Occurrence of pneumonitis following radiotherapy of breast cancer – A prospective study

Cancer therapy-triggered lung impairment interferes with quality of life. Since locoregional radiotherapy is state-of-the-art treatment in breast cancer, radiation pneumonitis (RP) still has to be accepted as an early to intermediate toxicity. Generally, pneumonitis is classified in stages I–IV [1] corresponding to various pathophysiological alterations in lung tissue. The first symptoms are observed during an exudative phase of increased capillary permeation and leukocyte infiltration, followed by an organizing or proliferating phase, which potentially leads to permanent fibrotic lung damage associated with extended pneumocyte death [2]. Clinical manifestation is mainly characterised by newly developed dyspnoea, usually accompanied by dry cough. Other symptoms can be fever and malaise [3]. In the case of permanent respiratory symptoms the alterations can lead to RILD (radiation-induced lung disease) [4]. In some cases, severe tissue complications such as bronchiolitis obliterans organizing pneumonia (BOOP) [5, 6] or chronic eosinophilic pneumonia can evolve [7]—even in the “nonirradiated” lung [8]. Pneumonitis normally occurs between 6 and 21 weeks [1, 9, 10] after radiotherapy. In the past, telecobalt therapy [11] caused pneumonitis in up to 35% of all treated breast cancer patients, whereas incidences reported for modern CT(computer tomography)-based photon therapy range from 1% [12] and 4% [13], 12% [14, 15] to 21% [16]. However, the correlation between normal tissue complication probabilities (NTCPs) and dosimetry parameters, such as extent of co-irradiated lung tissue [17], central lung distance [18], and mean lung dose [19, 20], is still discussed.

Sequential and concomitant radiochemotherapy can further increase the risk for pulmonary complications depending on the composition and temporal placing of systemic therapies [21]. Early reports show that classical chemotherapy alone can induce pneumonitis without contributing ionizing irradiation [22].

Among commonly used chemotherapeutics in breast cancer treatment, especially taxanes are controversially discussed as inductors of pneumonitis independently of subsequent radiotherapy [23‐27]. Another example is given by a randomized study on fluorouracil, epirubicin, cyclophosphamide (FEC) and FEC plus high-dose platinum-based chemotherapy, which demonstrated an increased risk for changes in pulmonary function [28].

Recall phenomena in the target area of previously performed radiotherapy are exemplarily mentioned for intrapleural administration of doxorubicin [29].

At present, therapy of early breast cancer is increasingly individualised. As a consequence, multimodal systemic neo-adjuvant therapy contains combinations of classical and novel drugs. Such regimens applied together with locoregional radiotherapy (RT) have to be investigated for their influence on the occurrence of pneumonitis. Especially novel adjuvants including targeted agents might enhance promotion of pneumonitis after sequential radiotherapy. Antibody therapy is normally well tolerated in sequential and concomitant regimens [30, 31]. By contrast, antibody-based checkpoint (PD[program death]-1, PDL[program death ligand]-1) inhibitors are known to induce immune-mediated adverse events, such as pneumonitis [32‐35]. Hormonal therapy in an adjuvant setting can be used sequentially as well as concurrently with RT. Among these substances, the widely used tamoxifen combined with chemoradiotherapy has been shown to act as a possible enhancer of radiation-induced pneumonitis [36], lung fibrosis [37], or occasionally even of lethal pneumonia [38].

For efficient organisation of follow-up visits and supportive care management it is of crucial interest to delimit as best possible the timeframe for potential pneumonitis onset (approximately 6–16 weeks) [10, 39]. Once diagnosed, it is essential to immediately treat pneumonitis at onset in order to prevent irreversible lung damage. Prompt corticosteroid administration has been successfully established as standard medication of radiation-induced pneumonitis. Its efficacy highly depends on in-time detection of latent early and first clinical symptoms. It is reported [1] that functional impairment as a consequence of radio-induced pneumonitis might be clinically reversible within 21 weeks post-irradiatio. Tokatli et al. [39] described a subsequent partial recovery detected by repeated Tc99m-DTPA and high-resolution computed tomography (CT) imaging within 52 weeks. However, Otani et al. [40] demonstrated that steroid treatment might increase predisposition and recurrence of radiation-induced organising pneumonia. Thus, reliable predictive factors are needed that might facilitate detection of higher-risk patients to be channelled into a surveillance program consisting of proposed DLCO [diffusing capacity or transfer factor of the lung for carbon monoxide; 41] monitoring or other pulmonary function analyses [42].

Another chance to finally overcome normal tissue complications resulting from partial lung-tissue irradiation is given by the ongoing research on substances acting as specific radioprotectors. Soy isoflavones (phytoestradiols, like the ERβ-antagonist genistein) have already been introduced in clinics because of their double action as radiosensitizers in tumour cells, and as radioprotectors in normal lung cells [43]. Myrtol standardised [44, 45] and cerium oxide nanoparticles have been shown to act as potent protectors against high-dose radiation-induced pneumonitis [46], as strong inhibitors of inflammatory cytokines [47] and of ROS, ameliorating early pneumonitis and protecting healthy tissue [48].

Recent findings suggest that peroxisome proliferator-activated receptor (PPARγ) agonists (thiazolidinediones, TZD) attenuate pulmonary injury induced by single-dose lung irradiation in mice [49, 50]. These PPARγ agonists could be selectively used to control the inflammatory response even in special anatomical districts with potential therapeutic applications, such as inhibition of normal tissue co-irradiation-induced pneumonitis and radiotriggered lung fibrosis [51].

Among a variety of other substances, statins are also discussed as potential lung tissue protectors. A significant reduction of inflammation responses in lung tissue has been demonstrated using statins in experimental murine models [52]. Shyamsundar et al. also reported a 35% reduction in inflammatory reactions observed in healthy volunteers stimulated by lipopolysaccaride (LPS) inhalation [53]. Little is known about the molecular mechanisms induced by statins to act as radioprotective agents. A radioprotective effect of simvastatin in total-body irradiated mice is possibly related to the inhibition of apoptosis and an improvement in oxygen-carrying and antioxidant activities [52]. At least for radiation treatment of murine salivary glands, Zhao et al. [54] reported remitting collagen deposition, as well as a reduction and a delay in the elevation of TGFβ1 expression as a consequence of statin action.

In the present study, pertuzumab [55] and the meanwhile phased-out vaccine tecemotide [56, 57] were candidates for scarcely investigated potential enhancers of posttherapeutic lung tissue complications in sequential chemoradiation. A small cohort of patients received premedication based on statins to treat hypercholesterolemia, potentially acting as pneumonitis protectors. In addition to the influence of chemoirradiation, patient smoking history and former lung diseases were investigated as possible co-modulators of pneumonitis incidences.

Case report forms were completed at each of the three visits per patient. The initial planning CT before radiotherapy served as a reference in comparison to the CT scan follow-up performed after 12 and 25 weeks. Potential radiologic abnormalities in the lung were classified into two categories: mild—followed by translucent turbidity, and severe—assessed by increased diffuse interstitial structure, as well as spotted or confluent infiltrates. Symptomatic pneumonitis was diagnosed from the appearance of dry cough, dyspnoea, breathlessness, fever and/or thoracic pain manifested within two weeks before visits. These symptoms were subclassified into various degrees of severity.

Diagnosis of pneumonitis was evidenced by clinical symptoms correlated with thoracic CT scans. Treatment was immediately started by administering high-dose dexamethasone followed by continuous dose reduction based on a taper-schedule lasting 6 weeks.

CT-based 3D treatment planning (Pinnacle V.14; Philips Medicals, Fitchburg, WI, USA) was performed for each patient following field delineation according to the consensus definition of radiation therapy oncology group (RTOG; Breast Cancer Atlas for Radiation Therapy Planning, 2012). Treatment was planned while respecting institutional dose constraints to avoid lung toxicity. For this purpose, cumulative dose–volume histograms (DVH) were calculated restricting the predominantly involved ipsilateral lung volume during 2 field radiotherapy (2FRT) breast and chest wall irradiation to a V_20Gy of ≤30% and a V_30Gy of ≤20% of lung volume. In the case of 3FRT and 4FRT, the cumulative dose–volume histograms of co-irradiated lung tissue were restricted to the same constraints as set for 2FRT. All performed treatment plans applied met the mentioned dose–volume constraints. Comparative analysis of DVHs was normalized by using involved total lung volume (TLV) for all three techniques. In addition, mean lung dose (MLD) was recorded.

For fractionated irradiation, all patients were placed in an identical fixed supine position during the CT planning sessions and treatment (Mamma-Step®, IT-V, Innsbruck, Austria). Patients with pretreated carcinoma in situ were given a total dose of 50 Gy (Synergy, Precise, and Versa HD linear accelerators; iViewGT™ electronic portal imaging; Elekta Instrument AB, Stockholm, Sweden), whereas the total dose for an invasive carcinoma of the breast was set at 56 Gy.

2FRT after breast-conserving surgery: The targeted volume was defined as the complete breast including the chest wall and treated with two tangential photon beams (energy range: 6 MV). A total dose of 56 Gy with 2 Gy/day, at five fractions per week was used. A boost of 4 Gy was optional, and its application depended on the histological findings. Central lung distance (CLD) was set individually to maintain the depth doses below 3.0 cm. Two outliers were tolerated regarding CLD values above the preset limit of 3.0 cm (3.1 cm and 3.5 cm, respectively) due to anatomical necessities (field-inclusion of larger breast-tissue volumes). Single fractions were reduced to 1.8 Gy/day if neoadjuvant chemotherapy including taxanes was administered. In this case, the total dose was 55.8 Gy. The same dose adaptation was indicated in cases of larger breast tissue volumes.

3FRT: The target for LRRT (locoregional radiotherapy) with one involved lymph node was defined as the complete breast and the adjacent chest wall supplemented by the supraclavicular region. The added field was treated using a 6 MV photon beam until reaching a total dose of 50 Gy at 2 Gy/day (five fractions per week). Taxane pretreatment necessitated a single-dose reduction to 1.8 Gy (TD: 50.4 Gy) as described for 2FRT.

4FRT including the IMN (internal mammary nodes): In the case of more than one positively diagnosed lymph node, a parasternal irradiation field was added. In order to limit normal tissue complication probabilities (NTCP) to the adjacent heart, an oblique electron treatment (20 Gy/10 fractions or 19.8 Gy/11 fractions) was combined with photon therapy (30 Gy/15 fractions or 30.6 Gy/17 fractions).

Bilateral parasternal lymphatic irradiation was an institutional adaptation following a consensus agreement of all members of the local tumour board and of the TAKO (Tyrolean Oncology Group) to prevent metastases in cases of large tumour size, medial or central tumour localisation, and of evidenced IMN involvement (≥1 involved axillary LN).

Epirubicin and cyclophosphamide (EC), EC and 5‑fluorouracil (FEC), docetaxel, paclitaxel, pegylated doxorubicin hydrochloride, and combinations of these substances were used for neoadjuvant chemotherapy. Trastuzumab and pertuzumab were used for neoadjuvant and adjuvant antibody therapy. In addition, tamoxifen, letrozole, anastrozole, exemestane, goserelin, and combinations of these therapeutics were administered during neoadjuvant and adjuvant regimens, respectively. Indication and dosage setup of all administered therapeutics was performed according to generally approved recommendations derived from the actualised versions of the S3 guidelines and the ABCSG.

SPSS software (version 21, IBM Corp., Armonk, NY, USA) was used for statistical analysis. For all variables of interest descriptive analysis was performed: absolute or relative frequencies were reported for categorical parameters. Such numerical parameters as age and BMI were reported as mean ± standard deviation (SD), median and range. Relationships between occurrence of pneumonitis and various categorical parameters were analysed using Fisher’s exact test. Differences in age or BMI between patients with or without pneumonitis were examined by means of T tests if data were normally distributed or Mann–Whitney U test if data were not normally distributed. Normality of numerical data was assessed with the help of Kolmogorov–Smirnov test. Spearman’s correlation coefficients were computed to examine the strength of associations between variables of interest.

Central lung distance (CLD) within the irradiated field has traditionally been used to define the risk for radiation pneumonitis. Since the early 1990s it has repeatedly been reported that lung depth measured at the central axis is the best predictor of the proportion of ipsilateral lung volumes included in the tangential fields of breast cancer radiotherapy. However, these findings are still discussed controversially, especially if nodal involvement requires 3FRT or 4FRT [58]. The results obtained in this prospective study including 100 breast cancer patients confirmed the absence of a correlation between increasing CLD and the occurrence of radiation pneumonitis showing its limited value whenever treatment becomes more complex [59]. Thus, other predictors were identified that better estimate risk for radiation pneumonitis. Hence, evaluation of directly assessed lung DVH is known to more reliably estimate normal tissue complication probabilities after adjuvant breast radiotherapy. As suggested by others, three-dimensional CT planning allows improved evaluation of target volume coverage and also of lung volumes co-irradiated during breast cancer treatment [60, 61]. Our findings support the usefulness of such suggestions, since total involved lung volumes derived from DVH data clearly increased whenever 3FRT or 4FRT was applied to cover breast-adjacent nodal regions. A strong relationship between detected RP incidences and more fields used can be confirmed. Therefore, although CLDs of <3 cm were maintained in 98% of irradiated breast cancer patients, significantly more RP cases were detected in patients receiving 4FRT (41.2%, N = 17) than in those receiving 2FRT (6.7%, N = 75).

As a consequence, mean lung dose appeared as the most efficient predictor of RT-related pneumonitis incidences in a nonlinear dose-effect relationship. Using standard 3D-RT, mean lung dose >10 Gy increased the risk of developing symptomatic RP from 20% (MLD 7.5–10 Gy) up to 67%, and of asymptomatic RP from 50–100%. Thus, MLD provides a valuable and cumulative predictor for the incidence of symptomatic and asymptomatic RP in our cohort of patients by suggesting a “threshold” MLD of ≤10 Gy, which still accounts for a residual risk of approximately 20% to develop moderate symptomatic RP (corresponding to <50% risk of asymptomatic RP). As a consequence of this finding, treatment planning at our institution has been adapted to best possible comply with this constraint.

Compared to prior studies, the recorded incidence of moderate pulmonary complications in our cohort of patients is within the reported range by others for 3D planned breast cancer RT with or without IMN inclusion [12‐16, 42]. Irrespective of interstudy comparability and IMN cotreatment, if referred to a widely accepted V_20Gy of less than 30% total lung volume to exclude Grade ≥ 3 severe pulmonary toxicity (RTOG score), and a V_20Gy of less than 22% TLV involvement to circumvent moderate Grade ≥ 2 complications, according to our data also a further reduction of V_20Gy to less than 15% TLV could restrict the risk of developing such moderate complications only to less than 15% (see Fig. 3). Thus, it is not possible to concretise a certain threshold dose delivered to a given TLV percentage, since V_20Gy of seven out of 13 patients developing moderate pneumonitis during this study ranged from 3.2 to 14.1% TLV involvement (V_20Gy ≤ 15%; N = 82).

As observed by others [62], adverse effects of chemotherapy and neo-adjuvant treatment schedules including monoclonal antibodies, anti-hormones or aromatase inhibitors could not be evidenced or attributed to the systemic treatment as a significant co-inductor of pneumonitis. This might be due to the large variation within the applied systemic schedules and, therefore, to the small number of cases characterised by identical systemic regimens. However, our findings clearly demonstrate a predominant role of treatment-involved lung volume and dose distribution in co-irradiated lung tissue. Only for three patients, who all developed RP after 4FRT and combined systemic pretreatment including chemotherapy, antibody and anti-hormone therapy, could a significant correlation be assessed between increasing complexity of systemic treatment and the risk for radiation pneumonitis. However, these data have to be validated by investigating larger cohorts of analogously treated patients. Therefore, also the poorly investigated potential enhancers of posttherapeutic lung tissue complications in sequential chemoradiation pertuzumab [55], and the meanwhile phased-out vaccine tecemotide [56], did not deliver valid information about their real riskiness in more complex sequential regimens.

In addition to the influence of chemoirradiation, smoking history and patients’ former lung diseases were investigated as possible co-modulators of pneumonitis. Even though our findings regarding the protective effect of smoking habits were not statistically significant as compared to non-smokers, patients with a history of smoking tend to develop symptomatic pneumonitis to a lesser extent than non-smokers, as observed by others [63]. Possible mechanisms might be related to depressed inflammatory response, moderate hypoxia, or altered lung sensitivity of smokers versus non-smokers. However, the repeatedly reported effect is not yet investigated to clearly address its underlying biomolecular mechanism. In addition, reports of clear clinical symptoms defining moderate RP might less likely be attributed exclusively to this side effect, since patients with smoking history frequently are habituated to dry cough and dyspnoea. Thus, CT scans should routinely be performed to prevent overlooking RP in smokers.

We also show that former lung diseases—such as a past pneumonia—significantly increase the risk for developing radiation-induced pneumonitis. However, other lung infections or chronic diseases seem not to play a critical role as pneumonitis enhancers following radiotherapy of breast cancer.

Interestingly, a small cohort of 11 patients, who received statin premedication to treat hypercholesterolemia, did not incur pneumonitis. The potential to act as pneumonitis protectors, so far demonstrated only in experimental murine models [54] and healthy volunteers stimulated by LPS inhalation [53], is supported by our preliminary findings. These studies address the (radio)protective mechanism mainly to immune-modulating and anti-inflammatory effects of statins in lung tissue via reduction of expressed myeloperoxidase, tumour necrosis factor-α, matrix metalloproteinases 7, 8, and 9, NF-κB in monocyte-derived macrophages, and plasma C-reactive protein. Similar results were observed in vitro with pravastatin decreasing pro-inflammatory cytokines including IL-6 and IL-8 in irradiated lung endothelial cells [64]. Of course, further investigations in larger cohorts of patients cotreated with statins are needed to prove the clinical value of these molecules also in the prevention of radiation-induced pneumonitis.

DVH-derived 3D data are reliable in estimating lung tissue complication probabilities for locoregional radiotherapy of breast cancer. A MLD of ≥10 Gy was the best predictive factor to determine an escalating risk of developing RP in our cohort of patients, thereby implicating the necessity to comply with this constraint. Especially if nodal involvement requires the addition of supraclavicular (3FRT) and parasternal fields (4FRT) the incurrence of pneumonitis following locoregional radiotherapy is significantly enhanced. Anamnestic pneumonia might represent the major non-treatment-related pneumonitis-promoting factor. A protective effect of tobacco smoke was detectable, however, not in a significant extent. The use of novel and more complex systemic regimens seems to be of minor importance in promoting pneumonitis after radiotherapy. Finally, we identified statins—used by a smaller cohort of our patients to treat hypercholesterolemia—as possible protectors against radiation pneumonitis. In order to also validate the effectiveness of statins in lowering the risk for radiation pneumonitis for breast cancer patients, we currently continue the presented prospective study with special attention dedicated to this preliminary finding—as well as to a primarily MLD based risk assessment in treatment planning.