The 5-year cancer survival rate for all children 0 to 19 years old who were diagnosed between 2001 and 2007 has increased to 82.6% in the United States, as estimated by the SEER program.1
Nonetheless, we found continued disparities in survival between black and white patients registered in the SEER program, beginning in an earlier treatment period (1992 to 2000) and persisting through a recent period (2001 to 2007). This trend appears to include all childhood cancers, with the exception of nephroblastoma, germ cell tumors, osteosarcoma, rhabdomyosarcoma, and certain brain tumors (). By contrast, the same analysis applied to data from our pediatric cancer center indicated similar survival rates for black and white patients, regardless of treatment period and cancer type, possibly excluding ependymoma and retinoblastoma in the earlier period and nonrhabdomyosarcoma soft tissue sarcoma in the recent period ().
The well-recognized lower incidence of ALL, brain tumors, germ cell tumors, hepatoblastoma, and especially Ewing sarcoma and melanoma in black compared with white children25
was readily apparent in the racial distribution by diagnosis in the SEER data and in our own patient population ( and ). Less clear was the higher proportion of black children in certain diagnostic categories within our cohort compared with the SEER database. One explanation is that SJCRH has a different catchment area and is a tertiary referral center with a policy of accepting patients regardless of their ability to pay. Any comparison of outcome data between the SEER program and SJCRH must take into account the severity of the cancer at diagnosis. Because the acceptance policy of our institution is based mainly on protocol availability, and because many of our protocols are designed for cancers that are difficult to treat, our study population would be expected to comprise disproportionately large groups of high-risk patients. For example, with the development of the Pediatric Brain Tumor Consortium in 1999, we have not only attracted a higher proportion of patients with brain tumors compared with the SEER program (29% v
< .01) but we also have an increased referral of patients with high-grade glioma. Among the 254 patients with high-grade glioma in our institution, 109 (42.9%) had brainstem glioma and 77 (30.0%) had glioblastoma multiforme—diseases associated with a dismal prognosis—with survival rates ranging from 10% to 20%.21
The proportion of patients with either subtype of brain tumor is not apparent in the SEER program but is almost certainly lower than that of our patient population because of the relatively high 5-year survival rate. In this regard, brain and other nervous system tumors have been difficult to diagnose pathologically, especially when classified as high-grade glioma.26,27
All of our patients included in the high-grade glioma category had their histologic diagnosis confirmed at SJCRH before treatment, excluding patients with diffuse pontine glioma, for which the diagnosis was made by imaging and clinical criteria, because biopsy was not performed in these patients. The overall survival of our patients with high-grade glioma is consistent with results recently reported by the COG.28,29
Thus, direct comparison of outcome for high-grade gliomas in SEER versus SJCRH patients would not yield reliable results. Similarly, of our 66 patients with nonrhabdomyosarcoma soft tissue sarcoma, 20 (30%) presented with metastatic disease, a proportion substantially higher than that of other reported series.30,31
Of the four black patients, two presented with metastatic disease, and one each had unresectable or high-grade tumor. Hence, the poor outcome is not entirely unexpected in this small cohort of patients.
Why, then, do black and white children with cancer tend to have the same outcome when treated at SJCRH? The most straightforward explanation is that both groups receive the same effective risk-directed therapy and supportive care, which can abolish the prognostic impact of many clinical and biologic variables. This capacity has been clearly demonstrated for patients with ALL, the most common childhood cancer. Despite having a higher frequency of unfavorable prognostic features such as high leukocyte count, T-cell immunophenotype, and chromosomal translocation t(1;19), and a lower frequency of favorable hyperdiploid karyotype, the 68 black patients treated in our Total XIII studies in the 1990s fared as well as the 338 white counterparts (5-year survival rates, 86.2% [95% CI, 77.2% to 95.2%] v
85.0% [95% CI, 80.9% to 89.1%]).18
There was also a parallel improvement in outcome for the 79 black and 340 white patients with ALL enrolled in our Total Therapy XV study between 2000 and 2007, with 5-year survival rates of 88.3% ± 6.2% (SE) and 94.8% ± 2.0%, respectively.32
In this regard, our recent collaborative study with the COG showed that Native American ancestry was associated with an increased risk of relapse in ALL, but this ancestry-related relapse hazard could be abrogated with an additional course of delayed intensification therapy among patients treated on a COG protocol.33
Undoubtedly, recent improvements in the COG studies for ALL have narrowed the survival gap between black and white patients with ALL registered in the SEER program (), since most of these patients were treated on COG protocols.
There have also been notable improvements in risk-directed treatment and supportive care for patients with acute myeloid leukemia, both at our institution34,35
and in COG studies.36,37
Immunotherapy with the monoclonal antibody anti-GD2 with granulocyte-macrophage colony-stimulating factor, interleukin-2, and isotretinoin has also significantly improved the survival of patients with high-risk neuroblastoma treated in a recent COG study.38
These therapeutic gains have benefited black and white children equally at SJCRH, although among patients registered in the SEER program and treated on COG protocols, their impact has been limited almost exclusively to whites. In fact, the survival gap has widened between black and white patients with acute myeloid leukemia or neuroblastoma in the SEER program. Conceivably, this discrepancy in outcome is due to a lack of equal access to effective treatment and supportive care among some patients represented by the SEER data. For patients without sufficient insurance coverage, access to costly procedures such as stem-cell transplantation may be particularly restricted, even though it is a proven life-saving measure for refractory or relapsed leukemia, non-Hodgkin's lymphoma, and neuroblastoma. In this regard, we have demonstrated recently that white and nonwhite children with high-risk leukemia had similarly high survival rates after transplantation at our institution, where there are no barriers posed by a lack of health insurance or the availability of a matched donor.35
Whether our findings can be generalized to patients treated at other major pediatric cancer centers is unknown.
Our study has several limitations. First, the number of black patients treated at SJCRH is relatively small for certain disease categories, preventing firm conclusions for some types of cancer. Second, specific disease characteristics and follow-up observations were not available for individual patients registered in the SEER program, precluding comparisons based on stratified analyses with adjustment for known risk factors and censoring bias. Third, it is uncertain what proportion of patients in the SEER program received treatment in tertiary medical centers with a multidisciplinary team and extensive psychosocial support, a combination that generally yields superior results.39
Fourth, it is well recognized that protocol-directed therapy is the best treatment for patients with cancer. In this regard, for a variety of reasons, the proportion of children age younger than 20 years in the SEER program who were enrolled in a COG protocol has been low (57%), especially for patients 15 to 19 years of age (24%).40
At SJCRH, 66% to 72% of the patients were enrolled in a therapeutic protocol and 97% in a therapeutic or nontherapeutic protocol. Although there was no substantial difference in the protocol registration rate by race or ethnicity in the SEER program,40
the possibility that the adverse impact of failure of protocol enrollment was more pronounced in blacks cannot be excluded. Finally, the socioeconomic status of patients registered in the SEER program cannot be determined with any degree of certainty. Since this variable clearly affects the outcome of children with cancer, as has been convincingly demonstrated in studies of adults,3,5,6
future research on this topic will need to consider insurance status or ability to pay as potential confounding factors.