Cancer in adolescent and young adults (AYAs) aged 15–30 years accounts for approximately 2% of all malignancies, compared to <1% for children under 15 years of age.1, 2 In the year 2000, almost 25,000 AYAs were diagnosed with cancer in the USA, and almost 2,250 in Australia. Despite major improvements in outcomes for children and older adults with cancer over the last 3 decades,1, 3 recent data from the USA indicates there has been little or no improvement in survival for AYAs.1 For many cancers, survival for AYAs is significantly worse than for children.4, 5 The reasons behind the meagre improvements in survival for AYA cancer are poorly understood. Low participation rates for AYAs in clinical trials, particularly in comparison to paediatric patients, correlates with outcomes.6–8 Survival has been linked to treatment on paediatric protocols,4 or to treatment at specialized paediatric centers.6 It has been widely speculated that clinicians believe that adults tolerate treatment less well than children,9–11 and chemotherapy dose intensity in adult acute lymphoblastic leukaemia (ALL) protocols was lower than in paediatric protocols.4 Other contributing factors may include delayed clinical presentation and compliance issues.12, 13 Both host and tumor biology of AYA cancers are emerging areas of research, which are likely relevant to survival.14 Most of the available data is based on outcomes in ALL, and there is a paucity of data in other cancer types that have markedly different biology, treatments and outcomes.
To identify possible points of intervention, we analyzed Australian outcomes data for patients diagnosed with the major cancers of adolescence and young adulthood since 1982, followed by a detailed hypothesis-testing study in 179 AYAs with cancer treated at a single institution. The aims of the study are to evaluate whether a survival discrepancy exists between AYAs and children with cancer in Australia, including whether any subgroups may be particularly affected, and to identify possible explanations for any differences in the outcomes observed.
Material and methods
Approval for this study was obtained from the relevant institutional review boards. National population-based cancer registry data from the National Cancer Statistics Clearing House at the Australian Institute of Health and Welfare15 for AYAs (aged 15–30 years) and children (<15 years) between 1982 and 2002 were evaluated, focusing on cancers affecting both young people and children for purposes of comparison. In total, 14,824 patients with ALL, rhabdomyosarcoma (RMS), Ewing sarcoma (ES), osteosarcoma (OS), Hodgkin lymphoma (HL), malignant CNS tumors (CNS), thyroid cancer and melanoma were analyzed. Notification of all malignant tumors to cancer registries is mandatory in all Australian States and Territories. Cases are obtained from public and private hospitals, pathology laboratories and from State Registrars of Births, Deaths and Marriages. Tumor incidence is determined according to the multiple primary rules of the International Association of Cancer Registries. Regular linkage is made to the National Death Index to identify deaths occurring within any state or territory of Australia.
On the basis of findings from the national survey, we undertook a detailed single institution retrospective analysis of 179 AYA patients with HL, ES or OS undergoing treatment with curative intent at the Peter MacCallum Cancer Centre between 1996 and 2006. The Peter Mac is the major referral center for adult bone tumors in the state of Victoria. All patients with OS underwent treatment with chemotherapy and surgical resection. Patients with ES received chemotherapy, and surgical resection and/or radiotherapy. For OS, chemotherapy comprised doxorubicin, cisplatin, high-dose methotrexate, ifosfamide and etoposide.16–18 For ES, treatment used doxorubicin, vincristine, ifosfamide, cyclophosphamide, etoposide and actinomycin-D.5, 19 Patients with HL received treatment with doxorubicin, vincristine, bleomycin and dacarbazine, given every 2 weeks.20 All patients received granulocyte colony-stimulating factor support.
For both OS and ES, response was assessed via radiological imaging and, for patients having surgery, on histopathological assessment of the percentage tumor necrosis seen within the resected specimen. Histopathological response was defined as “good” for >90% tumor necrosis and “poor” for ≤90% tumor necrosis. For HL, radiological response was classified using the Response Evaluation Criteria in Solid Tumors.21 Toxicity measurements (nadir neutrophil and platelet counts) were obtained from full blood counts obtained in all cases between days 8 and 15 of the first cycle of chemotherapy. Data was obtained from medical files, and recorded onto standardized case report forms. Information collected included baseline and demographic data, treatment details, and follow-up and outcome data.
Survival was estimated via the Kaplan-Meier method with the log-rank test used to assess differences. Survival times were defined as the time from first diagnosis to first relapse or death from any cause for relapse free survival, and to death for overall survival. Survival analyses for patients with ES excluded those with desmoplastic small round cell tumors, which is more common in males, and whose outcomes are known to be worse. Statistical analyses were performed using STATA version 9 (StataCorp, College Station, TX). A p value of <0.05 was considered significant. Group comparisons were assessed using Pearson's χ2 test for categorical variables and a two-sided t test for continuous variables. The Cox proportional hazards model was utilized for multivariate analyses with the proportional hazards assumption examined via Schoenfeld residuals.
To study survival rates, we extracted poulation-based incidence data from a national registry (Australian Institute of health and Welfare). A total of 14,824 children and AYAs with ALL, RMS, ES, OS, HL, CNS tumors, thyroid cancer and melanoma were diagnosed between 1982 and 2002. Of these, 6188 were children <15 years of age and 8,636 were AYAs aged 15–30 years (Table I). Cancers were more common in males than females for all cancer types other than melanoma, OS in children <15 years of age and thyroid cancer in AYAs. ALL and RMS were much more common in children; HL, thyroid cancer, melanoma and OS were more common in AYAs; and ES and CNS malignancies were approximately equally distributed between the 2 age groups. For all cancers combined, overall survival (OS) from cancer was similar for children <15 compared to AYAs 15–30 years of age, and improvements in survival over the period were similar in both groups (data not shown). Differences in survival favored children for ALL, RMS, ES, OS and HL, all of which are cancers treated primarily by chemotherapy (Table II). By contrast, no significant difference in survival was seen for CNS tumors, whereas survival was better in AYAs than in children for thyroid cancer and melanoma.
Table I. Australian Cancer Cases by Age, Gender and Tumor Type 1982–2002
ALL, acute lymphoblastic leukaemia; RMS, rhabdomyosarcoma; ES, Ewing family of tumors; OS, osteosarcoma; HL, Hodgkin lymphoma; CNS, brain tumors; Thyroid, thyroid cancers.
Table II. Survival for Children (<15 Years) and Young People (15–30 Years) in Australia, Stratified by Cancer Type
5 year survival (%)
95% Confidence interval
Significance was estimated using Cox regression.
ALL, acute lymphoblastic leukaemia; RMS, rhabdomyosarcoma; ES, Ewing family of tumors; OS, osteosarcoma; HL, Hodgkin lymphoma; CNS, brain tumors.
Since puberty is a key event demarcating childhood from adolescence, we analyzed the impact of gender on survival in each group. Strikingly, for ES, OS and HL, the excess mortality seen in AYAs occurred almost entirely in males, an effect not seen in ALL or RMS. Whilst female AYAs with these cancers had comparable outcomes to children (see Table III), hazard ratios (HR) for death comparing male AYAs to male children were 2.1 (95% CI 1.6–2.9, p < 0.01) in ES, 1.6 (95% CI 1.1–2.4, p = 0.01) in OS and 1.8 (95% CI 1.2–2.8, p < 0.01) in HL. The corresponding HRs comparing male AYAs to female AYAs with ES, OS and HL were 1.8 (95% CI 1.3–2.5, p < 0.01), 1.4 (95% CI 1.0–2.0, p = 0.03) and 1.5 (95% CI 1.2–2.0, p < 0.01) respectively. Gender had no effect on survival in children <15 years of age for any cancer type except ALL (HR for males versus females of 1.3; 95% CI 1.1–1.5, p < 0.01). No significant effect of gender on outcomes was seen for CNS or thyroid cancers. The impact of gender and age on survival in AYAs with these cancers is shown in Figure 1. In addition, there was no significant variation in results seen when analyzed by time cohort, with the impact of gender remaining consistent throughout the entire observation period.
Table III. Effect of Gender on Hazard Ratios for Death from Cancer in Australian Children and Young Adults
Age (15–30 years vs. <15 years)
Gender (Males vs. Females)
Significance was estimated using Cox regression and Breslow method for ties. Data presented include the hazard ratios and the 95% confidence intervals.
ALL, acute lymphoblastic leukaemia; RMS, rhabdomyosarcoma; CNS, brain tumors; ES, Ewing family of tumors; OS, osteosarcoma; HL, Hodgkin lymphoma.
To identify factors contributing to poor survival in males with chemo-sensitive cancers of adolescence and young adulthood, a total of 179 AYA patients who underwent treatment for HL, ES or OS at Peter MacCallum Cancer Center from 1996 to 2006 were evaluated (Table IV). Factors studied included gender-related differences in stage at presentation, treatment intensity, toxicity and efficacy. Patients were followed up for a median of 37 months. There were no significant differences between the sexes in baseline features such as age, duration of symptoms at presentation, treatment adherence, and stage of disease at diagnosis or site of their primary lesion for either HL, ES or OS. Male AYAs with bone sarcomas tended to present with larger primary tumors compared to females. Other features such as past medical history, family history, smoking history, renal function and baseline alkaline phosphatase were also comparable between the 2 genders for all 3 groups (data not shown).
Table IV. Characteristics and Treatment Outcomes for Adolescents and Young Adults With Hodgkin Lymphoma, Ewing Sarcoma or Osteosarcoma Treated at Peter Maccallum Cancer Centre from 1996–2006
Age (mean ± standard error; range)
27.7 ± 1 (14–60)
29.5 ± 1.4 (16–69)
Osteosarcoma and Ewing sarcoma
Intended dose intensity (%; mean ± standard error)
87 ± 2
83 ± 2
Nadir neutrophil count in cycle 1 (×109/L; mean ± standard error)
1.52 ± 0.23
0.88 ± 0.09
2.25 ± 0.94
1.50 ± 0.62
2.24 ± 1.08
0.94 ± 0.64
1.77 ± 0.29
1.00 ± 0.19
Tumor response (%)
5 year relapse-free survival (%; 95%CI)
5 year overall survival (%; 95%CI)
Unlike brain tumors and thyroid cancer, chemotherapy is critical to cure in HL, OS and ES. There were no significant differences in the total overall intended chemotherapy dose and dose intensity received between male and female AYAs for any disease type or overall for the entire group, which was if anything greater in males (Table IV). In our cohort, the nadir neutrophil count after the first cycle of chemotherapy was significantly lower in females than males (p = 0.029) and this observation remained consistent amongst the different chemotherapy regimens. Female AYAs had an average of 2.5 episodes of neutropenic fever compared to 1.7 for males, though this did not reach statistical significance. On average, 3.3 cycles of chemotherapy were complicated by the requirement for blood transfusions in female AYAs compared to 1.9 in the male AYA group (p = 0.027), but notably female AYAs had lower baseline haemoglobin levels (126 vs. 144 g/L; p < 0.0001). No significant gender-related differences were seen in platelet levels.
For all cancers combined, the tumor response to chemotherapy favored females, with 48% of patients achieving a good response compared to 27% of males (p = 0.027). Thirty-one percent of male AYAs with OS and ES achieved a good histopathological response (defined by >90% tumor necrosis) compared to 67% of female AYAs (p = 0.008), with an average percentage tumor necrosis in males of 70.5% compared to 83.4% for females. Following stratification for regimen, the HR for relapse for males compared to females was 2.54 (95% CI 1.27–5.1, p = 0.008). Relapse-free survival data are shown in Figure 2, favoring females for all cancers (p = 0.03, Log-rank test, Mantel-Cox), ES (p = 0.06) and OS (p = 0.04). There were too few events in the HL cohort to reach significance. Multivariate analysis of gender, age and treatment regimen identified gender as significantly correlated with nadir neutrophil count (p = 0.034) and relapse (p = 0.047).
The key finding in this study is that, for chemosensitive cancers which arise predominantly in young adulthood, almost all the excess mortality seen in AYAs compared to children occurs in males. Female AYAs actually had similar outcomes to children of either sex, and no effect was seen in cancer types treated primarily surgically. The second important observation is that the gender-related differences in outcomes correlate with toxicity, a surrogate marker of pharmacodynamic effect. In our patients, the gender-related differences in survival could not be explained by stage or presentation of disease, compliance with therapy or treatment deviations. The obvious inference is that chemotherapy appears to be more toxic but also more effective in females.
It is important to put these findings in context, given the not uncommon perception that adults tolerate chemotherapy less well than children.9–11 An important recent retrospective analysis by Juergens et al. (2006) of the EURO-Ewing 99 protocol in 851 patients with ES included 355 patients aged 12–18 years and 224 patients aged 19–49 years.22 This study demonstrated that both increasing age and male gender were associated with significantly less toxicity using VIDE chemotherapy. In HL, females were reported to experience both greater toxicity and better survival than males, and the degree of neutropenia correlated with eventual outcome.23 Females experience worse toxicity than males following treatment for colorectal cancer.24–26 Although not linked to pharmacodynamic measures, better outcomes for females have also been described for ES27, 28 and OS.18, 29
Gender-related difference in outcomes may be pharmacologic in origin, provided we accept that chemotherapy is critical to cure for these cancers. Females may clear drugs more slowly, for a variety of reasons, including differential distribution into fat, slower metabolism, or slower excretion of drug. Interestingly, the AUC for fluorouracil appears higher in females,24 and doxorubicin, which is a key component of chemotherapy for HL, ES and OS, has demonstrated significant gender- and body composition-related differences in pharmacokinetic profile.30–34 Our studies add to these findings by suggesting that almost all of the survival gap in AYAs may be linked to relative underdosing of males based on current dosing algorithms.
Given the weight of evidence in the literature, the length of usage of these drugs, as well as the obvious and testable interventions that could be implemented, it is hard to understand why this hypothesis has not been addressed. It is possible that biases in study populations may be important. Gender is obviously less relevant before puberty, whereas the non-linear relationship between body surface area (BSA) and volume may also limit our ability to extrapolate from children to AYAs. BSA increases less than body volume (or mass) with growth, which is obviously a feature of adolescence and current dosing formulations in oncology are almost entirely based on derivatives of total weight and estimates of BSA. In addition, many studies have traditionally recruited a minority of AYA patients.5, 19, 35–39 In one randomized controlled trial in ES,5 only 12.6% of patients were over the age of 18, whereas combining trials in ES which reported no impact of gender on outcomes, only approximately 25% of patients included were in the AYA age group.5, 19, 35 The rarity of ES and OS in particular may present challenges of a statistical kind to stratification of outcomes by both age AND gender.
Second, there may be fundamental biases in study design regarding dose modification rules. This is particularly relevant where studies are dominated historically by the common, fundamentally chemo-insensitive, adult cancer types in which the therapeutic goals may be categorically distinct from the curable cancers seen in childhood and young adulthood. In these situations, dose modification rules are always designed to avoid toxicity, rather than achieving cure. This is potentially important where tumor response is dose-dependent, as seen in many sarcomas for doxorubicin and ifosfamide. Finally and ironically, we may be limited by our success. Perhaps the strongest evidence comes from HL,23 where the numbers are large and studies have been done. Here, it may be precisely that outcomes are generally so good that there is a lack of incentive to test the most basic of assumptions regarding the use of long-established regimens.
It important to note that our retrospective and cancer registry-based study is hypothesis-generating rather than definitive. Although sexual differentiation and body growth are fundamental to puberty, additional biological factors, such as hormonal and metabolic changes, and age-dependent changes in cancer biology may also be important. The factors identified here may also not be pertinent for all cancer types, as exemplified by RMS, ALL and melanoma. However, the importance of the hypotheses raised here is that chemotherapy dosing presents immediately testable implications for clinical trial design in AYAs and in cancers. Improving survival for AYA cancer may not solely revolve around novel treatments, but also more effective use of our current armamentarium. If correct, the long-advocated concept of individualized dosing23, 40–43 may find significant application in cancers which affect AYAs. Finally, our report highlights the relative lack of research specifically within the AYA population and the under-representation of this age group in clinical trial participation, which runs parallel to the lack of improvement in survival in the United States of America.8 We strongly advocate for efforts to increase recruitment of AYA patients onto clinical trials, so that these and other hypotheses can be tested.
The authors acknowledge the invaluable assistance of the Australian Institute of Health and Welfare (AIHW) and the Australasian Association of Cancer Registries (AACR) in providing the national data for analysis. D.M.T. is supported by a Victorian Cancer Agency Clinician Scientist fellowship.