• rhabdomyosarcoma;
  • survival;
  • population-based;
  • surveillance


  1. Top of page
  2. Abstract


Evidence from clinical trials has documented improvements in event-free survival from childhood rhabdomyosarcoma (RMS) since the 1970s; however, the survival experience of children enrolled on cancer clinical trials may not reflect the full range of patients treated in community settings. The current study evaluated 5-year survival and 10-year conditional survival for RMS from U.S. population-based cancer registry data.


Public-use data from the Surveillance, Epidemiology, and End Results (SEER) program were used in life table and Cox regression analyses to evaluate RMS survival by patient age at diagnosis, gender, tumor histology, tumor site and stage, and major treatment eras among 848 children who were age < 20 years at the time of diagnosis, were a resident of 1 of 9 geographic reporting regions, and were diagnosed between 1973 and 2000.


The 5-year survival probabilities were found to be highest for younger-age children (ages 1–4 years: 77%), patients with localized disease (83%), those whose tumors had an embryonal histology (67%), and patients with orbital (86%) and genitourinary (80%) tumor sites. Poor prognosis was associated with diagnosis during infancy (47%) and adolescence (48%); metastatic disease at the time of presentation (31%); alveolar histology (49%); and tumors of the extremities (50%), retroperitoneum (52%), and trunk (52%). Conditional 10-year survival probabilities among those who survived ≥ 5 years were 85% or higher. The probability of survival by stage at the time of diagnosis increased with each successive treatment era, suggesting a stage shift phenomenon over time.


Large variations in 5-year survival were evident depending on patient age and tumor characteristics. However, children who survived the first 5 years after diagnosis were found to have an excellent long-term prognosis. The patterns in RMS survival noted from the current population-based evaluation did not appear to differ substantially from those previously reported by major clinical trials. Cancer 2005. © 2005 American Cancer Society.

Rhabdomyosarcomas (RMS) are a heterogeneous group of malignancies of mesenchymal cell origin that arise primarily in striated muscle tissues. RMS accounts for approximately 3% of childhood malignancies, with approximately 350 new cases diagnosed each year among children ages ≤ 19 years in the U.S.1, 2 Results from clinical trials since the 1970s have documented increasing 5-year event-free survival for children and adolescents treated on risk-based treatment protocols by the Intergroup Rhabdomyosarcoma Study Group (IRSG).1 Clinical trials protocols, however, are not uniformly accessible to all pediatric and adolescent cancer patients, and some children who are eligible elect not to participate. Study inclusion criteria may restrict participation on the basis of extent of disease, previous history of treatment, comorbidity, insurance limitations, psychosocial considerations, and other factors. Variations in clinical trial participation by age group,3–8 cancer type,7, 8 institution,6, 7 and geographic region5 have been noted previously. Therefore, the survival experience among children on protocol may not reflect the wide range of patients who are treated in community settings. For example, two recent studies9, 10 of childhood acute lymphocytic leukemia (ALL) found that survival rates were similar by race/ethnicity for those treated in a specialized pediatric cancer treatment center10 whereas, in contrast, ALL survival was lower among minority children compared with white children based on population-based data, even in the modern treatment era.9

In light of these potential differences, we undertook an analysis of population-based data collected by the Surveillance, Epidemiology, and Ends Results (SEER) program of the National Cancer Institute to describe the survival of U.S. children diagnosed with RMS based on demographic characteristics; tumor site, histology, and stage; and major IRSG treatment eras.


  1. Top of page
  2. Abstract

Data Source and Study Population

The SEER program has been described in detail elsewhere.11 Briefly, since 1973, SEER program registries have collected demographic, diagnostic, and treatment data on all newly diagnosed cancer patients residing within specific U.S. geographic regions. Case reporting is mandated by law within each region and must meet completeness standards of at least 98%.12 The vital status of registrants is ascertained with active follow-up on all living patients and abstraction of death certificates whenever cancer is listed as a cause of death.11 Data for this analysis were obtained from the 2002 submission of the SEER Public-Use dataset13 and are based on the nine SEER geographic regions in operation since 1973. These 9 regions represent approximately 10% of the U.S. population, and are comprised of the states of Connecticut, Hawaii, Iowa, New Mexico, and Utah and the metropolitan areas of Detroit, Atlanta, San Francisco-Oakland, and Seattle-Puget Sound.

Children diagnosed with malignant, first primary RMS between 1973 and 2000 at ages ≤ 19 years were eligible for this analysis. Primary tumor sites and morphologies collected by SEER were coded according to the International Classification of Diseases for Oncology, Version 2 (ICD-O-2).14 Eligible cases were grouped according to the International Classification of Childhood Cancer (ICCC) coding system,15 which classifies pediatric cancers on the basis of morphology rather than tumor site. For this analysis, eligible RMS cases had ICCC code IXa, with ICD-O-2 morphology codes 8900, 8901, 8902, 8910, 8920, and 8991; only tumors coded as malignant were included in the study. RMS cases were excluded from the analysis if a diagnosis was based solely on an underlying cause of death (i.e., death certificate only) or the morphology was not confirmed microscopically.

Data Analysis

We used life table methods16 to compute survival function estimates of RMS survival probabilities, both overall and stratified by demographics and cancer characteristics that have been previously reported to predict survival.17, 18 Because of the relatively low frequency of competing causes of death in this young population,19, 20 we analyzed observed survival, which represents the cumulative probability of surviving through a specified time interval, without adjusting for other, noncancer causes of death.21 To evaluate the probability of long-term RMS survival, 10-year conditional survival19 was calculated using life table methods as the probability of surviving at least an additional 5 years given that subjects had already survived at least 5 years since their diagnosis of RMS.

Survival time (i.e., follow-up time) was measured in years, starting from the date of diagnosis and ending with either the date of death from any cause or on a censoring date from the date the patients was last known to be alive or December 31, 2000 (i.e., the data file cutoff date). Survival functions were compared based on the log-rank test for grouped data22 and the Greenwood method16 was used to calculate 95% confidence intervals (95% CI).

The demographic variables considered in the analyses were gender, race/ethnicity, and age at diagnosis. Age at diagnosis was assessed as a categoric variable with 4 levels: age < 1 year, ages 1–4 years, ages 5–9 years, and ages 10–19 years. Race/ethnicity variables, derived from the joint distribution of the SEER “Race recode A” and “Origin recode” variables, were classified as a categoric variable with two levels (nonwhite and white) and four levels (white, non-Hispanic; black, non-Hispanic; Hispanic; and other or unknown.) Diagnostic information included primary tumor site, histologic tumor subtype, stage at diagnosis, and treatment era. Primary RMS site codes were grouped into nine categories: orbit, parameningeal region, other head and neck sites, genitourinary system, bladder/prostate, extremity, retroperitoneum, trunk, and other or unknown sites. Histology was classified as embryonal, alveolar, and other histologic subtypes, including not otherwise specified (NOS). Treatment eras corresponded approximately to the years in which the first four major IRSG clinical trials were performed: 1973–1978, 1979–1984, 1985–1991, and 1992–2000. The most recent era includes the first 2 years of the fifth IRSG clinical trial.

Stage at diagnosis was based on the SEER historic staging system, which broadly defines the spread of disease at the time of diagnosis using categories that have been fairly constant over time.11 This scheme, which is not synonymous with IRS clinical group or staging systems, defines cancer stage as: 1) localized (confined entirely to the organ of origin), 2) regional (extension beyond the organ of origin directly into surrounding organs or tissues, lymph nodes via the lymphatic system, or a combination of tissue and lymphatic extension), 3) distant (spread to body remote from the primary tumor by direct extension or distant metastasis), and 4) unstaged (insufficient information to assign a stage). In our analyses, all localized and regional prostate RMS cases were collapsed into a single “localized/regional” category to reflect changes in the scheme for staging prostate tumors that occurred in 1988. Because the reporting of SEER historic stage by all registries of the nine SEER regions was not complete until 1975, our analyses of the effects of stage on prognosis were restricted to data from the years 1975–2000 for nonbladder/prostate RMS tumors.

Cox proportional hazard regressions23 were performed to assess the combined effects of the demographic and cancer-related covariates on survival. Predictor variables were modeled as indicator (dummy) variables with reference groups as indicated in the results. Likelihood ratio tests were used to assess the statistical significance of the predictor variables. The assumption of proportional hazards was confirmed by graphing the negative log of survival time versus the log of time for each covariate in the final model.24

Statistical Analysis Software (version 8.2; SAS Institute Inc., Cary, NC) was used to perform all analyses. All tests were two-sided and, by convention, a P value < 0.05 was considered statistically significant.


  1. Top of page
  2. Abstract

The SEER study base included 855 cases diagnosed with malignant, first primary RMS between 1973 and 2000. Approximately 99% of these subjects (848 cases) were included in the analyses, with 7 exclusions that included 1 case with a death certificate only, 2 cases in which the diagnoses were not microscopically confirmed, and 3 cases with an unknown method of diagnostic confirmation. The median age at the time of diagnosis was 7 years (range, birth–19 years) and survival time ranged from 0–25 years. As seen in Table 1, the majority of cases were male (60%) and white, non-Hispanic (73%) and were diagnosed before the age of 10 years (63%). Approximately 34% of cases had RMS primary tumors in the head or neck. The predominant histologic subtypes of RMS were embryonal (60%), alveolar (19%), and RMS, NOS (16%). Although not shown, the distribution of stage varied by tumor site, with proportionately more cases who had orbital RMS presenting with localized disease at the time of diagnosis and more cases with RMS of the retroperitoneum or “other/unknown” primary tumor sites presenting with distant disease at the time of diagnosis.

Table 1. Characteristics of 848 U.S. Children and Adolescents (Age < 20 Years) Diagnosed with RMS, SEER 1973–2000a
  • RMS: rhabdomyosarcoma; SEER: Surveillance, Epidemiology, and End Results program of the National Cancer Institute; ICD-O-2: International Classification of Diseases for Oncology, Version 2; NOS: not otherwise specified.

  • a

    Microscopcially confirmed, with International Classification of Childhood Cancer (ICCC) code IXa.

  • b

    1975–2000 rhabdomyosacoma diagnoses.

 White, non-Hispanic61873
 Black, non-Hispanic11914
Age at diagnosis (yrs)  
 < 1567
Vital status  
Histology (ICD-O-2 morphology code)  
 Embryonal (8910.3)50960
 Alveolar (8920.3)16219
 Rhabdomyosarcoma, NOS (8900.3)13616
 Pleomorphic (8901.3)132
 Mixed type (8902.3)91
 Embryonal sarcoma (8991.3)192
Primary tumor site group  
 Other head/neck sites13516
 Parameningeal region8310
 Genitourinary system12014
SEER historic stageb  
 Localized/regional (prostate only)132
Treatment era (based on year of diagnosis)  

Life Table Analyses

Table 2 displays the results of life table analyses at 5 years and 10 years after diagnosis. The overall probability of survival remained fairly constant at approximately 60% at both the 5-year and 10-year time points. There were no significant differences noted in the estimated survival functions by gender (P = 0.83) or race (P = 0.87). However, there were significant differences in the estimated probabilities of survival by age at diagnosis (P < 0.0001). Survival was consistently higher for children diagnosed between ages 1–4 years compared with infants (age < 1 year of age at the time of diagnosis) and children ages 10–19 years at the time of diagnosis (data not shown). The (unadjusted) 5-year and 10-year life table survival estimates for older children (ages 5–9 years) was similar to those for younger children (ages 1–4 years). Long-term prognosis was poor for infants or children ages 10–19 years at the time of diagnosis (< 50% at 10 years) compared with children who were ages 1–4 years or ages 5–9 years at the time of diagnosis (72% and 61%, respectively, at 10 years).

Table 2. 5-Year and 10-year Survival Probabilities for RMSa in Childhood and Adolescence, 1973–2000 SEER
 5-year10-year10-year conditionalb
  • RMS: rhabdomyosarcoma; SEER: Surveillance, Epidemiology, and End Results program of the National Cancer Institute; NA: not available.

  • a

    Microscopically confirmed, with International Classification of Childhood Cancer (ICCC) code IXa.

  • b

    Ten-year survival among children who survived at least 5 years from the time of the initial diagnosis of rhabdomyosarcoma.

  • c

    1975–2000 rhabdomyosarcoma diagnoses, excluding 56 cases of rhabdomyosarcoma in the prostate.

All cases0.61 (0.57–0.64)0.58 (0.54–0.61)0.95 (0.93–0.97)
 Male0.61 (0.56–0.65)0.59 (0.55–0.64)0.97 (0.95–0.99)
 Female0.61 (0.55–0.66)0.56 (0.50–0.62)0.92 (0.88–0.97)
 White, non-Hispanic0.61 (0.57–0.65)0.58 (0.54–0.62)0.95 (0.92–0.97)
 Black, non-Hispanic0.57 (0.48–0.67)0.57 (0.48–0.67)NA
 Hispanic0.55 (0.40–0.69)0.55 (0.40–0.69)NA
 Others/unknown0.68 (0.55–0.81)0.59 (0.44–0.73)NA
Age at diagnosis (yrs)   
 <10.47 (0.33–0.60)0.47 (0.33–0.60)NA
 1–40.77 (0.72–0.83)0.72 (0.66–0.78)0.94 (0.90–0.98)
 5–90.65 (0.58–0.72)0.61 (0.54–0.68)0.94 (0.90–0.99)
 10–190.48 (0.42–0.53)0.46 (0.40–0.52)0.97 (0.93–1.00)
Primary tumor site   
 Head and neck sites0.69 (0.63–0.75)0.65 (0.59–0.70)0.94 (0.90–9.98)
  Orbit0.86 (0.77–0.94)0.84 (0.75–0.93)0.98 (0.94–1.00)
  Other head/neck0.67 (0.59–0.75)0.59 (0.50–0.68)0.88 (0.80–0.96)
  Parameningeal region0.58 (0.47–0.69)0.57 (0.45–0.68)0.97 (0.92–1.00)
 Nonhead and neck sites0.60 (0.56–0.65)0.58 (0.53–0.62)0.96 (0.93–0.98)
  Genitourinary0.80 (0.72–0.87)0.80 (0.72–0.87)1.00 (1.00–1.00)
  Bladder/prostate0.66 (0.53–0.78)0.66 (0.53–0.78)1.00 (1.00–1.00)
  Extremity0.50 (0.41–0.60)0.44 (0.34–0.54)0.87 (0.77–0.98)
  Retroperitoneum0.52 (0.43–0.60)0.49 (0.40–0.57)0.94 (0.87–1.00)
  Trunk0.52 (0.37–0.67)0.49 (0.34–0.64)0.95 (0.85–1.00)
 Other/unknown0.37 (0.26–0.48)0.37 (0.26–0.48)1.00 (1.00–1.00)
 Embryonal0.67 (0.63–0.72)0.64 (0.60–0.69)0.96 (0.93–0.98)
 Alveolar0.49 (0.41–0.58)0.42 (0.34–0.51)0.86 (0.76–0.96)
 Others0.52 (0.44–0.59)0.52 (0.44–0.59)NA
SEER historic stagec   
 Localized0.83 (0.78–0.88)0.79 (0.74–0.85)0.95 (0.92–0.99)
 Regional0.66 (0.59–0.72)0.62 (0.55–0.69)0.94 (0.90–0.99)
 Distant0.31 (0.24–0.38)0.28 (0.21–0.35)0.90 (0.81–0.99)
 Other/unknown0.52 (0.38–0.66)0.52 (0.38–0.66)NA
Treatment era   
 1973–19780.47 (0.39–0.55)0.43 (0.36–0.51)0.92 (0.85–0.98)
 1979–19840.64 (0.57–0.71)0.60 (0.52–0.67)0.93 (0.89–0.98)
 1985–19910.64 (0.58–0.70)0.62 (0.56–0.68)0.97 (0.94–1.00)
 1992–20000.64 (0.58–0.71)NANA

The probability of survival also varied significantly according to tumor histology, stage at diagnosis, and primary tumor site. Throughout follow-up, the probability of survival was lowest for children with alveolar RMS compared with those diagnosed with either embryonal RMS or the other histologic RMS subtypes (P < 0.0001). Only 42% of children diagnosed with alveolar RMS survived at least 10 years after diagnosis compared with 60% of children diagnosed with embryonal RMS and approximately 50% of children with other histologic subtypes of RMS (Table 2). Distant metastases at the time of diagnosis imparted a significantly poorer prognosis than local or regional stage disease, with an estimated 30% (95% CI, 23–36%) of children with distant disease surviving 5 years after the time of diagnosis compared with 82% (95% CI, 72–87%) of children with localized stage disease and 65% (95% CI, 59–71%) of children with regional stage disease. The prognosis for children diagnosed with RMS in the head and neck region was generally better than that for children diagnosed with RMS in other anatomic sites (i.e., nonhead and neck sites and other or unknown sites). The 5-year survival was highest for children diagnosed with orbital and genitourinary RMS and lowest for children diagnosed with RMS of the extremities, retroperitoneum, trunk, and other or unknown sites. Less than half of this latter group of children had survived at least 10 years after diagnosis (Table 2) (Fig 1A, 1B).

thumbnail image

Figure 1. (A) Rhabdomyosarcoma survival among children and adolescents age < 20 years at the time of diagnosis by tumor site–orbit, head and neck, parameningeal region, genitourinary, and bladder/prostate (Surveillance, Epidemiology, and End Results [SEER] program 1973–2000). (B) Rhabdomyosarcoma survival among children and adolescents age < 20 years at the time of diagnosis by tumor site–extremity, retroperitoneum, trunk, and other/unknown (SEER 1973–2000).

Download figure to PowerPoint

Survival function curves by treatment era show that the probability of survival increased substantially between the first two treatment eras, but there was no evidence of significant improvement in survival occurring in any of the subsequent treatment periods (Table 2). However, stage-specific survival increased with time (Table 3). For example, the 5-year probability of surviving localized disease increased from 75% in 1975–1978 to 92% in 1992–2000. Similarly, the probabilities of survival from regional and distant disease generally increased. Furthermore, although not statistically significant, the distribution of stage at the time of diagnosis by treatment era demonstrates that, over the study period, the proportion of cases with localized disease tended to decrease whereas the proportion of cases with regional and distant disease increased slightly (Table 3), suggesting a stage shift with time (chi-square test for trend, 1 degree of freedom = 2.2, P = 0.14).

Table 3. 5-Year Survival Probablitiesa by Disease Stage at Diagnosis and Treatment Era, SEER 1975–2000
Treatment eraLocalizedRegionalDistant
No (%)Survival (95% CI)No. (%)Survival (95% CI)No. (%)Survival (95% CI)
  • SEER: Surveillance, Epidemiology, and End Results program of the National Cancer Institute 95% CI: 95% confidence interval.

  • a

    Excludes 56 cases of bladder-prostate rhabdomyosarcoma.

1975–197836 (40)0.75 (0.61–0.89)30 (33)0.45 (0.27–0.63)24 (27)0.13 (0.00–0.26)
1979–198460 (42)0.81 (0.71–0.91)48 (33)0.65 (0.51–0.78)36 (25)0.36 (0.20–0.52)
1985–199166 (35)0.81 (0.72–0.91)67 (35)0.72 (0.61–0.82)58 (30)0.33 (0.21–0.45)
1992–200088 (34)0.92 (0.85–0.98)97 (37)0.69 (0.57–0.80)75 (29)0.33 (0.21–0.44)

Cox Regression Models

As shown in Table 4, children diagnosed between ages 10–19 years had a significantly higher mortality rate than children diagnosed between ages 1–9 years, after adjusting for the other variables in the model. The all-cause mortality rate also was found to be higher among children with alveolar RMS compared with those with embryonal RMS (P < 0.01), and among children with regional (P = 0.0002) and distant disease (P < 0.0001) compared with those with localized disease. Children with RMS in the nonorbital head and neck regions, extremities, and all other sites combined were found to have higher mortality rates than those diagnosed with genitourinary RMS (Table 4). Finally, the all-cause death rate was significantly lower among those diagnosed in treatment eras subsequent to 1975–1978, after adjusting for all the other prognostic variables in the model. The variables for race or ethnicity and gender were not found to be significantly associated with survival time and were not included in the final model.

Table 4. Multivariate Risk of Death among Children and Adolescents with RMS, SEER 1975–2000
VariableRR95% CIP value
  1. RMS: rhabdomyosarcoma; SEER: Surveillance, Epidemiology, and End Results program of the National Cancer Institute; RR: relative risk; 95% CI: 95% confidence interval.

Age at diagnosis (yrs)   
 1–40.50.3–0.7< 0.001
Histologic subtype   
 Alveolar1.51.1–2.1< 0.01
Primary tumor site group   
 Genitourinary system1.0Reference 
 Head/neck sites1.91.1–3.30.01
 Parameningeal region1.81.0–3.30.04
 Other/unknown3.11.8–5.5< 0.0001
Stage at diagnosis   
 Distant4.83.4–6.9< 0.0001
Treatment era   
 1985–19910.50.3–0.7< 0.0001
 1992–20000.40.3–0.6< 0.0001

Long-Term Survival

Table 2 shows that, among those in the SEER cohort who survived at least 5 years, the overall and stratified 10-year conditional survival probabilities were at least 90% for most prognostic factors, except for cases diagnosed with RMS of the extremities (87.5%) and other head or neck sites (88%), and alveolar RMS (85.9%). The 5-year conditional survival probabilities differed only slightly between males and females (97% vs. 92%; P = 0.15).


  1. Top of page
  2. Abstract

The current analysis evaluated survival among patients who were newly diagnosed with childhood RMS from a population-based national data registry. These results provide an analysis that reflects the full spectrum of RMS and the full spectrum of children with RMS. Unlike most clinical studies, the current analyses included all pediatric RMS cases residing within a specific geographic region, including both those treated and not treated on formal protocols, regardless of comorbidity, disease stage, or other prognostic or personal factors.

We found age at diagnosis, stage of disease, histologic tumor subtype, and primary tumor site to be important predictors of RMS survival. A good prognosis was found to be associated with younger age at diagnosis (i.e., ages 1–9 years), localized disease, embryonal RMS, and orbital and genitourinary primary tumor sites, whereas poor prognosis appeared to be associated with diagnosis during infancy (age <1 year) and adolescence (ages 10–19 years); distant or metastatic disease; alveolar histology; and tumors of the extremities, retroperitoneum, trunk, and other or unknown primary tumor sites. These findings are consistent with previous clinical investigations.17, 18

Differential survival by age at diagnosis has been noted previously in both clinical25 and population-based analyses.26 These differences reflect, at least in part, underlying age-related biologic differences in RMS, as evidenced by the differing distributions of disease characteristics among infants (age < 1 year), children (ages 1–9 years), and adolescents (ages ≥ 10 years).25 Although adolescents tend to be diagnosed with specific characteristics (e.g., alveolar histology, RMS of the extremities, and distant disease at the time of diagnosis) that are correlated with poor prognosis,26 Joshi et al. found that survival rates were lower among adolescents, even after adjusting for these adverse prognostic factors.25 The low participation rates of older adolescents in clinical trials in the U.S. (approximately 20%3) also might account for some of the reduced survival observed on a population level in this underserved age group.27 Biologic differences with regard to drug metabolism, immune system function, and susceptibility to adverse radiation effects might contribute to age-related RMS survival differentials, especially among infants.25 Our finding that survival between younger (ages 1–4 years) and older (ages 5–9 years) children differed only slightly is consistent with the theory that there are 3 age-related risk categories.

Consistent with the results of previous clinical trials, we found no substantial differences with regard to RMS survival by minority status.28 Baker et al. conducted a retrospective analysis of RMS survival by race and ethnicity using data from the IRS-III (1984–1991) and IRS-IV (1991–1997) clinical trials, and they also found that race was not predictive of survival.28

Although 5-year survival estimates from univariate life table analyses remained unexpectedly constant at approximately 64% from 1985 onward, the long-term trends in RMS survival appeared promising in analyses that adjusted for the effects of other prognostic variables. In life table analyses stratified by disease stage, the probability of 5-year survival increased with each successive treatment era, and in multivariate Cox regression analyses, the risk of all-cause mortality was found to decrease with treatment era after adjustment for the effects of disease stage, tumor histology, patient age at the time of diagnosis, and tumor site group. We also found that the prospects of long-term survival were good, both overall and stratified by various prognostic factors, with a probability of surviving 10 years of > 85% reported among those who had already survived at least 5 years after their diagnosis. In a recent study, Sung et al. found low rates of late adverse events (death, disease recurrence, or second malignant neoplasms) among childhood RMS survivors who had a failure-free survival period of at least 5 years after successful treatment.29 Survivors at the greatest risk of failure after 10 years were those who had large primary tumors or advanced disease at the time of diagnosis.

We note that temporal trends in survival over long follow-up periods are difficult to interpret because, along with changes in treatment, there are changes in the criteria used to classify and characterize the disease (e.g., pathology and clinical and pathologic staging systems), in the methods and technologies used to diagnose and treat patients, in the relative distributions of prognostic factors within a group of cases (e.g., case mix),30, 31 and in supportive care for cancer patients. Therefore, trends in survival reflect more than just the effects of improved treatment approaches.

Evaluating long-term trends in survival is further complicated by shifts in the overall distribution of disease stage as a result of more accurate diagnostic techniques and characterizations of the extent of disease at the time of diagnosis occurring during recent treatment eras. As Feinstein et al.32 observed, advances in diagnostic technologies and methods have enabled the detection of microscopic tumor extensions and metastases, which over time results in proportionately fewer patients being diagnosed with localized disease and more patients being diagnosed with regional and distant stages of disease in the later versus earlier eras. Consequently, survival probabilities stratified by stage of disease can erroneously suggest an improvement in survival with time, even though overall survival has not increased during the follow-up period. Stage-specific survival rates reported during the later eras will show improvement because cases with microscopic disease extensions and metastases have been shifted out of the group with localized disease (thus improving the survival rates for cases with localized disease) into the group with overt regional and distant disease (thus improving the overall prognosis for cases with regional and distant disease).32 Consistent with clinical findings,31 the current analyses showed that such a shift in the stage distribution of RMS was likely to occur during the follow-up of the cohort of RMS cases, but without detailed clinical information regarding how stage was assessed, we are unable to evaluate the impact of stage migration on survival definitively.

The selection of participants for randomized clinical trials can limit the generalizability of study results and lead to inconsistent findings between studies, as mentioned previously. However, such concern may not be as pertinent to pediatric RMS as for other cancer types because RMS has been the focus of national, collaborative research efforts since approximately 1972. We found that the overall patterns and trends in survival noted in our population-based evaluation did not differ substantially from those previously reported by the IRSG. Given the high participation rates (i.e., approximately 75–85%) of pediatric patients in some prior IRSG trials,33, 34 and in spite of the differences between clinical and registry coding systems, methods, and conventions,11 the concordance of findings from these two widely different study designs might suggest the close coordination of specialized care that is provided to pediatric RMS patients in the U.S.. Alternatively, at least some of the similarity noted in the findings might be explained by an overlap of the nine SEER geographic regions on which this study was based with the geographic locations of the facilities in which IRSG clinical trials were implemented.

RMS manifests wide biologic heterogeneity in children, making this cancer difficult to diagnose and subtype. Unlike IRSG clinical trial operations, the SEER program does not employ a central review of diagnoses (pathology, primary tumor site, extent of disease), increasing the likelihood of disease misclassification. Furthermore, the RMS coding and staging systems utilized in IRSG clinical studies differ from those in the SEER program. Therefore, comparative analyses must be made with caution. The advantage of using SEER data in these analyses is that the patients represent an unselected sample, and the data are of high quality and are collected in a uniform manner with uniform data standards. Through the routine implementation of data quality control and improvement programs, the SEER Program has documented a relatively low rate of errors associated with the coding and abstracting of data.

The current analysis of population-based data demonstrates that, despite the poor prognosis of certain subgroups, the overall survival rates for RMS are relatively high and have been improving with time. Analyses of 5-year conditional survival indicate that long-term survivors have an excellent chance of a cure but, as recent research suggests, survivors will require ongoing medical surveillance for many years after their diagnosis, given that the majority of late deaths among long-term survivors is due to the original RMS or the diagnosis of a second malignant neoplasm.35


  1. Top of page
  2. Abstract
  • 1
    Wexler LH, Crist WM, Hilman LJ. Rhabdomyosarcoma and the undifferentiated sarcomas. In: PizzoPA, PoplackDG, editors. Principles and practices of pediatric oncology. Philadelphia: Lippincott Williams & Wilkins, 2001: 939971.
  • 2
    Gurney JG, Young JL, Roffers SD, Smith MA, Bunin GR. Soft tissue sarcomas. In: RiesLAG, SmithMA, GurneyJG, et al., editors. Cancer incidence and survival among children and adolescents: United States SEER Program 1975–1995, National Cancer Institute, SEER Program. NIH Pub. No. 99-4649. Bethesda, 1999: 111124. Available from URL: [accessed July 22, 2004].
  • 3
    Bleyer WA. Cancer in older adolescents and young adults: epidemiology, diagnosis, treatment, survival, and importance of clinical trials. Med Pediatr Oncol. 2002; 38: 110.
  • 4
    Bleyer WA, Tejeda H, Murphy SB, et al. National cancer clinical trials: children have equal access; adolescents do not. J Adolesc Health. 1997; 21: 366373.
  • 5
    Ross JA, Severson RK, Pollock BH, Robison LL. Childhood cancer in the United States. A geographical analysis of cases from the Pediatric Cooperative Clinical Trials groups. Cancer. 1996; 77: 201207.
  • 6
    Shochat SJ, Fremgen AM, Murphy SB, et al. Childhood cancer: patterns of protocol participation in a national survey. CA Cancer J Clin. 2001; 51: 119130.
  • 7
    Bernstein L, Sullivan-Halley J, Krailo MD, Hammond GD. Trends in patterns of treatment of childhood cancer in Los Angeles County. Cancer. 1993; 71: 32223228.
  • 8
    Krailo MD, Bernstein L, Sullivan-Halley J, Hammond GD. Patterns of enrollment on cooperative group studies. An analysis of trends from the Los Angeles County Cancer Surveillance Program. Cancer. 1993; 71: 33253330.
  • 9
    Kadan-Lottick NS, Ness KK, Bhatia S, Gurney JG. Survival variability by race and ethnicity in childhood acute lymphoblastic leukemia. JAMA. 2003; 290: 20082014.
  • 10
    Pui CH, Sandlund JT, Pei D, et al. Results of therapy for acute lymphoblastic leukemia in black and white children. JAMA. 2003; 290: 20012007.
  • 11
    National Cancer Institute. Surveillance, Epidemiology and End Results [website]. Available from URL: [accessed July 22, 2004].
  • 12
    National Cancer Institute. Surveillance, Epidemiology and End Results [website]. Available from URL: [accessed July 22, 2004].
  • 13
    National Cancer Institute. SEER cancer incidence public-use database, 1973–2000. November 2002 submission [CD ROM]. April 2003. Available from URL: [accessed July 22, 2004].
  • 14
    Percy C, Van Holten V, Muir C. International classification of diseases for oncology. 2nd edition. Geneva: World Health Organization, 1990.
  • 15
    Kramarova E, Stiller CA. The international classification of childhood cancer. Int J Cancer. 1996; 68: 759765.
  • 16
    Lawless JF. Life tables, graphs, and related procedures. In: LawlessJF. Statistical models and methods for lifetime data. New York: John Wiley & Sons, 1982: 5281.
  • 17
    Rodary C, Gehan EA, Flamant F, et al. Prognostic factors in 951 nonmetastatic rhabdomyosarcoma in children: a report from the International Rhabdomyosarcoma Workshop. Med Pediatr Oncol. 1991; 19: 8995.
  • 18
    Raney RB, Anderson JR, Barr FG, et al. Rhabdomyosarcoma and undifferentiated sarcoma in the first two decades of life: a selective review of Intergroup Rhabdomyosarcoma Study Group experience and rationale for Intergroup Rhabdomyosarcoma Study V. J Pediatr Hematol Oncol. 2001; 23: 215220.
  • 19
    Henson DE, Ries LA. On the estimation of survival. Semin Surg Oncol. 1994; 10: 26.
  • 20
    Myers MH, Ries LA. Cancer patient survival rates: SEER program results for 10 years of follow-up. CA Cancer J Clin. 1989; 39: 2132.
  • 21
    Berrino F, Micheli A, Sant M, Capocaccia R. Interpreting survival differences and trends. Tumori. 1997; 83: 916.
  • 22
    Kalbfleisch JD, Prentice RL. The statistical analysis of failure time data. New York: John Wiley & Sons, 1980.
  • 23
    Cox DR. Regression models and life tables. J R Stat Soc [B]. 1972; 34: 187220.
  • 24
    Cantor AB. Proportional hazards regression. In: CantorAB. SAS survival analysis techniques for medical research. 2nd edition. Cary, NC: SAS Institute, Inc., 2003: 136138.
  • 25
    Joshi D, Anderson JR, Paidas C, Breneman J, Parham DM, Crist W. Age is an independent prognostic factor in rhabdomyosarcoma: a report from the Soft Tissue Sarcoma Committee of the Children's Oncology Group. Pediatr Blood Cancer 2004; 42: 6473.
  • 26
    Stiller CA, Stevens MC, Magnani C, Corazziari I, Group EW. Survival of children with soft-tissue sarcoma in Europe since 1978: results from the EUROCARE study. Eur J Cancer. 2001; 37: 767774.
  • 27
    Michelagnoli MP, Pritchard J, Phillips MB. Adolescent oncology-a homeland for the “lost tribe.” Eur J Cancer. 2003; 39: 25712572.
  • 28
    Baker KS, Anderson JR, Lobe TE, et al. Children from ethnic minorities have benefited equally as other children from contemporary therapy for rhabdomyosarcoma: a report from the Intergroup Rhabdomyosarcoma Study Group. J Clin Oncol. 2002; 20: 44284433.
  • 29
    Sung L, Anderson JR, Donaldson SS, Spunt SL, Crist WM, Pappo AS. Late events occurring five years or more after successful therapy for childhood rhabdomyosarcoma: a report from the Soft Tissue Sarcoma Committee of the Children's Oncology Group. Eur J Cancer 2004; 40: 18781885.
  • 30
    Berrino F, Esteve J, Coleman MP. Basic issues in estimating and comparing survival of cancer patients. In: BerrinoF, SantM, VerdecchiaA, CapocacciaR, HakulinenT, EsteveJ, editors. Survival of cancer patients in Europe. IARC Scientific Pub. No. 132. Lyon, France: IARC Scientific Publications, 1995: 113.
  • 31
    RiesLAG, EisnerMP, KosaryCL, et al., editors. SEER Cancer Statistics Review, 1975–2001, National Cancer Institute. Bethesda, 2004. Available from URL: [accessed July 22, 2004].
  • 32
    Feinstein AR, Sosin DM, Wells CK. The Will Rogers phenomenon. Stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer. N Engl J Med. 1985; 312: 16041608.
  • 33
    Pappo AS. Rhabdomyosarcoma and other soft tissue sarcomas of childhood. Curr Opin Oncol. 1995; 7: 361366.
  • 34
    Ruymann FB, Grufferman S. Introduction and epidemiology of soft tissue sarcomas. In: MaurerHM, RuymannFB, PochedlyC, editors. Rhabdomyosarcoma and related tumors in children and adolescents. Boca Raton, FL: CRC Press, Inc., 1991: 318.
  • 35
    Mertens AC, Yasui Y, Neglia JP, et al. Late mortality experience in five-year survivors of childhood and adolescent cancer: the Childhood Cancer Survivor Study. J Clin Oncol. 2001; 19: 31633172.