Tel.: 905-521-2100x73464, Fax: 905-521-1703
Surveillance and survival among adolescents and young adults with cancer in Ontario, Canada
Article first published online: 29 MAR 2012
Copyright © 2012 UICC
International Journal of Cancer
Volume 131, Issue 11, pages 2660–2667, 1 December 2012
How to Cite
Furlong, W., Rae, C., Greenberg, M. L. and Barr, R. D. (2012), Surveillance and survival among adolescents and young adults with cancer in Ontario, Canada. Int. J. Cancer, 131: 2660–2667. doi: 10.1002/ijc.27523
- Issue published online: 25 SEP 2012
- Article first published online: 29 MAR 2012
- Accepted manuscript online: 7 MAR 2012 04:27AM EST
- Manuscript Accepted: 20 FEB 2012
- Manuscript Received: 14 DEC 2011
- Public Health Agency of Canada
- young adults
Gains in survival rates among adolescents and young adults (AYA) are reported from the USA to be lower than in both younger and older patients. Limiting factors include low accrual to clinical trials related to the type of institutional care. This study aimed to determine the incidence of cancer in the 15–29 age group in Ontario, and the 5-year survival of these cases by disease class, age at diagnosis group and highest level of institutional complexity of care. The primary data source was Cancer Care Ontario (CCO). Diseases were classified according to an AYA-specific system. Age at diagnosis was grouped as 15–19, 20–24 and 25–29 years; and institutional site of care was categorized as pediatric oncology group of Ontario (POGO) centers, regional cancer centers (RCC—tertiary care centers associated with CCO), RCC affiliate and satellite institutions and other institutions having no specialized cancer services. More than 10,000 incident cases were identified during 1990–2001. Carcinomas and lymphomas each accounted for >20% of the total. Overall 5-year survival rate was 83%; significantly higher for lymphomas at POGO centers and RCC than elsewhere. About 40% of eligible AYA cases were treated at a POGO center and 25% of those were accrued to clinical trials. The low proportion of adolescents referred to pediatric cancer centers may result in a survival disadvantage for this group. All AYA, especially with lymphomas, should be referred to specialized centers. Accrual of AYA to clinical trials must be improved substantially.
Among adolescents and young adults (AYA) with cancer, it has been reported that the average annual percent change in 5-year survival, of patients diagnosed between 1975 and 1997 in the United States, is a fraction of that for all other age groups.1
Adolescents are defined as aged 15–19 years and young adults as 20–29 years of age by the surveillance, epidemiology and end results (SEER) program of the National Cancer Institute (NCI) in the USA2 and by the Canadian Cancer Society.3 The upper age bound for young adults is more contentious and a case has been made for flexibility in this regard.4 The Progress Review Group of the NCI and the Lance Armstrong Foundation settled on 39 years5 while Cancer Care Ontario (CCO) has taken 44 years as the upper limit.6
The distribution of diseases in AYA shifts progressively with increasing age from typically pediatric neoplasms to those more characteristic of older adults.2 Indeed, a distinct classification system for malignant diseases in the AYA age group has been reported.7
Although there has been some recent improvement in the survival rate for AYA with cancer in the USA, at least for certain diseases,8 and the overall 5-year survival rate for this age group in Canada has climbed to 83%,3 cancer remains the most common cause of disease-related death in both countries among 15–29 year olds.2, 9 In recent years, the factors contributing to the relative lack of improvement in survival of AYA with cancer in the USA were explored by the NCI5 and the Children's Oncology Group (COG).1 These include delay in the recognition/reporting of symptoms and limited compliance of AYA with therapeutic regimens. Additional factors of importance are inadequate health insurance10 and low accrual to clinical trials.11 Differences in tumor and host biology, by comparison with similar diseases occurring in younger and older persons, may also play a role.
The great majority of children with cancer in North America are treated in children's hospitals, >90% at cooperative group institutions and 55–65% are enrolled in clinical trials, mainly through the agency of COG.1 Among adults over 29 years of age, the estimated participation rate in clinical trials is 3–5%.1 Among 15–19 year olds in the USA only 20–35% are seen at cooperative group institutions and only about 10% are entered into clinical trials. For those in the 20–29 year age group, only 10% receive care at cooperative group institutions and no >1% are enrolled on clinical trials.1, 12 These rates are thought to have been similar in the United Kingdom,13 but have improved substantially in recent years.14
The low level of participation by AYA in clinical trials is problematic for several reasons. There are correlations between participation and both duration of survival and reduction in mortality rates,15, 16 at least for some diseases, although some investigators have challenged the quality of the supporting evidence.17
A similar trend has been recorded from the USA with respect to a transition of referrals from pediatric to adult treatment centers, starting approximately from the age of 15 years, although the decision-making process regarding such referrals is poorly understood.1 In Ontario a progressive decline has been observed in referrals to pediatric cancer centers, between the ages of 15 and 19 years of age; due in part to the upper age limit of 18 years for admission to children's hospitals.18
The site of care (institutional specialty level) has implications for clinical outcomes. Pediatric treatment protocols, that are usually more intensive than those used in adults, may offer survival advantages in the AYA group, as exemplified by experience with acute lymphoblastic leukemia (ALL) in North America, France, Italy and the Netherlands. Furthermore, when a single protocol is used across the age spectrum, there may be a survival advantage if treatment is delivered in a pediatric rather than an adult center, as reported from Germany in patients with Ewing sarcoma.19
This project was designed to determine the incidence of new diagnoses of cancer in AYA (15–29 years of age as in the SEER analysis) in Ontario, Canada and the 5-year survival rates for the incident cases by disease class, age group at diagnosis and site of care category. The study also investigated sources of clinical trial participation data in the Province and the relationships among sets of these variables.
Material and Methods
Summary results were obtained from CCO and the Ontario Cancer Registry (OCR) for counts of incident cases during the period of January 1, 1990 through December 31, 2001 and deaths for the incident cases; defined by diagnostic disease classes, age at diagnosis (age) groups, institutional site of care categories and calendar year of diagnosis. Validation of the incidence results was accomplished using SEER incidence rates,2 which are very similar to those published recently in Canada,3 and Ontario population sizes from the 2001 Canadian census. The diagnostic disease classes were the 10 main diagnostic groups in the AYA classification system.7 The age groups were 15.0 through 19.9 (15–19), 20.0 through 24.9 (20–24) and 25.0 through 29.9 (25–29) years. The calendar year of diagnosis factor was excluded from more detailed analyses because of concerns about confidentiality and statistical power associated with small cell sizes. This study was approved by the research ethics board (REB) at McMaster University and Hamilton Health Sciences (REB# 09-367-C).
The site of care factor was conceptualized originally as representing the type of institution at which a diagnosis was made and treatment administered. However, the process of generating the preliminary summary results made it clear that available data sets did not support this level of detail and therefore the site of care concept was revised to represent the institutional site of care with the greatest oncological specialization used by each incident case.
Sites of care were defined as Pediatric Oncology Group of Ontario (POGO) centers and non-POGO centers. Non-POGO centers were further categorized by a new CCO facility classification system; The Regional Systemic Treatment Program Provincial Plan September 2009 Classification System of Institution/Facility Levels Based on Complexity of Care (Fig. 6 and Table 4 of http://www.cancercare.on.ca/common/pages/UserFile.aspx?fileld=58067). The site of care for each patient was identified based on records of institutions providing health care services in the Canadian Institutes of Health Information Discharge Abstract Database. Each case was classified into one site of care category using a two-step process. First, each case was identified as POGO or Regional Cancer Center (RCC) or other site; POGO center by registration in POGO's Networked Information System (POGONIS) and RCC as any record of care at an RCC. Second, each “other site” case was identified as RCC affiliate/satellite or Remaining based on the name of the service institution for the case. The site of care was determined in large part from the contemporary list of Ontario Regional Systemic Treatment Program RCC affiliate and satellite hospitals (RCC affiliates/satellites). Use of the contemporary list results in some incorrect site categorizations because the number of institutions providing specialized cancer care in the Greater Toronto Area (GTA) had increased over the study time period, but in non-GTA parts of Ontario there had been a reduction in the number of such individual institutions as a result of progressive centralization of cancer care and the amalgamation of hospital corporations. Subsequently, all incident cases were assigned to one of four mutually exclusive site of care categories: POGO centers, RCC, RCC affiliates/satellites and Remaining (institutions not providing specialized cancer care).
Finally, the study PI requested counts for the total number of newly diagnosed patients aged 15–29 years, and the total number of these patients enrolled on therapeutic trials, from POGO and the Ontario Institute for Cancer Research (OICR).
Data Management and Analyses
Anonymous records of incident cases were provided by CCO to the study group in Excel spreadsheet data files. Records for each case included date of birth, date of diagnosis, date of death if applicable, disease class, POGO/RCC/Other site of care category and name of treating institution. Survival status at 5 years after diagnosis was determined for each case based on date of death, if applicable. Statistical analyses were completed using Minitab 13.20 and SPSS 15.0.1.
Incidence and mortality were described using frequency distribution counts and percentages by three major study design factors: disease class, age at diagnosis group and site of care category. Differences in proportions were assessed using chi square and Fisher exact tests, as appropriate. Statistical significance was set at the 5% level with Bonferroni adjustment for multiple testing. Minimum incident counts to support future research of the effects on survival of additional independent factors, such as clinical trial enrolment, was defined as 30 or more cases Province-wide in cells defined by each of disease classes or age at diagnosis groups or site of care categories.
A total of 10,075 incident cases were recorded for the 12-year study period. Two diagnostic groups (carcinomas and lymphomas) accounted individually for >20% of the total incident cases (Table 1) The expected numbers of incident cases, in total and by 5-year .age categories, are displayed in Table 2. The overall observed-to-expected incidence ratio is 0.99 and the distribution of diseases is very similar to that reported by SEER for the 15–29 year old age group.2 The frequency counts and percentages of all incident cases by disease class and age group are shown in Table 3. The total number of incident cases increased across the 5-year age intervals and there was a significant shift (p < 0.001) in the proportion of cases in disease classes, excluding miscellaneous and unspecified diseases, among age groups; the proportion of lymphomas declined while the proportion of carcinomas increased with age, to >40% in the 25–29 year age category. Of note, there was a statistically significant (test for trend, p < 0.001) increase of unspecified diagnoses with increasing age (Table 3), with 15–19 less than 20–24 and 25–29 (both p < 0.001) but no difference between the 20–24 and 25–29 age groups.
The distributions of all AYA incident cases by disease class and site of care category are shown in Table 4. There was significant variability in the proportions of cases within disease classes among site of care categories (p < 0.001). Most disease classes had one or more site of care categories with <30 total cases; considered too small for the purposes of future investigations into the effects of clinical trial participation on survival rates. The majority of cases at POGO institutions were lymphomas and leukemias, at RCC were lymphomas and carcinomas, at RCC affiliates/satellites were carcinomas and at remaining institutions were carcinomas and melanoma. The patterns of incident cases, by disease class and site of care category, were similar for the subpopulation of 15–19 year olds (Table 5). Fewer than 10% of AYA with cancer are seen outside the system of specialized care (in the remaining category of institutions). Among the adolescent population (15–19 years of age), fewer than 25% were referred to POGO institutions but <5% were seen outside the cancer system.
The proportions of cases in POGO centers versus RCC and their affiliates/satellites in the 15–19 year old group varied among the disease classes (p < 0.001). For carcinomas, the proportions were 7% POGO versus 81% RCC and affiliates/satellites, for leukemias 46% versus 51%, for lymphomas 28% versus 70% and for bone tumors 38% versus 62%. It was expected that 60% of 15–19 year old incident cases would be eligible for treatment at a POGO center, so the observed proportions of <50% indicate that many eligible adolescents are not receiving care at a POGO center.
There was a significant association (p < 0.001) between 5-year survival proportions and each of the three study factors; age group, disease class and site of care category. Survival rates by disease class are listed in Table 6, by age group in Table 7 and by site of care category in Table 8. There is considerable overlap of cases in the 15–19 year old age group with the POGO site of care category, with the potential for important confounding of the survival effects attributable to each of these variables. The survival rate for all diseases combined was lower for the 15–19 age group than for the 20–24 (p < 0.001) and 25–29 (p = 0.002) age groups. There were no significance differences in alive/deceased proportions among age groups for seven of the eight disease classes, the exception being for lymphomas. For lymphomas, post-hoc pair wise analyses (adjusted for multiple testing, level of significance p < 0.05/3 or p < 0.017) showed no significant difference in survival between 15–19 (86%) and 20–24 (88%) as well as 25–29 (83%) age groups; but a significant (p = 0.013) 5% difference between 20–24 and 25–29 age groups.
Table 9 presents the disease class 5-year survival rates by specific disease classes and site of care categories for the entire AYA cohort. There were significant differences in 5-year survival rates among site of care categories for six of the eight specific disease classes: carcinomas; CNS tumors; germ cell tumors; lymphomas; melanomas and osseous tumors. Germ cell tumors have the only significant difference in survival rates between POGO centers and RCC with POGO rates being lower; and carcinomas have lower rates at POGO centers and RCC than at RCC affiliates/satellites and remaining institutions. With respect to carcinomas, the differences may relate in large measure to the proportionality of metastatic disease, although information on this issue was not available. In a further analysis limited to the 15–19 year age group, an additional significant difference (p = 0.005) was revealed between the 5-year survival proportions for soft tissue tumors at POGO centers (57%) and RCC (83%). This may represent a predominance of patients with non-rhabdomyosarcomas at RCC and the predominance of the alveolar variant of rhabdomyosarcoma in adolescents attending POGO centers, although no such detail is available. For lymphomas, post-hoc analyses of differences between all pairs of site of care categories (adjusted for multiple testing, level of significance p < 0.05/6 or p < 0.008) showed that the survival rates at POGO centers and RCC were significantly higher (16% difference in proportions) than at RCC affiliates and satellites. For melanoma the reverse was true, with survival rates being lower in the RCC than in RCC affiliates/satellites (p < 0.005) and in the remaining category of institutions (p < 0.001). For CNS tumors, the significant variability in survival rates among the four complexity of care levels was not sustained in post-hoc analysis.
Clinical trial enrolment
Information on clinical trial enrolment for cases ages 15.0–17.9 years at POGO centers were recorded in POGONIS from 1995. Data for cases treated at all RCC in Ontario were collected by OICR for the period 2004–2009 but were not divisible by age groups.
Of the 1,022 adolescents registered in POGONIS in the 16-year period January 1995 to December 2010, there were 254 (24.9%) enrolled in multi-center clinical trials, 151 (14.8%) who followed a clinical trial protocol but were not enrolled and 617 (60.4%) who were not identified as following a multi-center clinical trial protocol as a treatment plan.
For the 12 RCC in Ontario, 275,004 incident cases of all ages were recorded by OICR in 6 years. Of these 28,118 (10.2%) were enrolled on clinical trials. Enrolment proportions for all RCCs combined varied among years (p < 0.001): increasing from 2004 (8.9%) to a maximum in 2007 (12.4%) and falling thereafter to a minimum in 2009 (8.5%). Of all cases treated during years 2004 through 2009, the enrolment rate varied by RCC (p < 0 001) from 2.9% to 22.1%. The five centers with annual new case loads each exceeding 5,000 patients had higher clinical trial enrolment rates than the smaller RCC.
The total number of incident cases in the OCR, for patients of all ages, during the study period was 526,445. According to SEER,2 cases in the 15–29 year age group account for 2% of all invasive cancers leading to an expected study population for Ontario of 10,529 based on census data from the end of the study period, so providing the highest number for the population in that interval. The expected study population estimate is within 5% of the observed population size. Again according to SEER,2 15–19 year olds account for 18% of the cases in the 15–29 year age group, yielding an expected subpopulation of 1,814 which is within 1% of that observed. The average of 840 incident AYA cases per year in Ontario from this study is 40% of the average number of 2,075 incident AYA cases for Canada during 1992–20053 and similar to the 38% of the 2001 Canadian general population living in Ontario.
The proportion of 15–19 year old patients treated in POGO centers during 1990–2001 (24%) is plausible but lower than the 32% reported previously for 1997–2001,3 perhaps indicative of the impact of recent efforts by POGO to increase the referral of adolescents with cancer to its centers. Assuming that 18 and 19 year olds (representing two-fifths of all 15–19 year old incident cases) are not eligible for admission to children's hospitals in Ontario associated with POGO centers, this means that 59% of eligible 15–17 year olds (641 of 1077 incident cases) were not seen at POGO centers. This “leakage” estimate is similar to that described previously by Greenberg et al.20 who estimated that 55% of the 15–17 year old cases were not referred to POGO centers. Apart from the issue of lower clinical trial enrolment for non-POGO patients, this population may be at a survival disadvantage for some diseases, given the high survival rates reported recently for adolescents with ALL who were treated in pediatric centers.21, 22
The 5-year observed survival rates in this Ontario study were similar to those reported for AYA in Canada, excluding Quebec, in a later cohort (1997–2004).3 For example, the comparative survival rates were 83% and 83% overall, 58% and 61% for leukemias and 67% and 66% for CNS tumors for cases in Ontario and Canada respectively. In the Canada-wide study, death was commonest in patients with leukemias and CNS tumors,3 but in this Ontario study the leading causes of death were carcinomas followed by lymphomas then leukemias and CNS tumors. There is no obvious explanation for this disparity.
The relatively low survival rate in POGO centers reflects the relatively high incidence of diseases with low survival rates in the young age group (CNS tumors, leukemias, osseous tumors and soft tissue tumors), while the high survival rate in the remaining category of institutions may reflect a predominance of non-metastatic disease (carcinomas, osseous tumors and sarcomas).
The high survival rates in the remaining category of institutions for CNS and soft tissue tumors suggest that these were low grade malignancies. Similarly, the 5-year survival rate exceeding 95% for carcinomas in these institutions may be related to a predominance of thyroid and testicular tumors as well as non-melanoma skin cancers (basal cell and squamous cell carcinomas). Again, the high survival rate for melanoma in the RCC affiliates/satellites and remaining category of institutions suggests that the great majority of these cases were of low stage disease.
For lymphomas, the most prevalent category of diseases in the AYA age group after carcinomas, the 5-year survival rates in POGO centers and RCC are so much better (16% difference) than those in RCC affiliates/satellites that consideration should be given to ensuring that all lymphoma cases be treated in RCC, or POGO centers if less than 18 years of age. More than 90% of lymphoma cases were served in this way during the 1990–2001 period.
The population sizes in site of care categories are a limitation for selecting disease classes for the purposes of future investigations into the detailed effects of clinical trial participation on survival rates. Age at diagnosis grouping is not a limiting factor for such purposes. There are only two diseases with 30 or more incident cases in each of the three age groups and in each of the four care levels; lymphomas and CNS tumors. Therefore these are the only classes of disease for future assessment of clinical trial enrolment in AYA in Ontario. These two disease classes also have large differences in 5-year survival rates among site of care categories.
Given the 7.5 fold difference in clinical trial enrolment rates across RCC in Ontario, studies of the effects of clinical trial enrolment on survival rates should focus on centers treating relatively large numbers of patients and having relatively high rates of clinical trial enrolment. There is considerable scope for improvement in the accrual of AYA with cancer to clinical trials in Ontario, including adolescents at POGO centers.
Comparisons with experience elsewhere are informative. In Victoria, Australia a progressive decline in clinical trial accrual rates after the age of 15 years has been reported, falling to 6.9% for those aged 16–19 and 4.2% for the age group of 20–24 years.23 In the United Kingdom accrual rates to National Cancer Research Institute (NCRI), clinical trials were 24.5, 10.6 and 8.0% in the age groups 15–19, 20–24 and 25–29, respectively.24 The reports from Australia and the UK highlight particular deficits in accrual rates to brain tumor trials; 2% and zero in 20–24 year age group respectively. The Australian study also noted the importance of place of treatment in determining clinical trial accrual, being an order of magnitude higher for adolescents (ages 10–19 years) treated in children's hospitals (38%) than for those treated in adult centers (3%).23 Even when age restrictions are removed and clinical trials are designed for diseases especially prevalent in the AYA age group, accrual rates may decline with increasing age. This is exemplified by the EURO-Ewings and EURAMOS-1 trials, respectively for Ewing sarcoma and osteosarcoma, to which accrual rates of 72.7, 20.6 and 7.1% have been reported from the UK for the age groups 15–19, 20–24 and 25–29 years, respectively.14
Considerable efforts have been made to address such deficits. The COG formed an AYA committee in 2000 and has increased recruitment of patients under 40 years of age by more than three fold; from 5.1% in 1998–1999 to 16.2% in 2005–2005.25 In the UK, the NCRI formed the Teenage and Young Adult Clinical Studies Development Group and joint ventures have been started between adult and pediatric brain tumor study groups.26 The national Task Force on AYA Cancer in Canada was established in 2008 and is expected to promote these advances, among others.27, 28
Production of this report has been made possible through funding from the Public Health Agency of Canada. The views expressed herein do not necessarily represent the view of the Public Health Agency of Canada. The authors thank C. Ackland, D. Burns, B. DiMonte, W. Evans, C. Gosnell, T. Hudson, K. Milnes and C. Sagoe for their contributions.
- 1Introduction. In: W Archie Bleyer, Ronald D. Barr, eds. Cancer in adolescents and young adults. Heidelberg: Springer-Verlag, 2007. 1–26., , , .
- 2Cancer epidemiology in older adolescents and young adults 15 to 29 years of age, including SEER incidence and survival: 1975–2000. Bethesda, MD: National Cancer Institute, 2006., , , .
- 3Canadian Cancer Society's Steering Committee. Canadian Cancer Statistics 2009. Toronto: Canadian Cancer Society, 2009.
- 4What should the age range be for AYA oncology? J Adolesc Young Adult Oncol 2011; 1: 4., , .
- 5Adolescent and Young Adult Progress Review Group. Closing the gap: research and cancer care imperatives for adolescents and young adults with cancer. Bethesda MD: Department of Health and Human Services, National Institutes of Health, National Cancer Institute and LIVESTRONG Young Adult Alliance, August 2006. Available at: http://www.planning.cancer.gov/library/AYAO_PRG_Report_2006_FINAL.pdf. Accessed on December 14, 2011.
- 6Cancer Care Ontario: Cancer in Young adults in Canada. Toronto, Canada, 2006. Available at: http://www.cancercare.on.ca. Accessed on December 14, 2011.
- 8Latest estimates of survival rates of the 24 most common cancers in adolescent and young adult Americans. J Adolesc Young Adult Oncol 2011; 1: 37–42..
- 19Cure rates in Ewing tumor patients aged over 15 years are better in pediatric oncology units. Results of GPOH CESS/EICESS studies. Proc ASCO 2003; 22: 816(abstract 3279)., , .
- 24Rates of inclusion of teenagers and young adults in England into National Cancer Research Network clinical trials: report from the National Cancer Research Institute (NCRI) Teenage and Young Adult Clinical Studies Development Group. Br J Cancer 2008; 99: 1967–74., , , , , , , , , , .
- 26National Institute for Clinical Excellence Guidance on Cancer Services. Improving outcomes in children and young people with cancer. London: NICE, 2005.
- 28Principles and recommendations for the provision of health care in Canada to adolescent and young adult-aged cancer patients and survivors. J Adolesc Young Adult Oncol 2011; 1: 53–9., , , , , , , .