4 | DISCUSSION
In this retrospective population-based cohort study of AYA with OGS or EWS, we found that despite significant differences in clinical and treatment exposures between patients treated at pediatric versus adult centers (Table 1 ), the location of treatment did not impact survival for either disease.
Previous studies have shown that AYA with ALL or Hodgkin lymphoma have superior survival when treated at pediatric centers.7,10 This has been attributed to disparities in treatment regimens and chemotherapy dose intensity, clinical trial enrollment, and treatment team experience11,12,14However, there are few studies that have compared the impact of clinical and treatment differences between pediatric and adult centers on the outcomes in AYA with bone tumors.12,13,16
Howell et al. investigated the impact of LOC on survival outcomes in a retrospective study of 1,751 cancer patients (including 76 patients with OGS or EWS) aged 0 to 19 years, treated at one of five US pediatric cancer centers (Children’s Oncology Group [COG] members) or in a non-COG (“adult”) institution.13 Only 36% of patients, aged 15 to 19 years were treated at a COG (pediatric) institution. In AYA patients with “pediatric type” cancers (i.e., cancers that are more common in pediatric age groups: OGS, EWS, Hodgkin lymphoma, ALL, neuroblastoma, rhabdomyosarcoma), those treated at pediatric centers had superior survival outcomes than those who received care at non-pediatric institutions. However, these differences were not statistically significant, possibly due to the small number of the patients.13 In a subsequent report, Bleyer et al. re-analyzed the same data to assess the likelihood of treatment of the type of cancer (“pediatric vs adult-type”) by oncologists with pediatric versus adult cancer experience.12 The results revealed that AYA with “pediatric-type” cancers had better survival at COG (pediatric) centers and those with adult-type tumors (e.g. non-Hodgkin lymphoma) showed a better survival at non-COG (adult) institutions.12 The authors concluded that the AYA patients with pediatric types of cancer (including OGS and EWS) fare better if their care is delivered by pediatric oncologists.12
In an institution-based retrospective study of 53 patients (29 pediatric and 24 adult) with localized EWS, Gupta et al. studied survival outcomes among those treated at an adult or a pediatric tertiary cancer care center in Ontario between 1990 and 2005.16 The median age for the pediatric and adult patients was 13.4 years (range, 0.29-16.2 years) and 26.1 years (range, 16.7-66.5 years), respectively. The total cumulative doses of ifosfamide and cyclophosphamide were lower (p<0.0001) and the time-to-local therapy was longer (7.4 months vs 3.7 months; p= 0.0003) in the adult center compared to the pediatric center. The 3-year EFS rates in pediatric and adult patients were 70% and 43% (p=0.1) and the 3-year OS rates were 81% and 59% (p=0.02), respectively. The authors concluded that the inferior survival outcome of EWS patients treated at an adult center may be related to lower cumulative doses of alkylating agents and a delay in local therapy in adult-center patients.16 In our localized EWS cohort, none of LOC, primary cancer site or time to local therapy was associated with survival.
In our study, the reason for superior EFS in patients with metastatic disease who received higher cumulative doses of methotrexate, doxorubicin, and cisplatin in OGS or cisplatin in EWS is unclear. One possibility is that the higher dose of chemotherapy effectively impacts the survival of patients with metastasis. The second, more likely possibility is that patients with more aggressive disease may have received lower doses of chemotherapeutic agents with a goal of controlling symptoms and avoiding toxicities, or that more aggressive tumors progressed leading to discontinuation of primary therapy, thus confounding the relationship between chemotherapy dose and survival. Unfortunately, the relevant data needed to classify metastatic patients according to their disease severity and aggressiveness were not available to test this hypothesis. The prognostic impact of cumulative doses of chemotherapeutic agents are not separately described in papers that have investigated OGS and EWS prognostic factors. In some studies, the response to neoadjuvant chemotherapy, measured by necrosis extent in surgical specimens, has been considered as a prognostic factor.27,28,29 We did not have access to pathology data from excised tumors, so could not evaluate this relationship.
Some studies have shown superior outcomes for patients who treated at pediatric centers with high rate of participation in clinical trials.30-33 But some authors argue that patients enrolled in clinical trials might not represent the general population and thus the results might be biased.11 In our study, about half of pediatric-center patients with OGS and EWS and only a few adult-center patients were enrolled in clinical trials. Neither EFS nor OS rates significantly differed in patients who were enrolled versus those who were not enrolled in clinical trials.
In our study and several other reports19-21, there was no association between DIL and survival in either OGS or EWS. Goedhart et al. reported that prolonged DIL does not necessarily result in inferior survival in OGS and EWS and concluded that the impact of tumor location and its resectability on survival is more than the effect of delay in diagnosis.20
The strength of this study, as one of the few studies in literature that compares outcomes of AYA OGS and EWS in pediatric versus adult centers, was the use of health administrative data along with population-based databases. This combination of datasets provided us with an opportunity to capture the relevant health data of the entire provincial population and to assess a wide range of demographic, disease- and treatment-related variables and their impact on outcome as well as with access to possible interactions with the health care system. Using this type of data, improved the generalizability of the results and ensured that our findings were not susceptible to recall bias and poor documentation seen in studies based on chart abstraction or questionnaires.
One of the issues in evaluating the prognostic effect of age in OGS, EWS, and other cancers is disparities in age ranges and in definitions of pediatric, adolescent and young adult age groups in various studies. The limited age range of AYA (15 to 21 years) in our study reduced its sample size and hence its power and generalizability and made it difficult to compare our results with those of other studies that included a wider age ranges of AYA patients. Another limitation was missing data in some recognized prognostic factors such as tumor size, response to neoadjuvant chemotherapy (tumor necrosis) and surgical margin status; so that we had to exclude these variables from the study.
In conclusion, survival outcomes in AYA with OGS and EWS in our study did not differ by LOC. This stands in contrast to AYA cancers such as ALL and Hodgkin lymphoma, where studies have shown superior survival associated with treatment at pediatric centers. In this study, even though AYA with bone tumors treated at pediatric centers received higher cumulative doses of chemotherapy than in adult centers, the differences did not impact outcomes. Lack of outcome differences by LOC in our study might be explained by the similarities of treatment team expertise and therapeutic regimens for bone tumors and also biologic characteristics of the patients (given the narrow AYA age range) between pediatric and adult cancer centers. Furthermore, although data was not collected, all bone tumors in Ontario are operated on by specialized orthopedic oncologists; the quality of surgical resection is known to be an important prognostic factor in those with localized bone tumors.34,35
Future studies can work on extending the AYA age range to above 21 years by either chart abstraction or linkage to other administrative databases in order to provide a larger sample size and to address the uncertainties about the prognostic impact of LOC, age, cumulative doses of chemotherapeutic agents, and other covariates.
ABBREVIATIONS : AYA, adolescents and young adults; CI, confidence interval; CIC, the Citizenship and Immigration Canada database; COG, Children’s Oncology Group; Cox PH survival analysis, Cox proportional hazard survival analysis; EFS, event-free survival; EWS, Ewing sarcoma; HR, hazard ratio; ICES, Institute for Clinical Evaluative Sciences; IMPACT, the Initiative to Maximize Progress in Adolescent and Young Adult Cancer Therapy; IQR, interquartile range; OCR, Ontario Cancer Registry; OGS, osteogenic sarcoma; POGONIS, Pediatric Oncology Group of Ontario Networked Information System; SCC, specialized cancer center; SMN; subsequent malignant neoplasm