Original Article

Analysis of Prognostic Factors in EwingSarcoma Using a Population-Based CancerRegistry

Joe Lee, MD1; Bang H. Hoang, MD2; Argyrios Ziogas, PhD3; and Jason A. Zell, DO, MPH4

BACKGROUND: Ewing sarcoma is a high-grade malignancy that most often occurs in children. Because its occur- rence in adults has been historically low, few studies have been published on the epidemiology of Ewing sarcoma in this group of patients. By using data from a large, population-based cancer registry, the authors designed the present study to examine the outcome of children and adult patients with Ewing sarcoma and relevant prognostic factors. METHODS: A retrospective analysis of Ewing sarcoma patient cases in the California Cancer Registry database was performed to identify incident patient cases diagnosed between 1989-2007. Comparisons were made to examine dif- ferences in demographics, disease characteristics, treatment, and survival. Survival analyses were performed using Kaplan-Meier method with log-rank tests and Cox proportional hazards models. RESULTS: Seven hundred and twenty-five incident patient cases of Ewing sarcoma were identified, including 372 (51.3%) children and 353 (48.7%) adults. Hispanic race was associated with young age (P .001) and lower socioeconomic status (SES; P .0001). Pelvic involvement was associated with large tumor size (>8 cm; P < .0001), an increased incidence of metastasis (P < .0002), and poorer survival (P < .0001). After adjusting for clinically relevant factors, statistically significant decreased overall survival was seen with adults (hazard ratio [HR], 1.71; 95% confidence interval [CI], 1.35-2.17), His- panics (HR, 1.33; 95% CI, 1.01-1.75), metastatic disease (HR, 2.74; 95% CI, 2.14-3.49), large tumor size (HR, 1.65; 95% CI,1.17-2.34), no surgical treatment, and low SES. CONCLUSIONS: The authors determined that adult age, Hispanic race, metastatic disease, large tumor size, and low SES are poor prognostic factors for overall survival among Ewing sar- coma patient cases. Cancer 2010;116:196473. VC 2010 American Cancer Society.

KEYWORDS: Ewing sarcoma, epidemiology, survival, cancer.

Ewing sarcoma refers histologically to a collection of small, round cell tumors and clinically manifests with a variety of presentations.1,2 Despite the substantial amount of work achieved in the last few decades to improve the treatment and survival for Ewing sarcoma patients, little is known about the epidemiology of the disease.3,4 This is especially true when studying adult patients, as Ewing sarcoma has historically been seen mostly in children and adolescents. Our current understanding about the epidemiology of Ewing sarcoma comes mainly from relatively small, retrospective series.1,3,5-7The population-based studies that have been done focused on pediatric patients, or report only on incidence and survival, without analyses on outcome or prognostic factors.4,8-11In 1985, population-based cancer reporting became required in the state of California. This established the Califor- nia Cancer Registry (CCR), which is now recognized as one of the leading cancer registries in the world.12-15 The size and diversity of the California population make the CCR particularly valuable to study the epidemiology of rare diseases such as Ewing sarcoma. By using the CCR, we designed the present study to examine the outcome of children and adult patients with Ewing sarcoma and to determine the relevant prognostic factors for survival.

Corresponding author: Jason A. Zell, DO, MPH, Department of Epidemiology, University of California, Irvine, 224 Irvine Hall, Irvine, CA 92697-7550; Fax: (949)824-1343; jzell@uci.edu1Department of Orthopaedic Surgery, University of California at Irvine, Irvine, California; 2Department of Orthopaedic Surgery, Chao Family Comprehensive Cancer Center, University of California at Irvine, Irvine, California; 3Department of Epidemiology, School of Medicine, and the Genetic Epidemiology Research Institute, University of California at Irvine, Irvine, California; 4Division of Hematology/Oncology, Department of Medicine, Chao Family Comprehensive Cancer Center, Department of Epidemiology, School of Medicine, and the Genetic Epidemiology Research Institute, University of California at Irvine, Irvine, California

Presented in part at the Musculoskeletal Tumor Society Annual Meeting, November 6-9, 2008, Phoenix, Arizona.

DOI: 10.1002/cncr.24937, Received: April 4, 2009; Revised: June 28, 2009; Accepted: July 31, 2009, Published online February 11, 2010 in Wiley InterScience(www.interscience.wiley.com)

Cancer1964April 15, 2010

Cancer1964April 15, 2010

Original Article

Prognostic Factors in Ewing Sarcoma/Lee et al

MATERIALS AND METHODSStudy PopulationWe performed a retrospective, case-only analysis of Ewing sarcoma patient cases in the CCR database. CCR is part of the National Cancer Institutes Surveillance, Epidemiology, and End Results (SEER) program, and is the largest contiguous-area, population-based, cancer registry in the world15; standardized data collection and quality control procedures have been in place since1988.12-16 Case reporting is estimated at 99% for theentire state of California,15 with follow-up completion rate greater than 95% for most tumors.15 Data were abstracted from medical and laboratory records bytrained tumor registrars.14 Tumor site and histology were coded according to World Health Organization criteria in International Classification of Diseases (ICD) for On-cology.16 Patient cases were identified and extracted from the CCR by using the ICD-O-3 histology code 9260 for Ewing sarcoma. We then excluded patient cases with thefollowing unknown or missing variables: nonspecific or unknown SEER extent of disease classification and patients whose diagnosis dates and follow-up dates were inconsistent.Data were obtained on 725 incident Ewing sarcoma patient cases during 1989-2007 in CCR. Patient cases were identified through high-quality reporting sources (hospital inpatient/outpatient centers, oncology treat- ment centers, laboratories, private practitioners, or nurs- ing home/convalescent home/hospice facilities). Recorded data included demographic information, pres- ence of metastasis, treatment during the first course of therapy, socioeconomic status (SES), and vital status. Adults were defined as patient cases age 18 and over at di- agnosis. SES is denoted as a single index variable in CCR using statewide measures of education, income, and occu- pation from census data, as described previously by Yost et al.17 The SES index used is based on principle compo- nent analysis of census block-level CCR data linked to census data assessing the following areas: education level, median household income, proportion below 200% pov- erty level, median house value, median rent, percentage employed, and percentage with blue-collar employ- ment.17-21 Of note, proportion below 200% poverty level was used because of its greater sensitivity and is defined as all those in poverty plus those who have income above poverty but less than 2 times their poverty thresh- old.22 Essentially, each aspect mentioned serves as a proxy for SES and is weighted together to derive at a score. Scores were divided into 5 categories, with the higher

scores representing higher SES. Quintiles for the SESscore were analyzed.Cause of death was recorded according to ICD crite- ria in effect at the time of death. Death from all causes was used to define overall survival (OS). Ewing-specific survival was defined as death from muskuloskeletal can- cer. Hospital registrars contacted patient cases annually, and CCR staff annually reviewed state death certificates to identify deceased registry patient cases. The last date of follow-up was either the date of death or the last date of contact. In 2006-2007, complete follow-up was available for 681 (94%) patients.

Statistical AnalysisThe clinical characteristics, including age, gender, race/ ethnicity, presence of metastasis, tumor size, anatomic tumor site, SES quintile, and treatment were analyzed with Pearson chi-square test or Fisher exact test for cate- gorical and dichotomous variables, and the nonparamet- ric Kruskall-Wallis test for comparison of continuous variables (age) for more than 2 groups. Life tables and Kaplan Meier curves were generated for each category, and curves were compared with the log-rank test. Multi- variate survival analysis was used to calculate overall sur- vival and Ewing-specific survival using Cox proportional hazards ratios. All statistical analyses were conducted using SAS 9.1 statistical software (SAS Insti- tute, Cary, NC). Statistical significance was assumed for a 2-tailed P < .05.

Ethical ConsiderationsThis study involved analysis of existing data from CCR database with no subject intervention. No identifiers were linked to subjects. This study was approved by the Uni- versity of California Irvine Institutional Review Board (IRB) under the category exempt status (IRB#2008-6181).

RESULTSDemographic DataSeven hundred and twenty-five patient cases of Ewing sar- coma were identified between 1989-2007. The demo- graphic and clinicopathologic data are shown in Table 1. The median age was 17, with 372 (51.3%) children and353 (48.7%) adults (age 18 and older). Male to female ra- tio of patient cases was 3 to 2. Distribution by race/ethnic- ity revealed 414 whites (57.1%), 241 Hispanics (33.2%),20 blacks (2.76%), 44 (6.1%) Asians/Pacific Islanders,

Table 1. Patient Characteristics and Univariate Analysis for Ewing Sarcoma Patient Cases 1989-2007a

ChildrenAdultsChi-SquareP for OSP for Ewing-SpecificSurvival

Sex0.21.93.21

Male217 (58.3)222 (62.9)

Female155 (41.7)131 (37.1)

Race/ethnicityWhite197 (53.0)217 (61.5)0.001.03.24

Hispanic146 (39.3)95 (26.9)Worse survivalWorse survival

Other29 (7.8)41 (11.6)

SES 0.36.002.005

Lowest81 (21.8)67 (19.0)Worse survivalWorse survival

Second lowest75 (20.2)73 (20.7)

Middle78 (21.0)83 (23.5)

High70 (18.8)52 (14.7)

Highest68 (18.3)78 (22.1)

Stage 0.18