Conditional relative survival among long-term survivors of adolescent and young adult cancers

Abstract

Background

Many studies have examined long-term outcomes after childhood cancer, but few address outcomes for adolescent and young adult (AYA, age 15-39) cancer survivors. Conditional survival reflects changing mortality risk with time since cancer diagnosis and is a useful measure for planning long-term follow-up care.

Methods

Using the Surveillance, Epidemiology, and End Results (SEER9) database, we identified a cohort of AYAs diagnosed with a first malignant cancer during 1973-2009 and followed through 2014. We estimated 5-year relative survival at diagnosis, and at each additional year survived up to 25 years post-diagnosis, conditional on being alive at the beginning of that year.

Results

A total of 205,954 AYA cancer patients were identified. Thyroid, melanoma, testicular, breast, lymphoma, leukemia, and central nervous system tumors (CNS) made up 67% of all cancers. For all cancers combined, 5-year relative survival was 84.5% (95% CI: 84.3, 84.7) at 1 year post-diagnosis and 94.0% (95% CI: 93.9, 94.2) at 5 years. Relative survival first exceeded 95%, reflecting minimal excess mortality compared to the general population, at 7 years. Patients with thyroid, testicular, melanoma, and breast cancers reached >95% relative survival at diagnosis, 1, 3, and 18 years post-diagnosis, respectively. Estimates for Hodgkin lymphoma and leukemia were >95% at 6 and 13 years, respectively, but declined to <95% at 20 years. AYAs with CNS tumors did not reach 95% by 25 years post-diagnosis.

Conclusions

For AYA survivors of breast cancer, CNS tumors, and hematologic malignancies, long-term excess mortality should be considered when planning follow-up care.

Introduction

Over 70,000 adolescents and young adults (AYAs, ages 15-39 years) are diagnosed with cancer each year in the United States.¹ Although considerable variability exists according to cancer site and stage, five-year relative survival at diagnosis has been estimated at over 80% for all AYA cancer patients combined.² Extensive survivorship research in recent decades has focused on patients diagnosed with cancer as children and older adults, yet few studies to date have reported outcomes specifically for patients diagnosed as AYAs. With most AYA cancer patients expected to become long-term survivors, there is a critical need for research efforts to inform the survivorship care of this growing population.

Estimates of conditional survival represent clinically useful measures for cancer patients and their physicians when planning appropriate follow-up care after initial cancer treatment, and provide survivors with relevant indicators of prognosis at various stages of survivorship. Conditional survival is the probability of surviving for a specified time interval, given that one has already survived a certain length of time after cancer diagnosis, and, when plotted over time, reflects the changing risk of death with increasing time since diagnosis. Examining conditional survival among cancer patients relative to expected survival among comparable groups in the general population can reveal excess mortality remaining among cancer patients at a given point after diagnosis. Excess mortality, identified by a conditional relative survival of <100%, may reflect death from cancer recurrences, second primary cancers, or treatment-related late effects. Long-term conditional relative survival estimates have been published for various cancer sites, particularly those most common in older adults,^3–8 but have not been reported specifically among AYAs with cancer. Using population-based data from the Surveillance, Epidemiology, and End Results (SEER) program, the objective of this study was to estimate five-year relative survival for AYAs with cancer, at diagnosis and for each additional year survived (i.e. conditional five-year relative survival) up to 25 years post-diagnosis.

Methods

Using the SEER 9 database,⁹ we identified all patients diagnosed with a cancer at ages 15-39 years during 1973-2009 and followed through 2014. We included patients diagnosed with a first primary cancer with malignant behavior. Patients diagnosed with Kaposi sarcoma, death certificate or autopsy only cases, and patients recorded as alive with no survival time were excluded. Cancer site was classified according to AYA-specific recodes of ICD-O-3 site and histology codes, available through SEER.¹⁰ For testicular cancer, which does not have an AYA-specific recode, we used the regular SEER Site recode (ICD-O-3/WHO 2008). We classified stage using SEER Historic Stage A for applicable sites. Other patient characteristics obtained from SEER included age at diagnosis and sex. This study was considered exempt by the Institutional Review Board of the University of North Carolina.

Statistical analysis

Relative survival was calculated as the ratio of observed survival among cancer patients to expected survival in the general population with a similar distribution of age, sex, and race during the same time period. The actuarial method was used for survival calculations. General population mortality data, provided by the National Center for Health Statistics and accessed through SEER, was used to generate expected survival using the Ederer II method.¹¹ We estimated five-year relative survival among AYA cancer patients at diagnosis, and for each additional year survived up to 25 years post-diagnosis, conditional on being alive at the beginning of that year. Relative survival estimates >100% were adjusted and are reported as 100%. Estimates were generated for all cancer sites combined, and for selected common cancers diagnosed among AYAs. We also report the years at which conditional relative survival exceeded 90% and 95%, considered to reflect little and minimal excess mortality, respectively.³ In subgroup analyses, we examined site-specific estimates according to diagnosis year, sex, age at diagnosis, and stage. Patients with distant stage disease were excluded from analyses according to stage due to small numbers with extended follow-up for most cancer sites. Analyses were performed using SEER*Stat version 8.3.4.

Results

A total of 205,954 AYA cancer patients contributed to these analyses. Thyroid, melanoma, testicular, breast, lymphoma, leukemia, and central nervous system tumors (CNS) made up 67% of all cancers. For all cancers combined, 5-year relative survival was 84.5% (95% CI: 84.3, 84.7) and 94.0% (95% CI: 93.9, 94.2) at 1 year and 5 years post-diagnosis, respectively (Table 1; Supplemental Table 1). Relative survival first exceeded 95%, reflecting minimal excess mortality, by 7 years after diagnosis. Patients with thyroid and testicular cancers exhibited >95% five-year relative survival beginning at diagnosis and 1 year, respectively, while those with melanoma and cervical/uterine cancers reached >95% by 3 years. Relative survival increased rapidly for colorectal cancer, from 69.9% at 1 year, to 97.0% at 10 years. In contrast, gains were more gradual for female breast cancer, with >95% relative survival not achieved until 18 years after diagnosis. Estimates exceeded 95% by 7 years for soft tissue sarcomas, and 9 years for Non-Hodgkin lymphoma. Those for Hodgkin lymphoma and leukemia were >95% at 6 and 13 years, respectively, but were slightly less than 95% at 20 years. Across types of leukemia, 95% relative survival was exceeded at 8 years for patients with acute lymphoid leukemia and 9 years for those with acute myeloid leukemia, but was not achieved by 25 years post-diagnosis for those with chronic myeloid leukemia (data not shown). For patients with central nervous system (CNS) tumors, five-year relative survival did not exceed 90% by 25 years post-diagnosis. In analyses according to year of diagnosis, relative survival was generally greater for patients diagnosed in 1988-2009 compared to those diagnosed in 1973-1987. Improvements over time were observed for most cancer types, but were especially apparent for leukemias, female breast cancer, lymphomas, and CNS tumors within the first 10 years of diagnosis.

Table 1.

Five-year relative survival among AYAs with cancer, conditional on surviving to 1, 5, 10, and 20 years after diagnosis, overall and according to year of diagnosis

	Conditional 5-year relative survival
	At 1 year		At 5 years		At 10 years		At 20 years
	N survived to 1 year	5-year relative survival, % (95% CI)	N survived to 5 years	5-year relative survival, % (95% CI)	N survived to 10 years	5-year relative survival, % (95% CI)	N survived to 20 years	5-year relative survival, % (95% CI)	>90% from year:	>95% from year:
All sites	185,236	84.5 (84.3, 84.7)	155,465	94.0 (93.9, 94.2)	118,273	96.5 (96.4, 96.6)	64,148	97.1 (96.9, 97.3)	3	7
1973-1987	60,779	80.3 (80.0, 80.7)	49,224	92.6 (92.4, 92.9)	44,784	95.9 (95.7, 96.1)	39,517	97.0 (96.8, 97.3)	4	8
1988-2009	124,457	86.6 (86.4, 86.8)	106,241	94.8 (94.6, 94.9)	73,489	97.0 (96.8, 97.1)	24,631	97.2 (96.8, 97.6)	2	6
Female breast	30,421	78.3 (77.8, 78.8)	24,194	87.6 (87.2, 88.1)	17,392	92.8 (92.3, 93.2)	8,836	95.6 (95.0, 96.2)	7	18
1973-1987	10,456	71.7 (70.9, 72.6)	7,750	84.9 (84.0, 85.7)	6,483	92.1 (91.3, 92.7)	5,379	95.7 (94.9, 96.4)	8	18
1988-2009	19,965	81.7 (81.2, 82.3)	16,444	89.0 (88.5, 89.6)	10,909	93.3 (92.7, 93.8)	3,457	95.3 (93.9, 96.3)	6	17
Thyroid	20,060	99.7 (99.6, 99.8)	19,486	99.8 (99.6, 99.9)	14,804	99.7 (99.5, 99.9)	8,004	99.9 (90.9, 100.0)	0	0
1973-1987	5,540	99.4 (99.1, 99.7)	5,436	99.5 (99.1, 99.7)	5,322	99.6 (99.1, 99.8)	5,050	99.9 (94.2, 100.0)	0	0
1988-2009	14,520	99.8 (99.6, 99.9)	14,050	99.9 (99.5, 100.0)	9,482	99.8 (99.3, 100.0)	2,954	100.0	0	0
Melanoma	24,061	93.3 (93.0, 93.6)	21,999	96.9 (96.7, 97.2)	17,332	98.4 (98.1, 98.6)	9,494	99.5 (98.9, 99.8)	0	3
1973-1987	7,684	89.1 (88.4, 89.8)	6,823	95.6 (95.0, 96.1)	6,384	97.5 (97.0, 97.9)	5,808	99.2 (98.6, 99.6)	2	5
1988-2009	16,377	95.3 (94.9, 95.6)	15,176	97.6 (97.3, 97.9)	10,948	98.9 (98.6, 99.2)	3,686	99.8 (99.1, 99.9)	0	1
Non-Hodgkin lymphoma	9,794	81.9 (81.1, 82.7)	7,970	93.0 (92.4, 93.6)	5,717	95.2 (94.5, 95.9)	2,614	95.4 (94.0, 96.4)	3	9
1973-1987	2,549	75.9 (74.2, 77.6)	1,973	87.8 (86.2, 89.2)	1,698	92.5 (91.0, 93.7)	1,418	95.0 (93.4, 96.2)	7	16a
1988-2009	7,245	84.1 (83.2, 84.9)	5,997	95.0 (94.3, 95.6)	4,019	96.6 (95.8, 97.3)	1,196	95.6 (92.7, 97.3)	2	6
Hodgkin lymphoma	13,721	90.5 (90.0, 91.0)	12,409	94.8 (94.4, 95.2)	9,814	96.0 (95.6, 96.5)	5,603	93.7 (92.8, 94.4)	1	6a
1973-1987	5,119	86.6 (85.6, 87.5)	4,490	92.3 (91.4, 93.1)	4,084	94.1 (93.2, 94.8)	3,470	93.4 (92.4, 94.3)	3	–
1988-2009	8,602	92.9 (92.3, 93.4)	7,919	96.4 (95.9, 96.9)	5,730	97.7 (97.1, 98.1)	2,133	94.7 (92.9, 96.0)	0	3a
Leukemias	6,492	58.9 (57.7, 60.1)	3,883	86.9 (85.7, 88.0)	2,506	94.1 (92.9, 95.1)	1,028	94.5 (92.3, 96.0)	7	13a
1973-1987	2,073	39.7 (37.5, 41.8)	890	74.0 (70.9, 76.8)	647	89.9 (87.2, 92.1)	516	94.5 (91.6, 96.5)	11	22a
1988-2009	4,419	68.1 (66.7, 69.5)	2,993	91.3 (90.1, 92.4)	1,859	95.8 (94.5, 96.8)	512	94.2 (89.9, 96.7)	5	10a
Colorectal	7,513	69.9 (68.8, 70.9)	5,275	91.6 (90.7, 92.4)	3,759	97.0 (96.2, 97.7)	1,878	96.6 (95.1, 97.7)	5	7
1973-1987	2,144	66.1 (64.0, 68.1)	1,438	92.3 (90.7, 93.7)	1,302	97.5 (96.1, 98.4)	1,164	96.7 (94.8, 97.9)	5	7
1988-2009	5,369	71.4 (70.2, 72.6)	3,837	91.2 (90.1, 92.2)	2,457	96.6 (95.5, 97.5)	714	96.8 (93.5, 98.4)	5	8
Cervical/uterine	14,201	89.5 (89.0, 90.1)	12,481	97.3 (97.0, 97.6)	10,387	98.1 (97.8, 98.5)	6,416	97.3 (96.6, 97.8)	2	3
1973-1987	5,580	89.2 (88.3, 90.0)	4,933	97.1 (96.5, 97.6)	4,679	97.8 (97.3, 98.3)	4,231	97.2 (96.4, 97.8)	2	3
1988-2009	8,621	89.8 (89.1, 90.4)	7,548	97.5 (97.0, 97.9)	5,708	98.4 (97.9, 98.8)	2,185	97.7 (96.1, 98.6)
Testicular	15,606	96.2 (95.9, 96.6)	14,744	99.5 (99.2, 99.6)	11,930	99.3 (98.9, 99.5)	6,839	98.6 (97.9, 99.0)	0	1
1973-1987	4,774	93.0 (92.2, 93.8)	4,406	99.2 (98.7, 99.6)	4,305	98.4 (97.8, 98.9)	3,993	98.3 (97.5, 98.9)	1	2
1988-2009	10,832	97.7 (97.3, 98.0)	10,338	99.6 (99.2, 99.7)	7,625	99.7 (99.4, 99.9)	2,846	99.2 (97.2, 99.8)	0	0
CNS	8,173	66.4 (65.4, 67.5)	5,631	80.8 (79.6, 81.9)	3,657	86.5 (95.2, 87.7)	1,596	89.1 (87.1, 90.8)	–	–
1973-1987	2,707	58.8 (56.9, 60.7)	1,690	76.3 (74.2, 78.3)	1,270	85.3 (83.1, 87.2)	905	89.7 (87.3, 91.7)	–	–
1988-2009	5,466	70.2 (69.0, 71.4)	3,941	83.0 (81.6, 84.2)	2,387	87.2 (85.6, 88.7)	691	88.0 (83.8, 91.1)	15b	–
Soft tissue sarcomasc	6,794	81.4 (80.4, 82.3)	5,492	93.8 (93.0, 94.4)	4,310	97.3 (96.7, 97.9)	2,516	98.0 (96.9, 98.7)	3	7
1973-1987	2,371	81.4 (79.7, 82.9)	1,942	94.7 (93.5, 95.7)	1,808	97.6 (96.5, 98.3)	1,638	98.3 (97.0, 99.0)	3	6
1988-2009	4,423	81.4 (80.2, 82.5)	3,550	93.1 (92.1, 94.0)	2,502	97.2 (96.2, 97.9)	878	97.2 (94.3, 98.6)	3	7

Abbreviations: CNS, central nervous system tumors

^aExceeds 95% at indicated year but decreases to <95% before 25 years after diagnosis

^bExceeds 90% at indicated year but decreases to <90% before 25 years after diagnosis

^cExcludes Kaposi sarcoma patients

For most cancer sites, including melanoma, non-Hodgkin lymphoma, Hodgkin lymphoma, leukemia, colorectal cancer, and soft tissue sarcomas, conditional relative survival estimates were somewhat better for females than males, particularly in the years closest to diagnosis (Supplemental Table 2). Estimates were generally similar according to age at diagnosis, but were consistently slightly higher among those aged 15-29 than those aged 30-39 (Figure 1; Supplemental Table 3), particularly for central nervous system tumors and leukemia. Within the first few years of diagnosis, the opposite pattern was observed for female breast cancer, with slightly higher estimates for patients diagnosed at ages 30-39, though differences between age groups appeared negligible by about 7 years post-diagnosis. For colorectal cancer and soft tissue sarcomas, five-year relative survival estimates among patients with regional stage disease approached those among patients with localized disease by or before 10 years post-diagnosis (Figure 2; Supplemental Table 4). Estimates for patients with regional stage melanoma and cervical/uterine cancers exceeded 95% before 10 years, but remained slightly lower than estimates for patients with localized disease throughout the study period. Among women with regional stage breast cancer, five-year relative survival remained consistently lower than among those with localized stage, and did not exceed 95% by 25 years post-diagnosis.

Figure 1.

Five-year relative survival at diagnosis and at each additional year survived after diagnosis among AYA cancer patients according to age at diagnosis: a) all cancer sites, b) thyroid cancer, c) testicular cancer, d) melanoma, e) cervical/uterine cancers, f) colorectal cancer, g) leukemias, h) non-Hodgkin lymphoma, i) Hodgkin lymphoma, j) central nervous system tumors, k) soft tissue sarcomas (excluding Kaposi sarcoma), l) female breast cancer. Dotted line indicates 95% relative survival, or minimal excess mortality relative to the general population.

Figure 2.

Five-year relative survival at diagnosis and at each additional year survived after diagnosis among AYA cancer patients according to stage: a) thyroid cancer, b) testicular cancer, c) melanoma, d) cervical/uterine cancer, e) colorectal cancer, f) soft tissue sarcomas (excluding Kaposi sarcoma), g) female breast cancer. Dotted line indicates 95% relative survival, or minimal excess mortality relative to the general population. Distant stage disease not shown due to small sample sizes for most cancer sites.

Discussion

In this study, we examined conditional relative survival up to 25 years after diagnosis among AYAs with cancer, an understudied patient subgroup with respect to long-term health outcomes. For all cancer types combined, five-year relative survival exceeded 95% by 7 years post-diagnosis, suggesting minimal excess mortality beyond this point, and there was some evidence of a survival improvement over time. However, for specific cancer types, including central nervous system tumors, female breast cancer, Hodgkin lymphoma, and leukemia, our analyses suggested evidence of excess mortality risk that persisted, or re-emerged, more than 10 years after cancer diagnosis.

In recent years, AYAs with cancer have increasingly been distinguished from childhood and older adult cancer patients in their cancer type distribution, disease biology, treatment efficacy, and psychosocial barriers to care, and short-term relative survival estimates among AYA patients have been reported.^{2, 12} However, little research exists to inform the follow-up care of AYAs who have already survived several years beyond diagnosis. Potential contributors to excess mortality among long-term cancer survivors include cancer recurrences, second primary malignancies, and non-cancer treatment-related late effects, but long-term survival measures among AYAs with cancer have rarely been described.

Although our overall analyses indicated minimal survival deficits for AYAs with cancer after 7 years post-diagnosis, patterns of relative survival gains over time were not uniform across cancer types. Much of the high relative survival close to diagnosis was driven by a few cancer types, particularly thyroid and testicular cancers, with both high incidence and high survival in this age group. Cervical/uterine cancers and melanoma also quickly achieved >95% relative survival, though other cancer types still exhibited some excess mortality beyond five years post-diagnosis.

The slight decline in relative survival after 10 years among Hodgkin lymphoma patients is of note, and may largely reflect late effects of cancer treatment in this population.¹³ In the modern era of radiation and chemotherapy, Hodgkin lymphoma is generally considered a curable malignancy, with relatively high survival in the years immediately following diagnosis.¹³ However, studies have consistently documented significantly elevated risks of second cancers and cardiovascular disease among Hodgkin lymphoma survivors, with three- to six-fold increases in risk of coronary heart disease, valvular heart disease, and heart failure,^13–15 and two- to greater than twenty-fold increases in risk of various types of second malignancies,^{13, 16–22} compared to the general population. Anthracycline-based chemotherapies and mediastinal radiation have often been implicated in the development of cardiovascular disease in Hodgkin lymphoma survivors, while alkylating chemotherapies and radiation likely account for most of the excess in second cancers.¹³

In our analyses, conditional relative survival was lowest among AYA patients with CNS tumors. Although neither age group exceeded 95% relative survival within 25 years of diagnosis, our results demonstrated poorer long-term survival for patients diagnosed with CNS tumors at ages 30-39 than at 15-29 years. Differences according to age were also apparent for AYAs with leukemia. Patients in the 15-29 year age group exhibited minimal excess mortality after 10 years post-diagnosis, while those aged 30-39 at diagnosis showed evidence of a slight decline in relative survival after 10 years, and remained below 95% relative survival at 25 years, potentially reflecting differences in pediatric versus adult treatment regimens for leukemia.^{23, 24} These findings are informative for comparison with prior studies of long-term outcomes among childhood cancer survivors (diagnosed before age 21), which have reported somewhat higher survival at 20 years post-diagnosis for CNS and leukemia patients than we observed in older AYAs.²⁵ Our results further highlight the critical importance of long-term surveillance for patients diagnosed with these malignancies at the upper end of the AYA age range.

For all stages combined, five-year relative survival in AYA breast cancer patients remained below 95% until 18 years after diagnosis. Although differences according to stage decreased over time, women with regional breast cancer continued to have poorer survival than those with localized breast up to 25 years post-diagnosis. Late recurrences, even 10 to 20 years after diagnosis, are likely contributors to the slower gains in relative survival in breast cancer patients, particularly in those with hormone receptor-positive tumors.²⁶ However, the risk of a second primary cancer is also increased among younger breast cancer survivors,²⁷ and several studies have suggested elevated cardiovascular morbidity and mortality among women with a breast cancer history, especially those treated with radiation prior to the 1980s or with high doses of anthracycline-based chemotherapies.^28–32 Continued follow up of patients treated in the last two to three decades will be needed to gauge how the recent evolution of therapies for breast cancer, including improvements in radiation techniques,³³ decreased use of anthracyclines,³⁴ and the introduction of trastuzumab for HER2-positive breast cancers,³⁵ will affect long-term relative survival in AYA patients.

Strengths of this study include the long follow-up period and the large sample of AYA patients from population-based registries across the U.S. Our study also has limitations to consider. Recurrence information is not available in registry data, but would help to further stratify patients for evaluation of conditional survival estimates. Detailed information on cancer treatments would also be informative for assessing the potential impact of treatment-related late effects on long-term survival, but is not captured in SEER. Lastly, recent changes in cancer treatments would not be reflected in our estimates of relative survival at 20+ years post-diagnosis, since these analyses were necessarily restricted to patients diagnosed in the 1970s and 1980s. Further research will be needed to monitor trends in long-term conditional survival for AYA patients diagnosed in more recent years.

Overall, our analyses suggest that most AYAs with cancer who have survived at least 7 years after diagnosis will experience little difference in survival from the general population. However, for cancer types such as leukemia, CNS tumors, Hodgkin lymphoma, and breast cancer, the potential for long-term excess mortality should be considered when planning follow-up care.