Medical Assessment of Adverse Health Outcomes in Long-term Survivors of Childhood Cancer FREE

The introduction of more effective treatments for childhood cancer has dramatically improved survival rates, implying that childhood cancer survivors comprise a rapidly growing group of young adults.^1- 2 Unfortunately, improved prognosis has been accompanied by the occurrence of late, treatment-related complications such as second neoplasms, organ dysfunction, and psychosocial and cognitive problems.^3- 4 Late treatment sequelae will increase the incidence of chronic diseases in survivors and ultimately reduce their life expectancy. Therefore, the need for long-term follow-up of survivors of childhood cancer is uniformly recognized.^5- 8

Research focusing on identification and characterization of high-risk populations is an essential foundation on which to build evidence-based recommendations for long-term follow-up.^6,9 Although many studies have reported on late effects of treatment in survivors of childhood cancer,^10- 21 most of these focused on only 1 late effect, had incomplete follow-up, or were performed in a survivor group of limited size. The aim of the present study was to assess the total burden of adverse health outcomes (clinical and subclinical disorders, hereafter termed “adverse events”) following childhood cancer in a cohort of childhood cancer survivors treated at the Emma Children's Hospital/Academic Medical Center (EKZ/AMC) in the Netherlands. While called adverse events, these events sometimes may not be related to the childhood cancer or its treatment. After long-term and near-complete follow-up, we assessed the prevalence of adverse events based on medical information and graded all events for severity. Subsequently, we evaluated treatment-related risk factors for a high burden of disease in survivors of childhood cancer.

All patients who were treated for childhood cancer in the EKZ/AMC between 1966 and 1996 and who survived for at least 5 years were included in the study cohort. They were identified using the Childhood Cancer Registry of the EKZ/AMC, which was established in 1966. In total, 1362 of 2596 patients survived their primary malignancy for 5 years or more. Complete data concerning cancer diagnosis and therapy were extracted from the registry.

In 1996, the EKZ/AMC started an outpatient clinic (Polikliniek Late Effecten Kindertumoren) for the assessment of late effects of childhood cancer treatment. Special attempts were made to trace and invite all 5-year survivors.

Of the 1362 survivors, 1078 visited the late-effects outpatient clinic between January 1, 1996, and January 1, 2004, and 107 were seen by their own pediatric or medical oncologist. Medical follow-up data of 93 survivors were obtained from other treating physicians, and 6 survivors filled out a mailed health questionnaire. We therefore obtained medical follow-up data until January 1, 2004, for 1284 (94.3%) of the 1362 survivors. Of the 78 survivors with missing medical follow-up data, 57 had died (with known cause of death) before the study started; 1.5% of all survivors were lost to follow-up.

In 1185 survivors who visited the late-effects clinic or their own oncologist (87% of all survivors), a full medical assessment was performed by a physician according to standardized follow-up protocols based on previous treatment modalities including medical history, physical examination, additional radiological and functional investigations, and blood analysis. Furthermore, these survivors were seen at least once by a psychologist or specialized nurse. All clinical and subclinical disorders were registered in a specially designed database. The EKZ/AMC institutional review board reviewed and approved the collection of data used for the analyses presented. Written informed consent was obtained from all participants.

All adverse events were graded by 1 of the authors (M.M.G.) according to the Common Terminology Criteria for Adverse Events version 3.0 (CTCAEv3.0, available at http://ctep.cancer.gov/forms/CTCAEv3.pdf), a scoring system developed through the US National Cancer Institute by a multidisciplinary group. The CTCAEv3.0 instrument can be used to score both acute and chronic conditions in patients with cancer and distinguishes grades 1 through 5 with unique clinical descriptions of the severity for each event (grade 1, mild; grade 2, moderate; grade 3, severe; grade 4, life-threatening or disabling; grade 5, adverse event–related death).

To investigate interobserver variability, 3 authors (M.M.G., C.vd.B., P.J.M.B.) graded 240 adverse events among randomly chosen survivors based on the CTCAEv3.0 scoring. Originally, there were 40% differences between the 3 observers. Specific events not listed in CTCAEv3.0, as well as those responsible for major inconsistencies, were discussed, and detailed coding rules were made (available on request from the authors). After these sessions, interobserver variability was assessed again, with only 5% differences observed. The remaining variation mainly concerned grade 1 or 2 psychosocial items and fatigue. There were almost no differences in grade 3 or higher adverse events.

To evaluate the total burden of adverse events, we classified survivors into 4 different groups based on the total number of events and the grade of each (Table 1). Survivors with 1 or more grade 1 event were classified as having a low burden; those with 1 or more grade 2 and/or 1 grade 3 event, a medium burden; those with 2 or more grade 3 events, or 1 grade 4 event and at most 1 grade 3 event, a high burden; and those with more grade 3/4 events or a grade 5 event, a severe burden.

The outcome of interest was the prevalence of adverse events. Multivariable logistic regression was used to evaluate treatment-related risk factors for the occurrence of any adverse event, a high or severe burden of such events, and selected events. All logistic analyses were adjusted for follow-up duration, age at diagnosis, and sex. Since the prevalence of events was high, the odds ratios from logistic regression analyses could not be interpreted as relative risks (RRs) and therefore were translated to study population–averaged RRs.²²

The 78 survivors with unknown adverse events were excluded from all multivariable analyses. Analyses were performed using SPSS version 12.0.1 (SPSS Inc, Chicago, Ill) and SAS version 8.2 (SAS Institute Inc, Cary, NC). A user-written SAS macro was used to perform the study population−averaged calculations.

The large majority (94.3%) of our 1362 survivors was diagnosed with childhood cancer before age 15 years. At the end of follow-up the median attained age of the survivors was 24.4 years, with 1194 (88%) of survivors younger than 35 years (Table 2). The median follow-up time was 17.0 (interquartile range, 11.6-23.3) years. The distribution of primary childhood cancer diagnoses by treatment category is shown in Figure 1. As of January 1, 2004, 121 of the 1362 survivors had died, the majority (72 [59.5%]) due to the primary cancer. The overall excess risk of death (excluding death from primary cancer) was increased 5.1-fold compared with the general population.²³

Of the 1362 survivors, 269 (19.8%) had no adverse events and 1015 (74.5%) had 1 or more events, whereas no medical follow-up (except for cause of death) was available for the remaining 78 (5.7%) (Table 3). The majority of survivors with medical follow-up data had more than 1 event (757/1284 [59%]), and 316/1284 (24.6%) had 5 or more events. Furthermore, 473 (36.8%) of the survivors had at least 1 severe or life-threatening or disabling disorder, and 41 (3.2%) died due to an adverse event.

Among the 1015 survivors with at least 1 adverse event, a total of 3751 events were observed. Figure 2 gives an overview of the events most frequently observed in our cohort. Almost 22% of events were severe, life-threatening or disabling, or caused death. Of those events, orthopedic disorders occurred most often (14.2%), followed by second tumors (11.9%), obesity (9.4%), fertility disorders (8.9%), psychosocial or cognitive disorders (7.9%), neurologic disorders (7.7%), and endocrine disorders (5.3%).

Of all survivors with medical follow-up data, 301 (23.4%) had a high or severe burden of adverse events, defined as at least 2 severe events or 1 or more life-threatening or disabling event. Figure 3 shows the distribution of adverse-event burden scores according to primary childhood cancer diagnosis. Survivors of bone tumors most often had a high or severe burden of events (64%), while survivors of leukemia or Wilms tumor least often had a high or severe burden of events (12% each). The distribution of adverse-event burden scores according to treatment is depicted in Figure 4. Of all patients treated with radiotherapy only, 55% had a high or severe burden of events, compared with 15% and 25% among patients treated with chemotherapy only and surgery only, respectively.

Table 4 shows treatment-specific risk factors for the development of an adverse event (excluding mild events) and a high or severe burden of disease. Survivors who received radiotherapy as part of their treatment had a significantly increased risk of an event of at least moderate severity, compared with survivors treated with surgery only, with the highest risk for survivors in the radiotherapy-only group (RR, 1.49 [95% confidence interval {CI}, 1.27-1.74]). Survivors who received radiotherapy only were also most likely to develop a high or severe burden of events (RR, 2.18 [95% CI, 1.62-2.95] vs surgery only). Survivors treated with chemotherapy only were significantly less likely to develop a high or severe burden of events than survivors in the surgery-only group (RR, 0.65 [95% CI, 0.46-0.90]). Female survivors were significantly more likely than male survivors to develop an event of at least moderate severity (RR, 1.10 [95% CI, 1.03-1.18]) and also had a greater risk of a high or severe burden of disease. Similar results were obtained when we assessed the risk of developing at least 1 severe event (grade 3-5 vs grade 0-1), although RRs for radiotherapy tended to be slightly higher than in other models.

Table 4 also shows the effects of specific chemotherapy and radiotherapy regimens on the development of adverse events. Treatment with anthracyclines, alkylating agents, or both did not affect the risk of a high or severe burden of events, compared with treatment without chemotherapy. Survivors who were treated with other chemotherapy carried a significantly lower risk (RR, 0.59 [95% CI, 0.45-0.76]), compared with treatments not involving chemotherapy. When the 3 treatment groups (anthracyclines, alkylating agents, or both) were combined and compared with the other chemotherapy group, the RR was 1.50 (95% CI, 1.10-1.89). The analysis was repeated for patients treated only with chemotherapy (without radiotherapy, central nervous system surgery, or amputation of the limb, n = 600). When the 3 different treatment groups (anthracyclines, alkylating agents, or both) were compared with treatment with other chemotherapy, treatment with alkylating agents carried the highest risk of developing a high or severe burden score (RR, 3.12 [95% CI, 1.41-6.92]). All radiation areas were associated with a significantly higher risk (about 2-fold) of developing a high or severe burden of adverse events, in comparison with treatments not involving radiation. Surgery was also associated with a significant 1.70-fold (95% CI, 1.41-2.05) increased risk of a high or severe burden of events.

Table 5 shows the treatment-specific risks of events in survivors of childhood cancer. The risk of a cardiovascular adverse event was increased in survivors treated with anthracyclines (RR, 3.53 [95% CI, 1.52-8.20] vs no chemotherapy), and an even higher risk was found for survivors treated with both anthracyclines and alkylating agents (RR, 7.41 [95% CI, 3.60-15.3]). The RRs for a cardiovascular or peripheral vascular event were also increased after radiotherapy to the thorax, abdomen, or both and radiotherapy to the extremities only (RRs of 2.36 [95% CI, 1.69-3.29] and 2.07 [95% CI, 1.20-3.56], respectively, compared with no radiotherapy). Radiotherapy involving the head and neck area was the most important risk factor for obesity (RRs from 1.63 [95% CI, 0.94-2.83] to 1.97 [95% CI, 1.05-3.70]). In addition, the risks for endocrine and neurologic events were significantly increased in survivors treated with radiotherapy involving the head and neck area (RRs from 4.25 [95% CI, 2.97-6.09] to 8.38 [95% CI, 6.00-11.7] for endocrine events and from 3.06 [95% CI, 2.20-4.26] to 3.20 [95% CI, 2.18-4.71] for neurologic events, compared with no radiotherapy). The risks of developing an endocrine or neurologic event were also increased in survivors treated with total body irradiation (RRs of 4.30 [95% CI, 3.08-6.00] and 3.28 [95% CI, 2.04-5.26], respectively). The risk of developing an endocrine event was also increased in survivors treated with radioactive iodine (iodine-131-meta-iodobenzylguanidine [MIBG]) (RR, 5.57 [95% CI, 4.21-7.38]). The risk of nephrologic events was associated with chemotherapy (RR, 5.07 [95% CI, 2.25-11.40]), especially cisplatin (data not shown).

In male survivors, the risk of fertility problems was increased in survivors treated with anthracyclines, alkylating agents, or combined treatment with these agents (RRs of 5.20 [95% CI, 1.65-16.30], 9.03 [95% CI, 3.32-24.5], and 10.6 [95% CI, 3.86-29.40], respectively, vs treatment without chemotherapy) and in survivors treated with total body irradiation (RR, 2.15 [95% CI, 1.37-3.36]) (Table 5). Since the majority of female survivors used oral contraceptives, their risk of fertility problems could not be assessed. The risk of a second malignancy was elevated more than 2-fold in survivors treated with radiotherapy. The risk of fatigue was increased in female compared with male survivors (RR, 2.77 [95% CI, 1.94-3.94]); radiotherapy involving the head and neck area was also a risk factor (RRs from 1.76 [95% CI, 1.14-2.71] to 2.43 [95% CI, 1.54-3.82]). The risk of developing any psychosocial or cognitive event was increased more than 2-fold in survivors treated with radiotherapy involving the head and neck area. Similar results were found when analyses were restricted to grade 3-5 events (radiotherapy of the head and neck: RR, 3.11 [95% CI, 1.71-5.67] [Table 6]; and radiotherapy of the head and neck and the thorax and abdomen: RR, 2.55 [95% CI, 1.07–6.11]).

After a median follow-up of 17 years, nearly 75% of childhood cancer survivors had experienced at least 1 adverse event, 40% at least 1 severe or life-threatening or disabling event, and 23.4% a high or severe burden of events. This is a high burden of disease considering the young age of our survivor population; 88% were younger than 35 years at end of follow-up. Treatment with radiotherapy only was associated with the highest risk of developing a high or severe burden of disease, while chemotherapy only was associated with significantly lower risk than surgery only.

Unique features of our study include near-complete medical follow-up (94%), with 87% of survivors being seen in our specialized late-effects clinic. In addition, all adverse events were graded for severity in a standardized manner using the CTCAEv3.0 instrument, which represents a comprehensive, multimodality grading system for reporting acute and late effects of cancer treatment. The low interobserver variability in our study, after applying more detailed coding rules, shows that CTCAEv3 is a reliable method of assessing the severity of adverse events.

Only a few studies have addressed the overall burden of adverse events in long-term survivors.^20,24- 26 Although these studies used different methods to assess such events, they all reported at least 1 event in more than 40% of survivors. Limitations of these reports include the small study size^24- 26 and the lack of a uniform scoring system²⁴ or a control group.^24- 26 In the United States, the Childhood Cancer Survivor Study (CCSS) addressed several of these limitations by assessing late adverse events in a well-characterized cohort of 5-year survivors (N = 10 397), using various methods to score events.^20,27 According to a recent study²⁰ that used the same scoring system (CTCAEv3.0) as ours did, 62.3% of the survivors had at least 1 chronic condition and 27.5% had a severe or life-threatening condition. These values are slightly lower than ours (74.5% and 36.8%, respectively). Furthermore, the proportion of survivors with 3 or more adverse events was 24% in the CCSS and 45% in our study.

Several methodological aspects of the CCSS and our study may explain these differences. Only 70% of the eligible survivor population participated in the CCSS. On the one hand, incomplete follow-up may result in an overestimation of the risk of late complications, since healthy survivors may be less likely to be traced and to participate.^28- 29 On the other hand, survivors with severe chronic conditions may have declined participation because of their poor health status, which would lead to underestimation of risk. Furthermore, information on adverse events in the CCSS was based on self-reported conditions, which were medically confirmed only for second malignancy. This may have led to both overestimation and underestimation of the incidence and severity of various chronic health conditions.²⁰ By contrast, the results of our study are based on medical information; we also determined overall burden of disease, based on the number of adverse events and the severity of each. The slightly higher proportion of events in our survivor population compared with the CCSS population may be due partly to the fact that we included adverse psychosocial outcomes and that we had much more complete information on morbidity in 5-year survivors who had already died before medical assessment of adverse events. However, when we excluded psychosocial outcomes, 35.9% of our survivors still had a severe or life-threatening condition (vs 36.8% originally), due to the fact that most survivors with a severe or life-threatening psychosocial condition also had other such conditions.

Another difference between the 2 studies is that 94% of our survivors were treated before age 15 years, whereas the CCSS also included survivors treated at ages 18 to 21 years. Furthermore, differences in cancer diagnoses and treatment could explain some differences in adverse health outcomes between the CCSS and our study. For example, the CCSS included slightly more survivors of central nervous system tumors and leukemia and slightly fewer survivors of nephroblastoma. Also, the proportions of patients receiving any radiotherapy were different: 72.9% in the CCSS and 44.6% in our study. Chemotherapy was administered to 79.0% of patients in the CCSS and to 85.6% of survivors in our study. These treatment differences may be explained partly by different treatment protocols used in the 2 studies and partly by less complete follow-up in the CCSS and a relatively high proportion of CCSS patients with unknown treatment data (15%).

A limitation of our study is that we could not compare the prevalence of adverse events in our survivors with that in a healthy population. Most of the specific events included in our analyses also occur in the general population, but the population prevalence of most events is unknown. Therefore we could only compare the prevalence of adverse events and the total burden of disease between treatment groups. Nonetheless, it is clear that the prevalence is strongly increased for most events graded 3 or higher, since they rarely occur in persons younger than 35 years.²⁰ In the CCSS report, the risk of a grade 3 or 4 event was increased 8.2-fold (95% CI, 6.9-9.7) compared with sibling controls.²⁰

Our results show that radiotherapy is the most important risk factor for a high burden of adverse events. It is important to note that 43% of survivors treated with radiotherapy only were survivors of brain or central nervous system tumors and who had received high doses of cranial radiotherapy. Radiotherapy was associated with more than 2-fold increased risks of cardiovascular, endocrine, and neurologic events, second malignancies, and psychosocial and cognitive events. Previous studies that focused on radiation-induced injury also showed prominent effects on the development of these events,^{10- 11,30- 33} but total burden of disease was not considered in these reports. A long-term mortality study in the CCSS also showed that the risk of death was increased in survivors who had received radiotherapy (RR, 2.5 [95% CI, 1.6-3.9]),³⁴ although to a lesser extent than in our study.²³ The recent report by Oeffinger et al²⁰ showed that chest irradiation combined with either bleomycin, anthracyclines, or subdiaphragmatic irradiation increased by at least 10-fold the risk of developing a severe or a life-threatening or disabling adverse event, compared with sibling controls. In interpreting the risk increase associated with radiotherapy, it should be considered that the median follow-up for irradiated patients is longer than that for those receiving chemotherapy. Although we adjusted for follow-up time, we cannot exclude the possibility that yet-unknown late effects of chemotherapy may emerge with longer follow-up.

Treatment with chemotherapy only was associated with a much lower burden of disease than radiotherapy only but also carried a lower risk than surgery only. It is important to note that surgery in pediatric oncology is associated with important long-term sequelae, for example, due to amputation and brain surgery. Although chemotherapy was associated with a relatively low burden of disease, elevated risks were noted for cardiovascular adverse events, fertility problems in male survivors, and nephrologic events. The risk of cardiovascular events was increased following anthracycline-containing chemotherapy, but alkylating agents appeared to add to this effect. While the cardiotoxicity of anthracyclines has been established,^12,35- 36 the role of alkylating agents is still controversial.³⁷ The risk of nephrologic events was especially elevated after platinum-containing chemotherapy.

When we considered adverse events of any grade, psychosocial and cognitive events were most frequently observed. Thirteen percent of these were graded 3 or higher, implying serious impact on daily life. Radiotherapy to the head and neck area was strongly associated with a high risk of developing psychosocial and cognitive events (RR, 2.07 [95% CI, 1.64-2.61]), and this result became even more pronounced when the analysis was restricted to the most serious events. Of all grade 3 or higher events, orthopedic events were most frequent, as was also observed in the recent CCSS report.²⁰

In our study, survivors of bone tumors had the highest burden of disease at end of follow-up. This is partly due to the high proportion of amputations and the occurrence of peripheral vascular events following radiotherapy to the extremities. Furthermore, patients with bone tumors were often treated with anthracyclines and cisplatin, predisposing them to cardiovascular and nephrologic adverse events. Other studies also have reported on the high risk of late effects in survivors of bone tumors.^20,27,38 More than 80% of survivors of brain tumors had a medium, high, or severe burden score of adverse events. Neurocognitive and endocrine events have been intensively studied in patients with central nervous system tumors, and adverse outcomes are generally associated with whole-brain irradiation.^{31- 32,39- 40} The increased risk of a high or severe burden of disease in survivors of irradiated brain tumors in our study is supported by CCSS data.^11,20 Forty-three percent of survivors of brain tumors in the CCSS study had 1 or more endocrine disorders, and 18% reported a cardiovascular problem. The risk of developing an endocrine disorder was increased in patients treated with MIBG (RR, 5.57 [95% CI, 4.21-7.38]), mostly due to thyroid problems following treatment. One other study also reported on endocrine disorders following treatment with MIBG.⁴¹

In conclusion, childhood cancer survivors are at increased risk of many severe health problems, resulting in a high burden of disease during young adulthood. This will inevitably affect the survivors' quality of life and also will ultimately reduce their life expectancy. Therefore, we feel that risk-stratified lifelong medical surveillance of childhood cancer survivors is needed to allow early detection of adverse events that are amenable to intervention. Future studies should focus on the efficacy of follow-up programs and other intervention strategies for adverse events, to further improve health outcomes in survivors of childhood cancer.