Original Article
Varying Recurrence Rates and Risk Factors Associated With Different Definitions of Local Recurrence in Patients With Surgically Resected, Stage I Nonsmall Cell Lung Cancer John M. Varlotto, MD1; Abram Recht, MD2; John C. Flickinger, MD3; Laura N. Medford-Davis, BA4; Anne-Marie Dyer, MS5; and Malcolm M. DeCamp, MD6
BACKGROUND: The objective of this study was to examine the effects of different definitions of local recurrence on the reported patterns of failure and associated risk factors in patients who undergo potentially curative resection for stage I nonsmall cell lung cancer (NSCLC). METHODS: The study included 306 consecutive patients who were treated from 2000 to 2005 without radiotherapy. Local recurrence was defined either as ‘‘radiation’’ (r-LR) (according to previously defined postoperative radiotherapy fields), including the bronchial stump, staple line, ipsilateral hilum, and ipsilateral mediastinum; or as ‘‘comprehensive’’ (c-LR), including the same sites plus the ipsilateral lung and contralateral mediastinal and hilar lymph nodes. All recurrences that were not classified as ‘‘local’’ were considered to be distal. RESULTS: The median follow-up was 33 months. The proportions of c-LR and r-LR at 2 years, 3 years, and 5 years were 14%, 21%, and 29%, respectively, and 7%, 12%, and 16%, respectively. Significant risk factors for c-LR on multivariate analysis were diabetes, lymphatic vascular invasion, and tumor size; and significant factors for r-LR were resection of less than a lobe and lymphatic vascular invasion. The proportions of distant (nonlocal) recurrence using these definitions at 2 years, 3 years, and 5 years were 10%, 12%, and 18%, respectively, and 14%, 19%, and 29%, respectively. Significant risk factors for distant failure were histology when using the c-LR definition and tumor size when using the r-LR definition. CONCLUSIONS: Local recurrence increased nearly 2-fold when a broad definition was used instead of a narrow definition. The definition also affected which factors were associated significantly with both local and distant failure on multivariate analysis. Comparable definitions must be used when analyzing different C 2010 American Cancer Society. series. Cancer 2010;116:2390–400. V KEYWORDS: lung cancer, disease recurrence, stage I, surgical resection.
Lung cancer is the second most common cancer in both men and women, occurring in 92,700 and 87,770 individuals in 2006, respectively.1 However, it remains the number 1 cancer killer in both sexes, resulting in 90,330 and 81,770 deaths, respectively.1 Surgery remains the standard of care for patients with stage I and II nonsmall cell lung cancer (NSCLC) and for some patients with stage III disease. Approximately 20% of patients who present with NSCLC are candidates for potentially curative resection.2 Understanding the patterns of failure after surgery is critical to assessing whether adjuvant therapies might be of benefit. However, there is no standard definition for ‘‘local recurrence.’’ Different investigators have defined local recurrence to include different sites, such as the ipsilateral mediastinum3; the bronchial stump, ipsilateral hilum, and/or mediastinum4,5; the ipsilateral hilum, mediastinum, and/or primary site6; the ipsilateral lung and/or mediastinum7-9; and anywhere inside the thorax.10 Local recurrence was not defined explicitly in 3 phase 3 trials of adjuvant radiotherapy,11-13 although 1 trial reported that adjuvant radiotherapy reduced local recurrences in the irradiated areas among patients in the experimental arm.12
Corresponding author: John M. Varlotto, MD, Penn State Milton S. Hershey Medical Center, 500 University Avenue, Cancer Institute-Radiation Oncology, H063, Hershey, PA 17033; Fax: (717) 531-0882;
[email protected] 1 Division of Radiation Oncology, Penn State Hershey Cancer Institute, Hershey, Pennsylvania; 2Department of Radiation Oncology, Beth Israel Deaconess Medical Center, Boston, Massachusetts; 3Department of Radiation Oncology, Pittsburgh Cancer Institute, Pittsburgh, Pennsylvania; 4Harvard Medical School, Boston, Massachusetts; 5Department of Public Health Sciences, Penn State University, Hershey, Pennsylvania; 6Division of Cardiothoracic Surgery, Robert H. Lurie Comprehensive Cancer Center, Beth Israel Deaconess Medical Center, Boston, Massachusetts
DOI: 10.1002/cncr.25047, Received: July 21, 2009; Revised: September 11, 2009; Accepted: September 14, 2009, Published online March 11, 2010 in Wiley InterScience (www.interscience.wiley.com)
2390
Cancer
May 15, 2010
LR in Stage I NSCLC After Resection/Varlotto et al
The objective of our current investigation was to demonstrate how using 2 different definitions of local recurrence (1 broader than the other) affected the observed patterns of failure of patients with resected stage I NSCLC. We demonstrate that the choice of definition has a substantial impact on the observed incidence of both local and distant recurrence and on which factors are associated significantly with these events.
MATERIALS AND METHODS This retrospective study was conducted with approval of the institutional review boards of the Beth Israel Deaconess Medical Center (Boston, Mass) and the Denver and Boston Department of Veterans Affairs (VA) hospitals. Hospital and departmental tumor registries were used to identify all patients who underwent potentially curative resection of stage I through IIIA NSCLC from 2000 to 2005. Patients were included in the study population if they had postoperative follow-up 3 months, no second primary cancer diagnosed within 5 years of the index lung cancer, and no neoadjuvant or adjuvant radiotherapy. Medical records were reviewed for each patient to ensure eligibility. Two hundred twenty-nine of 402 patients who potentially were available for our study at Beth Israel Deaconess Medical Center were eligible. Corresponding rates of eligible and available patients from the Denver and Boston VA hospitals were 62 of 94 patients and 82 of 116 patients, respectively. Of the 373 eligible patients, 306 patients had stage I disease and were included in our current investigation. The pathologic stage was IA for 193 patients (63%) and IB for 113 patients (37%). Twenty-one patients (7%) received chemotherapy (2 patients received neoadjuvant chemotherapy, and 19 patients received adjuvant chemotherapy) with carboplatin and paclitaxel (14 patients), cisplatin and navelbine (1 patient), cisplatin and etoposide (2 patients), other cisplatin-containing regimens (2 patients), and other regimens (2 patients). Patients who received chemotherapy included 7 patients (4%) with stage IA disease and 14 patients (12%) with stage IB disease. Multiple factors that potentially were related to outcome were abstracted from the available medical records. The patient-related and staging factors that were recorded were age, weight loss, preoperative hemoglobin, body mass index (BMI), presenting symptoms (cough, hemoptysis, dyspnea, or none), pulmonary function test results (forced expiratory volume in 1 second [FEV1], FEV1%, Cancer
May 15, 2010
and carbon monoxide diffusion capacity percentage), smoking status(never, quit for >2 years, quit for 1-2 years, quit for 6-12 months, quit 30 days to 6 months before surgery, or currently smoking), drinking of alcoholic beverages (never, quit >2 years ago, quit 1-2 years ago, quit 6 months to 1 year ago, quit 30 days to 6 months ago, currently drinking), number of drinks per week, chronic use of steroids, history of intercurrent illnesses (diabetes, hypertension, myocardial infarction, coronary artery disease, cardiac rhythm disturbance, coronary artery bypass graft, cerebrovascular accidents, renal failure, thromboembolic disease), performance of staging positron emission tomography (PET)-computed tomography (CT) images at diagnosis, medication use at diagnosis (aspirin, statins, and nonsteroidal anti-inflammatory drugs), and hemoglobin and albumin values within 30 days before surgery. Surgery-related factors that were recorded included operative time, fluid balances during surgery (urine output, estimated blood loss, fluids infused, total excess intraoperative fluids [assuming that 1 unit of packed erythrocytes ¼ 300 mL]); transfusions (total number of units required within 90 days of surgery, number of units required within the first day, number of units required within 2-4 days, and number of units required after 4 days), type of resection (wedge resection, segmentectomy, lobectomy/bilobectomy, pneumonectomy), type of lymph node staging procedure, intraoperative and postoperative complications (pulmonary infection or pneumonia, adult respiratory distress syndrome, arrhythmias and their type [supraventricular tachycardia, ventricular tachycardia, or nonspecific], fistula formation, stump leak, prolonged air leak, anastomosis leak, mediastinitis/ empyema, myocardial infarction, death, pulmonary embolus, deep venous thrombosis, and brachial plexus injury), length of hospital stay, and surgeon. Histopathologic factors that were recorded included tumor size, tumor grade (categorized as grade ‘‘1.5’’ when recorded as well to moderately differentiated and grade ‘‘2.5’’ when recorded as moderately to poorly differentiated), performance/positivity of previous fine-needle aspiration biopsy, lymphatic or vascular invasion (LVI), perineural invasion, margin status, extracapsular lymph node involvement, details of lymph node resection (number resected, lymph node level resected, number of N1 lymph nodes resected, number of N2 lymph nodes resected, number of lymph node stations sampled at mediastinoscopy, number of lymph node stations examined during resection), histology (NSCLC not otherwise specified [NSCLC-NOS], squamous cell, adenocarcinoma, large cell carcinoma,
2391
Original Article
neuroendocrine carcinoma, bronchioloalveolar carcinoma, adenocarcinoma with bronchioloalveolar features), tumor lobe location, and invasion of the visceral/parietal pleura. Details of chemotherapy recorded included type of chemotherapy (carboplatin and paclitaxel, cisplatin and navelbine, cisplatin and etoposide, other cisplatin-containing regimens, gemcitabine-containing combinations, other regimen), treatment sequencing (giving chemotherapy preoperatively, postoperatively, or both), and the number of cycles. Patterns of failure were determined using physician clinical assessment, radiographic reports, and/or review of imaging studies. Patients were diagnosed with a lymph node recurrence when a new or enlarging lymph node that measured >1 cm in the short axis was identified on follow-up CT scans. For patients who underwent PET scanning at the time of recurrence, all sites of abnormal uptake that correlated with a lymph node or soft tissue mass were scored, regardless of size. When available, data from bronchoscopy, biopsy, or mediastinoscopy were used to supplement the radiographic findings. Only the initial sites of recurrence were scored.14 Patients with simultaneous local and distant failures were scored as having both types of failure for both bivariate and multivariate analyses. Failure information generally was taken from the assessment of the individual patient’s treating or following physician but was reviewed by 2 of the authors (J.M.V. and L.N.M.-D.) to ensure accuracy of their assessments with radiologic, bronchoscopic, and/or pathologic studies. CT scans were available for review from the Boston VA hospital and from the Beth Israel Deaconess Medical Center. Because all follow-up visits were at established cancer programs with appropriate subspecialists, >98% of all follow-up visits were accompanied by CT or CT/PET scans. Local failure was defined in 2 different ways. A ‘‘comprehensive’’ local recurrence (c-LR) included any recurrence within the ipsilateral lung, bronchial stump, staple line, and the N1–N3 nodal groups (Fig. 1a). A ‘‘radiation’’ local recurrence (r-LR) was restricted only to those sites that typically would be included within the postoperative radiotherapy fields defined by Trodella and colleagues,10 namely, the bronchial stump, the staple line, the ipsilateral hilum, and the ipsilateral mediastinum (Fig. 1b). All failures that were not ‘‘local’’ were defined as ‘‘distant recurrence’’ and were designated as comprehensive distant recurrence (c-DR) and radiation distant recurrence (r-DR) in relation to the definitions for c-LR and rLR, respectively. Therefore, all failures that were not
2392
Figure 1. These diagrams depict the definitions of (a) comprehensive local recurrence and (b) radiation local recurrence.
defined as r-LR were classified as r-DR, and all failures that were not defined as c-LR were classified as c-DR. For example, contralateral mediastinal lymph nodes (N3) would be classified as distal recurrence (r-DR) using the rLR definition but would be classified as a local recurrence (c-LR) using the c-LR definition. Ipsilateral pulmonary lesions were classified as a local recurrence if the involved physicians were certain that this tumor was not a new primary. In equivocal cases, we considered the tumor a second primary cancer and censored the patient from our study group. The distinction between a new primary cancer or a metastasis was established as described previously.15,16 Patients were classified as having multiple primary lesions if both lung tumors had a different histology or if the histology was the same but the second lesion originated from a carcinoma in situ or was located in a different lung or lobe with no evidence of lymphatics common to both and with no extrapulmonary metastases at the time of diagnosis. Recurrence and survival estimates at 2 years, 3 years, and 5 years were calculated by using the Kaplan-Meier method.17 Bivariate proportional hazards regression analyses were run to assess associations between the risk factors, local and distal recurrence, and survival. Those variables with a bivariate significance of P < .10 were included in a multivariate regression model, and backwards selection was run with a selection criteria of P < .05.18
RESULTS The median length of follow-up was 33 months (range, 4-98 months). The characteristics of the population with regard to selected patient-related, treatment-related, and Cancer
May 15, 2010
LR in Stage I NSCLC After Resection/Varlotto et al
Table 1. Patient Factors and Patterns of Failure
No. of Patients (%) Variable
All Patients
All patients Median age, y Median Hgb, g/dL
a
c-LR
r-LR
c-DR
r-DR
Both c-LR and c-DR
Both r-LR and r-DR
306 (100) 70.5 13.50
47 (15) 69.0 13.40
25 (8) 69.0 13.30
21 (7) 69.0 12.60
52 (17) 70.5 13.30
18 (6) 73.5 13.50
9 (3) 73.0 13.50
20 87 102 54 43
2 12 18 11 4
BMI, kg/m2 30 Unknown
(7) (28) (33) (18) (14)
(10) (14) (18) (20) (9)
2 6 9 6 2
(10) (7) (9) (11) (5)
2 8 5 3 3
(10) (9) (5) (6) (7)
2 16 18 9 7
(10) (18) (18) (17) (16)
0 3 9 2 4
(0) (3) (9) (4) (9)
0 1 5 1 2
(0) (1) (5) (2) (5)
Cough No Yes
213 (70) 93 (30)
33 (16) 14 (15)
19 (9) 6 (6)
11 (5) 10 (11)
30 (14) 22 (24)
11 (5) 7 (8)
6 (3) 3 (3)
267 (87) 39 (13)
43 (16) 4 (10)
24 (9) 1 (3)
16 (6) 5 (13)
42 (16) 10 (26)
15 (6) 3 (8)
8 (3) 1 (3)
20 133 40 86 27
3 23 3 14 4
2 14 3 5 1
2 20 7 19 4
1 10 2 4 1
1 6 1 1 0
Hemoptysis No Yes
Smoking Never Quit ‡2 y Quit
