipsilateral lung volume [12]. We have also reported an association between age and both radiological [13], [14] and clinical pneumonitis [12]. Other groups have found relations between tamoxifen intake [17] and chemotherapy exposure [18], [19], and RT-induced lung damage.

In the present study, we report on RT-related radiological lung density changes and rates of clinical radiation pneumonitis according to CTC-criteria [20] in a cohort of 66 prospectively followed women with BC who received local-regional adjuvant RT based on 3-D planning aimed at a V20 of < 30% of the ipsilateral lung volume. The radiological sequelae were assessed with repeated computed tomography (CT) scans of the thorax and analyzed by use of a modification and CT-adaptation of the Arriagada classification system for RT-induced lung changes [14], [21]. Furthermore, the lung subvolumes with density changes were contoured separately and their mean doses were calculated and compared to the dose distribution in the entire ipsilateral lung volume.

Patients and methods

This study was approved by the local ethics committee. All participating women gave informed consent before study enrolment.

Patient population

All women who were referred to the Radiotherapy Department at Stockholm Söder Hospital during 2003–2005 for adjuvant local-regional RT after surgery for early breast cancer were asked to participate in this study. Of the 71 included patients, 66 completed both the pre-RT and the post-RT CT 4 months after the cessation of treatment and the 7 months follow-up for signs and symptoms of clinical pneumonitis. Five patients withdrew their consent during the follow-up period due to early recurrence of disease and were, thus, not evaluable. Of the evaluable cases, 48 had undergone a modified radical mastectomy and the remaining were referred after lumpectomy with axillary dissection. Data on patient- and therapy characteristics were collected prospectively. The mean age of the participants was 56 years (range: 33–81) and limited co-morbidities were registered at base line. Of the total population, no patient had a history of severe heart disease but three cases of hypertension or mild heart disease without cardiac failure were reported. Furthermore, three cases of mild pulmonary disease, i.e. asthma, were registered at base line. Fourteen of the women were active smokers during the period of RT and 45% of the total population reported a reduced level of function prior to RT, i.e. not being able to climb three flights of stairs without a rest due to shortness of breath.

Of the total, 51 patients were treated with pre-RT chemotherapy which generally was concluded 3–4 weeks prior to RT. Concurrent chemotherapy was never given. A large variety of pre-RT chemotherapy combinations were used during the study period but the regimens typically consisted of doxorubicin, cyclophosphamide, and 5-fluorouracil. In eighteen patients, the adminstered pre-RT chemotherapy included the taxane docetaxel and five received trastuzumab. The latter therapy was administered during RT in seven cases and concurrent intake of tamoxifen or anastrozol was recorded in a total of 38 patients, evenly split.

Radiotherapy treatment techniques

The RT treatment techniques used have been described in detail in an earlier publication [16]. Local-regional RT after mastectomy (n = 48) was delivered with an anterior electron beam covering the chest wall and the internal mammary lymph nodes (IMN) (range: 6–12 MeV) and with 6 MV photon beams covering the supraclavicular fossa and axilla (total dose 46 Gy). The field arrangement is shown in Figure 1A, B. Local-regional RT after lumpectomy without coverage of the IMN (n = 10) consisted of two tangential photon beams of 4 or 6 MV which included the breast parenchyma (50 Gy) and regional lymph node coverage similar to the previous description (46 Gy) (Figure 2A, B). In four additional cases, an additional oblique electron beam was added to include the IMN in the target volume (46 Gy). An equal number of patients (n = 4) were treated with miscellaneous techniques. The treatment was given in daily fractions of 2 Gy, 5 days per week.

Figure 1A.  Frontal image (Beam's-eye-view (BEV)) depicting the position of the frontal 6 MV photon beam covering the supraclavicular fossa and axilla. Barred areas are shielded. Green areas represent regions with post-RT density changes.

Treatment planning and dose-volume histogram analyses

Treatment planning (Pinnacle; version 6.2b) was performed with the aim of an acceptable target coverage and avoidance of a dose in excess of 20 Gy to more than 30% of the ipsilateral lung volume. The cumulative dose-volume histograms were calculated and the ipsilateral lung volumes receiving ≥13 Gy (V13), ≥20 Gy (V20), ≥30 Gy (V30), and mean lung dose (MLD) [12] were determined. The mean values of V13, V20, V30 and MLD in the entire population were 32% (range: 8–44), 25% (range: 6–32), 16% (range: 4–27) and 12 Gy (range: 3–15), respectively, and for the subgroup of patients treated after mastectomy (n = 48) 34%, 25%, 16%, and 13 Gy.

Evaluation with CT and a modification of Arriagada's classification

Computerized tomography (CT; Siemens Somatom Volume Access) of the thorax was performed prior to and 4 months following the cessation of RT. Care was taken to position the patient identically on both occasions to facilitate the matching of the examinations. On both occasions, slices of 0.5 cm were examined for radiological changes throughout the entire lung volume by use of the standard lung window (mean −600; width 1 000 Hounsfields's Units). An increase in density was graded with a CT-adapted modification of Arriagada's classification [14] (i.e. 0 = no change; 1 = low opacity in linear streaks; 2 = moderate opacity; 3 = complete opacity), which has been described in detail in our earlier publication [14]. The lung was, furthermore, divided into the three regions suggested by Arriagada (i.e apical-lateral (A-L), central-parahilar (C-P), and basal-lateral (B-L)) [14], [21]. The border between the A-L and B-L regions was set at the level of the pulmonary artery. The width of the CP-region was set to 5 cm. The highest density grade in each region was added together. Total scores of 1–3 were considered to represent mild radiological reactions while scores from 4–9 represented moderate to severe reactions.

Definition of the clinical radiation pneumonitis

The patients were followed for signs and symptoms of clinical radiation pneumonitis (i.e. cough and/or dyspnoea, with or without fever) at 1, 4, and 7 months after the conclusion of RT. All patients attended every pre-planned visit at our out patient ward. Cases of radiation pneumonitis were classified according to modified CTC-criteria (version 2.0) [20] into three groups as follows:

1. No reaction: no registered respiratory symptoms or symptoms judged by the clinicians not to be caused by RT.

2. Mild reaction (CTC grade 1): respiratory symptoms judged by the clinician to be caused by RT but not treated with corticosteroids.

3. Moderate reactions (CTC grade 2): same as 2 but impairing daily function and treated with corticosteroids.

No case of CTC grade 3–4 toxicity was diagnosed. Chest radiography was recommended as a complimentary diagnostics in patients with respiratory symptoms.

Statistical methods

The relation between clinical and radiological pneumonitis was tested with Gamma-statistics. The relation between radiological pneumonitis and the studied dosimetric factors and covariates was analyzed with univariate and multivariate logistic regression (Wald-Enter method). The relation among the studied dosimetric factors was tested with Pearson Correlation. The Mann-Whitney test was used to compare differences in subvolumes with density changes and calculated ipsilateral V20-values in symptomatic vs. asymptomatic patients. The Wilcoxon Matched Pairs Test was used when the mean dose of the radiologically changed subvolumes were compared with the MLD of the entire ipsilateral lung. All reported results were based on two-sided tests and p-values < 0.05 were considered statistically significant.

Results

Moderate clinical radiation pneumonitis was diagnosed in only one woman and mild reactions were registered in three cases, i.e. a total incidence of < 10%. The average ipsilateral V20 was higher in patients affected by symptomatic pneumonitis compared to asymptomatic individuals, 29% vs. 24% (p = 0.04). The rates of radiological and clinical radiation pneumonitis are depicted in Table I. The majority of patients developed mild radiological changes on CT (score 1–3) but very few patients experienced symptomatic pneumonitis, when the pre-planned lung dose volume constraints were applied. The correlation test between clinical and radiological pulmonary toxicity did not attain statistical significance (Table I).

Table I.  Results of correlation test between CT scores and clinical radiation pneumonitis:

	Radiation pneumonitis
CT scores^1	None (Grade 0)^2	Mild (Grade 1)	Moderate (Grade 2)
0	14	0	0	G^3=0.66 (p = 0.1)
1–3	35	2	0
4–9	13	1	1

¹Modified and CT adapted Arriagada Classification. ²Common Toxicity Criteria. ³Gamma statistics.

A relation between post-RT radiological density changes (none and mild (0–3) vs. moderate to severe (4–9)) and patient/therapy specific factors was tested (Table II). According to univariate logistic regression analyzes, fewer cases of moderate-severe radiological density changes were seen in younger patients and in patients undergoing pre-RT chemotherapy. Pre-RT docetaxel containing chemotherapy was, furthermore, not related with more cases of radiological lung toxicity. No relation was seen between moderate-severe radiological changes and ongoing smoking or concurrent tamoxifen, anastrozol, or trastuzumab therapy (Table II).

Table II.  Results of tests of relation between co-variates and moderate/severe radiological changes and multivariate (MVA) modeling based on the four studied dosimetric factors.

Co-variates	Univariate analysis^1	MVA^2 V13	MVA V20	MVA V30	MVA MLD
Dosimetric factor	–	p = 0.1 (NS)	p = 0.1 (NS)	p = 0.08 (NS)	p = 0.09 (NS)
Age	p = 0.017	NS	NS	NS	NS
Pre-RT chemotherapy	p = 0.015	NS	NS	NS	NS
Pre-RT docetaxel	NS	–	–	–	–
Ongoing smoking	NS	–	–	–	–
Concurrent tamoxifen	NS	–	–	–	–
Concurrent anastrozol	NS	–	–	–	–
Concurrent trastuzumab	NS	–	–	–	–

¹Univariate logistic regression. ²Multivariate logistic regression (Wald-Enter). NS – not significant.

Four different multivariate models were tested which included the studied dosimetric factors separately, i.e. V13, V20, V30, and MLD, and the factors age and pre-RT chemotherapy (Table II). None of the multivariate models could detect a statistically significant association, however, there appeared to be a trend for a relation between V30 and density changes of grade 4–9 (p = 0.08). The results of the other three multivariate models performed in parallel were MLD; p = 0.09, V13; p = 0.1, and V20; p = 0.1 (Table II). Age and pre-RT chemotherapy exposure did not affect the occurrence of moderate-severe post-RT radiological changes according to multivariate modeling. There was strong interrelations among all the studied dosimetric factors (i.e. range for correlation coefficients: 0.62–0.97; p < 0.01 for all tested correlations).

An example of the location of a contoured lung subvolume with density changes in relation to the beam arrangement is shown in Figure 1A, B in a patient who received RT following mastectomy. The cumulative ipsilateral lung dose volume histograms for the entire lung and for the affected lung subvolumes in the same individual are depicted in Figure 1C. The mean V20-value with SD for all patients treated following mastectomy and the average mean dose in damaged subvolumes of patients with radiological changes are also inserted in Figure 1C. Another example of a contoured affected subvolume is shown in Figure 2A, B in a woman who was treated after lumpectomy. The cumulative dose volume histograms for this patient are shown in Figure 2C. The mean V20-value with SD for all patients treated following lumpectomy and the average mean dose in damaged subvolumes of patients with radiological changes are also inserted in Figure 2C. The average mean dose in the subvolumes with density changes in all affected patients was 37 Gy (SEM: 1.5), which was significantly higher than the average ipsilateral MLD of 13 Gy (p < 0.001). The affected mean volume was 20 ml (SEM: 2.8). There was no statistical difference between the symptomatic_n = 4 vs. the asymptomatic patients_n = 48 with CT changes with respect to the average volume of altered lung tissue, 30 ml vs. 16 ml.

Figure 1B.  Frontal image (BEV) depicting the position of the frontal electron beam covering the chest wall and the IMN. Barred areas are shielded. Green areas represent regions with post-RT density changes.

Figure 1C.  Cumulative dose volume histograms for the entire ipsilateral lung (red) and for regions with post-RT density changes (green-dashed) (Normalized volumes; typical total lung volume about 1 200 ml and volume with density changes about 20 ml). The red vertical line illustrates the mean V20 value (±SD) for all 48 irradiated, mastectomized patients. The green horizontal line depicts the average mean dose (± SD) of the damaged lung subvolumes among the mastectomized patients with post-RT lung density changes.

Figure 2A.  Frontal image (BEV) depicting the position of the frontal photon beam covering the supraclavicular fossa and axilla. Green areas represent regions with post-RT density changes.

Figure 2B.  Lateral image (BEV) depicting the medial tangential photon beam. Green areas represent regions with post-RT density changes.

Figure 2C.  Cumulative dose volume histograms for the entire ipsilateral lung (blue) and for regions with post-RT density changes (green-dashed red). The teal line depicts the dose to the contralateral lung. The blue vertical line illustrates the mean V20 value (±SD) for all 14 irradiated, lumpectomized patients. The light brown horizontal line depicts the average mean dose (±SD) of the damaged lung subvolumes among the lumpectomized patients with post-RT lung density changes.

Discussion

Short-term clinical radiation pneumonitis was a rare event in this cohort of women who had undergone 3-D treatment planning aimed at minimizing V20 to < 30% of the ipsilateral lung volume, i.e. only one case of moderate and three cases of mild pneumonitis in 66 prospectively followed women. We could still detect a statistical difference between the average V20-value in symptomatic vs. asymptomatic individuals, i.e. 29% vs. 24%. In our previously reported cohort of patients receiving local-regional RT with similar techniques when lung dose volume constraints where not applied, the incidence of moderate clinical pneumonitis was about 10% [12]. The average V20 and MLD for women receiving local-regional RT following mastectomy in the latter study were 35% and 16 Gy, respectively, which should be compared with the mean values of the present series of 25% and 13 Gy. Thus, our goal to reduce the rate of clinically significant symptomatic pneumonitis, i.e. CTC grade ≥2, by applying the pre-planned dose volume constraints appears to have been reached.

The rates of radiological pneumonitis with the used semi-quantitative evaluation technique appeared, however, similar to our earlier treatment series [14]. One should, however, observe that the entire lung volume was inspected for radiological changes in the present cohort of patients rather than only three CT-slices, which was the done in our earlier report [14]. Limited changes could, thus, have been missed in the previous treatment series. Furthermore, we did not have volumetric information concerning the density changes in our first treatment series and some of the subvolumes in the present series were relatively small. We can, however, conclude that local-regional RT with the described techniques and dose volume constraints frequently results in structural lung changes in regions receiving average doses in excess of 30 Gy which are not accompanied by short-term symptoms.

Is it of interest to continue studying radiological changes in women with BC receiving adjuvant local-regional RT as the short-term consequences appear very limited when lung dose volume constraints are applied? Firstly, limited data are available about the long-term effects of the observed radiological changes. We are presently executing a long-term follow-up of our previously studied cohort of 500 women who underwent RT 1994–1998 [12] and we will shortly be able to add information on this issue. Secondly, recent reports suggest an increased risk of secondary lung cancer in irradiated women with BC compared to healthy controls [22], [23]. It is reasonable to believe that such a risk could be related to the amount of damaged lung tissue and that we should try to reduce/limit the development also of asymptomatic post-RT radiological changes in this group of patients with a long expected mean survival.

We could not detect a confounding effect of concurrent trastuzumab or tamoxifen on RT-related radiological changes. Bentzen et al. have reported a relation between the latter factor and post-RT lung changes on chest x-rays [17]. Furthermore, we were unable to detect a protective effect of smoking against RT-induced lung toxicity, which was suggested by Johansson and co-workers in an earlier report [24]. Carmustine- [18] and paclitaxel-containing [19] chemotherapy regimens have been associated with increased post-RT lung morbidity according to previously published data. We could not observe a similar effect for the chemotherapy drugs and timing used in the current report. In summary, both the complexity and variation of the pre- and post-RT therapies in patients with BC are increasing, e.g. by the introduction of targeted therapies. We need to continue to gather data on the potential confounding effect of concurrent therapies and patient characteristics on the development of post-RT morbidity in future large treatment series.

In conclusion, we found few cases of short-term symptomatic radiation pneumonitis when a dose volume constraint of V20 < 30% of the ipsilateral lung was applied to women undergoing local-regional RT for BC. Mild and moderate post-RT radiological sequelae were, however, still frequently observed on chest-CT in subvolumes of the lung receiving an average dose > 30 Gy. We were not able to detect a relation between the development of RT-induced radiological changes and any of the studied co-variates. Presently, it is not fully known if the radiological changes will translate into late morbidity or whether there could be a relation between the amount of acute radiological lung damage and an increased risk of secondary lung cancer. For the latter reasons and due to the frequent introduction of new adjuvant therapies in BC we feel that RT-induced early and late pulmonary morbidity should be monitored also in future treatment series where patients with BC receive local-regional RT.

This work was supported by the Swedish Cancer Foundation (Cancerfonden). We are grateful for the work that the staffs of the Radiotherapy Department and the Breast Cancer Outpatient Ward at Stockholm Söder Hospital have put into this study.