Predicting the need for palliative thoracic radiation after first-line chemotherapy for advanced nonsmall cell lung carcinoma


  • Presented at the 5second Annual Meeting of the American Society for Therapeutic Radiation and Oncology (ASTRO); October 30 to November 4, 2010; San Diego, California.



The objective of this secondary analysis was to identify patients with selected stage IIIB/IV nonsmall cell lung carcinoma and good performance status who were at high risk for requiring subsequent palliative thoracic radiotherapy after initial treatment with first-line chemotherapy.


The authors conducted a pooled analysis of patients at a single institution who enrolled onto 10 prospective phase 2 and 3 clinical trials that involved first-line, platinum-based chemotherapy. Baseline lung-related characteristics before trial enrollment were analyzed as possible prognostic factors for freedom from pulmonary events (defined either as subsequent thoracic radiation or as a new collapsed lung, which is an indication for thoracic radiation).


Of 244 consecutive patients who were reviewed, 42 patients received a palliative course of thoracic radiation, 40 exhibited evidence of new lobar collapse on follow-up chest imaging, and 14 received thoracic radiation for lobar collapse. On univariable analysis, pulmonary symptoms (P = .043) or pneumonia at presentation (P = .0001), increasing size of hilar disease (P < .0001), and evidence of obstruction of major bronchi or vessels (P = .0003) were associated with subsequent pulmonary events. On multivariable analysis, hilar disease measuring >3 cm (hazard ratio, 1.8; P = .003) and prechemotherapy pneumonia (hazard ratio, 2.1; P = .009) were associated with pulmonary events; patients who had both risk factors or hilar disease >5 cm in greatest dimension exhibited a >50% risk of subsequent events.


Patients with bulky hilar disease and a history of pneumonia at presentation were at high risk for requiring palliative thoracic radiation. The authors propose studying these patients to determine whether early thoracic radiation may be beneficial by preserving quality of life and performance status. Cancer 2011. © 2011 American Cancer Society.


For patients with selected stage IIIB or IV nonsmall cell lung carcinoma (NSCLC), palliative chemotherapy as initial therapy has demonstrated improvements in overall survival and quality of life in multiple randomized clinical trials compared with supportive care alone.1, 2 For patients with a good Eastern Cooperative Oncology Group performance status (ECOG PS) (0 or 1), the standard of care for most patients consists of a platinum doublet regimen for 4 to 6 cycles followed by either maintenance or immediate second-line chemotherapy3-6 or delayed second-line chemotherapy at the time of progression.7-10

Palliative thoracic radiotherapy (RT) usually is reserved for patients who develop significant pulmonary symptoms, such as dyspnea, hemoptysis, cough, or chest pain, or for patients who exhibit radiographic signs of bronchial obstruction and parenchymal lobar collapse.11-15 A course of thoracic RT can be delivered at diagnosis or at the time of symptomatic or radiographic progression.14 Although thoracic RT is efficacious in alleviating symptoms,11-14 to our knowledge, no modern randomized trial has been conducted comparing thoracic RT versus supportive care alone for patients with advanced stage IIIB/IV disease.16 A meta-analysis of a heterogeneous group of randomized trials revealed an improvement in overall survival for high doses versus low doses of thoracic RT in patients with good performance status.15

The objectives of the current study were to examine the clinical and radiographic factors available at diagnosis for patients with advanced stage IIIB/IV NSCLC to identify predictive factors for subsequent pulmonary events (defined as receipt of palliative thoracic RT or development of lobar collapse on follow-up imaging, which is an indication for palliative thoracic RT). The ultimate goal is to define a subgroup of patients in whom early thoracic RT may be of benefit before severe pulmonary symptoms or collapsed lung parenchyma arise. We performed a pooled analysis of patients who received treatment on 11 prospective trials at a single institution involving first-line, platinum-based chemotherapy.


Patient Selection

This study was an institutional review board-approved, pooled analysis of previously conducted clinical trials in which patients were enrolled and treated at the University of North Carolina Lineberger Comprehensive Cancer Center. All patients who received treatment between the years 1997 and 2010 who met the following criteria were included in this study: 1) new diagnosis of advanced stage IIIB or IV NSCLC; 2) enrollment onto a prospective, phase 2 or 3 trial involving first-line, platinum-based chemotherapy for patients with high performance status (ECOG PS 0 or 1); 3) no prior thoracic surgery or RT; and 4) available prechemotherapy chest computed tomography (CT). For a list of the 10 trials that were included in this study, see Table 1. Of 318 patients identified, 2 were excluded from the analysis because no follow-up information was available. Patients who developed metastatic disease after previous thoracic surgery (27 patients) or radical RT (10 patients) for stage I, III, or III NSCLC were eligible for enrollment onto some of the included trials but were excluded from our analysis, because previous treatment may affect the ability to receive subsequent palliative RT. Sixteen patients received palliative thoracic RT immediately before enrollment onto a trial and, thus, were excluded; and 31 patients did not have available chest CT scans and were excluded. In total, 244 patients met all criteria for the current analysis.

Table 1. Characteristics of Clinical Trials Included in the Pooled Analysis
Trial NameNo. of PatientsaPhaseChemotherapy AgentsStudy PopulationTrial Status (Reference)Study Dates
  • Abbreviations: ±, with or without; CALGB, Cancer and Leukemia Group B; ECOG PS, Eastern Cooperative Oncology Group performance status; LCCC, Lineberger Comprehensive Cancer Center; q3wk, every 3 weeks; Rand, randomized.

  • a

    The number of patients enrolled in the specified trial at the University of North Carolina Lineberger Comprehensive Cancer Center.

  • b

    Only patients receiving first-line chemotherapy were included.

LCCC9719473Carboplatin/paclitaxel: Continuous until progression vs 4 cyclesECOG PS 0-1Published (Souquet 19952)1997-2001
LCCC2003662Carboplatin/taxol: Weekly vs q3wkECOG PS 0-1Published (Fidas 20093)2001-2004
A1-99002L373Gemcitabine/carboplatin or paclitaxel vs paclitaxel/carboplatinECOG PS 0-1Published (Stinchcombe & Socinski 20114)2001-2006
CA225081112Carboplatin/cetuximabECOG PS 0-1Published (Ciuleanu 20095)2005-2006
CALGB3040626Rand, 2Erlotinib±carboplatin/paclitaxelLight or never smokers, ECOG PS 0-1Completed (Cappuzzo 20106)2005-2008
CA2250582Rand, 2Two dose schedules of carboplatin/paclitaxel/cetuximabECOG PS 0-1Published (Hanna 20047)2006
H6Q-US-S00423Rand, 2Carboplatin/pemetrexed±enzastaurin hydrochloride, or carboplatin/docetaxelECOG PS 0-1Published (Shepherd 20008)2006-2008
AVF3752G122Bevacizumab in combination with first-line or second-line therapybPatients with treated brain metastasesPublished (Shepherd 20059)2007-2008
CA03163Carboplatin/ABI-007 vs carboplatin/paclitaxelECOG PS 0-1Completed (Socinski 200210)2008-2009
H3E-MC-JMHD33Carboplatin/pemetrexed/ bevacizumab+maintenance pemetrexed/bevacizumab vs carboplatin/paclitaxel/ bevacizumab+maintenance bevacizumabECOG PS 0-1Open to accrual2009


Table 2. Baseline Clinical and Pathologic Characteristics
 Patients (N = 244)
Clinical and Pathologic Characteristics at EnrollmentNo.%
  • Abbreviations: AJCC, American Joint Committee on Cancer; SVC, supraclavicular; NSCLC, non-small cell lung carcinoma; ECOG, Eastern Cooperative Oncology Group

  • a

    Disease was staged was according to the 7th edition of the AJCC Cancer Staging Manual (2010).

Median age (range), y60 (23-86) 
 IIIB (advanced SCV adenopathy)73
 Unspecified NSCLC4418
 Nonsquamous cell carcinoma16166
 Squamous cell carcinoma3916
ECOG performance status  
 Unknown (0-1 by trial enrollment criteria)11848
Smoking history  
 Light smoker (≤15 pack-years)2912
 Heavy smoker (>15 pack-years)16768
Clinical presentation  
 Weight loss (>10% body mass)4117
 Weight loss (any)9439
 Chest pain5919
 Pneumonia (antibiotics prescribed)7122

Data Collection

Baseline clinical data

Patient demographic information, the date of study enrollment, and the date of either death or last clinical follow-up were obtained through the Lineberger Comprehensive Cancer Center clinical trial databases and publicly available resources. Clinical stage was consistently redefined using the American Joint Committee on Cancer seventh edition guidelines17 and was obtained through a review of history and physical examination records and imaging reports. Symptoms reported on the pretrial history and physical examinations were recorded, including new cough (chronic cough not scored), hemoptysis, chest pain, and dyspnea above baseline (chronic dyspnea not scored). Pneumonia was defined by the prescription of antibiotics for pulmonary-related symptoms (cough, dyspnea, hemoptysis, chest pain) within 60 days before enrollment (See Table 2).

Baseline radiographic data

For all 244 patients, chest CT scans were obtained within 6 weeks of trial enrollment and were reviewed and reported by the University of North Carolina Department of Radiology. Of the 244 scans, 231 (95%) were contrast enhanced. The CT reports were reviewed to collect information regarding the presence or absence of hilar disease (defined as hilar nodes or masses that originate from the parenchyma or mediastinum and extend into the hilum, as described in the reports) and the size of hilar disease as estimated by the maximum dimension in the axial plane. Bronchial/vascular compression was defined as the narrowing, obstruction, or obliteration of the trachea or mainstem, lobar, or segmental bronchi or major vessels. The presence or absence of any atelectasis, atelectasis of 1 or more lobe, and pleural effusion also was noted (See Table 3).


A pulmonary event was defined in this study as the use of palliative thoracic RT or radiographic evidence of a new collapsed lobe on follow-up chest imaging. Any course of palliative radiation targeted toward disease in the lung parenchyma, hila, or mediastinum that was delivered after the date of enrollment onto trial was scored as a pulmonary event.

The reports of all chest CT scans and chest x-rays were reviewed. A collapsed lobe was defined as any new atelectasis after enrollment of more than half of the parenchyma in a lobe as described in the CT reports and confirmed by a second review of the images by the first author (D.S.H.). Atelectasis caused by either central bronchial obstruction or compressive pleural effusion was scored without distinction, because, in a minority of cases (2 of 40), such a distinction was unclear. If a patient manifested atelectasis of a whole lobe before the date of enrollment, then the development of a separate, new collapsed lobe was required to be scored as a pulmonary event.

Statistical Analysis

Freedom from pulmonary events was defined as the interval between the date of trial enrollment and the date of first pulmonary event. If a patient received thoracic RT and developed a collapsed lobe, then only the first event was scored. By using the method of Kaplan and Meier,18 patients were censored from the analysis if they died, were lost to follow-up, or were alive as of September 1, 2010 with no pulmonary event scored. We hypothesized that baseline lung-related factors, such as disease stage (IIIB, IVA, IVB), the presence and size of hilar disease, bronchial/vascular compression, atelectasis, malignant pleural effusions, pulmonary symptoms, and pneumonia, may be associated with pulmonary events.

Univariable analyses of freedom from pulmonary events were conducted using a log-rank test for categorical variables and Cox proportional hazards regression for continuous variables. All factors that demonstrated a significance level of P < .1 in univariable analyses were included in multivariable analysis using Cox proportional hazards regression. All statistical calculations were conducted with STATA version 11 (StataCorp LP, College Station, Tex).


Patient Characteristics

The median age of the study population was 60 years (range, 23-86 years), and 242 patients (99%) had an ECOG PS 0 or 1 (see Table 2). In total, 231 patients (95%) received doublet or triplet, carboplatin-based chemotherapy; 36 patients (15%) received bevacizumab in addition to carboplatin; and 13 patients (5%) who were light smokers or never smokers with adenocarcinoma received chemotherapy with erlotinib only (see Table 1). At the time of this analysis, 220 patients (90%) had died, 21 patients (9%) were alive with a median follow-up of 12.1 months (range, 4.3-54.3 months), and 3 patients (1%) were lost to follow-up after a median of 20.1 months (range, 0.3-40.0 months). The median survival for all patients was 7.7 months after the date of trial enrollment. In total, 234 patients (96%) had subsequent follow-up chest imaging studies available after trial enrollment, with a median of 3 CT scans (range, 0-31 CT scans).

Pulmonary Events

In total, 68 patients (28%) developed a pulmonary event: Forty-two received subsequent thoracic RT, and 40 developed lobar collapse (14 of whom also received thoracic RT). Patients received palliative thoracic RT a median of 5.6 months (range, 0.2-34.6 months) after enrolling onto a first-line chemotherapy trial. Approximately 80% of palliative thoracic RT courses were targeted to obstructing central mediastinal or hilar disease. Thirteen patients (31%) did not finish their initially prescribed course of RT because of declining overall functional status and moved toward supportive care only. The reasons for decline were progressive pulmonary distress (11 patients), progressive fatigue (1 patient), and mental status changes (1 patient). The median survival after a course of palliative thoracic RT was 2.9 months (range, 0.3-19.3 months).

Lobar collapse was identified on follow-up imaging a median of 5.6 months (range, 0.06-40.7 months) from trial enrollment and a median of 24 days before death (range, 0-6.8 months). Two of the 40 collapsed lungs (5%) were caused by compressive atelectasis from pleural effusions, 36 (90%) were caused by central obstruction, and, in 2 patients (5%), the distinction was unclear.

On univariable analysis, the size of hilar disease at baseline; the presence of bronchial/vascular compression, cough, or any pulmonary symptoms; pneumonia, and whole lobe atelectasis were associated significantly with subsequent pulmonary events (Table 4). On multivariable analysis, the size of hilar disease (hazard ratio [HR], 1.30; P < .0001) and pneumonia (HR, 2.30; P = .0004) remained significant (see Table 4 and Figure 1).

Table 3. Baseline Radiographic Characteristics on Chest Computed Tomography
 Patients (N = 244)
Baseline Chest CT Characteristics Before EnrollmentaNo.%
  • Abbreviations: CT, computed tomography.

  • a

    N = 244 available reports.

  • b

    Hilar nodes or parenchymal masses extending to the hilum.

  • c

    Segmental, lobar, or mainstem bronchi, trachea, aorta, pulmonary arteries/veins, lobar arteries/veins.

Hilar diseaseb  
Narrowing of major bronchi or vesselsc  
 Partial lobe5322
 One or more whole lobes177
Malignant effusion  

In separate sensitivity analyses of thoracic RT and lobar collapse, the size of hilar disease and pneumonia remained the only 2 significantly predictive factors on multivariable analyses. For thoracic RT (42 events), both the size of hilar disease (HR, 1.19; P = .018) and pneumonia (HR, 2.24; P = .015) were significant; whereas, for lobar collapse (40 events), the size of hilar disease size (HR, 1.26; P < .0001) was the only significant factor. In a sensitivity analysis that included all consecutive patients (N = 316), even those who had received prior thoracic RT or surgery for earlier stage disease, the size of hilar disease (HR, 1.24; P < .0001) and prechemotherapy pneumonia (HR, 2.07; P = .005) again were the 2 significant factors on multivariable analysis.

Risk Groups

We examined the risk of pulmonary events by patient groups with zero, 1, or 2 risk factors identified in multivariable analysis (the size of hilar disease and prechemotherapy pneumonia). For this analysis, hilar disease size was dichotomized by the median disease size in the patient cohort (3 cm) for clinical convenience. The cumulative incidence of pulmonary events for the 3 groups was 20% (0 factors), 30% (1 factor), and 60% (2 factors); and median survival also differed between the 3 groups (11.3 months, 7.0 months, and 4.7 months, respectively; log-rank P = .0001) (See Figure 2). These low-risk, intermediate-risk, and high-risk groups constitute 53%, 36%, and 11% of the population, respectively.


The standard treatment for most patients with advanced stage IIIB or IV NSCLC and a good performance status is platinum-based chemotherapy.1, 2 Local treatment (either RT or endobronchial debulking) is not given commonly upfront but, instead, is given at the time of pulmonary symptom development or when follow-up chest imaging studies suggest lung collapse.11-14 Several randomized trials support the use of immediate3, 6 or delayed,8, 19 second-line, systemic therapy, leaving the role and timing of thoracic RT less defined. Thus, the objective of our study was to increase understanding of which patients are at the greatest risk for requiring thoracic RT after receiving first-line systemic therapy. In these patients, the role of early thoracic RT could be studied to determine whether there is a quality-of-life benefit for RT delivered before pulmonary collapse and/or severe pulmonary symptoms arise.

We observed that 20% of our cohort developed pulmonary events at a median of 5.6 months after initiating chemotherapy. The median survival after a collapsed lobe was only 24 days; and, of the 40 patients who developed collapsed lobes, only 14 (35%) received thoracic RT or other local therapy to the lung. Of those patients who received thoracic RT, 31% were unable to complete a course of palliative RT because of declining overall functional status. These figures underscore the difficulty in providing a course of thoracic RT at the time of pulmonary distress and/or lung collapse.

The size of hilar disease and pneumonia at baseline were associated significantly with pulmonary events. Clinically, increasing size of central hilar disease can lead to increased risk for future bronchial obstruction, lobar collapse, and pulmonary symptoms. Baseline prechemotherapy pneumonia may signify a predilection for future postobstructive infections. In our cohort, patients with both risk factors had a 60% cumulative incidence of events. We estimate that the actual number of pulmonary events in this population could be even higher, because many patients in our cohort, choosing supportive care measures, appropriately did not receive follow-up chest imaging in the few months before death, and additional patients may have received palliative thoracic RT at other institutions without our knowledge.

Our observation that 31% of patients could not complete a course of palliative thoracic RT is analogous to published findings for systemic therapy. Among the patients who complete first-line chemotherapy, only 60% to 70% receive second-line chemotherapy at the time of progression, in part because of performance status declines.4-6, 20 Because progressive disease and the development of collapsed lung often are associated with a decline in overall functional status, these findings suggest that early thoracic RT may be beneficial, especially in patients at high risk of developing pulmonary events. Reducing the risk of collapsed lung and/or severe pulmonary symptoms through early thoracic RT may have a quality-of-life benefit by delaying a decline in functional status and thereby reducing a barrier to completion of necessary systemic therapy. In addition, lung collapse caused by central obstruction can be a fatal event–the median survival after collapse of 1 or more lobes in our series was only 24 days. Delaying or preventing its occurrence with early thoracic RT may improve survival in selected patients. These outcomes can be studied in a prospective trial.

The median time to pulmonary event of 5.6 months suggests that early thoracic RT could be interdigitated with first-line chemotherapy or could be delivered soon afterward to minimize interference with planned systemic therapy. Only patients at high risk for pulmonary events should be selected for early RT, because 1 randomized trial conducted without systemic therapy indicated that immediate palliative RT was not necessary for all asymptomatic patients.21 In that study, asymptomatic patients were randomized between immediate and delayed thoracic RT when needed to control symptoms. In the delayed group, 42% of patients required thoracic RT at a median of 4 months after trial enrollment, and only 1% of patients received cytotoxic chemotherapy. In our cohort, fewer patients seemed to need RT (28%) at a longer time interval from enrollment (5.6 months), perhaps reflecting the efficacy of first-line systemic therapy. In a secondary analysis of another randomized trial comparing 3 dose and fractionation schedules for immediate thoracic RT, those with symptomatic disease had worse overall survival compared with minimally symptomatic patients (6.0 months vs 11.8 months).22 Quality-of-life scores for both groups (symptomatic and minimally symptomatic) deteriorated over time, whereas transient esophagitis increased in both groups, leading the authors to recommend delayed RT at the time when significant pulmonary symptoms develop.1 Thus, these prior studies suggest that early thoracic RT should be studied prospectively only in a selected group of patients whose baseline clinical and radiographic characteristics suggest a high likelihood of the need for delayed RT after first-line systemic therapy.

The current study has several limitations, including its retrospective design and its analysis of a single institution. The patients were drawn from patients accrued over 13 years; thus, there may be stage migration because of the use of positron emission tomography and improvements in radiation treatment delivery with 3-dimensional CT planning. Our results require validation by others. The strengths of this study include the patient cohort, which was drawn from prospective trials of chemotherapy and, thus, had relatively complete follow-up and repeated chest imaging. Furthermore, to our knowledge, this study is the first to identify a subgroup of patients with advanced NSCLC and good performance status who are at high risk of developing pulmonary events after first-line chemotherapy. The potential benefit of early thoracic RT in this group of patients should be studied prospectively. 1

Figure 1.

The effects of baseline clinical and radiographic pulmonary factors on subsequent pulmonary events after first-line chemotherapy are illustrated. (A) The size of hilar disease and subsequent pulmonary events are shown. (B) Patients who present with pneumonia at diagnosis also exhibit an increased risk of subsequent pulmonary events. (C) The cumulative incidence of subsequent pulmonary events increases as baseline hilar disease increases. (D) This chart illustrates the detection of bronchial/vascular compression on initial staging chest computed tomography and subsequent pulmonary events.


Figure 2.

(A) Risk groups for subsequent pulmonary events are illustrated after enrollment onto first-line chemotherapy trials. The maximum single dimension of hilar disease and a presenting history of pneumonia remained significant in multivariable analysis (see Table 4) constitute the risk factors in this model. For clinical convenience, the presence of hilar disease size greater than the median of 3 cm was chosen as a threshold to score as a risk factor. (B) The high-risk, intermediate-risk, and low-risk groups for pulmonary events also predict overall survival.

Table 4. Possible Prognostic Factors for Pulmonary Events
  Univariable AnalysisMultivariable Analysis
VariableNo. of Patients (%)Cumulative Incidence of Pulmonary Events, %Log-Rank PHR95% CIP
  • Abbreviations: CI, confidence interval; HR, hazard ratio; NSCLC-NOS, nonsmall cell lung carcinoma, not otherwise specified.

  • a

    Based on the greatest single tumor dimension in centimeters (the Cox proportional hazards method was used for a continuous variable).

  • b

    The number of patients and incidence rates are reported for the yes category. The P value is reported for log-rank analysis of yes vs no.

 IIIB7 (3)0.17   
 IVA36 (15)25    
 IVB201 (82)29    
 Squamous39 (16)31.44   
 Nonsquamous161 (66)26    
 NSCLC-NOS44 (18)30    
Hilar disease size (continuous variablea)0.00001.301.16-1.46.000
Bronchial or vascular obstruction      
 Yes90 (37)33.00031.350.75-2.42.32
 No153 (63)24    
Pulmonary symptomsb      
 Chest pain: Yes vs no59 (24)37.012   
 Cough: Yes vs no115 (47)32.0081   
 Hemoptysis: Yes vs no25 (10)32.30   
 Dyspnea: Yes vs no82 (34)29.20   
 Any pulmonary symptom: Yes vs no170 (70)29.0430.810.43-1.52.51
Pneumonia at presentation      
 Yes71 (29)41.00012.301.31-4.03.004
 No173 (71)11    
 Any atelectasis: Yes vs no70 (29)29.42   
 Whole lobe or more: Yes vs no17 (7)41.0630.740.30-1.84.52
Pleural effusion      
 Yes54 (22)20.34   
 No190 (78)29    


We thank Larry Potter, MS and Gregg Tracton, BSE in the Department of Radiation Oncology, University of North Carolina, Chapel Hill, North Carolina for their assistance in data management.


No specific funding was disclosed.


The authors made no disclosures.