Abstract
Purpose:
To compare cancer types, stages, times to treatment, and survival for adolescent and young adults (AYAs) 15–24 years of age with other cancer patients <40 years.
Methods:
New South Wales Cancer Registry and treatment data were linked to explore differences in cancer type, stage, time to treatment, and survival between AYAs, children, and adults. Multivariable logistic regression and competing-risk regression were adjusted for sociodemographic, diagnostic period, and clinical characteristics.
Results:
Most common cancers in AYAs and adults were carcinomas compared with leukemias in children. Advanced (regional and distant) stage applied to 33% of AYA solid cancers, which was similar to adult stages, but lower than the 40% for children (adjusted odds ratio 1.21, 95% confidence interval [CI] 1.01–1.47). Proportions starting treatment ≤60 days from diagnosis were 93% for AYAs and children, and 94% for adults, with higher adjusted odds of starting ≤60 days of 1.39 (95% CI 1.11–1.73) for children and 1.23 (95% CI 1.06–1.44) for adults. Five-year disease-specific survival was 90% for AYAs and adults, and 87% for children. The adjusted subhazard ratio for children compared with AYAs was 0.67 (95% CI 0.52–0.88). Age differences in cancer stage, treatment start, and cancer survival varied by cancer type.
Conclusions:
After adjusting for cancer type, diagnostic period, and sociodemographic characteristics, AYAs had less advanced solid tumors than children; fewer AYAs were treated within 60 days than children and adults; and AYA survival was lower than for children. The potential for residual confounding from leukemia type and other confounders needs further analysis with larger Australia-wide cohorts.
Introduction
Although cancers in adolescent and young adults (AYAs) 15–24 years of age account for <1% of cancers recorded in Australia, they contribute about 8% of the fatal burden of cancer. 1 They can disrupt transition from childhood to adulthood through cognitive, biological, and socioeconomical effects, and present unique care needs, 2 and impact on education and employment. 3 AYAs generally must adapt from intensive care to primary care settings early in their cancer journey. 4 Later they are at risk of secondary malignancy,5,6 and impaired fertility, hormonal disturbance, and adverse heart and lung function, hearing, and vision. 7
Age definitions for AYAs vary. 8 The Australia Institute of Health and Welfare (AIHW) has used 15–24 years, as in Europe, whereas in North America, the range extends to 39 years. We chose 15–24 years for consistency with national reporting in Australia. 1
National AYA 5-year relative survival in Australia is 89% compared with 84% for ages 0–14 years and 88% for 25–39 years. 1 Cancer stage is a strong predictor of AYA survival but is not recorded nationally.1,9 Stage (degree of spread) is recorded by the New South Wales (NSW) Cancer Registry, covering about a third of the AYA national population. A two-fold and seven-fold higher cancer mortality is reported for regional and distant stage, respectively, compared with localized stage. 9 National hospital data suggest AYAs have less treatment than children and ages 25–39 years. 1
Time to treatment is commonly used as a cancer care accessibility index, with early care normally favored for higher survival,10,11 although differences by age may be justifiable, as for leukemias, which mostly are acute in children and more commonly chronic in adults where “watchful waiting” or deferred treatment may be preferred. Also, for some cases, curative treatment might not be possible.
Despite this variability, the National Service Delivery Framework advocates timely access of AYAs to cancer treatments, 12 little data exist that quantify timeliness and impacts on outcomes. Cancer United Kingdom has indicated that for broader ages, treatment should ideally commence within 1 month of diagnosis but recommends commencement within 2 months as a realistic target. 13 Although a different health system, single hospital data from United States and Canada indicate multiple determinants of treatment delays, including socioeconomic factors and less service accessibility.14,15
Using linked cancer registry and routinely collected treatment data in NSW, we aimed to establish AYA benchmarks and compare cancer types, times to treatment, stages, and cancer survival with corresponding features for ages 0–14 and 25–39 years. These ages were the same as used nationally by the AIHW to cover the transition from childhood to adult care settings that are thought to be vulnerable to gaps in service continuity. 1
We consider this study to be an important and novel addition to AIHW data, which have already described AYA cancer types and unadjusted clinical and sociodemographic relationships compared with cancers for ages of 0–14 and 25–39 years. 1 We took the AIHW data further by indicating unexplained (i.e., post-adjustment) relationships.
Methods
Ethical statement
Approval for this study was provided by the NSW Population Health Services Research Ethics Committee (Ref: 2015/05/585), the AIHW Ethics Committee (Ref: EO2016/1/224), and the Aboriginal Health & Medical Research Council (AH&MRC) of NSW Ethics Committee (Ref: 1201/16).
Study design, participants, data sources
This study included 23,619 patients 0–39 years of age at diagnosis in 2003–2015 recorded by the population-based NSW Cancer Registry and linked with routinely collected treatment data. The Registry follows international standards, collecting data under legal mandate on all malignant cancers in NSW residents, except squamous and basal cell skin carcinomas.16,17 Primary cancer site, morphology, diagnosis date, residential area, and stage (degree of spread) are recorded. Death dates and causes are traced using death registrations Australia wide. 18
Hospital inpatient data and universal health insurance claims through the Medicare Benefits Schedule (MBS) and Pharmaceutical Benefits Scheme (PBS) were the main sources of treatment data. Hospital data included dates of admission and clinical procedures, whereas radiotherapy center data included dates of radiotherapy. Collectively, these data covered most treatments (as verified by Registry staff through multiple reporting), with MBS and PBS claims covering privately funded hospital and community treatments.19,20 For people receiving systemic anticancer therapy in public facilities, data coverage is ≥94% for common cancer sites in adults. 21 Completeness of systemic data from all sources is likely to be close to complete when combining in-hospital with MBS and PBS data for the community health sector.
Linkage was performed by the Center for Health Record Linkage for NSW-based datasets and the AIHW for Commonwealth-based health insurance data. Deidentified linked data were stored in the Secure Unified Research Environment facility, a remote access environment. 22
Study outcomes
Outcomes included cancer type, stage, and time to treatment start obtained or derived from the Registry or inpatient, MBS, and PBS data. Cancers were classified into 10 broad categories using the Surveillance, Epidemiology, and End Results AYA site recode, 23 that is, leukemias, lymphomas, central nervous system (CNS) cancers, bone cancers, soft-tissue sarcomas, germ cell cancers, melanomas, carcinomas, miscellaneous specified neoplasms, and neoplasms not otherwise specified, and unspecified. 1
Stages of solid cancers, classified as localized (confined to origin site), regional (invasion of adjacent tissues or regional nodes), or distant (spread to distant nodes or other organs), were derived from hospital and pathology notes, as specified in international guidelines,16,17,24 using the maximum spread recorded up to 4 months from diagnosis according to Australian registry protocols.
Treatment initiation was defined as the first treatment date recorded in our data sources, whether surgery, systemic, or radiotherapy. Time to treatment was calculated as the interval between pathological confirmation of diagnosis to cancer treatment start (days), categorized as <30, 31–60, 61–90, or >90 days, as in past studies. 25
Other descriptors
Other descriptors included: sex; diagnosis year; age at diagnosis; residential remoteness; and socioeconomic status derived from residential address. Country of birth was classified as Australia, other mainly English-speaking countries, or mainly non-English-speaking countries, as previously. 26 Residential area was classified as major city, inner regional, outer regional or remote, using the Australian Standard Geographical Classification Remoteness Index. 27 Socioeconomic status was determined by place of residence at census collection district level and coded using the Socio-Economic Index for Areas (SEIFA). 28 Diagnostic period was categorized as 2003–2007, 2008–2012, and 2013–2015.
Statistical analysis
Comparisons were made of sociodemographic characteristics, cancer profiles, time from diagnosis to treatment, and vital status at the end of the study by age using the Pearson Chi-square or Fisher's exact test for categorical variables. 29
Adjusted odds ratios (95% confidence interval) (aOR, 95% CI) by age were calculated for stage (regional/distant vs. localized stage) and treatment start (≤60 vs. >60 days), by including cancer type, sex, country of birth, remoteness of residence, SEIFA quintile, and diagnostic period in the multivariate logistic regression models. AYA ages were the reference category for children and adults. Treatment start cutoff of 60 days was used in binary analyses as potentially more accurate than finer breakdowns of PBS/MBS claims and public hospital data.13,21 Sensitivity analysis using cutoff of 30 days was conducted. 13
Cancer mortality was estimated using Kaplan–Meier disease-specific method. Follow-up was from diagnosis to death (from the primary cancer or other causes) or to censoring date of December 31, 2015 for survivors. Age differences in cancer-specific mortality were presented as adjusted subhazard ratios (aSHRs) with 95% CIs, using competing risk regression, with deaths from other causes as competing events, 30 adjusting for sociodemographic and cancer characteristics, and time to treatment start. AYAs were used as the reference for comparison. This methodology was considered best to capture risk of cancer death under real-world circumstances of multiple competing causes of death.
Age disparity in risk of advanced cancer stage, time to treatment start ≤60 days, and mortality was examined by cancer category and tested using interaction terms of age and cancer type.
STATA release 16 was used for analyses. 31
Results
Patient characteristics
Of 26,327 cases 0–39 years of age registered on the NSWCR, 23,619 (90%) had linked treatment records. Of these: 2452 (10%) were 0–14 years of age, 3451 (15%) AYAs, and 17,716 (75%) 25–39 years; 57% were females; 65% were Australian born; and 77% were residents of major cities.
Males comprised a higher proportion of AYAs (52%) and children (55%) than adults (40%). The proportion of AYAs born in Australia (85%) was lower than for children (93%) but higher than for adults (72%). Lower proportions of AYAs and children lived in major cities and most disadvantaged areas than adults. No significant difference in distribution of cases was found by diagnosis period (p = 0.370) (Table 1).
Patient and Cancer Profiles by Age at Diagnosis: New South Wales Cancer Registry Data 2003–2015 (N = 23,619)
Children as ages 0–14, AYAs as ages 15–24, and older adults as ages 25–39. Numbers in table as case number (%).
Derived from Pearson Chi-square test.
AYAs, adolescent and young adults; SEIFA, Socio-Economic Index for Areas.
Cancer type
The most common cancers in AYAs were: carcinomas (23%), lymphomas (21%) and melanomas (15%); in children: leukemias (39%), miscellaneous specified neoplasms (20%), and CNS cancers (13%); and in adults: carcinomas (52%), melanomas (20%), and lymphomas (8%) (Table 1).
Table 2 listed the top 10 specific cancers comprising 79% of all registered cases. Of these, the most common cancers included: acute lymphoid leukemia, neuroblastoma, and acute myeloid leukemia for children; melanoma (skin), Hodgkin lymphoma, and gonadal germ cell cancer for AYAs; and melanoma, breast carcinoma, and thyroid carcinoma for adults.
Leading 10 Malignancies for Adolescent and Young Adults, Children, and Older Adults: New South Wales 2003–2015
Cancer stage
Of the 17,344 solid cancers, 11,016 (64%) were localized, whereas 6328 (36%) had spread to regional or more distant sites at diagnosis (Table 1). The proportion with advanced stage (regional or distant) was 33% in AYAs, compared with 40% in children and 37% in adults. The odds of having advanced cancers remained higher in children (aOR 1.21, 95% CI 1.01–1.47), but not in adults after adjusting for sociodemographic factors and cancer type (Table 3). Factors associated with advanced cancers included male sex, overseas born, more recent diagnosis, and cancer types other than CNS cancers (p < 0.05).
Age Differences in Cancer Stage, Initial Times to Treatment, and Survival: New South Wales 2003–2015
Children as ages 0–14, AYAs as ages 15–24, and older adults as ages 25–39; n (%): case number and %.
Adjusted OR for advanced stage (regional and distance): from multivariate logistic regression analysis adjusted for sex, country of birth, residential remoteness, SEIFA quintile, diagnosis period, and solid cancer group.1,23
Adjusted OR for initial treatment starting within 60 days after diagnosis: from multivariate logistic regression analysis adjusted for sex, country of birth, residential remoteness, SEIFA quintile, diagnosis period, and cancer group.1,23
% of 1-, 5-, 10-year cancer survival: from Kaplan–Meier disease-specific estimates; date of censoring of live cases—December 31, 2015; SHR for cancer-specific mortality from competing risk regression analysis using death of other causes other than the index cancer as competing risk; adjusting SHR from multivariate competing risk regression adjusted for sex, country of birth, residential remoteness, SEIFA quintile, diagnosis period, cancer group,1,23 and initial treatment time.
CI, confidence interval; OR, odds ratio; SHR, subhazard ratio.
The proportion of advanced stage differed by age for bone cancers, germ cell cancers, melanomas, and carcinomas, but not for other cancers. AYAs had a higher proportion of advanced lesions for bone cancers (52%) than children (36%) and adults (38%). The proportion in AYAs was higher than in adults for germ cell cancers (33% vs. 26%) and melanoma (14% vs. 10%) but lower than in children at 45%, and 30%, respectively. As for carcinomas, the proportion of advanced lesions in AYAs (45%) was the same as in children (45%) but was lower than for adults (51%), with aORs (95% CI) for advanced lesion of carcinomas being 0.98 (0.64–1.52) for children and 1.28 (1.09–1.50) (Supplementary Table S1).
Time to treatment following diagnosis
Of 23,619 cases with linked treatment data, 21,224 (90.0%) started treatment within 30 days of diagnosis, 862 (4%) in 31–60 days, 270 (1%) in 61–90 days, and 1263 (5%) in >90 days (Table 1). The mean initial treatment time following diagnosis was 34.0 days for the most common 10 malignancies listed for AYAs in Table 2. The proportion having treatment ≤60 days from diagnosis was 93% for children and AYAs. After adjusting for sex, country of birth, residential remoteness, SEIFA quintile, and diagnosis period, the odds of treatment ≤60 days was higher for children (aOR 1.39, 95% CI 1.11–1.73) and adults (aOR 1.23, 95% CI 1.06–1.44) than AYAs (Table 3). Sensitivity analysis using a cutoff of 30 days showed aORs of 1.48 (95% CI 1.24–1.79) for children and 1.18 (95% CI 1.04–1.33) for ages 25–39 years, which were similar to results for the 60-day cutoff.
Other factors with higher proportions treated within 60 days were major city residence, advantaged socioeconomic status, recent diagnosis, distant cancer stage, and cancers other than CNS cancers (p < 0.05).
Age differences in time to treatment varied by cancer type for leukemias, lymphomas, and carcinomas but not for other cancer groups. The proportion of leukemia treatments starting within 60 days was 93% in AYAs, 97% in children, and 87% in adults, giving a higher aOR of 2.35 (95% CI 1.37–4.01) for children and lower aOR of 0.62 (95% CI 0.39–0.98) for adults compared with AYAs. More AYAs received treatment within 60 days for lymphomas than children (98% vs. 95%), while fewer starting within 60 days for carcinomas (89%) than adults (94%) (Supplementary Table S2).
Cancer survival
Of the 23,619 diagnosed cases in 2003–2015, 20,618 were alive at the end of 2015, 2325 (10%) had died from their primary cancers, whereas 676 (3%) from other causes. The proportion of deaths from primary cancers in AYAs was 9%, compared with 12% in children and 10% in adults (Table 1).
The 5- and 10-year survival probabilities for AYAs were 90% and 88%, respectively, higher than the 87% and 85%, respectively, in children, but not different from adults. The risk of cancer-specific mortality (SHR) for children and adults compared with AYAs was 1.37 (95% CI 1.17–1.61) and 1.05 (95% CI 0.93–1.18). However, after adjusting for sex, country of birth, residential remoteness, SEIFA quintile, diagnosis period, cancer group, and initial treatment time, children were at significantly decreased risk of cancer mortality (aSHR 0.67, 95% CI 0.52–0.88) compared with AYAs. The risk for AYAs were not significantly different from adults (Table 3).
Other predictors of increased cancer mortality were male sex, born in Australia or other English-speaking countries, more socioeconomically disadvantaged residential area, diagnosis in earlier calendar years, advanced cancer stage, having CNS cancers compared with other cancer groups, and receiving initial treatment within 90 days (p < 0.05).
The risk of cancer mortality was not found to differ in AYAs compared with children and adults for cancer groups, except leukemias (Supplementary Fig. S1) where the risk of leukemia death in children was lower than for AYAs at aSHR 0.52 (95% CI 0.37–0.71).
Discussion
AYA cancers reflect a transition from leukemias, lymphoma, CNS cancers, bone, and soft tissue sarcoma in childhood to epithelial tumors, including melanoma and carcinomas of the breast, thyroid, cervix, and colon/rectum in adulthood. The pattern is consistent with common cancers found in Europe, the United States, and Australia wide.1,32,33
Our results indicate that stages of solid cancers at diagnosis were generally more localized in AYAs/adults than children, except where the odds of advanced stage were higher in AYAs than in other age groups (bone and germ cell cancers), and higher than in adults (melanoma). As cancers diagnosed early may be more treatable, 33 it is considered important to detect these cancers earlier in AYAs.
Results indicated that 94% of patients started initial treatment ≤60 days from diagnosis. This percentage was lower for AYAs than children and adults who had an increased odds of treatment start ≤60 days after adjustment for covariables. This finding was not universal, with lymphomas having a higher proportion of treatment starts ≤60 days for AYAs than children, and a lower corresponding proportion for leukemias than for adults.
The mean initial time to treatment from diagnosis for the most common AYA cancers was 34 days, compared with 27 days in a U.S. cohort, 14 and a median of 32 days in a Canadian cohort, 15 although direct comparisons were complicated by differences in AYA age definition, cancers types, study periods, and data sources. Our data indicated delayed treatment was associated with multiple factors, such as residing outside major cities, being socioeconomically disadvantaged, having localized stage, and having a CNS cancer compared with most other types. This is consistent with differences reported in the United States and Canada,14,15 which suggest both personal and broader social influences. 34 The later start of treatment observed outside major cities may reflect less ready access to targeted services. Further study of time to treatment by place of residence and treatment center is needed.
Overall, the 5-year survival of 90% in AYAs in NSW is equivalent to the relative survival reported nationally of 89% 1 and higher than the 83%–86% reported by EUROCARE 35 and the United States during the same period. 36 The lower survival among AYAs than children was reflected by a higher mortality in AYAs for leukemia cases, as reported from Europe, the United States, and Australia wide.35–37 Future investigation of this survival gap by leukemia subtype is needed using larger national cohorts.
This is the first comprehensive exploration of cancer profile, time to treatment, and survival in AYAs in NSW compared with children and adults. The use of linked population-based registry and administrative data was a strength, enabling adjustments for cancer types and sociodemographic factors. Causes of death were checked by registry staff facilitating estimates of cancer-specific mortality using competing survival analysis.
Study limitations included a lack of data at person level on many clinical, socioeconomic, behavioral, and family factors that may contribute to AYA differences. Initial treatment in this study was the first recorded in our data sources. It would overstate duration to treatment if earlier treatments were not recorded in these sources. This might apply to a small number treated as outpatients in public hospitals or outside NSW where there was not an MBS/PBS claim. In addition, active surveillance was not recorded as a management option.
Other study limitations included broad cancer categories, potentially masking differences by subtype. Also, for rare cancers, small case numbers would have limited the precision of estimates.
Approximately 10% of registered cancer cases did not link to treatment records and this proportion was higher for adults, males, those born in mainly non-English-speaking countries, those living outside major cities, the socioeconomically disadvantaged, and those diagnosed with distant metastases. Potentially, some of them could have migrated out of NSW post-diagnosis, such that their treatments were not recorded.
Conclusions
This study complements unadjusted results from earlier Australian research, in: (1) showing unexplained differences in stage of solid AYA cancers, time to treatment, and survival after adjusting for sociodemographic measures, tumor type, and diagnostic period; (2) indicating the importance of further research into determinants not covered by this study, such as treatment practices, quality of care, care pathways, and age-related biological differences and biomarkers; and (3) demonstrating the value of large population-based linked data that bring AYA cancer, treatment, and outcome data together from service outlets that would lack the case numbers for independent research.
Legal Authorization for This Study
Data from this project were from the Cancer Institute NSW and AIHW, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. However, data are available from the authors upon reasonable request and with permission of the NSW Cancer Institute and AIHW.
Consent for Publication
The project used already collected NSW Cancer Registry data. This project was undertaken as part of the Cancer Institute NSW's responsibilities under the Cancer Institute Act 2003. This work followed a scientific review guided by the STROBE checklist for observational studies in epidemiology.
Availability of Data and Material
Data analyzed for this article are not to be shared on any publicly available repository due to NSW privacy laws. Researchers can use these data with research Ethics Committee and data custodian approvals.
Footnotes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
