Acute and Late Toxicities of Thoracic Radiotherapy: Pulmonary, Esophagus, and Heart

Pathophysiology

Radiation pneumonitis (RP) can be divided into three stages—latent, acute, and late. Following radiation there is a latent phase in which there is no apparent symptomatic or radiologic changes. However, microscopically there is degranulation and loss of type II pneumocytes, loss of surfactant, swelling of the basement membrane, and protein transudation into the alveolar spaces. Due to an influx of macrophages and fibroblasts, there is a release of cytokines such as transforming growth factor-beta (TGF-β), interleukin (IL)-2, fibronectin and growth factors such as IGF-1, platelet-derived growth factor, and tumor necrosis factor-alpha (TNF-α). During the acute phase there can be imaging changes and clinical symptoms. The chest x-ray or computerized tomography (CT) scan may show the typical changes of a diffuse infiltrative process corresponding to the radiation field ( Fig. 42.1 ). The symptoms are classically a nonproductive cough, fatigue, shortness of breath, and/or fever. This acute phase occurs within 6–7 months of the delivery of radiation, with a peak incidence of 2–3 months. Microscopically there is a continued inflammatory response with capillary obstruction and an increase in leukocytes, plasma cells, macrophages, fibroblasts, and collagen fibers. The alveolar septa appear thickened and the alveolar space smaller. In the late phase of RP, the imaging will show dense consolidation and volume loss of lung tissue. Some of the acute symptoms such as fever, cough, and fatigue often resolve although it would not be uncommon to have chronic shortness of breath. Pathologically, there is fibrosis of the endothelium and an increase in the thickness of the alveolar septa with obliteration of many alveolar spaces.

The trachea and bronchi are lined with pseudostratified ciliated columnar epithelial cells and mucus-producing goblet cells. A mild to moderate dry cough is common during the acute phase of lung radiation due to depletion of the mucosa. This cough usually resolves shortly after 60–66 Gy radiation therapy, and severe late complications are relatively unusual. In 88 stage III nonsmall cell lung cancer (NSCLC) patients treated to 66 Gy or higher, Lee et al. observed bronchial stenosis in three patients between 2–7 months, representing 11% of all late complications. Miller et al. reported an actuarial rate of treatment-related bronchial stenosis at 1 and 4 years of 7% and 38%, respectively. Radiation dose was suspected to be a factor, with bronchial stenosis observed in 4% and 25% of patients treated to 74 Gy and 86 Gy, respectively. The correlation of mainstem bronchial stenosis with doses of 73 Gy or more was confirmed in a further study by Kelsey et al. This narrowing of the bronchi occurred as early as 3 months following radiation. The caliber of the trachea on the other hand was unchanged following even high doses of radiation.

Grading of Lung Toxicity

Radiation-induced pulmonary toxicity after radical radiotherapy or chemoradiation is complex to score in lung cancer patients because it is difficult to distinguish from tumor progression and exacerbation of preexisting pulmonary comorbidities. The Common Terminology Criteria for Adverse Events (CTCAE) grading system, in common use worldwide for scoring of toxicity, has been modified over the years ( http://www.eortc.be/services/doc/ctc/ ).

The Radiation Therapy Oncology Group (RTOG) had its own grading system initially but has adopted the CTCAE over the last 10 years. An important difference between the RTOG and CTCAE grading system is that the RTOG differentiates between acute (pneumonitis) and late (fibrosis) toxicity based upon the time interval (90 days) after treatment. The choice of 90 days is somewhat arbitrary as the active inflammatory phase of pneumonitis can persist beyond 90 days. Tables 42.1 and 42.2 summarize the RTOG grading systems as well as the CTCAE version for pneumonitis and pulmonary fibrosis. In all these grading systems grade 5 toxicity is death ( http://ctep.cancer.gov/protocolDevelopment/electronic_applications/ctc.htm#ctc_40 ).

Changes in Pulmonary Function Tests

The RTOG grading system was heavily dependent on the prescription of drugs or oxygen, which varies greatly from physician to physician. The CTCAE grading systems did away with that in the later versions, but still require subjective assessment of symptom severity, its effect on activities of daily living, and the indication for oxygen. More objective criteria to measure pulmonary toxicity have been sought such as changes in pulmonary function tests (PFTs).

Miller et al. summarized the changes in PFTs in 13 patients who had been treated definitively with radiation therapy and were thought to have no tumor recurrence for 2 years or longer. These patients had been followed prospectively with PFTs approximately every 6 months. PFTs included spirometry with forced expiratory volume in 1 second (FEV ₁ ) and forced vital capacity (FVC), diffusing capacity for carbon monoxide (DLCO), and lung volumes. There was a decline in the 6-month median FEV ₁ , FVC, and DLCO that reverted to baseline by 1 year. This improvement at 1 year following radiation therapy was attributed to tumor response. However, there was then a yearly reduction in the median FEV ₁ , FVC, and the DLCO of 7%, 9.5%, and 3.5%, respectively. The changes in FEV1 and FVC were significant. Ten of these 13 patients developed some new respiratory symptoms that developed at 6 weeks to 21 months following treatment, with a median of 6 months. The authors attributed this progressive decline in PFTs to progressive/evolving radiation-induced lung injury, possibly due to the continuous cycle of tissue hypoxia resulting in an influx of proinflammatory cells that release cytokines that perpetuate the process.

Observations such as this of progressive decline in lung function have resulted in a reluctance of some clinicians to prescribe high-dose radiotherapy to patients with poor pulmonary reserve. However, there is evidence that the extent of decline is linked to pretreatment function such that patients with poor PFTs pretreatment are at less risk of symptomatic lung toxicity or reduction in FVC or FEV ₁ compared with patients with good spirometry. These somewhat paradoxic observations suggest that poor PFTs should not by themselves be a contraindication to high-dose radiotherapy.

One prospective study of 185 patients given PFTs before and after thoracic RT found that the median maximal percentage reductions in FEV ₁ , uncorrected DLCO (not corrected for hemoglobin levels), and DLCO were 11.5%, 14.9%, and 15.3%, respectively. Reductions in corrected and uncorrected DLCO were larger than that observed for FEV1. Reductions in uncorrected DLCO were associated with the mean lung dose (MLD) and percentage of perfused lung (measured by SPECT) receiving 30 Gy or more (V30). However, the changes in the SPECT scans were relatively smaller than the observed changes in PFTs in many patients. In a later analysis in an expanded group of patients, the correlation between the predicted drop in DLCO and subsequent pneumonitis was not considered high enough to be very clinically useful.

Others have found a significant correlation between preradiation and postradiation therapy DLCO and pulmonary toxicity. In one large institutional retrospective study 85% of patients with NSCLC had a postradiation reduction of DLCO. The mean reduction in pretreatment and posttreatment DLCO was 20%. The proportional drop in DLCO differentiated between patients who developed RP grade 1 versus grade II (common toxicity criteria [CTC] v3). This reduction in DLCO correlated with RP grade ≤1 versus ≥2 in patients who were age 65 years or more, had advanced stage (III–IV vs. I–II), were smokers, received chemotherapy, had V20 Gy ≥ 30% (the percentage of normal lung receiving 20 Gy or more), and had baseline DLCO or FEV ₁ equal to or more than 60% of predicted. Patients who had higher proportional reductions in DLCO had significant higher rates of severe RP. However, the reductions in DLCO still varied widely between the grades of RP, making it less useful to use in routine clinical practice.

Dosimetric Factors

Estimation of the probability of developing RP after treatment with high-dose RT is important for patients with inoperable lung cancer. An important question is could the lung toxicity be predicted from the delivered dose and historical toxicity models? The total dose for an organ-at-risk after conventional fractionated radiotherapy is calculated as the biologic effective dose or normalized total dose in 2 Gy per fraction (NTD2Gy) according to the linear–quadratic model. Studies of the risk of RP relative to the radiation dose received to a certain volume of “normal” lung have used various theoretic models. These studies were only possible with the adoption of CT planning. In two of these models the dose–volume histogram is first reduced to a single parameter, which is subsequently related to the incidence of radiation pneumonitis. Single parameters are the MLD and the volume of the lung receiving more than a certain threshold dose (Vx). The MLD is defined as the average dose throughout the lungs (minus the gross tumor volume). One of the seminal articles by Graham et al. that studied the risk for 99 lung cancer patients with the incidence of RP found that it was correlated with the percent of normal lung receiving 20 Gy or more (V20). In this analysis Graham subtracted the planning target volume (PTV) from the total bilateral lung volume. The PTV is the volume that contains the gross disease, areas thought to represent microscopic extension with an additional margin to account for daily set-up variation. This method of calculating the V20 was continued in several RTOG radiation dose escalation studies. However, RTOG 0617, which is a phase III study comparing 60 Gy with 74 Gy with concurrent carboplatin/paclitaxel, calculated normal irradiated lung as the total lung minus the clinical target volume, i.e., the volume thought to contain only gross tumor and microscopic extension. Wang et al. demonstrated that the differences in the calculated mean lung dose or other dosimetric parameters varied greatly depending on which formula was used to define “normal lung.” The choice of which of the various formulas to use can lead to significant variation in the prediction of subsequent pulmonary toxicity. The most conservative approach is to subtract only the gross tumor volume (or the internal target volume if using 4-D CT simulation technique). Multiple authors have recommended this approach since the choice of the clinical target volume or PTV margin is more variable from physician to physician.

Marks et al. reviewed over 70 articles that correlated dose–volume parameters with subsequent RP following conventionally fractionated radiation therapy of NSCLC. This review, part of the Quantitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC) initiative, cautioned that there was no MLD below which there was no risk of RP. There is also the issue of overestimating RP in these patients due to exacerbation of preexisting pulmonary comorbidities, or tumor progression producing similar complaints. The correlation of MLD and the risk of RP was not linear but had a mildly exponential increase that was more pronounced with increasing MLD. The risk of symptomatic RP with MLD of 20 Gy and 30 Gy was approximately 20% and 40%, respectively. A significant variation in the risk of RP with increasing normal lung volumes, i.e., V20, V30, etc., was reported within the 14 studies summarized by QUANTEC. For example, for V20 of 30% the reported symptomatic RP varied from less than 10% to approximately 50%. There was less variation among the 10 studies that correlated MLD and subsequent symptomatic RP.

In addition to looking at the radiation dose to the total normal lung, there may also be an association between the ipsilateral normal lung dose and the risk of RP. Ramella et al. found that if the V20, V30, and MLD of the total lung volume did not exceed 31%, 18%, and 20 Gy, respectively, there was additional predictive value to the ipsilateral (affected) lung constraints. For example, if the ipsilateral V20 was 52% or less, the risk of RP was 9%, whereas if it was greater than 52%, the risk of RP was 46%. These dosimetric values were calculated by subtracting the PTV from the ipsilateral affected lung.

The use of intensity modulated radiation therapy (IMRT) has been reported to be associated with less pneumonitis compared with 3-D conformal radiotherapy, but it is difficult to understand why if normal lung parameters such as MLD or V20 are similar. IMRT plans are characterized by a more conformal dose distribution and steeper dose fall off and result in a decrease of the V20 and MLD, although the volume of lung receiving a low dose (V5) will be higher. It has been recommended that contralateral lung V5 be kept below 60% to reduce the risk of potentially fatal pneumonitis.

RTOG 0617, a phase III study comparing 60 Gy versus 74 Gy for locally advanced NSCLC, stratified patients according to 3-D CT or IMRT treatment planning. Preliminary results of this study indicate that quality of life was superior in patients treated with IMRT. There is unlikely to be a phase III study directly comparing 3-D CT and IMRT treatment planning. A population-based study of comparative effectiveness found similar rates of early and late pulmonary toxicity for both techniques.

Fewer studies have been done to predict the risk of RP following treatment of limited SCLC. However, there is likely a similar correlation between the dosimetric factors and subsequent RP. For example, Tsujino et al. looked at the risk of RP in patients with limited small cell lung cancer (SCLC) who were treated with cisplatin-based chemotherapy and concurrent thoracic radiation of 45 Gy given two fractions daily, over 15 treatment days. The 12-month cumulative incidence of symptomatic RP was 0%, 7.1%, 25%, and 42.9% in patients with a V20 of <20%, 21% to 25%, 26% to 30%, and >31%, respectively.

Clinical Factors

The impact of nondosimetric factors in predicting the risk of RP has been an area of intense study. Appelt et al. analyzed the patient data used to make the QUANTEC recommendations to look for clinical risk factors. Factors found to significantly increase the risk of RP were preexisting pulmonary comorbidity, mid or inferior tumor location, current smoking, age older than 63 years, and sequential (as opposed to concurrent) chemotherapy.

Other studies have found that sequential or concurrent chemotherapy involving cisplatin/etoposide does not appear to increase the risk of RP. RTOG 8808/ECOG 4588 was a three-arm phase III study comparing radiation only, 60 Gy in 2 Gy daily fractions (considered “standard”), or 69.6 Gy in 1.2 Gy fractions twice daily, or sequential cisplatin/vinblastine chemotherapy and standard radiation therapy. The 18-month cumulative incidence of toxicity, primarily pulmonary, was almost 30% in the combined modality arm, as compared with 20% to 25% incidence in the radiation therapy–only arms. The statistical significance between the pulmonary toxicity in the various arms is not stated although the authors said the toxicity was acceptable. Most of the toxicity occurred within the first 3 to 6 months in all treatment arms, with a relative flattening of the incidence after 9 months.

Given the difference from study to study in the nondosimetric variables predicting the risk of pneumonitis, meta-analyses have been done. Vogelius et al. reviewed the English language lung cancer literature between 1990–2010 in which RP was correlated with some patient or treatment variables. In the 31 studies analyzed, the statistically significant risk factors for increased RP were older age, mid-lower lung disease location, and presence of pulmonary comorbidity. The use of sequential rather than concurrent chemotherapy was found to increase the risk of RP but the authors suspected this was likely due to patient selection rather than a “real” predictive factor. Interestingly, smoking at the time of treatment was found to protect against RP. Even a prior history of smoking was associated with a decreased risk of RP, although this did not reach statistical significance.

Patients with collagen vascular diseases (CVD) are likely at increased risk of RP. In an excellent review by Lee et al. it appears that cytokines such as TGF-β are chronically elevated in such patients. There is often preexisting lung fibrosis. Radiation therapy may cause a patient with quiescent CVD to develop active CVD. These patients are also noted to have increased late toxicity as compared with patients without CVD. Lung fibrosis may extend outside of the high-dose region, so treatment volumes should be reduced when possible, although there is no reason to omit radiation therapy if it is indicated. Interestingly, tumors developing in patients with CVD may be more radiosensitive, prompting consideration of a radiation dose reduction of at least 10%.

Combined Dosimetric and Clinical Factors

Dosimetric and nondosimetric factors can be combined to increase the ability to predict RP. Bradley et al. proposed a nomogram to predict RP based on mean lung dose and the tumor location, superior to inferior ( Fig. 42.2 ).

Taking it one step further, Palma et al. performed a meta-analysis of both dosimetric and nondosimetric factors based on individual patient data. The authors did a search of articles published between 1993 and 2010, in which clinical and dosimetric factors of patients treated with concurrent chemoradiation therapy were available to be correlated with RP. These data were then utilized to conduct a recursive partitioning analysis (RPA) from a 557-patient dataset, and then the RPA was validated using another 279-patient dataset. Factors predicting greater risk of pneumonitis were carboplatin/paclitaxel chemotherapy (vs. cisplatin/etoposide), age greater than 65 years, V20, and MLD ( Fig. 42.3 ).

Biomarkers

Inflammatory cytokines are made by many cells within the lung, including the alveolar macrophages, type II pneumocytes, T lymphocytes, and lung fibroblasts. The blood levels of these cytokines pre-RT, during, and post-RT have been an area of intense interest. For example, Chen et al. found that circulating IL-6 levels pre-RT, during, and post-RT have been correlated with an increased risk of RP. In that study, TNF, another inflammatory cytokine, was not associated with an increased risk of RP.

TGF-β1 is a cytokine that has been extensively studied as a marker predictive of RP. It was noted that patients with NSCLC have increased pretreatment levels of TGF-β1 and that increased levels were associated with a higher MLD and a higher incidence of RP. Several studies showed that the absolute levels of TGF-β1 were not as predictive of RP as the observation of an increasing ratio of pretreatment to intratreatment TGF-β1 for stage III NSCLC treated with definitive radiation, with or without chemotherapy. The predictive value of the TGF-β1 ratio was even higher when combined with the MLD. The combination of a MLD of more than 20 Gy and a TGF-β1 ratio of more than 1 was associated with a 66% incidence of RP. Similarly, IL-8 levels pretreatment and then at weeks 2 and 4 of radiation therapy were associated with RP. Patients with RP grade 2 or higher tended to have higher baseline IL-8 levels and these had a slight downward trend during radiation therapy, as opposed to low and stable IL-8 levels in patients without RP. Combining IL-8, TGF-β1, and MLD into a single model led to an improved ability to predict RP as compared with either variable alone.

Certain polymorphisms of the VEGF gene have been correlated with the incidence and severity of RP.

Recall Radiation Pneumonitis

Recall RP is when symptoms of RP are activated by administration of a drug sometime after completion of radiotherapy, with worse than expected RP. Recall RP has been associated with multiple chemotherapy drugs such as taxanes, gemcitabine, vinca alkaloids, doxorubicin, and epirubicin. Tyrosine kinase inhibitors such as erlotinib and sunitinib have also been associated with radiation recall pneumonitis and increased risk of severe RP following palliative or definitive radiation therapy.

Prevention and Management