Published Ahead of Print on February 18, 2014 as 10.1200/JCO.2013.50.5453 The latest version is at http://jco.ascopubs.org/cgi/doi/10.1200/JCO.2013.50.5453
JOURNAL OF CLINICAL ONCOLOGY
O R I G I N A L
R E P O R T
Prognostic Impact of Minimal Pleural Effusion in Non–Small-Cell Lung Cancer Jeong-Seon Ryu, Hyo Jin Ryu, Si-Nae Lee, Azra Memon, Seul-Ki Lee, Hae-Seong Nam, Hyun-Jung Kim, Kyung-Hee Lee, Jae-Hwa Cho, and Seung-Sik Hwang Jeong-Seon Ryu, Azra Memon, Seul-Ki Lee, Hae-Seong Nam, Hyun-Jung Kim, and Jae-Hwa Cho, Center for Lung Cancer, Inha University Hospital; Si-Nae Lee, Kyung-Hee Lee, and Seung-Sik Hwang, Inha University Hospital, Incheon, S. Korea; and Hyo Jin Ryu, University of California, Los Angeles, CA. Published online ahead of print at www.jco.org on February 18, 2014. Supported by Grant No. A110518 from the Korea Healthcare Technology R&D Project, Ministry for Health, Welfare & Family Affairs, Republic of Korea. Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article. Corresponding author: Jeong-Seon Ryu, MD, PhD, Center for Lung Cancer, Inha University Hospital, 7-206, 3-Ga, Shinheung Dong, Jung Gu, Incheon, 400-103, South Korea; e-mail: jsryu@ inha.ac.kr. © 2014 by American Society of Clinical Oncology
A
B
S
T
R
A
C
T
Purpose Minimal (⬍ 10 mm thick) pleural effusion (PE) may represent an early phase of malignant PE, but its clinical relevance has rarely been studied. Therefore, we examined the proportion of minimal PE in patients with non–small-cell lung cancer (NSCLC) and its impact on survival. We also considered possible accumulation mechanisms in our data set. Patients and Methods On the basis of PE status from chest computed tomography scans at diagnosis, 2,061 patients were classified into three groups: no PE, minimal PE, and malignant PE. Twenty-one variables associated with four factors—patient, stage migration, tumor, and treatment—were investigated for correlation with survival. Results Minimal PE presented in 272 patients (13.2%). Of 2,061 patients, the proportion of each stage was the following: 5.2% stage I, 10.9% stage II, 13.2% stage IIIA, 23.8% stage IIIB, and 13.9% stage IV. Minimal PE correlated significantly with shorter survival time than did no PE (median survival time, 7.7 v 17.7 months; log-rank P ⬍ .001), even after full adjustment with all variables (adjusted hazard ratio, 1.40; 95% CI, 1.21 to 1.62). Prognostic impact of minimal PE was higher in early versus advanced stages (Pinteraction ⫽ .001). In 237 patients (87.8%) with minimal PE, pleural invasion or attachment as a direct mechanism was observed, and it was an independent factor predicting worse survival (P ⫽ .03). Conclusion Minimal PE is a commonly encountered clinical concern in staging NSCLCs. Its presence is an important prognostic factor of worse survival, especially in early-stage disease.
0732-183X/14/3299-1/$20.00 DOI: 10.1200/JCO.2013.50.5453
J Clin Oncol 32. © 2014 by American Society of Clinical Oncology
INTRODUCTION
Malignant pleural effusion (PE) is defined by the presence of malignant cells in pleural fluid or biopsy. The accurate evaluation of PE became more important for tumor staging in the seventh edition of TNM classification, in which its status was changed from T4 to M1a.1 Thoracentesis and/or pleural biopsy are the first diagnostic steps in determining PE characteristics. However, these procedures are not recommended when PE presents as a minimal thickness (⬍ 10 mm) on lateral decubitus radiography or chest computed tomography (CT) scans.2,3 Alternatively, the more invasive approach of videoassisted thoracoscopic surgery can be considered to identify whether fluid contains malignant cells. However, this requires general anesthesia and raises concerns about mortality or morbidity, especially in patients with advanced-stage disease.
For staging in patients with minimal PE, guidelines have taken an ambiguous or different stand between small-cell lung cancer and non–small-cell lung cancer (NSCLC).1,4 The National Comprehensive Cancer Network guideline recommends that a PE too small to warrant thoracentesis should not be included in staging of small-cell lung cancers.4 Therefore, the presence of minimal PE is deemed limitedstage disease. Conversely, minimal PE is not covered in the recommendations for staging NSCLCs.1,4 Chest CT scans have been used in staging disease for more than two decades, and the incidence of adenocarcinoma is increasing in most countries. Although there have been no epidemiologic studies, the proportion of patients with minimal PE at diagnosis is expected to be significantly increasing. Furthermore, minimal PE might represent an early phase of malignant PE or severe comorbid disease that would confer prognostic impact. If so, then it © 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Copyright 2014 by American Society of Clinical Oncology
1
Ryu et al
Inha Lung Cancer Cohort 2002-2010 Non–small-cell lung cancer (N = 2,340)
Excluded No contrast-enhanced chest CT scan No brain MRI or CT scan No whole-body bone scan
Pleural effusion on CT scan?
(n = 105) (n = 117) (n = 54)
Yes (n = 666)
No (n = 1,398)
Max. thickness < 10 mm (n = 280)
Max. thickness ≥ 10 mm (n = 386) Excluded Transudate
No pleural effusion (n = 1,397)
Minimal pleural effusion (n = 272)
Fig 1. Flow chart for enrollment and classification of patients. (*) Nine patients were reclassified into malignant pleural effusion from minimal pleural effusion (pleural nodules in 5 patients; cytologically positive after pleural effusion rapidly increased in 3 patients) and no pleural effusion (multiple malignant pleural nodules found in surgical resection in 1 patient). CT, computed tomography; MRI, magnetic resonance imaging.
(n = 3)
Malignant pleural effusion* (n = 392)
Final study population for analysis (n = 2,061)
may be an important characteristic for clinicians in staging the disease and selecting optimal treatment. For such reasons, we set out to evaluate the proportion of minimal PE in patients with NSCLC from a retrospective cohort and the impact of this characteristic on prognosis of survival and to examine mechanisms involved in its accumulation. PATIENTS AND METHODS Study Cohort A cohort of 2,340 consecutive patients diagnosed histologically with stage I to IV NSCLC at Inha University Hospital (from 2002 to 2010) was considered initially in this study (Fig 1). The stage of all patients was defined according to the seventh edition of the TNM classification.1 To maintain quality of information on staging, patients who underwent noncontrast chest CT scans (n ⫽ 105) or who did not undergo brain imaging (n ⫽ 117) or whole-body bone scans (n ⫽ 54) were excluded. All patients underwent a staging work-up and were treated at the hospital. No patient received crizotinib or was enrolled onto clinical trials for tyrosine kinase inhibitors or other monoclonal antibodies. All information on prognostic variables was collected prospectively from the records of the Lung Cancer Cohort of Inha University Hospital. The study protocol was approved by the institutional review boards of the university hospital, and informed consent by the patient was waived. Selection of Prognostic Variables Patient-related variables included age, sex, smoking habit, Eastern Cooperative Oncology Group performance status (ECOG PS), any weight loss during the 6 months before diagnosis, and serum levels at diagnosis of hemoglobin, albumin, alkaline phosphatase, and calcium. The Charlson comorbidity index (CCI) score was calculated for each patient, with 19 diseases weighted by impact on mortality.5 The variable related to stage migration was rated according to whether or not positron emission tomography (PET) scans had been taken. Tumor-related variables were histology, epidermal growth factor receptor (EGFR) mutation, T stage, tumor size, N stage, M stage, and number of organs affected by metastasis. Finally, treatment-related variables consisted 2
© 2014 by American Society of Clinical Oncology
of curative treatment, palliative treatment, and no treatment. Curative treatment was defined by stages: surgery alone, (neo) adjuvant chemotherapy, or concurrent (or sequential) chemoradiotherapy therapy for stages I through IIIA; cytotoxic chemotherapy or concurrent (or sequential) chemoradiotherapy therapy for stage IIIB; and cytotoxic chemotherapy or targeted therapy for stage IV. Classification of Patients Results from PE analysis at diagnosis were evaluated by two pulmonologists (H.-S.N. and S.-N.L.). All chest CT scans at diagnosis were reviewed by a radiologist (K.-H.L.) to evaluate the presence of pleural fluid, any laterality, and thickness. Evaluation was blinded to data on the patients, including their survival status. According to the presence of pleural fluid and its thickness (judged relative to a criterion of 10 mm on chest CT scans), patients were classified initially into three groups: no PE, minimal PE (⬍ 10 mm thick), and malignant PE. Of 280 patients who initially met the criterion for minimal PE, five showed pleural nodules; in three others, the PE increased rapidly and malignant cells were found by cytology before treatment. These eight patients were reclassified into the malignant PE group. Multiple malignant pleural nodules were found in surgical resection in one patient initially staged as IB who was subsequently reclassified as malignant PE. No patients in the minimal PE group underwent any diagnostic evaluation for pleural fluid before commencement of treatment or at surgical resection. The presence of minimal PE was not taken into account in tumor staging. The malignant PE group consisted of 392 patients, with positive findings of malignant cells in 237 (61%), no malignant cells but exudative PE in 71 (18%), pleural nodule(s) in 24 (6%), and no diagnostic work-up in 60 (15%). A total of 2,061 patients with NSCLC were finally included in this study. Mechanism Involved in the Accumulation of Minimal PE Possible causes for accumulation of PE were determined by the radiologist and two pulmonologists from review of chest CT scans and bronchoscopy results, laboratory data, and comorbid diseases at diagnosis. They were classified as direct, indirect, or other mechanisms (Appendix Table A1, online only). Survival Measurements Overall survival was measured as an outcome and estimated from the time of diagnosis until death as a result of all causes. Only 21 patients died of JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Studies of Minimal Pleural Effusion in NSCLC
Table 1. Overall Survival by Status of PE: Cox Proportional Hazards Modeling Results No PE (n ⫽ 1,397)
Minimal PE (n ⫽ 272)
Malignant PE (n ⫽ 392)
Variable
HR
HR
95% CI
HR
95% CI
Ptrend
Unadjusted Patient-related variables Demographics: age, sex, smoking habit Tumor burden: ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium Comorbidity: CCI score Stage migration: PET Tumor-related variables: histology, EGFR mutation, tumor size, N stage, No. of organs affected by metastasis Treatment: no v yesⴱ Final fully adjusted model†
1 1 1
2.33 1.67 2.19
2.04 to 2.65 1.45 to 1.92 1.92 to 2.50
2.80 2.18 3.07
2.48 to 3.16 1.90 to 2.51 2.70 to 3.49
⬍ .001 ⬍ .001 ⬍ .001
1 1 1
1.83 1.98 2.30
1.59 to 2.11 1.73 to 2.26 2.02 to 2.63
2.14 2.40 2.75
1.87 to 2.45 2.12 to 2.72 2.44 to 3.11
⬍ .001 ⬍ .001 ⬍ .001
1 1 1
1.64 2.31 1.40
1.43 to 1.88 2.02 to 2.64 1.22 to 1.62
2.19 2.83 1.86
1.91 to 2.51 2.50 to 3.20 1.59 to 2.17
⬍ .001 ⬍ .001 ⬍ .001
Abbreviations: CCI, Charlson comorbidity index; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; PE, pleural effusion; PET, positron emission tomography. ⴱ First-line treatments for patients with stage I to IIIA: surgery alone, (neo) adjuvant chemotherapy, or concurrent (or sequential) chemoradiation; stage IIIB: concurrent (or sequential) chemoradiation or chemotherapy alone; stage IV: palliative chemotherapy or targeted therapy. †The model included sex, age, smoking habit, CCI score, ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium, histology, EGFR mutation, tumor size, N stage, No. of organs affected by metastasis, PET, and treatment.
other cancers or causes. There were 1,931 patients who died in our hospital. Otherwise, the date of death was obtained principally by contacting relatives of the patients (n ⫽ 529). For 59 patients who could not be contacted because they were lost to follow-up after hospital discharge, information on their survival was collected from the Korean Ministry of Security and Public Administration. Statistical Methods Distribution of the 21 variables according to the patients’ PE status was assessed by using 2 tests. The effect of an individual variable or PE status on survival was estimated by using the Kaplan-Meier method and log-rank testing. The hazard ratios (HRs) and 95% CIs were determined by using a Cox proportional hazard model. The hazards were calculated in an unadjusted model and then adjusted for each group of prognostic variables as potential
confounders; patient-related variables, stage migration, and tumor- or treatment-related variables were added separately to the unadjusted model. The proportion hazards assumption was examined with a plot of –log [⫺log (survival functions)] against log (follow-up time) and Schoenfeld residuals. The curves for minimal PE remained close together over the entire follow-up period and did not suggest nonproportionality (data not shown). Clinical stage and T and M stages were overlapped for the analysis with tumor size, number of organs in which metastasis occurred, and PE status, respectively, and then excluded. The fully adjusted model included 17 variables (Tables 1 and 2). Performance of the model was evaluated by Harrell’s c-index. To examine effect modification of the PE by stage, we added an interaction term to the Cox proportional hazard model. All significance testing was done at the two-sided P ⬍ .05 level. Analyses were performed by using the IBM SPSS
Table 2. Overall Survival by Disease Stage and PE Status: Cox Proportional Hazards Modeling Results Minimal PE
Variables Unadjusted Patient-related variables Demographics: age, sex, smoking habit Tumor burden: ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium Comorbidity: CCI score Stage migration: PET Tumor-related variables: histology, EGFR mutation, tumor size, N stage, No. of organs affected by metastasis Treatment: no v yesⴱ Final fully adjusted model†
Stage I (n ⫽ 20)
Stage II (n ⫽ 16)
HR
HR
95% CI
95% CI
Stage IIIA (n ⫽ 30) HR
95% CI
Stage IIIB (n ⫽ 59) HR
95% CI
Stage IV (n ⫽ 147) HR
95% CI
Malignant PE Stage IV (n ⫽ 392) HR
95% CI
3.37 1.93 to 5.89 2.10 1.14 to 3.88 2.12 1.39 to 3.23 1.65 1.22 to 2.21 1.38 1.16 to 1.63 1.37 1.19 to 1.57 2.51 1.30 to 4.84 1.97 0.95 to 4.08 1.55 0.95 to 2.53 1.36 1.00 to 1.85 1.22 1.02 to 1.46 1.28 1.10 to 1.49 2.35 1.32 to 4.17 1.98 1.05 to 3.73 2.93 1.25 to 2.98 1.55 1.15 to 2.09 1.34 1.13 to 1.59 1.45 1.26 to 1.67
3.38 1.78 to 6.44 1.78 0.88 to 3.61 1.74 1.09 to 2.77 1.37 1.01 to 1.86 1.21 1.01 to 1.44 1.15 0.99 to 1.33 3.36 1.92 to 5.87 2.38 1.28 to 4.44 2.12 1.38 to 3.25 1.62 1.20 to 2.18 1.33 1.12 to 1.58 1.39 1.21 to 1.59 3.54 2.02 to 6.19 2.14 1.16 to 3.95 1.86 1.22 to 2.84 1.64 1.22 to 2.20 1.37 1.16 to 1.63 1.37 1,19 to 1.57
3.44 1.96 to 6.03 2.36 1.22 to 4.55 2.41 1.57 to 3.70 1.98 1.44 to 2.73 1.23 1.03 to 1.47 1.61 1.39 to 1.87 2.90 1.66 to 5.09 2.21 1.19 to 4.09 1.86 1.22 to 2.84 2.04 1.51 to 2.76 1.40 1.18 to 1.66 1.35 1.18 to 1.55 2.07 1.06 to 4.05 2.24 1.02 to 4.94 1.62 0.95 to 2.94 1.57 1.08 to 2.28 1.16 0.98 to 1.39 1.45 1.22 to 1.71
NOTE. All HRs and 95% CIs were estimated by using patients with no PE in each stage as reference. Abbreviations: CCI, Charlson comorbidity index; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; PE, pleural effusion; PET, positron emission tomography. ⴱ First-line treatments for patients with stage I to IIIA: surgery alone, (neo) adjuvant chemotherapy, or concurrent (or sequential) chemoradiation; stage IIIB: concurrent (or sequential) chemoradiation or chemotherapy alone; stage IV: palliative chemotherapy or targeted therapy. †The model included sex, age, smoking habit, CCI score, ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium, histology, EGFR mutation, tumor size, N stage, No. of organs affected by metastasis, PET, and treatment.
www.jco.org
© 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
3
Ryu et al
Table 3. Baseline Characteristics of Patients With Non–Small-Cell Lung Cancer According to Type of PE Total (N ⫽ 2,061) Characteristic Median age, years Sex Male Female Smoking habit Never Past Current CCI score 0 1 ⱖ2 ECOG PS 0-1 ⱖ2 Weight loss, % None to ⬍ 5 ⱖ5 Hemoglobin, g/dLⴱ ⱖ 12 ⬍ 12 Albumin, g/dLⴱ ⱖ 3.1 ⬍ 3.1 Alkaline phosphatase, IU/Lⴱ ⱕ 335 ⬎ 335 Calcium, mg/dLⴱ ⱕ 10.8 ⬎ 10.8 PET No Yes Histology ADC SQC Others EGFR mutation No Yes Not tested Stage† I II IIIA IIIB IV T stage Tx T1 T2 T3 T4 Tumor size, cm 0-3 3.1-5 5.1-7 7.1-16 Unmeasurable
No.
No PE (n ⫽ 1,397) %
No.
67.0
Minimal PE (n ⫽ 272) %
No.
66.2
%
Malignant PE (n ⫽ 392) No.
69.1
%
2 P
68.2 ⬍ .001
1,514 547
73.5 26.5
1,064 333
76.2 23.8
212 60
77.9 22.1
238 154
60.7 39.3
459 576 983
22.7 28.5 48.8
282 390 696
20.6 28.5 50.9
50 87 133
18.5 32.2 49.3
127 99 154
33.4 26.1 40.5
683 997 373
33.3 48.6 18.2
493 689 208
35.5 49.5 15.0
69 130 72
25.5 48.0 26.5
121 178 93
30.9 45.4 23.7
1,272 641
66.5 33.5
959 310
75.6 24.4
145 119
54.9 45.1
168 212
44.2 55.8
1,196 766
61.0 39.0
878 446
66.3 33.7
139 124
52.9 47.1
179 196
47.7 52.3
1,443 582
71.3 28.7
1,037 334
75.6 24.4
147 120
55.1 44.9
259 128
66.9 33.1
1,782 240
88.1 11.9
1,283 86
93.7 6.3
209 58
78.3 21.7
290 96
75.1 24.9
1,761 259
87.2 12.8
1,219 148
89.2 10.8
223 44
83.5 16.5
319 67
82.6 17.4
1,976 43
97.9 2.1
1,345 23
98.3 1.7
257 10
96.3 3.7
374 10
97.4 2.6
1,088 965
53.0 47.0
765 624
55.1 44.9
142 130
52.2 47.8
181 211
46.2 53.8
1,004 863 194
48.7 41.9 9.4
600 653 144
43.0 46.7 10.3
106 139 27
39.0 51.1 9.9
298 71 23
76.0 18.1 5.9
314 154 1,593
15.2 7.5 77.3
222 105 1,070
15.9 7.5 76.6
39 16 217
14.3 5.9 79.8
53 33 306
13.5 8.4 78.1
381 147 227 248 1,058
18.5 7.1 11.0 12.0 51.3
361 131 197 189 519
25.8 9.4 14.1 13.5 37.2
20 16 30 59 147
7.4 5.9 11.0 21.7 54.0
0 0 0 0 392
100.0
8 231 638 330 840
0.4 11.3 31.2 16.1 41.0
5 213 548 258 366
0.4 15.3 39.4 18.6 26.3
1 11 65 43 151
0.4 4.1 24.0 15.8 55.7
2 7 25 29 323
0.5 1.8 6.5 7.5 83.7
445 713 424 248 231
21.6 34.6 20.6 12.0 11.2
351 524 281 130 111
25.1 37.5 20.1 9.3 7.9
31 89 78 58 16
11.4 32.7 28.7 21.3 5.9
63 100 65 60 104
16.1 25.5 16.6 15.3 26.5
⬍ .001
⬍ .001
⬍ .001
⬍ .001
⬍ .001
⬍ .001
.001
.08
.007
⬍ .001
.55
⬍ .001
⬍ .001
⬍ .001
(continued on following page)
4
© 2014 by American Society of Clinical Oncology
JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Studies of Minimal Pleural Effusion in NSCLC
Table 3. Baseline Characteristics of Patients With Non–Small-Cell Lung Cancer According to Type of PE (continued) Total (N ⫽ 2,061) Characteristic
No.
N stage N0 N1 N2 N3 M stage M0 M1a M1b No. of organs affected by metastasis 0 1 2 ⱖ3 Treatment No Yes‡ Survival Alive Death
No PE (n ⫽ 1,397) %
No.
Minimal PE (n ⫽ 272)
Malignant PE (n ⫽ 392)
%
No.
%
No.
% 24.2 3.1 23.1 49.6
2 P ⬍ .001
742 190 505 605
36.4 9.3 2.7 29.6
585 155 330 319
42.1 11.2 23.7 23.0
64 23 86 95
23.9 8.6 32.1 35.4
93 12 89 191
1,003 399 659
48.7 19.3 32.0
878 158 361
62.9 11.3 25.8
123 41 106
45.5 15.2 39.3
0 200 192
51.0 49.0
1,402 376 181 102
68.0 18.2 8.8 5.0
1,036 218 99 44
74.2 15.6 7.1 3.1
166 56 29 21
61.0 20.6 10.7 7.7
200 102 53 37
51.0 26.0 13.5 9.5
900 1,161
43.7 56.3
561 836
40.2 59.8
143 129
52.6 47.4
196 196
50.0 50.0
340 1,721
16.5 83.5
313 1,084
22.4 77.6
18 254
6.6 93.4
9 383
2.3 97.7
⬍ .001
⬍ .001
⬍ .001
⬍ .001
Abbreviations: ADC, adenocarcinoma; CCI, Charlson comorbidity index; ECOG PS, Eastern Cooperative Oncology Group performance status; PE, pleural effusion; PET, positron emission tomography; SQC, squamous cell carcinoma. ⴱ Dichotomized by cutoff of normal value. †Clinical stage at the time of initial diagnosis was determined according to seventh edition of the TNM classification. ‡First-line treatments for patients with stage I to IIIA: surgery alone, (neo) adjuvant chemotherapy, or concurrent (or sequential) chemoradiation; stage IIIB: concurrent (or sequential) chemoradiation or chemotherapy alone; stage IV: palliative chemotherapy or targeted therapy.
statistical software package version 19.0 (SPSS, Chicago, IL) and STATA version 12.1 (STATA, College Station, TX).
RESULTS
Patient Characteristics Baseline characteristics according to PE status of the 2,061 patients in the study population are shown in Table 3. Median age of patients was 67.0 years, and there were 547 women (26.6%). According to histopathology, 48.7% of patients had an adenocarcinoma. Minimal PE presented in 272 patients (13.2%), and there was no PE in 1,397 (67.8%). Deaths were observed in 1,721 patients (83.5%). The proportion of patients presenting with minimal PE was seen to increase according to stage: 5.2% (20) in stage I, 10.9% (16) in stage II, 13.2% (30) in stage IIIA, 23.8% (59) in stage IIIB, and 13.9% (147) in stage IV (2 test P ⬍ .001). Patients with minimal PE had shorter survival compared with those with no PE (median survival time, 7.7 v 17.7 months, respectively; log-rank P ⬍ .001). Distribution of Variables Related to Patient, Stage Migration, Tumor, and Treatment and Their Impact on Overall Survival Distribution of the variables between no PE and minimal PE groups indicates no differences among sex, smoking habit, histology, EGFR mutation, or use of PET scans. However, lower CCI score; better ECOG PS (0 to 1); absence of weight loss; normal levels of hemoglobin, albumin, and alkaline phosphatase; smaller tumor size; www.jco.org
low N stage; small number of organs affected by metastasis; and curative treatment were more likely in the no PE group (2 test P ⬍ .001). All variables related to patient, stage migration, tumor, and treatment exerted significant effects on overall survival (Appendix Table A2, online only). Mechanisms Involved in Accumulation of Minimal PE and Their Impact on Overall Survival There were 270 patients with minimal PE in which a cause of accumulation was found during data review. Of these, 237 (87.8%) had a direct mechanism of invasion or attachment to pleura by tumor (Fig 2). For indirect mechanisms, 183 patients (67.8%) had tumor involvement with the mediastinal lymph nodes. Other mechanisms were observed in 18 patients (6.7%): pneumonia (10) and uncontrolled heart (4), kidney (3), and liver (1) diseases. Both direct and indirect mechanisms were found in 177 patients (65.5%). No difference was found in survival according to laterality of minimal PE (log-rank P ⫽ .63). Patients with direct mechanism had worse overall survival than patients without direct mechanism, when adjusted with all the variables (adjusted HR, 1.75; 95% CI, 1.06 to 2.91; Appendix Table A3, online only). Among patients with minimal PE, involvement with the mediastinal lymph nodes, heart and major vessels, or two or more numbers of causes tended to indicate worse survival, but differences were not significant. Status of PE and Overall Survival Patients with minimal PE showed significantly worse survival than those with no PE (unadjusted HR, 2.33; 95% CI, 2.04 to 2.65; © 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
5
Ryu et al
No. of Patients With Minimal Pleural Effusion
A
250 200
100
No. of Patients With Minimal Pleural Effusion No. of Patients With Minimal Pleural Effusion
77
50
30
Pleura Mediastinum Obstruction
Heart, Major Vessel
17
18
Lymphangitic Others Metastasis
200
177
150 100 59
50 0
C
183
150
0
B
237
34
Direct Only
100
Indirect Only
Both
91 80
75
75 50 20
25
4
0
One
Two
Three
Four
Five
Fig 2. Mechanisms involved in the accumulation of minimal pleural effusion by (A) causes, (B) direct and indirect mechanisms, and (C) number of causes (n ⫽ 270).
Table 1 and Fig 3). Presence of minimal PE was a consistent and significant factor for predicting worse overall survival at every step of the analysis when adjusted in a stepwise fashion, with each group of variables related to patient, stage migration, tumor, or treatment. In the final fully adjusted model, presence of minimal PE was an independent prognostic factor of worse survival when compared with the no PE group (adjusted HR, 1.40; 95% CI, 1.22 to 1.62; P ⬍ .001). The Harrell’s c-index was 0.77 (95% CI, 0.75 to 0.78) for assessing accuracy of the final model. From analysis of the impact of minimal PE on survival by stages, association of minimal PE with risk of death increased as a function of decrease in stage, from stages IV through I (adjusted HRs: 2.07 for stage I, 2.24 for stage II, 1.62 for stage IIIA, 1.57 for stage IIIB, and 1.16 for stage IV; Table 2). Effect of modification of PE status according to disease stage and its relationship to survival was analyzed, revealing that the impact of minimal PE was stronger in early-stage disease (Pinteraction ⫽ .001). When stage I with no PE was used as reference group in estimating risk of death, stage I with minimal PE showed higher risk compared with stage IIIA with no PE (adjusted HRs were 2.97 and 2.60, respectively; Appendix Table A4, online only). In patients with stage IV disease, risk of death increased with change of PE 6
© 2014 by American Society of Clinical Oncology
status (from no PE through malignant PE; adjusted HR, 1.20; 95% CI, 1.11 to 1.31; Ptrend ⬍ .001). When divided into M1a and M1b subgroups, this trend was maintained (Appendix Table A5, online only; Appendix Fig A1, online only). For patients with malignant PE, there were no differences in overall survival between patients with or without pleural malignant cells, pleural nodules, or no diagnostic work-up (P ⫽ .18; data not shown). DISCUSSION
In this study, we report that minimal PE usually presents at diagnosis and is an independent prognostic factor of worse survival among patients with NSCLC. Although there have been no epidemiologic studies, previous studies have reported that malignant PE is present in 11% to 32% of patients with advanced NSCLC.5-8 However, no study has reported the proportion of patients with minimal PE in which thoracentesis and/or pleural biopsy was not recommended. In this study, we found that 13.2% of patients with NSCLC in the study population present with minimal PE. This high proportion may be due to the meticulous attention paid to detecting minimal PE in this study, but it also agrees with a recent increasing trend toward adenocarcinoma (48.7% in this study). The presence of minimal PE should be of concern regarding delayed and inaccurate staging of the disease or with respect to morbidity or mortality related to more invasive approaches. There are several explanations for PE accumulation.9-12 Postmortem studies have suggested that tumor involvement of mediastinal lymph nodes is essential for PE accumulation.10,13 By contrast, Light et al12 claimed that this hypothesis does not fully explain exudative PE or the presence of malignant cells in most cases of malignant PE, and they postulated direct invasion of pleura as the primary reason, or in combination with indirect mechanisms. Direct mechanism was seen in 87% of patients with minimal PE and both direct and indirect mechanisms in 65%. These findings suggest that minimal PE represents an early phase in the development of malignant PE and support the combination hypothesis.12 In addition, minimal PE also occurs in severe comorbid diseases. The CCI score is a validated method for evaluating individual comorbidity and is well known to have an effect on treatment decision and survival of patients with NSCLC.14-16 Patients with minimal PE had higher CCI score compared with those in the no PE group in this study. When minimal PE represented an early phase of malignant PE or had developed as a result of severe comorbid diseases, it is reasonable to assume that these could impair survival. In this study, 17 variables related to patient, stage migration, tumor, and treatment were considered as potential confounders. It is important to know whether consistency of the impact of minimal PE was maintained at each step in the stepwise analysis of adjusting for each group of variables. We demonstrated that having minimal PE was consistently predictive of worse survival, and the minimal PE group showed a 40% increase in risk of death when compared with the no PE group. The prognostic impact of minimal PE can be deduced from a recent meta-analysis. Lim et al17 reported that pleural lavage cytology positive for tumor cells during surgical JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Studies of Minimal Pleural Effusion in NSCLC
0.8
95% CI 16.0 to 19.4 6.4 to 9.0 4.3 to 6.7
0.6 0.4 0.2
B
1.0
1
0.6 0.4 0.2
2
3
4
5
0
No PE Minimal PE
MST 17.7 10.6
95% CI 14.0 to 21.4 4.7 to 16.5
0.8 0.6 0.4 0.2
E
2
3
4
5
1
2
3
No PE Minimal PE
4
5
95% CI 19.6 to 31.9 8.3 to 12.5
0.6 0.4 0.2
0
1
2
3
4
5
Time (years)
MST 14.5 7.8
95% CI 12.5 to 16.5 5.7 to 10.2
0.8 0.6 0.4 0.2
0
MST 25.8 10.3
0.8
F
1.0
MST No PE 8.1 Minimal PE 5.6 Malignant PE 5.5
0.8
95% CI 7.0 to 9.2 4.2 to 6.9 4.3 to 6.7
0.6 0.4 0.2 P < .0001
P < .0001
Time (years)
No PE Minimal PE
P = .0151
1.0
P = .0003
1
1.0
Time (years)
Overall Survival (probability)
Overall Survival (probability)
1.0
0
C
P < .0001
Time (years)
D
95% CI 63.3 to 96.1 8.3 to 47.5
0.8
P < .0001
0
MST 77.9 27.9
No PE Minimal PE
Overall Survival (probability)
MST 17.7 7.7 5.5
No PE Minimal PE Malignant PE
Overall Survival (probability)
1.0
Overall Survival (probability)
Overall Survival (probability)
A
1
2
3
Time (years)
4
5
0
1
2
3
4
5
Time (years)
Fig 3. Overall survival of the total study population by (A) status of pleural effusion (PE) and by disease stage and status of pleural effusion at (B) stage I, (C) stage II, (D) stage IIIA, (E) stage IIIB, and (F) stage IV. MST, median survival time.
resection is an independent factor in predicting worse survival of patients with a resectable-stage tumor. We also found that the effect of minimal PE on survival varies according to disease stage, with greatest effect in early disease. This indicates that the impact could be maximized in patients with earlystage disease through tumor stage shifting. However, as expected, the proportion of patients with minimal PE was lower in early-stage disease compared with advanced stage disease. As PE status changed from no PE through malignant PE in patients with stage IV disease, risk of death in minimal or malignant PE gradually increased. Patients with malignant PE had larger amounts of effusion, and this could have had a prognostic impact. This is supported by a recent study in patients with advanced-stage ovarian carcinomas, demonstrating that a moderate-to-large PE is an independent prognostic factor for predicting worse survival when compared with a small PE.18 In addition, there is evidence that increased tumor burden within the pleural space is associated with worse survival among patients with malignant PE.19,20 This study has several limitations. First, environmental factors, such as education level, socioeconomic status, access to care, insurance status, and quality of treatment or diagnosis, could be important and should be considered in identifying more robust prognostic markers.21,22 Unfortunately, most of these factors could not be considered in this study because it was based on a cohort of patients with lung cancer that did not include such environmentally related items. However, factors related to host, tumor, and treatment were considered in detail in this study. Second, our study www.jco.org
population was enrolled over 9 years. Therefore, advances in treatment or supportive care and the possibility of stage migration by PET could have been subject to selection bias in analyzing prognostic effects. However, enrollment was limited to patients staged with contrast-enhanced chest CT scans, bone scans, and brain imaging to maintain homogeneity of the population. Interestingly, we found no difference in stage-specific survival based on whether PET was used or not (data not shown). Third, the natural course of minimal PE should be valuable in understanding its nature and significance. However, strict follow-ups with chest CT scan or pleural fluid analysis, when fluid had increased in amount, were not attempted aggressively in this study. This was largely because the study design was retrospective in nature and partially because of the shorter survival of patients: 43% of the patients received either no treatment or palliative treatment alone. Treatment response, treatment-related untoward effects, comorbid diseases, or pattern of tumor progression can also complicate the natural course of disease. In our study, we paid meticulous attention to evaluating the mechanism of pleural fluid accumulation through reviewing chest CT scans, bronchoscopic examinations, and laboratory data and noting comorbid diseases at diagnosis. Finally, validation is needed in another data set before conclusions from this study can be considered as a subset of TNM stage criteria. Nevertheless, this is a systematic study examining the proportion of patients with minimal PE, the mechanisms involved in PE accumulation, and its prognostic impact among patients with cancer. © 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
7
Ryu et al
AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The author(s) indicated no potential conflicts of interest.
AUTHOR CONTRIBUTIONS Conception and design: Jeong-Seon Ryu, Hyo Jin Ryu Financial support: Jeong-Seon Ryu
REFERENCES 1. Goldstraw P, Crowley J, Chansky K, et al: The IASLC Lung Cancer Staging Project: Proposals for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM Classification of malignant tumours. J Thorac Oncol 2:706714, 2007 2. Light RW: Pleural Diseases (ed 4). Philadelphia, PA, Lippincott Williams and Wilkins, 2001 3. American Thoracic Society: Management of malignant pleural effusions. Am J Respir Crit Care Med 162:1987-2001, 2000 4. National Comprehensive Cancer Network: NCCN Clinical Practice Guidelines in Oncology, Non-Small Cell Lung Cancer-Version 2.2014 http:// www.nccn.org/professionals/physician_gls/pdf/ nscl.pdf 5. Sugiura S, Ando Y, Minami H, et al: Prognostic value of pleural effusion in patients with non-small cell lung cancer. Clin Cancer Res 3:47-50, 1997 6. Morgensztern D, Waqar S, Subramanian J, et al: Prognostic impact of malignant pleural effusion at presentation in patients with metastatic non-smallcell lung cancer. J Thorac Oncol 7:1485-1489, 2012 7. William WN Jr, Lin HY, Lee JJ, et al: Revisiting stage IIIB and IV non-small cell lung cancer:
Administrative support: Jeong-Seon Ryu, Azra Memon, Seul-Ki Lee, Hyun-Jung Kim Provision of study materials or patients: Jeong-Seon Ryu, Hyun-Jung Kim Collection and assembly of data: Jeong-Seon Ryu, Hyo Jin Ryu, Si-Nae Lee, Seul-Ki Lee, Hae-Seong Nam, Hyun-Jung Kim, Jae-Hwa Cho Data analysis and interpretation: Jeong-Seon Ryu, Azra Memon, Kyung-Hee Lee, Seung-Sik Hwang Manuscript writing: All authors Final approval of manuscript: All authors
Analysis of the surveillance, epidemiology, and end results data. Chest 136:701-709, 2009 8. Naito T, Satoh H, Ishikawa H, et al: Pleural effusion as a significant prognostic factor in nonsmall cell lung cancer. Anticancer Res 17:47434746, 1997 9. Sahn SA: The pathophysiology of pleural effusions. Annu Rev Med 41:7-13, 1990 10. Meyer PC: Metastatic carcinoma of the pleura. Thorax 21:437-443, 1966 11. Leff A, Hopewell PC, Costello J: Pleural effusion from malignancy. Ann Intern Med 88:532-537, 1978 12. Light RW, Hamm H: Malignant pleural effusion: Would the real cause please stand up? Eur Respir J 10:1701-1702, 1997 13. Chernow B, Sahn SA: Carcinomatous involvement of the pleura: An analysis of 96 patients. Am J Med 63:695-702, 1977 14. Asmis TR, Ding K, Seymour L, et al: Age and comorbidity as independent prognostic factors in the treatment of non small-cell lung cancer: A review of National Cancer Institute of Canada Clinical Trials Group trials. J Clin Oncol 26:54-59, 2008 15. Williams CD, Stechuchak KM, Zullig LL, et al: Influence of comorbidity on racial differences in receipt of surgery among US veterans with earlystage non-small-cell lung cancer. J Clin Oncol 31: 475-481, 2013
16. Charlson ME, Pompei P, Ales KL, et al: A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis 40:373-383, 1987 17. Lim E, Clough R, Goldstraw P, et al: Impact of positive pleural lavage cytology on survival in patients having lung resection for non-small-cell lung cancer: An international individual patient data metaanalysis. J Thorac Cardiovasc Surg 139:1441-1446, 2010 18. Mironov O, Ishill NM, Mironov S, et al: Pleural effusion detected at CT prior to primary cytoreduction for stage III or IV ovarian carcinoma: Effect on survival. Radiology 258:776-784, 2011 19. Sahn SA, Good JT Jr: Pleural fluid pH in malignant effusions: Diagnostic, prognostic, and therapeutic implications. Ann Intern Med 108:345349, 1988 20. Heffner JE, Nietert PJ, Barbieri C: Pleural fluid pH as a predictor of survival for patients with malignant pleural effusions. Chest 117:79-86, 2000 21. Herndon JE 2nd, Kornblith AB, Holland JC, et al: Patient education level as a predictor of survival in lung cancer clinical trials. J Clin Oncol 26:4116-4123, 2008 22. Oh DY, Choi KS, Shin HR, et al: Public awareness of gastric cancer risk factors and disease screening in a high risk region: A population-based study. Cancer Res Treat 41:59-66, 2009
■ ■ ■
8
© 2014 by American Society of Clinical Oncology
JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Studies of Minimal Pleural Effusion in NSCLC
Acknowledgment The corresponding author thanks Bong Wan Noh, MD; Jun Hyeok Lim, MD; Jung Min Lee, MD; and Gwang Seok Yoon, MD, for their generous support and cooperation. Appendix
Table A1. Classification of Mechanisms Involved in Accumulation of Minimal PE Mechanism
Classification
Direct Indirect
Tumor invasion or broad attachment to pleura Tumor involvement of mediastinal lymph node Complete obstruction in one or more lobar or main bronchus or atelectasis in more than one lobe Significant tumor infiltration or compression to heart or major vessels (superior vena cava, main trunk of pulmonary artery or vein) or significant pulmonary thromboembolism Lymphangitic metastasis to more than one lobe Pneumonia Uncontrolled heart, kidney, thyroid and liver disease
Other
Abbreviation: PE, pleural effusion.
www.jco.org
© 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
9
Ryu et al
Table A2. Prognostic Variables and Overall Survival: Univariable Analysis Variable PE No Minimal Malignant Age per year Sex Male Female Smoking habit Never Past Current CCI score 0 1 ⱖ2 ECOG PS 0-1 ⱖ2 Weight loss, % None to ⬍ 5 ⱖ5 Hemoglobin, g/dLⴱ ⱖ 12 ⬍ 12 Albumin, g/dLⴱ ⱖ 3.1 ⬍ 3.1 Alkaline phosphatase, IU/Lⴱ ⱕ 335 ⬎ 335 Calcium, mg/dLⴱ ⱕ 10.8 ⬎ 10.8 PET Yes No Histology SQC ADC Others EGFR mutation Yes No Not tested Stage† I II IIIA IIIB IV Tumor size, cm 0-3 3.1-5 5.1-7 7.1-16 Unmeasurable
MST (months)
95% CI
HR
95% CI
17.7 7.7 5.5
16.0 to 19.4 6.4 to 9.0 4.3 to 6.7
1 2.32 2.80 1.03
2.04 to 2.65 2.48 to 3.17 1.03 to 1.04
11.1 17.6
10.1 to 12.1 15.2 to 20.0
1 0.77
0.69 to 0.86
19.4 10.6 11.6
16.7 to 22.2 9.1 to 12.2 10.4 to 12.7
1 1.43 1.34
1.25 to 1.64 1.18 to 1.52
16.5 13.0 6.8
13.9 to 19.1 11.7 to 14.2 5.7 to 8.0
1 1.24 2.02
1.11 to 1.38 1.76 to 2.31
17.4 5.9
15.8 to 19.1 5.0 to 6.7
1 2.46
2.22 to 2.74
17.4 7.9
15.7 to 19.1 7.1 to 8.6
1 1.85
1.67 to 2.04
14.8 7.8
13.6 to 15.9 6.7 to 8.8
1 1.55
1.40 to 1.72
14.3 3.9
13.3 to 15.3 2.7 to 5.2
1 2.47
2.15 to 2.85
13.3 6.5
12.3 to 14.4 5.0 to 7.9
1 1.64
1.42 to 1.89
12.7 2.7
11.7 to 13.6 0.1 to 5.3
1 2.65
1.94 to 3.62
14.4 10.9
12.9 to 15.8 9.8 to 12.1
1 1.20
1.10 to 1.33
12.0 14.3 7.9
10.7 to 13.4 13.0 to 15.7 5.5 to 10.4
1 0.98 1.23
0.89 to 1.09 1.04 to 1.46
21.9 11.7 12.2
14.8 to 29.0 9.8 to 13.6 11.0 to 13.3
1 1.40 1.31
1.12 to 1,75 1.08 to 1.59
75.5 25.1 16.8 12.2 6.8
59.3 to 91.7 20.9 to 29.3 14.3 to 19.2 10.2 to 14.3 6.1 to 7.6
1 2.07 3.45 5.23 7.64
1.59 to 2.68 2.79 to 4.27 4.27 to 6.42 6.42 to 9.10
29.0 14.5 8.8 5.6 9.2
25.3 to 32.8 12.8 to 16.2 7.5 to 10.0 3.9 to 7.3 7.1 to 11.4
1 1.73 2.67 3.24 2.59
1.50 to 1.99 2.29 to 3.12 2.72 to 3.86 2.17 to 3.09
(continued on following page)
10
© 2014 by American Society of Clinical Oncology
JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Studies of Minimal Pleural Effusion in NSCLC
Table A2. Prognostic Variables and Overall Survival: Univariable Analysis (continued) Variable N stage N0 N1 N2 N3 No. of organs affected by metastasis 0 1 2 ⱖ3 Treatment No Yes
MST (months)
95% CI
HR
95% CI
27.7 16.7 10.1 6.1
24.3 to 31.1 13.1 to 20.2 8.7 to 11.5 5.3 to 6.8
1 1.41 2.45 3.87
1.18 to 1.69 2.16 to 2.79 3.41 to 4.39
17.6 6.8 5.0 2.7
16.0 to 19.2 5.6 to 8.1 3.7 to 6.3 2.0 to 3.4
1 2.38 2.89 4.15
2.11 to 2.68 2.46 to 3.41 3.37 to 5.12
7.0 17.0
6.1 to 7.9 15.4 to 18.7
1 0.53
0.48 to 0.58
Abbreviations: ADC, adenocarcinoma; CCI, Charlson comorbidity index; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; MST, median survival time; PE, pleural effusion; PET, positron emission tomography; SQC, squamous cell carcinoma. ⴱ Dichotomized by cutoff point of normal value. †Clinical stage at the time of initial diagnosis was determined according to the seventh edition of the TNM classification.
Table A3. Mechanisms Involved in Accumulation of PE and Overall Survival in Patients With Minimal PE (n ⫽ 270) Mechanism Direct Pleura Yes No Indirect Mediastinum Yes No Obstruction Yes No Heart, major vessels Yes No Lymphangitic metastasis Yes No No. of causes 1 ⱖ2
aHRⴱ
95% CI
Cox P
1 1.75 1
1.06 to 2.91
.03
1.47 1
0.41 to 5.19
.55 —
.61
0.88 —
0.61 to 1.27 —
.48 —
⬍ .01
1 1.18
— 0.73 to 1.91
— .49
4.4 to 11.1 5.8 to 9.2
.62
1 1.67
— 0.90 to 3.10
— .10
6.2 to 14.5 5.2 to 8.3
⬍ .01
1 1.46
0.89 to 2.40
.14
MST (months)
95% CI
Log-Rank P
7.4 10.4
6.1 to 8.8 3.1 to 17.7
.06
6.7 9.5
5.2 to 8.3 6.5 to 12.5
⬍ .01
7.4 7.6
4.5 to 10.4 5.8 to 9.4
5.4 8.0
3.1 to 7.8 6.4 to 9.5
7.7 7.5 10.4 6.7
Abbreviations: aHR, adjusted hazard ratio; HR, hazard ratio; MST, median survival time; PE, pleural effusion. ⴱ HR after adjustment with sex, age, smoking habit, Charlson comorbidity index score, Eastern Cooperative Oncology Group performance status, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium, histology, epidermal growth factor receptor mutation, tumor size, N stage, No. of organs affected by metastasis, positron emission tomography, and treatment.
www.jco.org
© 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
11
Ryu et al
Table A4. Adjusted HRs for Mortality on Minimal PE Modified by Stages No PE Stage
HR
I II IIIA IIIB IV
1.00 (reference) 1.54 2.60 3.53 5.14
Minimal PE 95% CI
HR
95% CI
1.11 to 2.14 1.89 to 3.59 2.64 to 4.72 3.85 to 6.79
2.97 2.70 3.92 5.52 5.55
1.65 to 5.35 1.46 to 5.01 2.41 to 6.38 3.80 to 8.02 4.02 to 7.65
NOTE. HRs were estimated after adjustment with sex, age, smoking habit, Charlson comorbidity index score, Eastern Cooperative Oncology Group performance status, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium, histology, epidermal growth factor receptor mutation, positron emission tomography, and treatment. Abbreviations: HR, hazard ratio; PE, pleural effusion.
Table A5. Overall Survival and Status of PE Within Subgroups of Stage IV (M1a and M1b): Cox Proportional Hazards Modeling Results Stage IV M1a
Variable
Minimal PE
Malignant PE
No PE (ref)
HR
95% CI
HR
1 1 1
1.54 1.30 1.58
1.13 to 2.10 0.93 to 1.82 1.15 to 2.16
1.47 1.36 1.56
1 1 1
1.26 1.51 1.52
0.91 to 1.75 1.11 to 2.06 1.12 to 2.07
1 1 1
1.45 1.53 1.33
1.06 to 1.98 1.12 to 2.09 0.92 to 1.91
Unadjusted Patient-related variables Demographics: age, sex, smoking habit Tumor burden: ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium Comorbidity: CCI score Stage migration: PET Tumor-related variables: histology, EGFR mutation, tumor size N stage, No. of organs affected by metastasis Treatment: no v yesⴱ Final fully adjusted model†
Stage IV M1b Minimal PE
Malignant PE
No PE (ref)
HR
95% CI
HR
1.18 to 1.82 .001 1.06 to 1.73 .019 1.24 to 1.95 ⬍ .001
1 1 1
1.31 1.21 1.26
1.07 to 1.61 0.97 to 1.50 1.02 to 1.55
1.50 1.46 1.68
1.25 to 1.80 ⬍ .001 1.19 to 1.80 ⬍ .001 1.39 to 2.03 ⬍ .001
1.18 1.53 1.32
0.94 to 1.50 .170 1.23 to 1.91 ⬍ .001 1.07 to 1.62 .001
1 1 1
1.20 1.25 1.32
0.97 to 1.49 1.02 to 1.54 1.07 to 1.62
1.25 1.50 1.50
1.03 to 1.53 .020 1.25 to 1.80 ⬍ .001 1.25 to 1.80 ⬍ .001
1.64 1.47 1.66
1.30 to 2.06 ⬍ .001 1.18 to 1.83 .001 1.26 to 2.20 ⬍ .001
1 1 1
1.20 1.36 1.11
0.97 to 1.48 1.11 to 1.68 0.90 to 2.03
1.33 1.48 1.44
1.10 to 1.62 ⬍ .001 1.23 to 1.78 ⬍ .001 1.15 to 1.79 .002
95% CI
Ptrend
95% CI
Ptrend
NOTE. All HRs and 95% CIs were estimated by using patients without PE in each stage as ref. Abbreviations: CCI, Charlson comorbidity index; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; PE, pleural effusion; PET, positron emission tomography; ref, reference. ⴱ First-line treatments for patients with stage I to IIIA: surgery alone, (neo) adjuvant chemotherapy, or concurrent (or sequential) chemoradiation; stage IIIB: concurrent (or sequential) chemoradiation or chemotherapy alone; stage IV: palliative chemotherapy or targeted therapy. †The model included sex, age, smoking habit, CCI score, ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium, histology, EGFR mutation, tumor size, N stage, No. of organs affected by metastasis, PET, and treatment.
1.0
MST 10.6 8.8 7.4
No PE Minimal PE Malignant PE
0.8
95% CI 7.7 to 13.4 7.3 to 10.3 6.0 to 8.7
0.6 0.4 0.2
B Overall Survival (probability)
Overall Survival (probability)
A
1.0 0.8
MST 7.0 4.9 4.1
95% CI 6.0 to 8.1 3.7 to 6.2 3.2 to 5.0
0.6 0.4 0.2 P < .0001
P = .0008
0
No PE Minimal PE Malignant PE
1
2
3
Time (years)
4
5
0
1
2
3
4
5
Time (years)
Fig A1. Overall survival and status of pleural effusion (PE) within subgroups of stage IV: (A) M1a and (B) M1b. MST, median survival time. 12
© 2014 by American Society of Clinical Oncology
JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
JOURNAL OF CLINICAL ONCOLOGY
O R I G I N A L
R E P O R T
Prognostic Impact of Minimal Pleural Effusion in Non–Small-Cell Lung Cancer Jeong-Seon Ryu, Hyo Jin Ryu, Si-Nae Lee, Azra Memon, Seul-Ki Lee, Hae-Seong Nam, Hyun-Jung Kim, Kyung-Hee Lee, Jae-Hwa Cho, and Seung-Sik Hwang Jeong-Seon Ryu, Azra Memon, Seul-Ki Lee, Hae-Seong Nam, Hyun-Jung Kim, and Jae-Hwa Cho, Center for Lung Cancer, Inha University Hospital; Si-Nae Lee, Kyung-Hee Lee, and Seung-Sik Hwang, Inha University Hospital, Incheon, S. Korea; and Hyo Jin Ryu, University of California, Los Angeles, CA. Published online ahead of print at www.jco.org on February 18, 2014. Supported by Grant No. A110518 from the Korea Healthcare Technology R&D Project, Ministry for Health, Welfare & Family Affairs, Republic of Korea. Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article. Corresponding author: Jeong-Seon Ryu, MD, PhD, Center for Lung Cancer, Inha University Hospital, 7-206, 3-Ga, Shinheung Dong, Jung Gu, Incheon, 400-103, South Korea; e-mail: jsryu@ inha.ac.kr. © 2014 by American Society of Clinical Oncology
A
B
S
T
R
A
C
T
Purpose Minimal (⬍ 10 mm thick) pleural effusion (PE) may represent an early phase of malignant PE, but its clinical relevance has rarely been studied. Therefore, we examined the proportion of minimal PE in patients with non–small-cell lung cancer (NSCLC) and its impact on survival. We also considered possible accumulation mechanisms in our data set. Patients and Methods On the basis of PE status from chest computed tomography scans at diagnosis, 2,061 patients were classified into three groups: no PE, minimal PE, and malignant PE. Twenty-one variables associated with four factors—patient, stage migration, tumor, and treatment—were investigated for correlation with survival. Results Minimal PE presented in 272 patients (13.2%). Of 2,061 patients, the proportion of each stage was the following: 5.2% stage I, 10.9% stage II, 13.2% stage IIIA, 23.8% stage IIIB, and 13.9% stage IV. Minimal PE correlated significantly with shorter survival time than did no PE (median survival time, 7.7 v 17.7 months; log-rank P ⬍ .001), even after full adjustment with all variables (adjusted hazard ratio, 1.40; 95% CI, 1.21 to 1.62). Prognostic impact of minimal PE was higher in early versus advanced stages (Pinteraction ⫽ .001). In 237 patients (87.8%) with minimal PE, pleural invasion or attachment as a direct mechanism was observed, and it was an independent factor predicting worse survival (P ⫽ .03). Conclusion Minimal PE is a commonly encountered clinical concern in staging NSCLCs. Its presence is an important prognostic factor of worse survival, especially in early-stage disease.
0732-183X/14/3299-1/$20.00 DOI: 10.1200/JCO.2013.50.5453
J Clin Oncol 32. © 2014 by American Society of Clinical Oncology
INTRODUCTION
Malignant pleural effusion (PE) is defined by the presence of malignant cells in pleural fluid or biopsy. The accurate evaluation of PE became more important for tumor staging in the seventh edition of TNM classification, in which its status was changed from T4 to M1a.1 Thoracentesis and/or pleural biopsy are the first diagnostic steps in determining PE characteristics. However, these procedures are not recommended when PE presents as a minimal thickness (⬍ 10 mm) on lateral decubitus radiography or chest computed tomography (CT) scans.2,3 Alternatively, the more invasive approach of videoassisted thoracoscopic surgery can be considered to identify whether fluid contains malignant cells. However, this requires general anesthesia and raises concerns about mortality or morbidity, especially in patients with advanced-stage disease.
For staging in patients with minimal PE, guidelines have taken an ambiguous or different stand between small-cell lung cancer and non–small-cell lung cancer (NSCLC).1,4 The National Comprehensive Cancer Network guideline recommends that a PE too small to warrant thoracentesis should not be included in staging of small-cell lung cancers.4 Therefore, the presence of minimal PE is deemed limitedstage disease. Conversely, minimal PE is not covered in the recommendations for staging NSCLCs.1,4 Chest CT scans have been used in staging disease for more than two decades, and the incidence of adenocarcinoma is increasing in most countries. Although there have been no epidemiologic studies, the proportion of patients with minimal PE at diagnosis is expected to be significantly increasing. Furthermore, minimal PE might represent an early phase of malignant PE or severe comorbid disease that would confer prognostic impact. If so, then it © 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Copyright 2014 by American Society of Clinical Oncology
1
Ryu et al
Inha Lung Cancer Cohort 2002-2010 Non–small-cell lung cancer (N = 2,340)
Excluded No contrast-enhanced chest CT scan No brain MRI or CT scan No whole-body bone scan
Pleural effusion on CT scan?
(n = 105) (n = 117) (n = 54)
Yes (n = 666)
No (n = 1,398)
Max. thickness < 10 mm (n = 280)
Max. thickness ≥ 10 mm (n = 386) Excluded Transudate
No pleural effusion (n = 1,397)
Minimal pleural effusion (n = 272)
Fig 1. Flow chart for enrollment and classification of patients. (*) Nine patients were reclassified into malignant pleural effusion from minimal pleural effusion (pleural nodules in 5 patients; cytologically positive after pleural effusion rapidly increased in 3 patients) and no pleural effusion (multiple malignant pleural nodules found in surgical resection in 1 patient). CT, computed tomography; MRI, magnetic resonance imaging.
(n = 3)
Malignant pleural effusion* (n = 392)
Final study population for analysis (n = 2,061)
may be an important characteristic for clinicians in staging the disease and selecting optimal treatment. For such reasons, we set out to evaluate the proportion of minimal PE in patients with NSCLC from a retrospective cohort and the impact of this characteristic on prognosis of survival and to examine mechanisms involved in its accumulation. PATIENTS AND METHODS Study Cohort A cohort of 2,340 consecutive patients diagnosed histologically with stage I to IV NSCLC at Inha University Hospital (from 2002 to 2010) was considered initially in this study (Fig 1). The stage of all patients was defined according to the seventh edition of the TNM classification.1 To maintain quality of information on staging, patients who underwent noncontrast chest CT scans (n ⫽ 105) or who did not undergo brain imaging (n ⫽ 117) or whole-body bone scans (n ⫽ 54) were excluded. All patients underwent a staging work-up and were treated at the hospital. No patient received crizotinib or was enrolled onto clinical trials for tyrosine kinase inhibitors or other monoclonal antibodies. All information on prognostic variables was collected prospectively from the records of the Lung Cancer Cohort of Inha University Hospital. The study protocol was approved by the institutional review boards of the university hospital, and informed consent by the patient was waived. Selection of Prognostic Variables Patient-related variables included age, sex, smoking habit, Eastern Cooperative Oncology Group performance status (ECOG PS), any weight loss during the 6 months before diagnosis, and serum levels at diagnosis of hemoglobin, albumin, alkaline phosphatase, and calcium. The Charlson comorbidity index (CCI) score was calculated for each patient, with 19 diseases weighted by impact on mortality.5 The variable related to stage migration was rated according to whether or not positron emission tomography (PET) scans had been taken. Tumor-related variables were histology, epidermal growth factor receptor (EGFR) mutation, T stage, tumor size, N stage, M stage, and number of organs affected by metastasis. Finally, treatment-related variables consisted 2
© 2014 by American Society of Clinical Oncology
of curative treatment, palliative treatment, and no treatment. Curative treatment was defined by stages: surgery alone, (neo) adjuvant chemotherapy, or concurrent (or sequential) chemoradiotherapy therapy for stages I through IIIA; cytotoxic chemotherapy or concurrent (or sequential) chemoradiotherapy therapy for stage IIIB; and cytotoxic chemotherapy or targeted therapy for stage IV. Classification of Patients Results from PE analysis at diagnosis were evaluated by two pulmonologists (H.-S.N. and S.-N.L.). All chest CT scans at diagnosis were reviewed by a radiologist (K.-H.L.) to evaluate the presence of pleural fluid, any laterality, and thickness. Evaluation was blinded to data on the patients, including their survival status. According to the presence of pleural fluid and its thickness (judged relative to a criterion of 10 mm on chest CT scans), patients were classified initially into three groups: no PE, minimal PE (⬍ 10 mm thick), and malignant PE. Of 280 patients who initially met the criterion for minimal PE, five showed pleural nodules; in three others, the PE increased rapidly and malignant cells were found by cytology before treatment. These eight patients were reclassified into the malignant PE group. Multiple malignant pleural nodules were found in surgical resection in one patient initially staged as IB who was subsequently reclassified as malignant PE. No patients in the minimal PE group underwent any diagnostic evaluation for pleural fluid before commencement of treatment or at surgical resection. The presence of minimal PE was not taken into account in tumor staging. The malignant PE group consisted of 392 patients, with positive findings of malignant cells in 237 (61%), no malignant cells but exudative PE in 71 (18%), pleural nodule(s) in 24 (6%), and no diagnostic work-up in 60 (15%). A total of 2,061 patients with NSCLC were finally included in this study. Mechanism Involved in the Accumulation of Minimal PE Possible causes for accumulation of PE were determined by the radiologist and two pulmonologists from review of chest CT scans and bronchoscopy results, laboratory data, and comorbid diseases at diagnosis. They were classified as direct, indirect, or other mechanisms (Appendix Table A1, online only). Survival Measurements Overall survival was measured as an outcome and estimated from the time of diagnosis until death as a result of all causes. Only 21 patients died of JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Studies of Minimal Pleural Effusion in NSCLC
Table 1. Overall Survival by Status of PE: Cox Proportional Hazards Modeling Results No PE (n ⫽ 1,397)
Minimal PE (n ⫽ 272)
Malignant PE (n ⫽ 392)
Variable
HR
HR
95% CI
HR
95% CI
Ptrend
Unadjusted Patient-related variables Demographics: age, sex, smoking habit Tumor burden: ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium Comorbidity: CCI score Stage migration: PET Tumor-related variables: histology, EGFR mutation, tumor size, N stage, No. of organs affected by metastasis Treatment: no v yesⴱ Final fully adjusted model†
1 1 1
2.33 1.67 2.19
2.04 to 2.65 1.45 to 1.92 1.92 to 2.50
2.80 2.18 3.07
2.48 to 3.16 1.90 to 2.51 2.70 to 3.49
⬍ .001 ⬍ .001 ⬍ .001
1 1 1
1.83 1.98 2.30
1.59 to 2.11 1.73 to 2.26 2.02 to 2.63
2.14 2.40 2.75
1.87 to 2.45 2.12 to 2.72 2.44 to 3.11
⬍ .001 ⬍ .001 ⬍ .001
1 1 1
1.64 2.31 1.40
1.43 to 1.88 2.02 to 2.64 1.22 to 1.62
2.19 2.83 1.86
1.91 to 2.51 2.50 to 3.20 1.59 to 2.17
⬍ .001 ⬍ .001 ⬍ .001
Abbreviations: CCI, Charlson comorbidity index; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; PE, pleural effusion; PET, positron emission tomography. ⴱ First-line treatments for patients with stage I to IIIA: surgery alone, (neo) adjuvant chemotherapy, or concurrent (or sequential) chemoradiation; stage IIIB: concurrent (or sequential) chemoradiation or chemotherapy alone; stage IV: palliative chemotherapy or targeted therapy. †The model included sex, age, smoking habit, CCI score, ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium, histology, EGFR mutation, tumor size, N stage, No. of organs affected by metastasis, PET, and treatment.
other cancers or causes. There were 1,931 patients who died in our hospital. Otherwise, the date of death was obtained principally by contacting relatives of the patients (n ⫽ 529). For 59 patients who could not be contacted because they were lost to follow-up after hospital discharge, information on their survival was collected from the Korean Ministry of Security and Public Administration. Statistical Methods Distribution of the 21 variables according to the patients’ PE status was assessed by using 2 tests. The effect of an individual variable or PE status on survival was estimated by using the Kaplan-Meier method and log-rank testing. The hazard ratios (HRs) and 95% CIs were determined by using a Cox proportional hazard model. The hazards were calculated in an unadjusted model and then adjusted for each group of prognostic variables as potential
confounders; patient-related variables, stage migration, and tumor- or treatment-related variables were added separately to the unadjusted model. The proportion hazards assumption was examined with a plot of –log [⫺log (survival functions)] against log (follow-up time) and Schoenfeld residuals. The curves for minimal PE remained close together over the entire follow-up period and did not suggest nonproportionality (data not shown). Clinical stage and T and M stages were overlapped for the analysis with tumor size, number of organs in which metastasis occurred, and PE status, respectively, and then excluded. The fully adjusted model included 17 variables (Tables 1 and 2). Performance of the model was evaluated by Harrell’s c-index. To examine effect modification of the PE by stage, we added an interaction term to the Cox proportional hazard model. All significance testing was done at the two-sided P ⬍ .05 level. Analyses were performed by using the IBM SPSS
Table 2. Overall Survival by Disease Stage and PE Status: Cox Proportional Hazards Modeling Results Minimal PE
Variables Unadjusted Patient-related variables Demographics: age, sex, smoking habit Tumor burden: ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium Comorbidity: CCI score Stage migration: PET Tumor-related variables: histology, EGFR mutation, tumor size, N stage, No. of organs affected by metastasis Treatment: no v yesⴱ Final fully adjusted model†
Stage I (n ⫽ 20)
Stage II (n ⫽ 16)
HR
HR
95% CI
95% CI
Stage IIIA (n ⫽ 30) HR
95% CI
Stage IIIB (n ⫽ 59) HR
95% CI
Stage IV (n ⫽ 147) HR
95% CI
Malignant PE Stage IV (n ⫽ 392) HR
95% CI
3.37 1.93 to 5.89 2.10 1.14 to 3.88 2.12 1.39 to 3.23 1.65 1.22 to 2.21 1.38 1.16 to 1.63 1.37 1.19 to 1.57 2.51 1.30 to 4.84 1.97 0.95 to 4.08 1.55 0.95 to 2.53 1.36 1.00 to 1.85 1.22 1.02 to 1.46 1.28 1.10 to 1.49 2.35 1.32 to 4.17 1.98 1.05 to 3.73 2.93 1.25 to 2.98 1.55 1.15 to 2.09 1.34 1.13 to 1.59 1.45 1.26 to 1.67
3.38 1.78 to 6.44 1.78 0.88 to 3.61 1.74 1.09 to 2.77 1.37 1.01 to 1.86 1.21 1.01 to 1.44 1.15 0.99 to 1.33 3.36 1.92 to 5.87 2.38 1.28 to 4.44 2.12 1.38 to 3.25 1.62 1.20 to 2.18 1.33 1.12 to 1.58 1.39 1.21 to 1.59 3.54 2.02 to 6.19 2.14 1.16 to 3.95 1.86 1.22 to 2.84 1.64 1.22 to 2.20 1.37 1.16 to 1.63 1.37 1,19 to 1.57
3.44 1.96 to 6.03 2.36 1.22 to 4.55 2.41 1.57 to 3.70 1.98 1.44 to 2.73 1.23 1.03 to 1.47 1.61 1.39 to 1.87 2.90 1.66 to 5.09 2.21 1.19 to 4.09 1.86 1.22 to 2.84 2.04 1.51 to 2.76 1.40 1.18 to 1.66 1.35 1.18 to 1.55 2.07 1.06 to 4.05 2.24 1.02 to 4.94 1.62 0.95 to 2.94 1.57 1.08 to 2.28 1.16 0.98 to 1.39 1.45 1.22 to 1.71
NOTE. All HRs and 95% CIs were estimated by using patients with no PE in each stage as reference. Abbreviations: CCI, Charlson comorbidity index; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; PE, pleural effusion; PET, positron emission tomography. ⴱ First-line treatments for patients with stage I to IIIA: surgery alone, (neo) adjuvant chemotherapy, or concurrent (or sequential) chemoradiation; stage IIIB: concurrent (or sequential) chemoradiation or chemotherapy alone; stage IV: palliative chemotherapy or targeted therapy. †The model included sex, age, smoking habit, CCI score, ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium, histology, EGFR mutation, tumor size, N stage, No. of organs affected by metastasis, PET, and treatment.
www.jco.org
© 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
3
Ryu et al
Table 3. Baseline Characteristics of Patients With Non–Small-Cell Lung Cancer According to Type of PE Total (N ⫽ 2,061) Characteristic Median age, years Sex Male Female Smoking habit Never Past Current CCI score 0 1 ⱖ2 ECOG PS 0-1 ⱖ2 Weight loss, % None to ⬍ 5 ⱖ5 Hemoglobin, g/dLⴱ ⱖ 12 ⬍ 12 Albumin, g/dLⴱ ⱖ 3.1 ⬍ 3.1 Alkaline phosphatase, IU/Lⴱ ⱕ 335 ⬎ 335 Calcium, mg/dLⴱ ⱕ 10.8 ⬎ 10.8 PET No Yes Histology ADC SQC Others EGFR mutation No Yes Not tested Stage† I II IIIA IIIB IV T stage Tx T1 T2 T3 T4 Tumor size, cm 0-3 3.1-5 5.1-7 7.1-16 Unmeasurable
No.
No PE (n ⫽ 1,397) %
No.
67.0
Minimal PE (n ⫽ 272) %
No.
66.2
%
Malignant PE (n ⫽ 392) No.
69.1
%
2 P
68.2 ⬍ .001
1,514 547
73.5 26.5
1,064 333
76.2 23.8
212 60
77.9 22.1
238 154
60.7 39.3
459 576 983
22.7 28.5 48.8
282 390 696
20.6 28.5 50.9
50 87 133
18.5 32.2 49.3
127 99 154
33.4 26.1 40.5
683 997 373
33.3 48.6 18.2
493 689 208
35.5 49.5 15.0
69 130 72
25.5 48.0 26.5
121 178 93
30.9 45.4 23.7
1,272 641
66.5 33.5
959 310
75.6 24.4
145 119
54.9 45.1
168 212
44.2 55.8
1,196 766
61.0 39.0
878 446
66.3 33.7
139 124
52.9 47.1
179 196
47.7 52.3
1,443 582
71.3 28.7
1,037 334
75.6 24.4
147 120
55.1 44.9
259 128
66.9 33.1
1,782 240
88.1 11.9
1,283 86
93.7 6.3
209 58
78.3 21.7
290 96
75.1 24.9
1,761 259
87.2 12.8
1,219 148
89.2 10.8
223 44
83.5 16.5
319 67
82.6 17.4
1,976 43
97.9 2.1
1,345 23
98.3 1.7
257 10
96.3 3.7
374 10
97.4 2.6
1,088 965
53.0 47.0
765 624
55.1 44.9
142 130
52.2 47.8
181 211
46.2 53.8
1,004 863 194
48.7 41.9 9.4
600 653 144
43.0 46.7 10.3
106 139 27
39.0 51.1 9.9
298 71 23
76.0 18.1 5.9
314 154 1,593
15.2 7.5 77.3
222 105 1,070
15.9 7.5 76.6
39 16 217
14.3 5.9 79.8
53 33 306
13.5 8.4 78.1
381 147 227 248 1,058
18.5 7.1 11.0 12.0 51.3
361 131 197 189 519
25.8 9.4 14.1 13.5 37.2
20 16 30 59 147
7.4 5.9 11.0 21.7 54.0
0 0 0 0 392
100.0
8 231 638 330 840
0.4 11.3 31.2 16.1 41.0
5 213 548 258 366
0.4 15.3 39.4 18.6 26.3
1 11 65 43 151
0.4 4.1 24.0 15.8 55.7
2 7 25 29 323
0.5 1.8 6.5 7.5 83.7
445 713 424 248 231
21.6 34.6 20.6 12.0 11.2
351 524 281 130 111
25.1 37.5 20.1 9.3 7.9
31 89 78 58 16
11.4 32.7 28.7 21.3 5.9
63 100 65 60 104
16.1 25.5 16.6 15.3 26.5
⬍ .001
⬍ .001
⬍ .001
⬍ .001
⬍ .001
⬍ .001
.001
.08
.007
⬍ .001
.55
⬍ .001
⬍ .001
⬍ .001
(continued on following page)
4
© 2014 by American Society of Clinical Oncology
JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Studies of Minimal Pleural Effusion in NSCLC
Table 3. Baseline Characteristics of Patients With Non–Small-Cell Lung Cancer According to Type of PE (continued) Total (N ⫽ 2,061) Characteristic
No.
N stage N0 N1 N2 N3 M stage M0 M1a M1b No. of organs affected by metastasis 0 1 2 ⱖ3 Treatment No Yes‡ Survival Alive Death
No PE (n ⫽ 1,397) %
No.
Minimal PE (n ⫽ 272)
Malignant PE (n ⫽ 392)
%
No.
%
No.
% 24.2 3.1 23.1 49.6
2 P ⬍ .001
742 190 505 605
36.4 9.3 2.7 29.6
585 155 330 319
42.1 11.2 23.7 23.0
64 23 86 95
23.9 8.6 32.1 35.4
93 12 89 191
1,003 399 659
48.7 19.3 32.0
878 158 361
62.9 11.3 25.8
123 41 106
45.5 15.2 39.3
0 200 192
51.0 49.0
1,402 376 181 102
68.0 18.2 8.8 5.0
1,036 218 99 44
74.2 15.6 7.1 3.1
166 56 29 21
61.0 20.6 10.7 7.7
200 102 53 37
51.0 26.0 13.5 9.5
900 1,161
43.7 56.3
561 836
40.2 59.8
143 129
52.6 47.4
196 196
50.0 50.0
340 1,721
16.5 83.5
313 1,084
22.4 77.6
18 254
6.6 93.4
9 383
2.3 97.7
⬍ .001
⬍ .001
⬍ .001
⬍ .001
Abbreviations: ADC, adenocarcinoma; CCI, Charlson comorbidity index; ECOG PS, Eastern Cooperative Oncology Group performance status; PE, pleural effusion; PET, positron emission tomography; SQC, squamous cell carcinoma. ⴱ Dichotomized by cutoff of normal value. †Clinical stage at the time of initial diagnosis was determined according to seventh edition of the TNM classification. ‡First-line treatments for patients with stage I to IIIA: surgery alone, (neo) adjuvant chemotherapy, or concurrent (or sequential) chemoradiation; stage IIIB: concurrent (or sequential) chemoradiation or chemotherapy alone; stage IV: palliative chemotherapy or targeted therapy.
statistical software package version 19.0 (SPSS, Chicago, IL) and STATA version 12.1 (STATA, College Station, TX).
RESULTS
Patient Characteristics Baseline characteristics according to PE status of the 2,061 patients in the study population are shown in Table 3. Median age of patients was 67.0 years, and there were 547 women (26.6%). According to histopathology, 48.7% of patients had an adenocarcinoma. Minimal PE presented in 272 patients (13.2%), and there was no PE in 1,397 (67.8%). Deaths were observed in 1,721 patients (83.5%). The proportion of patients presenting with minimal PE was seen to increase according to stage: 5.2% (20) in stage I, 10.9% (16) in stage II, 13.2% (30) in stage IIIA, 23.8% (59) in stage IIIB, and 13.9% (147) in stage IV (2 test P ⬍ .001). Patients with minimal PE had shorter survival compared with those with no PE (median survival time, 7.7 v 17.7 months, respectively; log-rank P ⬍ .001). Distribution of Variables Related to Patient, Stage Migration, Tumor, and Treatment and Their Impact on Overall Survival Distribution of the variables between no PE and minimal PE groups indicates no differences among sex, smoking habit, histology, EGFR mutation, or use of PET scans. However, lower CCI score; better ECOG PS (0 to 1); absence of weight loss; normal levels of hemoglobin, albumin, and alkaline phosphatase; smaller tumor size; www.jco.org
low N stage; small number of organs affected by metastasis; and curative treatment were more likely in the no PE group (2 test P ⬍ .001). All variables related to patient, stage migration, tumor, and treatment exerted significant effects on overall survival (Appendix Table A2, online only). Mechanisms Involved in Accumulation of Minimal PE and Their Impact on Overall Survival There were 270 patients with minimal PE in which a cause of accumulation was found during data review. Of these, 237 (87.8%) had a direct mechanism of invasion or attachment to pleura by tumor (Fig 2). For indirect mechanisms, 183 patients (67.8%) had tumor involvement with the mediastinal lymph nodes. Other mechanisms were observed in 18 patients (6.7%): pneumonia (10) and uncontrolled heart (4), kidney (3), and liver (1) diseases. Both direct and indirect mechanisms were found in 177 patients (65.5%). No difference was found in survival according to laterality of minimal PE (log-rank P ⫽ .63). Patients with direct mechanism had worse overall survival than patients without direct mechanism, when adjusted with all the variables (adjusted HR, 1.75; 95% CI, 1.06 to 2.91; Appendix Table A3, online only). Among patients with minimal PE, involvement with the mediastinal lymph nodes, heart and major vessels, or two or more numbers of causes tended to indicate worse survival, but differences were not significant. Status of PE and Overall Survival Patients with minimal PE showed significantly worse survival than those with no PE (unadjusted HR, 2.33; 95% CI, 2.04 to 2.65; © 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
5
Ryu et al
No. of Patients With Minimal Pleural Effusion
A
250 200
100
No. of Patients With Minimal Pleural Effusion No. of Patients With Minimal Pleural Effusion
77
50
30
Pleura Mediastinum Obstruction
Heart, Major Vessel
17
18
Lymphangitic Others Metastasis
200
177
150 100 59
50 0
C
183
150
0
B
237
34
Direct Only
100
Indirect Only
Both
91 80
75
75 50 20
25
4
0
One
Two
Three
Four
Five
Fig 2. Mechanisms involved in the accumulation of minimal pleural effusion by (A) causes, (B) direct and indirect mechanisms, and (C) number of causes (n ⫽ 270).
Table 1 and Fig 3). Presence of minimal PE was a consistent and significant factor for predicting worse overall survival at every step of the analysis when adjusted in a stepwise fashion, with each group of variables related to patient, stage migration, tumor, or treatment. In the final fully adjusted model, presence of minimal PE was an independent prognostic factor of worse survival when compared with the no PE group (adjusted HR, 1.40; 95% CI, 1.22 to 1.62; P ⬍ .001). The Harrell’s c-index was 0.77 (95% CI, 0.75 to 0.78) for assessing accuracy of the final model. From analysis of the impact of minimal PE on survival by stages, association of minimal PE with risk of death increased as a function of decrease in stage, from stages IV through I (adjusted HRs: 2.07 for stage I, 2.24 for stage II, 1.62 for stage IIIA, 1.57 for stage IIIB, and 1.16 for stage IV; Table 2). Effect of modification of PE status according to disease stage and its relationship to survival was analyzed, revealing that the impact of minimal PE was stronger in early-stage disease (Pinteraction ⫽ .001). When stage I with no PE was used as reference group in estimating risk of death, stage I with minimal PE showed higher risk compared with stage IIIA with no PE (adjusted HRs were 2.97 and 2.60, respectively; Appendix Table A4, online only). In patients with stage IV disease, risk of death increased with change of PE 6
© 2014 by American Society of Clinical Oncology
status (from no PE through malignant PE; adjusted HR, 1.20; 95% CI, 1.11 to 1.31; Ptrend ⬍ .001). When divided into M1a and M1b subgroups, this trend was maintained (Appendix Table A5, online only; Appendix Fig A1, online only). For patients with malignant PE, there were no differences in overall survival between patients with or without pleural malignant cells, pleural nodules, or no diagnostic work-up (P ⫽ .18; data not shown). DISCUSSION
In this study, we report that minimal PE usually presents at diagnosis and is an independent prognostic factor of worse survival among patients with NSCLC. Although there have been no epidemiologic studies, previous studies have reported that malignant PE is present in 11% to 32% of patients with advanced NSCLC.5-8 However, no study has reported the proportion of patients with minimal PE in which thoracentesis and/or pleural biopsy was not recommended. In this study, we found that 13.2% of patients with NSCLC in the study population present with minimal PE. This high proportion may be due to the meticulous attention paid to detecting minimal PE in this study, but it also agrees with a recent increasing trend toward adenocarcinoma (48.7% in this study). The presence of minimal PE should be of concern regarding delayed and inaccurate staging of the disease or with respect to morbidity or mortality related to more invasive approaches. There are several explanations for PE accumulation.9-12 Postmortem studies have suggested that tumor involvement of mediastinal lymph nodes is essential for PE accumulation.10,13 By contrast, Light et al12 claimed that this hypothesis does not fully explain exudative PE or the presence of malignant cells in most cases of malignant PE, and they postulated direct invasion of pleura as the primary reason, or in combination with indirect mechanisms. Direct mechanism was seen in 87% of patients with minimal PE and both direct and indirect mechanisms in 65%. These findings suggest that minimal PE represents an early phase in the development of malignant PE and support the combination hypothesis.12 In addition, minimal PE also occurs in severe comorbid diseases. The CCI score is a validated method for evaluating individual comorbidity and is well known to have an effect on treatment decision and survival of patients with NSCLC.14-16 Patients with minimal PE had higher CCI score compared with those in the no PE group in this study. When minimal PE represented an early phase of malignant PE or had developed as a result of severe comorbid diseases, it is reasonable to assume that these could impair survival. In this study, 17 variables related to patient, stage migration, tumor, and treatment were considered as potential confounders. It is important to know whether consistency of the impact of minimal PE was maintained at each step in the stepwise analysis of adjusting for each group of variables. We demonstrated that having minimal PE was consistently predictive of worse survival, and the minimal PE group showed a 40% increase in risk of death when compared with the no PE group. The prognostic impact of minimal PE can be deduced from a recent meta-analysis. Lim et al17 reported that pleural lavage cytology positive for tumor cells during surgical JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Studies of Minimal Pleural Effusion in NSCLC
0.8
95% CI 16.0 to 19.4 6.4 to 9.0 4.3 to 6.7
0.6 0.4 0.2
B
1.0
1
0.6 0.4 0.2
2
3
4
5
0
No PE Minimal PE
MST 17.7 10.6
95% CI 14.0 to 21.4 4.7 to 16.5
0.8 0.6 0.4 0.2
E
2
3
4
5
1
2
3
No PE Minimal PE
4
5
95% CI 19.6 to 31.9 8.3 to 12.5
0.6 0.4 0.2
0
1
2
3
4
5
Time (years)
MST 14.5 7.8
95% CI 12.5 to 16.5 5.7 to 10.2
0.8 0.6 0.4 0.2
0
MST 25.8 10.3
0.8
F
1.0
MST No PE 8.1 Minimal PE 5.6 Malignant PE 5.5
0.8
95% CI 7.0 to 9.2 4.2 to 6.9 4.3 to 6.7
0.6 0.4 0.2 P < .0001
P < .0001
Time (years)
No PE Minimal PE
P = .0151
1.0
P = .0003
1
1.0
Time (years)
Overall Survival (probability)
Overall Survival (probability)
1.0
0
C
P < .0001
Time (years)
D
95% CI 63.3 to 96.1 8.3 to 47.5
0.8
P < .0001
0
MST 77.9 27.9
No PE Minimal PE
Overall Survival (probability)
MST 17.7 7.7 5.5
No PE Minimal PE Malignant PE
Overall Survival (probability)
1.0
Overall Survival (probability)
Overall Survival (probability)
A
1
2
3
Time (years)
4
5
0
1
2
3
4
5
Time (years)
Fig 3. Overall survival of the total study population by (A) status of pleural effusion (PE) and by disease stage and status of pleural effusion at (B) stage I, (C) stage II, (D) stage IIIA, (E) stage IIIB, and (F) stage IV. MST, median survival time.
resection is an independent factor in predicting worse survival of patients with a resectable-stage tumor. We also found that the effect of minimal PE on survival varies according to disease stage, with greatest effect in early disease. This indicates that the impact could be maximized in patients with earlystage disease through tumor stage shifting. However, as expected, the proportion of patients with minimal PE was lower in early-stage disease compared with advanced stage disease. As PE status changed from no PE through malignant PE in patients with stage IV disease, risk of death in minimal or malignant PE gradually increased. Patients with malignant PE had larger amounts of effusion, and this could have had a prognostic impact. This is supported by a recent study in patients with advanced-stage ovarian carcinomas, demonstrating that a moderate-to-large PE is an independent prognostic factor for predicting worse survival when compared with a small PE.18 In addition, there is evidence that increased tumor burden within the pleural space is associated with worse survival among patients with malignant PE.19,20 This study has several limitations. First, environmental factors, such as education level, socioeconomic status, access to care, insurance status, and quality of treatment or diagnosis, could be important and should be considered in identifying more robust prognostic markers.21,22 Unfortunately, most of these factors could not be considered in this study because it was based on a cohort of patients with lung cancer that did not include such environmentally related items. However, factors related to host, tumor, and treatment were considered in detail in this study. Second, our study www.jco.org
population was enrolled over 9 years. Therefore, advances in treatment or supportive care and the possibility of stage migration by PET could have been subject to selection bias in analyzing prognostic effects. However, enrollment was limited to patients staged with contrast-enhanced chest CT scans, bone scans, and brain imaging to maintain homogeneity of the population. Interestingly, we found no difference in stage-specific survival based on whether PET was used or not (data not shown). Third, the natural course of minimal PE should be valuable in understanding its nature and significance. However, strict follow-ups with chest CT scan or pleural fluid analysis, when fluid had increased in amount, were not attempted aggressively in this study. This was largely because the study design was retrospective in nature and partially because of the shorter survival of patients: 43% of the patients received either no treatment or palliative treatment alone. Treatment response, treatment-related untoward effects, comorbid diseases, or pattern of tumor progression can also complicate the natural course of disease. In our study, we paid meticulous attention to evaluating the mechanism of pleural fluid accumulation through reviewing chest CT scans, bronchoscopic examinations, and laboratory data and noting comorbid diseases at diagnosis. Finally, validation is needed in another data set before conclusions from this study can be considered as a subset of TNM stage criteria. Nevertheless, this is a systematic study examining the proportion of patients with minimal PE, the mechanisms involved in PE accumulation, and its prognostic impact among patients with cancer. © 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
7
Ryu et al
AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The author(s) indicated no potential conflicts of interest.
AUTHOR CONTRIBUTIONS Conception and design: Jeong-Seon Ryu, Hyo Jin Ryu Financial support: Jeong-Seon Ryu
REFERENCES 1. Goldstraw P, Crowley J, Chansky K, et al: The IASLC Lung Cancer Staging Project: Proposals for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM Classification of malignant tumours. J Thorac Oncol 2:706714, 2007 2. Light RW: Pleural Diseases (ed 4). Philadelphia, PA, Lippincott Williams and Wilkins, 2001 3. American Thoracic Society: Management of malignant pleural effusions. Am J Respir Crit Care Med 162:1987-2001, 2000 4. National Comprehensive Cancer Network: NCCN Clinical Practice Guidelines in Oncology, Non-Small Cell Lung Cancer-Version 2.2014 http:// www.nccn.org/professionals/physician_gls/pdf/ nscl.pdf 5. Sugiura S, Ando Y, Minami H, et al: Prognostic value of pleural effusion in patients with non-small cell lung cancer. Clin Cancer Res 3:47-50, 1997 6. Morgensztern D, Waqar S, Subramanian J, et al: Prognostic impact of malignant pleural effusion at presentation in patients with metastatic non-smallcell lung cancer. J Thorac Oncol 7:1485-1489, 2012 7. William WN Jr, Lin HY, Lee JJ, et al: Revisiting stage IIIB and IV non-small cell lung cancer:
Administrative support: Jeong-Seon Ryu, Azra Memon, Seul-Ki Lee, Hyun-Jung Kim Provision of study materials or patients: Jeong-Seon Ryu, Hyun-Jung Kim Collection and assembly of data: Jeong-Seon Ryu, Hyo Jin Ryu, Si-Nae Lee, Seul-Ki Lee, Hae-Seong Nam, Hyun-Jung Kim, Jae-Hwa Cho Data analysis and interpretation: Jeong-Seon Ryu, Azra Memon, Kyung-Hee Lee, Seung-Sik Hwang Manuscript writing: All authors Final approval of manuscript: All authors
Analysis of the surveillance, epidemiology, and end results data. Chest 136:701-709, 2009 8. Naito T, Satoh H, Ishikawa H, et al: Pleural effusion as a significant prognostic factor in nonsmall cell lung cancer. Anticancer Res 17:47434746, 1997 9. Sahn SA: The pathophysiology of pleural effusions. Annu Rev Med 41:7-13, 1990 10. Meyer PC: Metastatic carcinoma of the pleura. Thorax 21:437-443, 1966 11. Leff A, Hopewell PC, Costello J: Pleural effusion from malignancy. Ann Intern Med 88:532-537, 1978 12. Light RW, Hamm H: Malignant pleural effusion: Would the real cause please stand up? Eur Respir J 10:1701-1702, 1997 13. Chernow B, Sahn SA: Carcinomatous involvement of the pleura: An analysis of 96 patients. Am J Med 63:695-702, 1977 14. Asmis TR, Ding K, Seymour L, et al: Age and comorbidity as independent prognostic factors in the treatment of non small-cell lung cancer: A review of National Cancer Institute of Canada Clinical Trials Group trials. J Clin Oncol 26:54-59, 2008 15. Williams CD, Stechuchak KM, Zullig LL, et al: Influence of comorbidity on racial differences in receipt of surgery among US veterans with earlystage non-small-cell lung cancer. J Clin Oncol 31: 475-481, 2013
16. Charlson ME, Pompei P, Ales KL, et al: A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis 40:373-383, 1987 17. Lim E, Clough R, Goldstraw P, et al: Impact of positive pleural lavage cytology on survival in patients having lung resection for non-small-cell lung cancer: An international individual patient data metaanalysis. J Thorac Cardiovasc Surg 139:1441-1446, 2010 18. Mironov O, Ishill NM, Mironov S, et al: Pleural effusion detected at CT prior to primary cytoreduction for stage III or IV ovarian carcinoma: Effect on survival. Radiology 258:776-784, 2011 19. Sahn SA, Good JT Jr: Pleural fluid pH in malignant effusions: Diagnostic, prognostic, and therapeutic implications. Ann Intern Med 108:345349, 1988 20. Heffner JE, Nietert PJ, Barbieri C: Pleural fluid pH as a predictor of survival for patients with malignant pleural effusions. Chest 117:79-86, 2000 21. Herndon JE 2nd, Kornblith AB, Holland JC, et al: Patient education level as a predictor of survival in lung cancer clinical trials. J Clin Oncol 26:4116-4123, 2008 22. Oh DY, Choi KS, Shin HR, et al: Public awareness of gastric cancer risk factors and disease screening in a high risk region: A population-based study. Cancer Res Treat 41:59-66, 2009
■ ■ ■
8
© 2014 by American Society of Clinical Oncology
JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Studies of Minimal Pleural Effusion in NSCLC
Acknowledgment The corresponding author thanks Bong Wan Noh, MD; Jun Hyeok Lim, MD; Jung Min Lee, MD; and Gwang Seok Yoon, MD, for their generous support and cooperation. Appendix
Table A1. Classification of Mechanisms Involved in Accumulation of Minimal PE Mechanism
Classification
Direct Indirect
Tumor invasion or broad attachment to pleura Tumor involvement of mediastinal lymph node Complete obstruction in one or more lobar or main bronchus or atelectasis in more than one lobe Significant tumor infiltration or compression to heart or major vessels (superior vena cava, main trunk of pulmonary artery or vein) or significant pulmonary thromboembolism Lymphangitic metastasis to more than one lobe Pneumonia Uncontrolled heart, kidney, thyroid and liver disease
Other
Abbreviation: PE, pleural effusion.
www.jco.org
© 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
9
Ryu et al
Table A2. Prognostic Variables and Overall Survival: Univariable Analysis Variable PE No Minimal Malignant Age per year Sex Male Female Smoking habit Never Past Current CCI score 0 1 ⱖ2 ECOG PS 0-1 ⱖ2 Weight loss, % None to ⬍ 5 ⱖ5 Hemoglobin, g/dLⴱ ⱖ 12 ⬍ 12 Albumin, g/dLⴱ ⱖ 3.1 ⬍ 3.1 Alkaline phosphatase, IU/Lⴱ ⱕ 335 ⬎ 335 Calcium, mg/dLⴱ ⱕ 10.8 ⬎ 10.8 PET Yes No Histology SQC ADC Others EGFR mutation Yes No Not tested Stage† I II IIIA IIIB IV Tumor size, cm 0-3 3.1-5 5.1-7 7.1-16 Unmeasurable
MST (months)
95% CI
HR
95% CI
17.7 7.7 5.5
16.0 to 19.4 6.4 to 9.0 4.3 to 6.7
1 2.32 2.80 1.03
2.04 to 2.65 2.48 to 3.17 1.03 to 1.04
11.1 17.6
10.1 to 12.1 15.2 to 20.0
1 0.77
0.69 to 0.86
19.4 10.6 11.6
16.7 to 22.2 9.1 to 12.2 10.4 to 12.7
1 1.43 1.34
1.25 to 1.64 1.18 to 1.52
16.5 13.0 6.8
13.9 to 19.1 11.7 to 14.2 5.7 to 8.0
1 1.24 2.02
1.11 to 1.38 1.76 to 2.31
17.4 5.9
15.8 to 19.1 5.0 to 6.7
1 2.46
2.22 to 2.74
17.4 7.9
15.7 to 19.1 7.1 to 8.6
1 1.85
1.67 to 2.04
14.8 7.8
13.6 to 15.9 6.7 to 8.8
1 1.55
1.40 to 1.72
14.3 3.9
13.3 to 15.3 2.7 to 5.2
1 2.47
2.15 to 2.85
13.3 6.5
12.3 to 14.4 5.0 to 7.9
1 1.64
1.42 to 1.89
12.7 2.7
11.7 to 13.6 0.1 to 5.3
1 2.65
1.94 to 3.62
14.4 10.9
12.9 to 15.8 9.8 to 12.1
1 1.20
1.10 to 1.33
12.0 14.3 7.9
10.7 to 13.4 13.0 to 15.7 5.5 to 10.4
1 0.98 1.23
0.89 to 1.09 1.04 to 1.46
21.9 11.7 12.2
14.8 to 29.0 9.8 to 13.6 11.0 to 13.3
1 1.40 1.31
1.12 to 1,75 1.08 to 1.59
75.5 25.1 16.8 12.2 6.8
59.3 to 91.7 20.9 to 29.3 14.3 to 19.2 10.2 to 14.3 6.1 to 7.6
1 2.07 3.45 5.23 7.64
1.59 to 2.68 2.79 to 4.27 4.27 to 6.42 6.42 to 9.10
29.0 14.5 8.8 5.6 9.2
25.3 to 32.8 12.8 to 16.2 7.5 to 10.0 3.9 to 7.3 7.1 to 11.4
1 1.73 2.67 3.24 2.59
1.50 to 1.99 2.29 to 3.12 2.72 to 3.86 2.17 to 3.09
(continued on following page)
10
© 2014 by American Society of Clinical Oncology
JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
Studies of Minimal Pleural Effusion in NSCLC
Table A2. Prognostic Variables and Overall Survival: Univariable Analysis (continued) Variable N stage N0 N1 N2 N3 No. of organs affected by metastasis 0 1 2 ⱖ3 Treatment No Yes
MST (months)
95% CI
HR
95% CI
27.7 16.7 10.1 6.1
24.3 to 31.1 13.1 to 20.2 8.7 to 11.5 5.3 to 6.8
1 1.41 2.45 3.87
1.18 to 1.69 2.16 to 2.79 3.41 to 4.39
17.6 6.8 5.0 2.7
16.0 to 19.2 5.6 to 8.1 3.7 to 6.3 2.0 to 3.4
1 2.38 2.89 4.15
2.11 to 2.68 2.46 to 3.41 3.37 to 5.12
7.0 17.0
6.1 to 7.9 15.4 to 18.7
1 0.53
0.48 to 0.58
Abbreviations: ADC, adenocarcinoma; CCI, Charlson comorbidity index; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; MST, median survival time; PE, pleural effusion; PET, positron emission tomography; SQC, squamous cell carcinoma. ⴱ Dichotomized by cutoff point of normal value. †Clinical stage at the time of initial diagnosis was determined according to the seventh edition of the TNM classification.
Table A3. Mechanisms Involved in Accumulation of PE and Overall Survival in Patients With Minimal PE (n ⫽ 270) Mechanism Direct Pleura Yes No Indirect Mediastinum Yes No Obstruction Yes No Heart, major vessels Yes No Lymphangitic metastasis Yes No No. of causes 1 ⱖ2
aHRⴱ
95% CI
Cox P
1 1.75 1
1.06 to 2.91
.03
1.47 1
0.41 to 5.19
.55 —
.61
0.88 —
0.61 to 1.27 —
.48 —
⬍ .01
1 1.18
— 0.73 to 1.91
— .49
4.4 to 11.1 5.8 to 9.2
.62
1 1.67
— 0.90 to 3.10
— .10
6.2 to 14.5 5.2 to 8.3
⬍ .01
1 1.46
0.89 to 2.40
.14
MST (months)
95% CI
Log-Rank P
7.4 10.4
6.1 to 8.8 3.1 to 17.7
.06
6.7 9.5
5.2 to 8.3 6.5 to 12.5
⬍ .01
7.4 7.6
4.5 to 10.4 5.8 to 9.4
5.4 8.0
3.1 to 7.8 6.4 to 9.5
7.7 7.5 10.4 6.7
Abbreviations: aHR, adjusted hazard ratio; HR, hazard ratio; MST, median survival time; PE, pleural effusion. ⴱ HR after adjustment with sex, age, smoking habit, Charlson comorbidity index score, Eastern Cooperative Oncology Group performance status, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium, histology, epidermal growth factor receptor mutation, tumor size, N stage, No. of organs affected by metastasis, positron emission tomography, and treatment.
www.jco.org
© 2014 by American Society of Clinical Oncology
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
11
Ryu et al
Table A4. Adjusted HRs for Mortality on Minimal PE Modified by Stages No PE Stage
HR
I II IIIA IIIB IV
1.00 (reference) 1.54 2.60 3.53 5.14
Minimal PE 95% CI
HR
95% CI
1.11 to 2.14 1.89 to 3.59 2.64 to 4.72 3.85 to 6.79
2.97 2.70 3.92 5.52 5.55
1.65 to 5.35 1.46 to 5.01 2.41 to 6.38 3.80 to 8.02 4.02 to 7.65
NOTE. HRs were estimated after adjustment with sex, age, smoking habit, Charlson comorbidity index score, Eastern Cooperative Oncology Group performance status, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium, histology, epidermal growth factor receptor mutation, positron emission tomography, and treatment. Abbreviations: HR, hazard ratio; PE, pleural effusion.
Table A5. Overall Survival and Status of PE Within Subgroups of Stage IV (M1a and M1b): Cox Proportional Hazards Modeling Results Stage IV M1a
Variable
Minimal PE
Malignant PE
No PE (ref)
HR
95% CI
HR
1 1 1
1.54 1.30 1.58
1.13 to 2.10 0.93 to 1.82 1.15 to 2.16
1.47 1.36 1.56
1 1 1
1.26 1.51 1.52
0.91 to 1.75 1.11 to 2.06 1.12 to 2.07
1 1 1
1.45 1.53 1.33
1.06 to 1.98 1.12 to 2.09 0.92 to 1.91
Unadjusted Patient-related variables Demographics: age, sex, smoking habit Tumor burden: ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium Comorbidity: CCI score Stage migration: PET Tumor-related variables: histology, EGFR mutation, tumor size N stage, No. of organs affected by metastasis Treatment: no v yesⴱ Final fully adjusted model†
Stage IV M1b Minimal PE
Malignant PE
No PE (ref)
HR
95% CI
HR
1.18 to 1.82 .001 1.06 to 1.73 .019 1.24 to 1.95 ⬍ .001
1 1 1
1.31 1.21 1.26
1.07 to 1.61 0.97 to 1.50 1.02 to 1.55
1.50 1.46 1.68
1.25 to 1.80 ⬍ .001 1.19 to 1.80 ⬍ .001 1.39 to 2.03 ⬍ .001
1.18 1.53 1.32
0.94 to 1.50 .170 1.23 to 1.91 ⬍ .001 1.07 to 1.62 .001
1 1 1
1.20 1.25 1.32
0.97 to 1.49 1.02 to 1.54 1.07 to 1.62
1.25 1.50 1.50
1.03 to 1.53 .020 1.25 to 1.80 ⬍ .001 1.25 to 1.80 ⬍ .001
1.64 1.47 1.66
1.30 to 2.06 ⬍ .001 1.18 to 1.83 .001 1.26 to 2.20 ⬍ .001
1 1 1
1.20 1.36 1.11
0.97 to 1.48 1.11 to 1.68 0.90 to 2.03
1.33 1.48 1.44
1.10 to 1.62 ⬍ .001 1.23 to 1.78 ⬍ .001 1.15 to 1.79 .002
95% CI
Ptrend
95% CI
Ptrend
NOTE. All HRs and 95% CIs were estimated by using patients without PE in each stage as ref. Abbreviations: CCI, Charlson comorbidity index; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio; PE, pleural effusion; PET, positron emission tomography; ref, reference. ⴱ First-line treatments for patients with stage I to IIIA: surgery alone, (neo) adjuvant chemotherapy, or concurrent (or sequential) chemoradiation; stage IIIB: concurrent (or sequential) chemoradiation or chemotherapy alone; stage IV: palliative chemotherapy or targeted therapy. †The model included sex, age, smoking habit, CCI score, ECOG PS, weight loss, hemoglobin, albumin, alkaline phosphatase, calcium, histology, EGFR mutation, tumor size, N stage, No. of organs affected by metastasis, PET, and treatment.
1.0
MST 10.6 8.8 7.4
No PE Minimal PE Malignant PE
0.8
95% CI 7.7 to 13.4 7.3 to 10.3 6.0 to 8.7
0.6 0.4 0.2
B Overall Survival (probability)
Overall Survival (probability)
A
1.0 0.8
MST 7.0 4.9 4.1
95% CI 6.0 to 8.1 3.7 to 6.2 3.2 to 5.0
0.6 0.4 0.2 P < .0001
P = .0008
0
No PE Minimal PE Malignant PE
1
2
3
Time (years)
4
5
0
1
2
3
4
5
Time (years)
Fig A1. Overall survival and status of pleural effusion (PE) within subgroups of stage IV: (A) M1a and (B) M1b. MST, median survival time. 12
© 2014 by American Society of Clinical Oncology
JOURNAL OF CLINICAL ONCOLOGY
Information downloaded from jco.ascopubs.org and provided by at University of Brighton on July 15, 2014 from Copyright © 2014 American Society of Clinical Oncology. All rights reserved. 194.81.203.94
