Excess mortality in young cancer survivors compared with the general population in Italy: a retrospective study from the Italian population-based cohort of adolescents and young adult cancer survivors

Background: Adolescents and young adults (AYA) cancer survivors experience increased morbidity and mortality from second cancers, cardiovascular, infectious, kidney, and other chronic diseases. We aim to calculate all-causes cancer and non-cancer excess mortality of young cancer survivors compared to the general population.

Methods: The AYA cohort includes cancer patients diagnosed between 1976 and 2013 and alive at 5 years after diagnosis in 30 population-based Cancer Registries and followed up until 31 December 2019. The standardised mortality ratios (SMRs) and absolute excess risks (AERs) per 100,000 for person-years were calculated.

Results: 58,387 5-year survivors were followed up for 427,130 person-years; the median follow-up time was 5.7 years beyond the 5^th year after diagnosis. During this time, 4,194 (7.2%) had died by the end of 2019, and only 1.6% were lost to follow-up. Compared with the general population, AYA survivors had higher mortality, overall, the SMR for all-cause mortality was 7.0 (95%CI: 6.8-7.2). The excess of mortality was higher in the first period after diagnosis (5–10 years), SMR 12.8 (95%CI 12.3-13.3), then it decreased, reaching an SMR of 2.2 (95%CI 1.6-3.2) after 30 years.

Conclusions: The excess mortality is mostly due to the malignancy of the primary tumour, but an about 2-fold excess of mortality is also appreciable for non-cancer causes. Young adult cancer survivors face a sevenfold increase in all-cause mortality compared to the general population, with a notable rise in both cancer-related and non-cancer deaths. Thirty years post-diagnosis, the excess risk from cancer and non-cancer causes becomes nearly equal.

1 Introduction

Cancers in adolescents and young adults (15–39 years at cancer diagnosis) are rare. About 5% of all new cases occurs among ages 15 to 39 (1, 2). More than 80% of AYAs diagnosed with cancer will survive their cancer for 5 years after diagnosis (3, 4). Nevertheless, cancer is the most common cause of disease-related death in adolescents and young adults (AYAs) in high-income countries (HICs) (5, 6).

The most common cancer type among AYAs in Europe is female breast cancer, with an age-adjusted rate of 24.4 new cases per 100,000 female AYAs, followed by testicular cancer (14.5 per 100,000 male AYAs), cervix (8.5, per 100,000 female AYAs), thyroid cancer (8.4 per 100,000 AYAs), and melanoma (8.0 per 100,000 AYAs) (2).

In the last 50 years, the survival of patients having cancers in their adolescence or early adulthood has improved in high-income countries thanks to earlier diagnosis, better therapies, and better organization of health care (7).

In AYAs, the incidence trend for all cancers increased by 1% per year (APC 1[95%CI 0.9-1.2]) from 1998 to 2006 and stabilized until 2019. Nevertheless, this overall increasing trend results from different trends of specific cancers, some of which had a stable incidence, some increased, and some decreased (8).

The high survival allowed large cohorts of AYA cancer patients to enter at older ages. However, AYA cancer survivors experience multiple long-term effects of cancer, which begin after completion of treatment, and significantly impact their quality of life and increase their risk of death (7). There are marked differences in risks and types of Subsequent Malignant Neoplasms (SMNs) among survivors, with solid tumours being notably prevalent (9). Cause-specific mortality (10) and hospitalization patterns indicate higher rates of complications from infections, respiratory conditions, and diseases related to the blood and hematopoietic system, particularly among leukaemia survivors (9).

Long-term effects, which begin during treatment and persist thereafter, may be due to several risk factors interacting together, including cancer treatment, lifestyle, genetic susceptibility, and environmental and occupational exposure (11, 12). Some of these factors are country-specific and so is their impact on cancer survivors. Thus, although cancers in AYAs are rare, country-specific research is also needed. Studies on long-term effects in AYA cancer survivors are available in different western countries (10, 11, 13), but are very limited in Italy (9, 14).

Thus, we leveraged the Italian AYA cancer survivor’s cohort (15) to describe the mortality in AYA cancer survivors and its trend across different periods, differences between primary cancer types, age at diagnosis and time since the initial diagnosis. Quantifying the excess mortality in AYA survivors compared to the general population may help in assessing their health needs, forecasting the evolution of the phenomenon, and generating hypotheses on underlying mechanisms.

2 Materials and methods

2.1 Cohort and variable definitions

An AYA 5-year cancer survivor’s cohort was established in Italy in 2018, pooling together data from 34 population-based Cancer Registries (CRs) which, altogether, cover a population of around 26 million people (43% of the Italian population).

The AYA cohort was established to study the pattern of clinical long-term effects, including mortality, estimate their excess risk, and create an infrastructure to analyse the causes of long-term effects.

The details of the inclusion criteria, the procedures for extracting and preparing the database, and the periods and geographic areas covered by the included CRs have been described elsewhere (15). Briefly, the AYA cohort has a retrospective, incident-based design. Each CR identified patients with a first cancer diagnosis between 15 and 39 years of age among their registered cases, linking them to hospital discharge records (HDRs), and the regional mortality registries. Then the cancer survivors were identified as those alive at least 5 years after the first cancer diagnosis. Primary tumours were grouped by adapting a previously used AYA cancer list (9), based on ICD-O-3 morphology and topography codes (Supplementary Table 1). The cohort included 93,291 AYAs diagnosed with cancer between 1976 and 2015, of which 67,692 were AYA cancer survivors (around 72.5% of the total incident cohort).

Since the cause of death was not available for all CRs and, for some CRs, it was not complete for some periods, to analyse the excess mortality risk of AYA cancer survivors, analyses were conducted including the periods of registration for each CR with a satisfactory level of completeness of death cause reporting, arbitrarily set to less than 20% of unknown causes of death. The resulting cohort included 58,387 AYA cancer survivors diagnosed between 1976 and 2013 from 24 CRs (Supplementary Figure 1). AYA survivors accrued 427,130 person-years of follow-up, with a median time of 6 years.

The outcomes of interest were all-cause mortality, mortality for cancer (ICD-9 codes, 140-239; ICD-10 codes, C00-D49, including recurrence of primary cancer as second cancers) and for non-neoplastic causes (ICD-9 codes, 001-139, 240-999; ICD-10 codes, A00-B99, D50-Z99). The cancer causes of death occurring in participants with a second cancer registered by the CR were manually checked and classified as probably due to the first cancer and possibly due to the subsequent cancer(s) according to cancer site and type.

All results were presented stratifying by age at diagnosis (15-19, 20-29, 30-39), sex, period of diagnosis (1976-1985, 1986-1995, 1996-2005, 2006-2013), cancer type, the time elapsed since diagnosis in years (5-10, 11-20, 21-30, 31-40), and cancer type (Supplementary Tables 2-4).

2.2 Excess risk analysis

For each AYA cancer survivor, the follow-up started in the fifth year after diagnosis and ended on 31 December 2019 or until death or migration out of the registry area, whichever occurred first.

The standardised mortality ratio (SMR) for the period of diagnosis 1976–2013 with a 95% confidence interval (95% CI) was estimated to compare the relative rate of observed deaths in patients to expected deaths in the general population. The overall excess of death was measured by calculating the Absolute Excess Risks (AERs) per 100,000 person-years by subtracting the expected number of cases from the number of observed, multiplying by 100,000, and finally dividing by person-years at risk. To calculate the expected number of deaths, the person-years (PY) at risk accrued by the AYA cancer survivors during the follow-up were matched, by sex, attained age and area covered by the CR, with the general population mortality rates, taken from the ISTAT website (16).

We estimated SMR for non-cancer causes of death by applying the sex, age, and calendar year-specific percentage of deaths for cancer causes from national statistics to the expected number of deaths. To estimate mortality from cancer and non-cancer causes, deaths with unknown causes were attributed proportionally to cancer and non-cancer deaths according to the proportion of cancer deaths in the period. SMR and AER were also stratified by the period and years since diagnosis for all causes and for cancer and non-cancer causes. Tests for trend comparing SMR and AER across different calendar periods were performed in additive Poisson regression models (17). The Nelson and Alan estimator was used to compute and plot overall all-cause cumulative mortality and stratified by years of diagnosis.

To help interpretation of SMR in survivors after 5 years, we plotted the SMR values of each cancer type according to the 5-year survival observed in a large European cohort. To make the graph meaningful, we plotted only tumour types representing a relatively homogeneous group of cancers, i.e. we excluded the group of “digestive organs”, “female genital tract”, “male genital tract”, “urinary organs”, and “head and neck tumours”. Furthermore, to make the graph less crowded and avoid the influence random fluctuations due to sparse data, we plotted only cancer types with more than 10 observed deaths in study follow-up (4).

The classification of cancer causes as probably due to the first cancer or possibly due to other cancers is presented only as a proportion of all cancer causes in a descriptive analysis, and no excess has been computed.

Analyses were performed using STATA 16.1 software.

2.3 Ethics

Ethics approval for this study was obtained from the Institutional Ethics Committee of Fondazione IRCCS Istituto Nazionale dei Tumori, study protocol number INT 0134/17.

3 Results

3.1 Description of the cohort

In the studied period, 58,387 5-year AYA cancer survivors were followed up for 427,130 person-years; the median follow-up time was 5.7 years beyond the 5^th year after diagnosis (maximum follow-up=33 years) and the mean follow-up was 7.3 (Supplementary Table 5). By the end of 2019, 4,194 (7.2%) had died, and only 1.6% were lost during the study follow-up.

Sixty-nine per cent of cases were in the age group 30 to 39, while 5.5% were in the age group 15 to 19. The number of included Cancer Registries increased over time; consequently, only 1.8% of cases were diagnosed in the 1976–85 period and 14.2% in the 1986–1995 period, while 41.6% and 42.4% were diagnosed in 1996–2005 and 2006–2013 periods, respectively. Considering the time at risk, the cases diagnosed in the first period accounted for 5.4% of person-years, those in the second period for 31.2%, and those diagnosed in the third and fourth periods for 48.2% and 15.1%, respectively. Most of the person-years (52.8%) were followed up in the period from 5 to 10 years since diagnosis, while only 0.4% of the person years were followed up more than 30 years since diagnosis.

The most frequent cancer types were breast and thyroid, with 9,823 (16.8%) and 9,163 (15.7%) cases, respectively, corresponding to 66,931 and 59,611 person-years. Other sites with high numbers of cases were lymphomas (8,561, 14.7%), melanomas (6,768, 11.6%), germ cells and trophoblastic tumours (5,639, 9.7%). The high proportion of breast and thyroid cancers, which are more common in females, means that women are 60.1% of the cohort. Moreover, these tumours and melanomas are more common in those aged 30–39 years, which represent 69% of the cases of the cohort. Despite the lower number of included CRs and consequently of cases, about one-half of the deaths (48.6%) occurred in people diagnosed in the first two study periods, i.e. 1976–1985 and 1986-1995. Breast cancers and lymphomas registered about one-half of the deaths that occurred in the cohort, with 36.2% and 13.7%, respectively. The Central Nervous System (CNS) and miscellaneous intracranial and intraspinal neoplasms had the highest proportion of patient deaths at the end of follow-up (25.3%) followed by thymic cancers (22%). Most of the deaths occurred in people diagnosed at 30 to 39 (79.4%) and 5 to 10 years after diagnosis (64.5%) (Supplementary Table 5). The all-cause cumulative mortality was 5.6% (95%CI 5.4% - 5.9%), 13.1% (95% CI 12.6% - 13.6%), and 24% (95%CI 22.2% - 26.0%) at 10, 20, and 30 years after diagnosis (Figure 1).

Figure 1

Figure 1. All-causes cumulative mortality.

In the age group 15-19, the most frequent cancer types were Lymphomas, including reticuloendothelial neoplasms, with 1,123 cases (35.4%) corresponding to 9,238.0 person-years (Supplementary Table 2). Also in the 20–29 age group, Lymphomas, including reticuloendothelial neoplasms, were the first cancer with 3,060 (20.4%) cases corresponding to 25,389 person-years. Still, Thyroid and other endocrine gland tumours with 2,673 (17.8%) cases corresponding to 18,675.1 person-years and Germ cell and trophoblastic cancer (excluding gonadal carcinomas) with 2,427 (16.2%) cases corresponding to 18,059.2 person-years were frequent (Supplementary Table 3). Breast tumours and Thyroid and other endocrine gland tumours were the most frequent cancer types in the age group 30-39, with 9,123 (22.7%) and 6,077 (15.1%) cases, respectively, corresponding to 62,088.7 and 38,165.9 person-years (Supplementary Table 4). Regarding gender, the most frequent cancer types in males were Germ cell and trophoblastic tumours (excluding gonadal carcinomas) and Lymphomas, including reticuloendothelial neoplasms, with 5,346 (22.9%) and 4,569 (19.6%) cases, respectively, corresponding to 39,402.9 and 35,898.5 person-years (Supplementary Table 6) while in females Breast tumours and Thyroid and other endocrine gland tumours were the most frequent cancer types with 9,789 (27.9%) and 7,054 (20.1%) cases, respectively, corresponding to 66,687.7 and 46,003.8 person-years (Supplementary Table 7).

3.2 Excess overall mortality

Compared with the general population, AYA survivors had higher mortality. The all-cause SMR was 7.0 (95%CI 6.8 – 7.2); this SMR corresponds to an AER of 842/100,000 (Table 1). In females, the SMR was higher than in males: 9.3 (95%CI 9.0-9.7) and 4.8 (95%CI 4.6-5.1), respectively. The SMR decreased with age at diagnosis, with 8.2 (95%CI 6.8-9.8), 8.1 (95%CI 7.5-8.7), and 6.8 (95%CI 6.6-7.0), in 15-19, 20–29 and 30–39 age groups, respectively; on the contrary, the AER increases with age going from 382, in patients aged 15-19, to 995/100,000, in those aged 30-39. The CNS and miscellaneous intracranial and intraspinal showed the highest SMR, 31.6 (95%CI 28.6-35.0). Among the most frequent cancer types, breast cancer (SMR 16.1, 95%CI 15.3-16.9) showed a higher SMR than the overall cohort, while thyroid and other endocrine glands cancers and germ cells and trophoblastic tumours showed the smallest SMRs (1.5, 95%CI 1.2-1.8 and 1.6, 95%CI 1.4-2.0, respectively). The SMR was higher in the first years after diagnosis (5 to 10 years), 12.8 (95%CI 12.3-13.3), then it decreased, reaching an SMR of 2.2 (95%CI 1.6-3.2) after 30 years since diagnosis. On the contrary, the AER had a U-shaped trend starting with the highest excess, 1106/100,000, decreasing to 548 and 517 in the mid-periods (11 to 30 years from diagnosis), and finally increasing again to 1002/100,000 after 30 years from diagnosis (Table 1).

Table 1

Table 1. Observed, expected cases and SMR with 95%CI and AER, by survivor’s characteristics at first diagnosis.

3.3 Distinguishing excess mortality due to cancer and non-cancer causes

SMRs for cancer causes reflect the overall excess mortality with similar patterns across all variables. Of note is that for AYA survivors of breast cancer the excess mortality for cancer causes is particularly high.

SMRs for non-cancer causes are quite homogeneous across all variables and cancer sites. Among cancer types with sufficiently large numbers of deaths to have a precise estimate, only two showed SMRs substantially higher than the average, lymphomas and CNS neoplasms. While the last ones have also the highest SMR for cancer causes, lymphomas have an SMR for cancer causes that is close to the average, if not lower (10.9, 95%CI 9.9-12.0) (Table 2). Among the cancer causes (3,429 deaths), those classified as probably attributable to a second cancer, and not to the cancer that led to the inclusion in the cohort, were 11.4%. Among the cancer causes (3,429 deaths), those classified as probably attributable to a second cancer, and not to the cancer that led to the inclusion in the cohort, were 11.4% (390). Looking at specific cancer sites, thyroid cancer has the highest proportion (50%) of cancer deaths due to cancers other than the first cancer, the proportion is high also for germ cell and trophoblastic (44%), male (40%) and female genital (22%), urinary tract (28%) and lymphomas (23%), while breast tumours had one of the lowest proportion (5.4%).The proportion increased dramatically with the increase of the time elapsed since diagnosis, with about 20%, 40% and over 50% after 11-20, 21–30 and 30+ years from diagnosis, respectively (Table 3).

Table 2

Table 2. Cohort of AYA cancer survivors.

Table 3

Table 3. Number survivors with a second cancer registered and deaths by cancer cause, by survivor’s characteristics at first diagnosis.

3.4 Time trends

To appreciate any time trend in the SMR it is necessary to stratify by the time elapsed since diagnosis (Table 4). The SMR in the period 5 to 10 years since diagnosis decreased from 14.9 (95%CI 12.4-17.8) in people diagnosed from 1976 to 1985 to 11.2 (95%CI 10.3-12.2) in those diagnosed in the period from 2006 to 2013 (p-value for trend <0.01), the corresponding decrease in AER is even more impressive with a three-fold reduction (p-value for trend <0.01). For the other strata, data from recent periods are sparse or not yet observed at all and there are no appreciable trends. Comparing the cumulative mortality according to the calendar period of diagnosis, we observe a decreasing trend in recent periods (Figure 1).

Table 4

Table 4. SMR and AER by period and years since diagnosis for all causes and for cancer and non-cancer causes.

3.5 Time trends of the excess in cancer and non-cancer mortality

Regarding the calendar time trend, the SMR for cancer causes shows a clear decreasing trend in the period 5 to 10 years since diagnosis (p-value for trend <0.01) (Table 4). For the deaths due to the first cancer, the trend by period can be distinguished since the beginning of follow-up, while for deaths due to second malignancies the cumulative mortality curves start to diverge later (Supplementary Figure 2). In the non-cancer causes, the calendar time trend is appreciable only when excluding the first period, i.e. the years before 1985, in which there is a large proportion of missing cause of deaths (Supplementary Figure 2). A non-significant increasing trend for the time elapsed since diagnosis is appreciable only in those diagnosed before 1985 (p-value for trend p=0.09). Similar trends occurred for AER: in the period 5–10 years since diagnosis, for cancer causes there was a decreasing trend (p-value for trend <0.01) but not in the non-cancer causes (p-value for trend p=0.19). It is worth noting that both SMR and AER for non-cancer deaths observed more than 30 years after diagnosis reaches the same magnitude (2.1 and 425/100,000, respectively) as the SMR and AER for cancer causes (2.4 and 579/100,000, respectively).

4 Discussion

Our results confirmed an excess risk of death in AYA cancer survivors compared to the general population. Although SMR decreases substantially as the years since diagnosis increase (from 13 to 2), we confirmed that even 30 years after cancer diagnosis, AYA cancer survivors have a 2-fold higher risk of death compared to the general population, i.e. who didn’t have cancer when they were young.

The SMR was higher between 5 and 10 years since diagnosis, and for those diagnosed at a younger age, but AER increased with age at diagnosis. As a rule, SMRs were higher when the mortality in the general population was low and consequently the number of expected cases is very small, as the case for younger age at diagnosis and female sex. On the contrary, the AERs are larger when the mortality in the general population is higher.

Among the cancer types, the excess mortality was higher for CNS neoplasms and breast, while the lowest was for thyroid and germ cell and trophoblastic tumours. CNS tumours are relatively rare in AYAs and very heterogeneous even across age groups: adolescents have a higher proportion of embryonal tumours and a lower proportion of high-grade gliomas than young adults (4). Histological heterogeneity and low incidence make the management of CNS tumours in AYAs difficult. Therefore, despite recent significant advances in neuro-oncology, CNS tumours continue to contribute significantly to AYA mortality (18). Thyroid and germ cell tumours are the AYA cancer with the highest survival and therefore the lowest SMR (Figure 1).

Among haematological cancers, SMR was higher for leukaemia than for lymphomas. This is likely the result of intensive treatments with immunosuppressive agents.

In lymphomas and soft tissue sarcomas survivors, the non-cancer causes were notably high despite the all-cause SMRs being close to the average. In breast cancer survivors, the opposite was observed, with low SMR for non-cancer causes and high SMR for cancer causes. Treatment for Hodgkin Lymphoma, the most common lymphoma in AYAs, often includes irradiation to the thyroid region, which increases the risk of thyroid diseases. In addition, evidence has shown that total body irradiation performed in preparation for bone marrow transplantation, which is required in some haematological cancers, results in high risks for gonadal dysfunction, thyroid dysfunction, and adrenal abnormalities, which could all contribute to worse health and non-cancer deaths. Also, for AYA sarcoma survivors, significantly increased mortality for both SMNs and non-cancer causes seemed associated with the initial receipt of chemotherapy or radiotherapy (19). Breast cancer behaviour is more aggressive in AYAs. The peculiar aggressiveness of luminal breast cancer in AYAs might partly be explained by AYA breast cancer genetic features, such as the common enrichment with GATA3 and ARID1A mutations (which predispose to endocrine resistance) and the lower prevalence of PIK3CA mutations (associated with better prognosis). Other contributing factors might be different host characteristics, such as higher basal estrogenic levels, restoration of ovarian function even after chemotherapy, and decreased compliance to hormonal therapy (20).

The SMR for cancer decreased with time elapsing from diagnosis. This could be due to the direct effects of cancer and recurrences, which decrease with time since diagnosis. For the non-cancer deaths, the trend, if any, goes in the opposite direction with increasing SMR when time elapses since diagnosis, most likely due to the long-term impact of cancer and cancer treatment on the general health conditions. In non-cancer mortality, the trend for AER is very strong and the excess mortality reaches almost the same magnitude as that due to cancer-causes 30 years after diagnosis. This observation is consistent with data from the SEER cohort (21).

4.1 Comparison with other studies

Our results are consistent with previous results that observed an excess overall mortality between 4 and 10 (13, 22). In all studies, excess cancer-cause mortality is higher in the follow up period closest to the diagnosis. On the contrary, non-cancer mortality increases mostly with age, with absolute risk of dying for non-cancer causes increasing dramatically when the survivors reach the age of fifty, for both cancer survivors and the general population (23). Therefore, the differences in the magnitude of the excess mortality, both relative and absolute, observed in different studies can be mostly explained by the different average lengths of follow-up of the studies and the relative number of person-years observed in the years close to the diagnosis and those observed in a long time (24). Furthermore, the sticky diagnosis bias, leading to overestimation of cancer causes, has been described (25, 26) and could act differently across ages, periods after diagnosis and countries.

In the analyses of calendar time trends, the SMR for cancer causes make evident the improvements in care and diagnosis across the years, consistent with previous studies (27–29). When we compare non-cancer mortality, the differences between AYAs and the general population are smaller than those observed for cancer mortality, and we do not appreciate any calendar time trend in the excess of mortality when measured as relative effect (SMR). Nevertheless, since the mortality in the general population from the ‘70s to the early 2000s decreased markedly, with an increase in life expectancy from 1976 to 2019 of 11 years (23), when we look at the absolute risk, we can appreciate a strong reduction with calendar time for both cancer and non-cancer deaths. These results are partially consistent with those of a large study conducted in the UK on children (<15 years old) cancer survivors that observed a reduction of the AERs for both cancer and non-cancer causes from the ‘60s to the early 2000s (27).

Part of the excess of cancer mortality in our cohort is due to second malignancies. Therefore, our estimate of excess mortality should be interpreted as the sum of the risk of dying from a recurrence of the primary cancer and the risk of dying of a new cancer. Not considering the deaths due to second cancers, which are 390 out of 3,664, the direction and magnitude SMRs and AERs estimates for cancer mortality would not change. Nevertheless, the proportion is higher with increased attained age and time elapsed since diagnosis, as already observed in the UK cohort (27). Thus, for cancer types with lower late mortality, the proportion of deaths due to second malignancies is close to one-half.

We showed that the excess mortality was mostly due to the malignancy in the 5–10 years following the diagnosis. Looking at the excess mortality by cancer type, it is worth looking at the long-term excess mortality in light of the 5-year survival. The population included in our study is the result of a selection that occurred in the first years after diagnosis, which differs for different types of cancers. In the case of CNS, a low 5-year survival, i.e. 61.6% (4), continues with a large excess in late mortality with an SMR larger than 30 (Figure 2). A symmetrically opposite, but consistent, behaviour can be observed for thyroid cancer and germ cell trophoblastic tumours, which have high survival in the first 5 years and small SMR for late mortality. Interestingly, for these cancers, there is no excess in non-cancer cause late mortality. This is also almost true for melanoma, which has a high 5-year survival, late overall excess mortality slightly lower than the average and almost no excess in non-cancer mortality. Finally, among those that show a similar behavior, lymphomas have both 5-year survival and late mortality close to the overall values. There are also cancers with different behaviors in 5-year survival and late mortality, as for Leukemia, with low 5-year survival and average late mortality. It is worth noting that the late excess for non-cancer causes is on the average for leukemia survivors, while it is large for lymphoma survivors, consistent with previous studies (21, 27, 30). The extreme example of this second behavior is breast cancer, with a high 5-year survival and high SMR for long-term cancer mortality. This finding is consistent with other studies (21). We cannot distinguish whether the death occurred due to a second tumour of the breast or due to relapses of the first cancer. Given the high proportion of BRCA1 and 2 mutation carriers in AYA breast cancers, second cancers are a relatively common event in these women reaching about 10% contralateral breast cancer cumulative incidence in 15 years (31). However, even for non-cancer causes and regardless of the time elapsed since diagnosis, an excess mortality has been reported that is double that of the general population.

Figure 2

Figure 2. Plot of the 5-year relative survival and late mortality SMR for cancer types.

4.2 Limitations

In Italy, Cancer Registries do not cover the entire population. Furthermore, the existing registries started their activities at different points. Therefore, despite this study including the vast majority of active cancer registries in Italy, the population under study is not the entire Italian population. In particular, very few Cancer Registries contribute to the estimates for the 1976–85 and, to a minor extent, 1986–95 periods. Time trends are surely influenced by the heterogeneity of registries included in different periods compared. Furthermore, despite including virtually all the suitable AYA cancers registered in Italy, the study has no power to investigate trends for specific cancer sites. During the study period, changes in treatment therapy efficacy and toxicity occurred, but their impact on cancer and non-cancer mortality is specific for each cancer type. In this study, we could only appreciate the average result on the population of all cancer patients.

We had to exclude a limited number of registries and short periods of registrations for some registries because the proportion of missing cause-of-death deaths was too high. This further selection had a limited impact on the overall numbers, and all the observed SMRs in the sub-cohort were very close to those observed in the whole cohort. The proportion of missing death causes in the sub-cohort is overall low (6%); nevertheless, it is not evenly distributed, and it is lower for deaths that occurred in patients diagnosed in older periods. If the missing causes are not randomly distributed between cancer and non-cancer, our cause-specific SMRs may be incorrect; the impact of this misclassification could be large on the non-cancer causes SMR in the early periods because the numbers are small, and even a few more observed cases could strongly affect the estimates.

We do not have the detailed cause-specific mortality rates for the reference population for each Cancer Registry in the older period. We could estimate cancer and non-cancer cause-specific mortality rates in the general population through the national statistics reporting the proportion of cancer causes in Italy by geographical macro-area. Thus, we cannot compute detailed cause-specific SMRs for non-cancer causes.

4.3 Implications for practice and research

In AYA cancer survivors, the excess mortality is still mostly due to cancer causes up to 30 years after diagnosis. Nevertheless, after 30 years of follow-up, when the survivors reach the age of 45 to 70, and the overall mortality rates in the general population become not negligible, the non-cancer causes equal the cancer causes. Most of these people will be cured when reaching the age when chronic diseases increase their prevalence, second cancer incidence increases, and mortality for non-cancer causes is not a rare event. Thus, particularly for those cancers with a good prognosis, therapeutic choices should be oriented at obtaining good cancer control in the short and medium term but thinking at the long-term survival that will be achievable for most patients, with a balance between the efforts for reducing oncologic risk and primary prevention of other diseases, including second malignancies, must be achieved.

The increase in survival for most cancer sites introduces new needs for AYA cancer survival, not only the issues related to reproductive health but also the healthy ageing issues must be considered in light of their fragility and high life expectancy. The survivorship plan should start from the beginning, when treatment is defined. It would be specific to and the risk factors, including the therapy, intrinsic of the disease and of the patient. The inclusion of non-cancer related specialists in the multidisciplinary tumor board could help obtain a better balance between all the aims of the therapy and survivorship (32, 33).

5 Conclusion

All-cause mortality is about 7 times higher in young adult cancer survivors than in the general population. This excess is largely due to mortality due to malignancies, but they also have 80% higher mortality for non-cancer causes. Cancer and non-cancer absolute excess risk become almost similar 30 years after diagnosis. This calls for a survivorship plan in which the prevention of recurrences and other diseases are considered together since the treatment plan though the follow up an after cure is reached.

The choice of the best survivorship plan model for each country calls for understanding the societal, geographic and cultural differences that could be achieved by a comprehensive evaluation of global and local needs. Thus, every country needs to adopt survivorship models that reflect its own health care system (34).

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement

Ethics approval for this study was obtained from the Institutional Ethics Committee of Fondazione IRCCS Istituto Nazionale dei Tumori, study protocol number INT 0134/17. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and institutional requirements.

Author contributions

PGR: Conceptualization, Data curation, Methodology, Writing – original draft, Writing – review & editing. FM: Formal analysis, Writing – original draft. PM: Data curation, Formal analysis, Writing – original draft. LM: Conceptualization, Data curation, Writing – original draft, Writing – review & editing. MV: Supervision, Visualization, Writing – review & editing. IB: Supervision, Visualization, Writing – review & editing. AB: Data curation, Formal analysis, Writing – review & editing. LB: Data curation, Formal analysis, Writing – review & editing. AT: Data curation, Funding acquisition, Writing – original draft, Writing – review & editing.

Ada working group

Manuel Zorzi, Anita Andreano, Paolo Contiero, Gianfranco Manneschi, Fabio Falcini, Marine Castaing, Rosa Angela Filiberti, Cinzia Gasparotti, Claudia Cirilli, Rosalba Amodio, Silvia Iacovacci, Maria Francesca Vitale, Fabrizio Stracci, Maria Adalgisa Gentilini, Rosario Tumino, Simona Carone, Giuseppe Sampietro, Anna Melcarne, Luciana Gatti, Lorenza Boschetti, Mariangela Corti, Magda Rognoni, Enzo Coviello, Maria Teresa Pesce, Giancarlo D’Orsi, Anna Clara Fanetti, Lucia De Lorenzis, Giuseppa Candela, Fabio Savoia, Cristiana Pascucci, Maurizio Castelli, Cinzia Storchi.

Funding

The author(s) declare that financial support was received for the research and/or publication of this article.This study was partially supported by the Italian Ministry of Health - Ricerca Corrente Annual Program 2026. This work was supported by 5×1000 Funds (2013) from the Italian Ministry of Health.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declare that no Generative AI was used in the creation of this manuscript.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc.2025.1580953/full#supplementary-material

References

1. SEER. Surveillance Epidemiology and End Results Program . Available online at: https://seer.cancer.gov/ (Accessed October 30, 2024).

Google Scholar

2. Trama A, Stark D, Bozovic-Spasojevic I, Gaspar N, Peccatori F, Toss A, et al. Cancer burden in adolescents and young adults in Europe. ESMO Open. (2023) 8:100744. doi: 10.1016/j.esmoop.2022.100744

Crossref Full Text | Google Scholar

3. SEER. Surveillance Epidemiology and End Results Program . Available online at: https://seer.cancer.gov/statfacts/html/aya.html (Accessed October 30, 2024).

Google Scholar

4. Trama A, Botta L, Stiller C, Visser O, Cañete-Nieto A, Spycher B, et al. Survival of European adolescents and young adults diagnosed with cancer in 2010-2014. Eur J Cancer. (2024) 202:113558. doi: 10.1016/j.ejca.2024.113558

Crossref Full Text | Google Scholar

5. GBD 2019 Adolescent Young Adult Cancer Collaborators. The global burden of adolescent and young adult cancer in 2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Oncol. (2022) 23:27–52. doi: 10.1016/S1470-2045(21)00581-7

PubMed Abstract | Crossref Full Text | Google Scholar

6. Barr RD, Ferrari A, Ries L, Whelan J, and Bleyer WA. Cancer in adolescents and young adults: A narrative review of the current status and a view of the future. JAMA Pediatr. (2016) 170:495–501. doi: 10.1001/jamapediatrics.2015.4689

PubMed Abstract | Crossref Full Text | Google Scholar

7. Armstrong GT, Chen Y, Yasui Y, Leisenring W, Gibson TM, Mertens AC, et al. Reduction in late mortality among 5-year survivors of childhood cancer. N Engl J Med. (2016) 374:833–42. doi: 10.1056/NEJMoa1510795

PubMed Abstract | Crossref Full Text | Google Scholar

8. National Cancer Institute. Childood Cancer Data Initiative. National Childood Cancer Registry Explorer. Available online at: https://nccrexplorer.ccdi.cancer.gov/application.html?site=1&data_type=4&grap_type=5&compareBy=sex&chk_sex_1=1&series=9&race=1&age_range=8&advopt_precision=1&advopt_show_ci=on&hdn_view=0&advopt_display=2resultsRegion0 (Accessed July 11, 2025).

Google Scholar

9. Trama A, Vener C, Lasalvia P, Bernasconi A, and Ada Working Group. Late mortality, subsequent Malignant neoplasms and hospitalisations in long-term survivors of adolescent and young adult hematological cancers. Front Oncol. (2022) 12:823115. doi: 10.3389/fonc.2022.823115

PubMed Abstract | Crossref Full Text | Google Scholar

10. Armenian SH, Xu L, Cannavale KL, Wong FL, Bhatia S, and Chao C. Cause-specific mortality in survivors of adolescent and young adult cancer. Cancer. (2020) 126:2305–16. doi: 10.1002/cncr.32775

PubMed Abstract | Crossref Full Text | Google Scholar

11. Ryder-Burbidge C, Diaz RL, Barr RD, Gupta S, Nathan PC, McKillop SJ, et al. The burden of late effects and related risk factors in adolescent and young adult cancer survivors: A scoping review. Cancers (Basel). (2021) 13:4870. doi: 10.3390/cancers13194870

PubMed Abstract | Crossref Full Text | Google Scholar

12. Travis LB, Rabkin CS, Brown LM, Allan JM, Alter BP, Ambrosone CB, et al. Cancer survivorship–genetic susceptibility and second primary cancers: research strategies and recommendations. J Natl Cancer Inst. (2006) 98:15–25. doi: 10.1093/jnci/djj001

PubMed Abstract | Crossref Full Text | Google Scholar

13. Zhang Y, Goddard K, Spinelli JJ, Gotay C, and McBride ML. Risk of late mortality and second Malignant neoplasms among 5-year survivors of young adult cancer: A report of the childhood, adolescent, and young adult cancer survivors research program. J Cancer Epidemiol. (2012) 2012:103032. doi: 10.1155/2012/103032

PubMed Abstract | Crossref Full Text | Google Scholar

14. AIRTUM Working Group. Italian cancer figures - report 2013 multiple tumours. Epidemiol Prev. (2013) 37:1–152.

PubMed Abstract | Google Scholar

15. Bernasconi A, Barigelletti G, Tittarelli A, Botta L, Gatta G, Tagliabue G, et al. Adolescent and young adult cancer survivors: design and characteristics of the first nationwide population-based cohort in Italy. J Adolesc Young Adult Oncol. (2020) 9:586–93. doi: 10.1089/jayao.2019.0170

PubMed Abstract | Crossref Full Text | Google Scholar

16. ISTAT. Istituto Nazionale di Statistica . Available online at: https://demo.istat.it/ (Accessed October 30, 2024).

Google Scholar

17. Breslow NE, Day NE, and Comparisons of exposure groups. Statistical Methods in Cancer Research, Vol. II: The Design and Analysis of Cohort Studies. Breslow NE, editor. Lyon: IARC Scientifc Publications (1987) p. 81–118.

Google Scholar

18. Zapotocky M, Ramaswamy V, Lassaletta A, and Bouffet E. Adolescents and young adults with brain tumors in the context of molecular advances in neuro-oncology. Pediatr Blood Cancer. (2018) 65. doi: 10.1002/pbc.26861

PubMed Abstract | Crossref Full Text | Google Scholar

19. Youn P, Milano MT, Constine LS, and Travis LB. Long-term cause-specific mortality in survivors of adolescent and young adult bone and soft tissue sarcoma: a population-based study of 28,844 patients. Cancer. (2014) 120:2334–42. doi: 10.1002/cncr.28733

PubMed Abstract | Crossref Full Text | Google Scholar

20. Kim HJ, Kim S, Freedman RA, and Partridge AH. The impact of young age at diagnosis (age <40 years) on prognosis varies by breast cancer subtype: A U.S. SEER database analysis. Breast. (2022) 61:77–83. doi: 10.1016/j.breast.2021.12.006

PubMed Abstract | Crossref Full Text | Google Scholar

21. Moke DJ, Song Z, Liu L, Hamilton AS, Deapen D, and Freyer DR. A population-based analysis of 30-year mortality among five-year survivors of adolescent and young adult cancer: the roles of primary cancer, subsequent Malignancy, and other health conditions. Cancers (Basel). (2021) 13:3956. doi: 10.3390/cancers13163956

PubMed Abstract | Crossref Full Text | Google Scholar

22. Armenian SH and Chao C. Burden of morbidity and mortality in adolescent and young adult cancer survivors. J Clin Oncol. (2024) 42:735–42. doi: 10.1200/JCO.23.01751

PubMed Abstract | Crossref Full Text | Google Scholar

23. ISTAT. Istituto Nazionale di Statistica . Available online at: https://www.istat.it/ (Accessed October 30, 2024).

Google Scholar

24. Reulen RC, Winter DL, Frobisher C, Lancashire ER, Stiller CA, Jenney ME, et al. Long-term cause-specific mortality among survivors of childhood cancer. JAMA. (2010) 304:172–9. doi: 10.1001/jama.2010.923

PubMed Abstract | Crossref Full Text | Google Scholar

25. Innos K, Paapsi K, Alas I, Baum P, Kivi M, Kovtun M, et al. Evidence of overestimating prostate cancer mortality in Estonia: a population-based study. Scand J Urol. (2022) 56:359–64. doi: 10.1080/21681805.2022.2119274

PubMed Abstract | Crossref Full Text | Google Scholar

26. Löffeler S, Halland A, Weedon-Fekjær H, Nikitenko A, Ellingsen CL, and Haug ES. High Norwegian prostate cancer mortality: evidence of over-reporting. Scand J Urol. (2018) 52:122–8. doi: 10.1080/21681805.2017.1421260

PubMed Abstract | Crossref Full Text | Google Scholar

27. Fidler MM, Reulen RC, Winter DL, Kelly J, Jenkinson HC, Skinner R, et al. Long term cause specific mortality among 34 489 five year survivors of childhood cancer in Great Britain: population based cohort study. BMJ. (2016) 354:i4351. doi: 10.1136/bmj.i4351

PubMed Abstract | Crossref Full Text | Google Scholar

28. Anderson C and Nichols HB. Trends in late mortality among adolescent and young adult cancer survivors. J Natl Cancer Inst. (2020) 112:994–1002. doi: 10.1093/jnci/djaa014

PubMed Abstract | Crossref Full Text | Google Scholar

29. Brewster DH, Clark D, Hopkins L, Bauer J, Wild SH, Edgar AB, et al. Subsequent mortality experience in five-year survivors of childhood, adolescent and young adult cancer in Scotland: a population based, retrospective cohort study. Eur J Cancer. (2013) 49:3274–83. doi: 10.1016/j.ejca.2013.05.004

PubMed Abstract | Crossref Full Text | Google Scholar

30. Henson KE, Reulen RC, Winter DL, Bright CJ, Fidler MM, Frobisher C, et al. Cardiac mortality among 200 000 five-year survivors of cancer diagnosed at 15 to 39 years of age: the teenage and young adult cancer survivor study. Circulation. (2016) 134:1519–31. doi: 10.1161/CIRCULATIONAHA.116.022514

PubMed Abstract | Crossref Full Text | Google Scholar

31. Agarwal S, Pappas L, Matsen CB, and Agarwal JP. Second primary breast cancer after unilateral mastectomy alone or with contralateral prophylactic mastectomy. Cancer Med. (2020) 9:8043–52. doi: 10.1002/cam4.3394

PubMed Abstract | Crossref Full Text | Google Scholar

32. Yarbrough A. Survivorship in adolescents and young adults with cancer. J Natl Cancer Inst Monogr. (2021) 2021:15–7. doi: 10.1093/jncimonographs/lgab003

PubMed Abstract | Crossref Full Text | Google Scholar

33. Osborn M, Johnson R, Thompson K, Anazodo A, Albritton K, Ferrari A, et al. Models of care for adolescent and young adult cancer programs. Pediatr Blood Cancer. (2019) 66:e27991. doi: 10.1002/pbc.27991

PubMed Abstract | Crossref Full Text | Google Scholar

34. Vaz-Luis I, Masiero M, Cavaletti G, Cervantes A, Chlebowski RT, Curigliano G, et al. ESMO Expert Consensus Statements on Cancer Survivorship: promoting high-quality survivorship care and research in Europe. Ann Oncol. (2022) 33:1119–33. doi: 10.1016/j.annonc.2022.07.1941

PubMed Abstract | Crossref Full Text | Google Scholar