- Email: [email protected]
Dose-Effect Relationships for Adverse Events After Cranial Radiation Therapy in Long-term Childhood Cancer Survivors
International Journal of
Radiation Oncology biology
physics
www.redjournal.org
Clinical Investigation: Pediatric Cancer
Dose-Effect Relationships for Adverse Events After Cranial Radiation Therapy in Long-term Childhood Cancer Survivors Irma W.E.M. van Dijk, MSc,* Mathilde C. Cardous-Ubbink, PhD,y Helena J.H. van der Pal, MD, PhD,y,z Richard C. Heinen, MSc,z Flora E. van Leeuwen, PhD,x Foppe Oldenburger, MD,* Rob M. van Os, MSc,* Ce´cile M. Ronckers, PhD,jj Antoinette Y.N. Schoutenevan Meeteren, MD, PhD,z Huib N. Caron, MD, PhD,y,z Caro C.E. Koning, MD, PhD,* and Leontien C.M. Kremer, MD, PhDy,z Departments of *Radiation Oncology and yMedical Oncology, Academic Medical Center; zDepartment of Pediatric Oncology, Emma Children’s Hospital/Academic Medical Center; xDepartment of Epidemiology, Netherlands Cancer Institute, Amsterdam; and jjDutch Childhood Oncology Group, Long-term Effects after Childhood Cancer, The Hague, The Netherlands Received Apr 24, 2012, and in revised form Jul 12, 2012. Accepted for publication Jul 13, 2012
Summary Adverse events (AEs) occur in the majority of long-term childhood cancer survivors, especially after cranial radiation therapy (CRT). In this retrospective study we converted CRT doses into the equivalent dose in 2-Gy fractions (EQD2) to account for fractionation dose. Using the EQD2 in multivariable models enabled us to assess dose-effect relationships for the prevalence and severity of AEs and for specific categories of AEs. Our analyses confirm that CRT
Purpose: To evaluate the prevalence and severity of clinical adverse events (AEs) and treatment-related risk factors in childhood cancer survivors treated with cranial radiation therapy (CRT), with the aim of assessing dose-effect relationships. Methods and Materials: The retrospective study cohort consisted of 1362 Dutch childhood cancer survivors, of whom 285 were treated with CRT delivered as brain irradiation (BI), as part of craniospinal irradiation (CSI), and as total body irradiation (TBI). Individual CRT doses were converted into the equivalent dose in 2-Gy fractions (EQD2). Survivors had received their diagnoses between 1966 and 1996 and survived at least 5 years after diagnosis. A complete inventory of Common Terminology Criteria for Adverse Events grade 3.0 AEs was available from our hospital-based late-effect follow-up program. We used multivariable logistic and Cox regression analyses to examine the EQD2 in relation to the prevalence and severity of AEs, correcting for sex, age at diagnosis, follow-up time, and the treatment-related risk factors surgery and chemotherapy. Results: There was a high prevalence of AEs in the CRT group; over 80% of survivors had more than 1 AE, and almost half had at least 5 AEs, both representing significant increases in number of AEs compared with survivors not treated with CRT. Additionally, the proportion of severe, life-threatening, or disabling AEs was significantly higher in the CRT group. The most frequent AEs were alopecia and cognitive, endocrine, metabolic, and neurologic events. Using the EQD2, we found significant dose-effect relationships for these and other AEs. Conclusion: Our results confirm that CRT increases the prevalence and severity of AEs in childhood cancer survivors. Furthermore, analyzing dose-effect relationships with the cumulative EQD2
Reprint requests to: Irma W. E. M. van Dijk, MSc, Department of Radiation Oncology, Academic Medical Center (AMC), University of Amsterdam, Postbox 22660, 1100 DD Amsterdam, The Netherlands, Tel: þ31-20-5666823, Fax: þ31-20-6091278; E-mail: [email protected] Int J Radiation Oncol Biol Phys, Vol. 85, No. 3, pp. 768e775, 2013 0360-3016/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ijrobp.2012.07.008
Supported by the KiKa Foundation, The Netherlands. Conflict of interest: none. Supplementary material for this article can be www.redjournal.org.
found
at
Volume 85 Number 3 2013
increases the risk for the development of AEs.
Late adverse events after cranial radiation therapy
769
instead of total physical dose connects the knowledge from radiation therapy and radiobiology with the clinical experience. Ó 2013 Elsevier Inc.
Introduction Owing to better treatment results, long-term childhood cancer survival has improved enormously. With the enhanced survival, however, the incidence of late tumor and treatment-related effects becomes more evident. Late effects occur in the majority of childhood cancer survivors, especially in those treated with cranial radiation therapy (CRT) (1, 2). For a long time, prophylactic brain irradiation (BI) has been part of therapy directed to the central nervous system (CNS) in patients with leukemia. Nowadays, prophylactic BI has been replaced by intrathecal and systemic chemotherapy (3). Nevertheless, patients with persisting or recurrent CNS disease still receive BI. Besides chemotherapy, CRT also is a major nonsurgical treatment method for high-grade CNS tumors; craniospinal irradiation (CSI) is indicated for medulloblastoma, and patients who receive total body irradiation (TBI) are also exposed to CRT. Late effects after CRT include neurocognitive and psychosocial outcomes, hearing loss, endocrine and metabolic disorders, fertility problems, and secondary tumors (4-11). The association between CRT and late effects is often analyzed by the use of dichotomous variables (ie, radiation therapy [RT] yes/no), dose categories, or total dose. However, fractionation studies have shown that late effects also depend on fractionation dose (12). Because treatment protocols have changed considerably over the years and fractionation doses vary within and between patients, the fractionation dose should also be considered in the evaluation of late effects. Our previous studies showed that the risk of secondary tumors (1, 13) was elevated in survivors treated with RT, and that RT involving the head-and-neck region increased the risk of several selected adverse events (AEs) (1). In the present study, we evaluated the prevalence and severity of AEs and treatment-related risk factors among survivors treated with CRT. Assessment of dose-effect relationships using the equivalent dose in 2-Gy fractions (EQD2) enabled us to more precisely examine the effect of CRT dose on specific categories of AEs.
Methods and Materials Patients The study population, previously described by Geenen et al (1), consisted of 1362 childhood cancer survivors diagnosed with a primary cancer at age of 0 to18 years between January 1, 1966, and January 1, 1996, in the Emma Children’s Hospital/Academic Medical Center (EKZ/AMC, Amsterdam, The Netherlands). They survived at least 5 years after diagnosis. Information on survivors’ characteristics, primary cancer diagnosis, and treatment was extracted from the EKZ/AMC Childhood Cancer Registry. From this registry, established in 1966, we identified 285 survivors who received CRT. Detailed CRT treatment data, including total dose, fractionation dose, and radiation schedule, were retrieved from their individual RT files. Since the establishment of the Outpatient Clinic for Late Effects of Childhood Cancer (Polikliniek Late Effecten
Kindertumoren; PLEK, Emma Children’s Hospital) in 1996, all living EKZ/AMC survivors were invited to this clinic for riskstratified, predefined screening and for the care of AEs. All clinically relevant AEs were extracted from the PLEK database and defined and graded as described in detail in our previous study (1) using the Common Terminology Criteria for Adverse Events (CTCAEv3.0) (14). This scoring system for both acute and late effects differentiates AEs as mild (grade 1), moderate (grade 2), severe (grade 3), and life-threatening or disabling (grade 4). Grade 5 refers to AE-related deaths.
Converting CRT dose into EQD2 CRT was defined as external beam radiation exposure of the brain resulting from these treatment types: brain irradiation (BI), craniospinal irradiation (CSI), and total body irradiation (TBI). All CRT doses including boost doses, delivered for the primary cancer or for metastatic or recurrent disease, were converted into the equivalent dose in 2-Gy fractions (EQD2) (15). The EQD2 was calculated with the formula based on the linear quadratic model: EQD2 Z D*(d þ a/b)/(2 þ a/b), in which the total dose D is the number of fractions multiplied by the fractionation dose d. The a/b ratio for late-responding brain tissue was estimated at 2 Gy (12). We calculated both the cumulative EQD2 and the EQD2 per treatment type (ie, EQD2_BI, EQD2_CSI, and EQD2_TBI). In case a survivor had undergone more than 1 treatment type, the cumulative EQD2 represents the total dose of all treatment types. For RT to other sites of the body, including all RT other than CRT, no EQD2 was calculated.
Statistics We described survivors’ characteristics and the prevalence and severity of AEs, comparing survivors who had received CRT with those who had not received CRT. Treatment-related risk factors for the prevalence and severity of AEs were evaluated with multivariable logistic regression models and reported as odds ratios (ORs). In model 1 we crudely analyzed CRT (yes/no) as a risk factor for the prevalence and severity of AEs and for specific AEs. In model 2 the cumulative EQD2 for CRT was used to assess dose-effect relationships for the prevalence and severity of AEs, and for the subset of specific AEs that showed a statistically significantly increased risk (P<.05) after CRT in model 1. In a third model, we attempted to asses dose-effect relationships per treatment type. Additionally, we performed stratified analyses for brain tumor survivors and survivors of other cancers to examine the role of primary cancer type. Besides CRT (y/n) or EQD2, all models included sex, age at diagnosis (years), follow-up time since diagnosis (years), and these treatment-related risk factors: surgery (y/n), chemotherapy (y/n), and RT other sites (y/n). We evaluated the assumptions for linearity of the EQD2; no violations of linearity assumptions were observed. Because the time-to-event for secondary tumors was known, we used Kaplan-Meier analyses to estimate the risk of both malignant and benign secondary tumors in the head-and-neck area 20 years after diagnosis. Additionally, Cox models were used to evaluate treatment-related risk factors for secondary tumors. All
770
International Journal of Radiation Oncology Biology Physics
van Dijk et al.
Table 1
Characteristics of the 1966-1996 EKZ/AMC cohort of 5 years childhood cancer survivors Number of survivors, (NZ1362) Characteristic
Sex M F Age at diagnosis childhood cancer, y Median (range) 0-5 >5-10 >10-15 >15 Primary cancer diagnosis Leukemia Lymphoma Kidney/Wilms tumor Bone tumor Soft tissue sarcoma Neuroblastoma Other* Brain/central nervous system Astrocytoma Ependymoma/medulloblastoma Intracranial germ cell tumor Primitive neuroectodermal tumor Sarcoma Calendar y of diagnosis 1966-1975 1976-1985 1986-1996 Overall treatment category Surgery only (S) Chemotherapy (CT) only, with/without S Radiation therapy (RT) only, with/without S CT þ RT with/without S first treatment, no recurrencey CT þ RT with/without S including recurrence treatmentz 131 I-MIBG with/without CT and/or RT and/or S Other Type of chemotherapy Anthracyclines with/without other chemotherapy Alkylating agents with/without other chemotherapy Anthracyclines and alkylating agents, with/without other chemotherapy Other chemotherapy only Specific agent Cisplatin Methotrexate Follow-up time, y Median (range) 5-15 >15-25 >25 Attained age at end of follow-up time, y Median (range) 5-15 >15-25 >25-35 >35
CRT, nZ285 (%)
No CRT, nZ1077 (%)
156 (54.7) 129 (45.3)
589 (54.7) 488 (45.3)
6.3 113 101 61 10
5.8 483 277 248 69
(0.0-17.6) (39.6) (35.4) (21.4) (3.5)
(0.0-17.8) (44.8) (25.7) (23.0) (6.4)
151 (53.0) 42 (14.7) 0 0 3 (1.1) 1 (0.4) 0 88 (30.9) 19 (6.7) 61 (21.4) 4 (1.4) 3 (1.1) 1 (0.4)
184 217 189 116 148 84 118 21 16 1
(17.1) (20.1) (17.5) (10.8) (13.7) (7.8) (11.0) (1.9) (1.5) (0.1) 0 4 (0.4) 0
49 (17.2) 161 (56.5) 75 (26.3)
163 (15.1) 383 (35.6) 531 (49.3)
0 0 41 (14.4) 151 (53.0) 93 (32.6) 0 0
102 650 49 175 85 15 1
(9.5) (60.4) (4.5) (16.2) (7.9) (1.4) (0.1)
33 (11.6) 61 (21.4) 77 (27.0)
77 (7.1) 199 (18.5) 375 (34.8)
73 (25.6)
155 (25.2)
2 (0.7) 177 (62.1)
75 (7.0) 285 (26.5)
19.6 102 119 64
(5.0-36.1) (35.8) (41.8) (22.5)
16.6 446 440 191
(5.0-38.1) (41.4) (40.9) (17.7)
26.3 36 92 121 36
(5.9-44.0) (12.6) (32.3) (42.5) (12.6)
24.0 169 421 355 132
(5.2-48.9) (15.7) (39.1) (33.0) (12.3)
(continued on next page)
Volume 85 Number 3 2013
Late adverse events after cranial radiation therapy
771
Table 1 (continued ) Number of survivors, (NZ1362) Characteristic
CRT, nZ285 (%)
Vital status at end of follow-up Alive Dead
No CRT, nZ1077 (%)
230 (80.7) 55 (19.3)
1011 (93.9) 66 (6.1)
Abbreviations: CRT Z cranial radiation therapy; EKZ/AMC Z Emma Children’s Hospital/Academic Medical Center; 131I-MIBG Z iodine-131-metaiodobenzylguanidine. * Including 43 gonadal germ cell tumors, 21 retinoblastomas, 13 thyroid carcinomas, 12 hepatoblastomas, 11 malignant histiocytoses, and 18 miscellaneous tumors. y Also treated with RT to other sites: mediastinum with/without abdomen (nZ10). z Also treated with RT to other sites: mediastinum with/without abdomen (nZ3), orbit (nZ4), testicles (nZ3), neck (nZ2), breast (nZ1), and brachytherapy of the mastoid (nZ1).
meta-iodobenzylguanidine [131I-MIBG]). The remaining 753 survivors had not received RT. We obtained detailed treatment information including fractionation doses for 95.8% (273/285) of survivors. This enabled us to calculate the EQD2_BI for 183 of 189 survivors who had received BI, the EQD2_CSI for 82 of 83 CSI survivors, and the EQD2_TBI for 23 of 29 TBI survivors. Sixteen of the 189 BI survivors additionally received CSI or TBI for a recurrence (Table 2).
data were analyzed with SPSS version 18.0 (SPSS, Chicago, IL) and the statistical program R version 2.13.0 (http://www.Rproject.org).
Results Table 1 shows the characteristics of the study cohort. The median follow-up time for all survivors was 17.0 years. The majority of the CRT group consisted of leukemia survivors, who were younger at diagnosis than were brain tumor survivors in the same group (median ages 4.7 and 8.3 years, respectively). Besides CRT, 24 survivors (8.4%) received RT to other sites of the body. Of the survivors not treated with CRT, 324 (30.1%) received RT to other sites of the body, including orbital RT, nasopharyngeal RT, brachytherapy, and treatment with radioactive iodine (iodine-131-
Table 2
Prevalence and severity of AEs Figure E1 presents an overview of all AEs. The most frequently scored AEs in the CRT group were cognitive disorders (14.4%), endocrine disorders (12.2%), neurologic disorders (9.2%), and
Cranial radiation therapy in the 1966-1996 EKZ/AMC cohort of 5 years childhood cancer survivors by primary tumor type. CRT, nZ285 (%) CRT dose; EQD2 (Gy) (median, range)
Treatment group Brain irradiation Whole brain Partial brain Whole brain with boost Unknown Craniospinal irradiation With boost Without boost Unknown Total body irradiation (TBI) TBI Unknown
Brain tumor, nZ88 (30.9) 31 (35.2) 0 21 (23.9) (51.3, 43.2-94.6) 6 (3.0) (50.2, 34.2-55.2) 4 (4.5) 57 (64.8) 53 (60.2) (49.3, 37.0-53.6) 3 (3.4) (42.9, 42.9-42.9) 1 (1.1) 0 0
Other cancer, nZ197 (69.1) 158* (80.2) 144 (73.1) (24.8, 13.6-48.3) 7 (3.6) (30.0, 9.6-54.1) 3 (1.5) (43.5, 22.3-47.9) 4 (2.0) 26 (13.2) 0 26 (13.2) (21.0, 13.1-45.6) 0 29 23 (11.8) 16.4 (0.8-14.3) 6 (3.0)
Abbreviations: CRT Z cranial radiation therapy; EKZ/AMC Z Emma Children’s Hospital/Academic Medical Center; EQD2 Z equivalent dose in 2-Gy fractions. * Includes survivors also treated with craniospinal irradiation (nZ11) or total body irradiation (nZ5).
772
International Journal of Radiation Oncology Biology Physics
van Dijk et al.
alopecia (9.0%). Table E1 shows that 6.3% of the survivors in the CRT group had no AEs, vs 23.3% in the non-CRT group. Of all survivors with known AEs, 81.2% of the CRT group and 53.4% of the non-CRT group had more than 1 AE. Furthermore, 46.3% of the CRT group had at least 5 events, vs 19.2% of the non-CRT group. Additionally, at least 1 severe, life-threatening, or disabling event was present in 45.1% of the CRT group, vs 34.8% of the non-CRT group. The proportion of AE-related deaths was twice as high in the CRT group as in the non-CRT group.
Treatment-related risk factors for AEs Figure E2, based on multivariable logistic regression models, shows the exponentially increasing risk for the development of any AE and of severity of AEs with increasing CRT doses in the whole cohort. Table 3 shows the results of the multivariable logistic regression analyses for treatment-related risk factors for the prevalence and severity of AEs, stratified by primary cancer type (brain tumor vs other cancers). CRT (y/n) significantly increased the risk of any AE in brain tumor survivors and in survivors of other cancers. CRT was also a significant risk factor for the severity of AEs (ie, for mild to moderate AEs vs no AEs, and for severe, lifethreatening, or disabling events and AE-related deaths vs no AEs [model 1]). Replacing CRT (y/n) by the cumulative EQD2 showed significantly increased risks of 1.07 Gy1 and 1.06 Gy1 for any AE in respectively brain tumor survivors and in survivors of other cancer types (model 2). The risk estimates were generally similar for different definitions of severity. We also tried to assess doseeffect relationships per treatment type using the EQD2_BI, EQD2_CSI, and EQD2_TBI, but each treatment group contained too few survivors and/or events (model 3, not shown).
Treatment-related risk factors for selected AEs In the whole cohort, CRT (y/n) significantly increased the risk of a dozen of outcome categories (results not shown). Dose-effect
relationships for these outcome categories were further analyzed with the cumulative EQD2 and stratified by primary cancer typedthat is, for brain tumor survivors and survivors of other childhood cancers separately. Figure 1 illustrates that the risk of selected AEs increases exponentially with increasing EQD2 among brain tumor survivors and survivors of other cancers, with different ORs in both groups. The analyses (Table E2) (model 2) showed significant dose-effect relationships for alopecia and for endocrine and cognitive events in both survivor groups. Survivors of childhood cancers other than brain cancer also had significantly increased risks of fatigue, headache, hearing loss, male fertility problems, short stature, and metabolic and neurologic events. The risk of male fertility problems and social-emotional problems was slightly higher in brain tumor survivors. The risk estimates for EQD2_BI, EQD2_CSI, and EQD2_TBI in model 3 (not shown) were generally similar to those of the cumulative EQD2 in model 2. However, for several selected AEs, it was not feasible to assess dose-effect relationships by treatment type because of the low number of survivors and/or AEs in each treatment group.
Secondary tumors in the head-and-neck area Of the secondary tumors in the CRT group, 29 occurred as solid tumors in the head-and-neck area (Fig. E1), counting 5 thyroid neoplasms, 6 basal cell carcinomas of the skin, 6 CNS tumors, 8 meningiomas, and 4 miscellaneous. The meningiomas occurred in survivors treated for leukemia (nZ6; median EQD2_BI Z 20.9 Gy), lymphoma (nZ1; EQD2_TBI Z 20.0 Gy), and medulloblastoma (nZ1; EQD2_CSI Z 45.9 Gy) after a median latency time of 17.8 years (range, 8.6-23.8 years). Figure 2 shows the Kaplan-Meier curves for secondary tumors in the head-and-neck area (including meningiomas) and for meningiomas separately. In survivors who had received CRT, the estimated cumulative risk for secondary tumors in the head-and-neck area 20 years after diagnosis was 10.2% (95% CI, 5.9%-14.5%), and 3.5% (95% CI,
Table 3 Treatment-related risk factors for severity of adverse events in the 1966-1996 EKZ/AMC cohort of 5 years childhood cancer survivors (NZ1362) stratified by primary cancer type Any adverse event by severity OR (95% CI) Grades 1-5 vs no AE Risk factor Model 1* CRT (y/n) RT-other-sites (y/n) Model 2* CRT; EQD2y RT-other-sites (y/n)
Grades 1-2 vs no AE
Grades 3-5 vs no AE
Brain tumor
Other cancers
Brain tumor
Other cancers
Brain tumor
Other cancers
30.7 (2.26-416) NAx
3.99 (2.26-7.06) 7.69 (4.42-13.4)
163 (1.25-2.14$104) NAx
2.99 (1.64-5.48) 5.99 (3.32-10.8)
27.8 (1.84-420) NAx
6.37 (3.33-12.2) 10.5 (5.91-18.8)
1.07 (1.01-1.13) NAx
1.06 (1.03-1.08) 7.61 (4.37-13.2)
1.11 (1.00-1.24)z NAx
1.04 (1.01-1.07) 5.93 (3.29-10.7)
1.07 (1.01-1.13) NAx
1.07 (1.04-1.10) 10.3 (5.79-18.4)
Abbreviations: AE Z adverse event; CI Z confidence interval; CRT Z cranial radiation therapy; EKZ/AMC Z Emma Children’s Hospital/Academic Medical Center; EQD2 Z equivalent dose in 2-Gy fractions; NA Z not applicable; OR Z odds ratio; RT Z radiation therapy. * Each model represents a multivariable logistic regression analysis including the variables listed for that particular model and corrected for sex, age at diagnosis, follow-up time, and these treatment-related risk factors: surgery and chemotherapy. y OR for EQD2 represents the increase in risk per Gray. Brain tumor survivors (nZ109): model 1 excluded 8 survivors with lack of information on AEs, model 2 excluded 12 survivors with lack of information on AEs and/or EQD2. Survivors of other cancers (nZ1253): model 1 excluded 70 survivors with lack of information on AEs, model 2 excluded 74 survivors with lack of information on AEs and/or EQD2. z P<.05. x An analysis for brain tumor survivors was not feasible because of the exceptionally low numbers in most of the cells.
Volume 85 Number 3 2013
Late adverse events after cranial radiation therapy
773
Fig. 1. Dose-effect relationships for selected adverse events (AEs) in the 1966-1996 EKZ/AMC cohort of 5 years childhood cancer survivors (nZ1362) stratified by primary cancer type. ORs result from multivariable logistic regression models, corrected for sex, age at diagnosis, follow-up time, surgery, chemotherapy, and radiation therapy to other sites of the body. ORs represent the increase in risk per Gray. Brain tumor survivors (nZ109); logistic regression excluded 12 survivors with lack of information on AEs and/or EQD2. Survivors of other cancers (nZ1253); logistic regression excluded 74 survivors with lack of information on AEs and/or EQD2. Male fertility; male brain tumor survivors (nZ66); logistic regression excluded 5 survivors with lack of information on AEs and/or EQD2, male survivors of other cancers (nZ679); logistic regression excluded 44 survivors with lack of information on AEs and/or EQD2. Abbreviations: OR Z odds ratio; CI Z confidence interval; EQD2 Z equivalent dose in 2-Gy fractions; SD Z standard deviation.
774
International Journal of Radiation Oncology Biology Physics
van Dijk et al.
Fig. 2. Kaplan-Meier curves for secondary tumors in the headand-neck area (including meningioma), and for meningioma separately. Abbreviation: CRT Z cranial radiation therapy. 0.8%-6.2%) for meningiomas. Cox regression analyses (Table 4) showed that both CRT and RT to other sites increased the risk of secondary tumors in the head-and-neck area (HREQD2, 1.04 Gy1 [CI, 1.02-1.05]; HRRTothersites, 2.83 [CI, 1.55-5.15]). CRT was the only significant risk factor for meningiomas (HREQD2, 1.04 Gy1 [CI, 1.01-1.08]).
Discussion Our results show a high prevalence of AEs in long-term childhood cancer survivors treated with CRT. Over 80% of survivors in the
CRT group had more than 1 AE, and almost half had at least 5 AEs, both representing significant increases in the number of AEs compared with survivors not treated with CRT. AEs after CRT were also significantly more often severe, life-threatening, or disabling. In previous studies (1, 13) we investigated the role of radiation therapy without assessing dose-effect relationships. In this study we particularly focused on CRT, with the aim of examining late effects in relation to CRT dose. We demonstrated that CRT increased the risk of alopecia; endocrine disorders; hearing loss; fatigue; male fertility problems; metabolic, neurologic, and ophthalmologic events; headache; social-emotional, and cognitive problems; short stature; and secondary tumors in the head-andneck area. Moreover, using the cumulative EQD2, we assessed significant dose-effect relationships for these selected AEs. Unexpectedly, the analyses stratified on primary cancer type showed slightly higher ORs for dose-effect relationships in survivors treated for other cancers than in survivors treated for brain tumor. A possible explanation may be that most survivors of other cancers were leukemia survivors who had received their diagnoses at younger ages than brain tumor survivors. Second, we did not include irradiated volume in our analyses, and brain tumor survivors received high doses in relatively small fields. Third, it seems that the brain tumor itself and/or brain surgery levels off the effect of radiation dose. Meningiomas occurred after a median interval of almost 18 years, which is similar to median latency times as reported in other studies (10, 11). Factors that increase the risk of meningioma include younger age at time of treatment, longer follow-up time, and higher CRT dose. In our cohort, meningiomas mainly occurred in leukemia survivors, who generally received lower CRT doses than did brain tumor survivors, but who were treated at younger ages and had longer follow-up times.
Table 4 Cox regression analyses for treatment-related risk factors of secondary tumors in the 1966-1996 EKZ/AMC cohort of 5 years childhood cancer survivors (NZ1362). Secondary tumor, HR (95% CI) Risk factor Model 1* Sex (F/M) Age at diagnosis Surgery (y/n) Chemotherapy (y/n) Surgery and/or chemotherapy (y/n) CRT (y/n) RT-other-sites (y/n) Model 2* Sex (F/M) Age at diagnosis Surgery (y/n) Chemotherapy (y/n) Surgery and/or chemotherapy (y/n) CRT; EQD2y RT-other-sites (y/n)
Head-and-neck area (incl. meningioma)
Meningioma
1.38 0.97 0.81 0.63
(0.80-2.35) (0.91-1.03) (0.42-1.54) (0.31-1.28) NA 5.08 (2.66-9.72) 3.07 (1.68-5.63)
0.30 (0.06-1.45) 1.00 (0.84-1.18) NA NA 1.34 (0.25-7.15) 34.6 (3.48-344) 1.71 (0.32-9.25)
1.45 0.96 0.52 0.72
0.33 (0.07-1.59) 0.98 (0.84-1.15) NA NA 0.49 (0.10-2.44) 1.04 (1.01-1.08) 0.98 (0.18-5.43)
(0.84-2.49) (0.90-1.02) (0.29-0.94) (0.35-1.47) NA 1.04 (1.02-1.05) 2.83 (1.55-5.15)
Abbreviations: CRT Z cranial radiation therapy; EKZ/AMC Z Emma Children’s Hospital/Academic Medical Center; EQD2 Z equivalent dose in 2-Gy fractions; HR Z hazard ratio; NA Z not applicable; RT Z radiation therapy. * Each model represents a multivariable Cox regression analysis including all variables listed for that particular model. y HR for EQD2 represents the increase in risk per Gray. Model 1 excluded 78 survivors with lack of information on adverse events; model 2 excluded 86 survivors with lack of information on adverse events and/or EQD2.
Volume 85 Number 3 2013 We consider the use of EQD2 to be the major strength of our study in which one of the aims was to assess the relationship between CRT dose and AEs. Obtaining detailed information about CRT treatment schedules enabled us to calculate the EQD2 for almost 96% of survivors who had received CRT. The EQD2 is useful in assessing dose-effect relationships because it includes both total dose and fractionation dose. As shown in fractionation studies, the latter is also important in determining late effects (12). The EQD2 accounts for the difference in biologic effect (ie, the biologic effect of 1 times 8 Gy differs from the biologic effect of 4 times 2 Gy, even though the total physical dose is equal). Most studies reporting on late effects after RT or CRT do not include fractionation dose (4-11), although the biologic effective dose has been analyzed in relation to late effects for similar reasons as we used the EQD2 (16). However, we prefer the EQD2 because it is easier to relate to everyday clinical practice, inasmuch as 2 Gy is a commonly used fractionation dose. Furthermore, we used the continuous cumulative EQD2, not categoric dose variables, so as not to lose power in our analyses. Other strengths of our study include the completeness of the cohort and the near complete inventory of CTCAE3.0 graded AEs from our hospital-based late-effect follow-up program. Our study also has some limitations. We evaluated a broad spectrum of selected AEs in a cohort that was heterogeneous with respect to primary tumor diagnosis and treatment modalities. Consequently, for the current study it was not feasible to address the role of irradiated volume. Focusing on a particular outcome would allow for reconstruction of dose-volume histograms in addition to the assessment of dose-effect relationships. Furthermore, we could not distinguish risk estimates for the EQD2 of the treatment types BI, CSI, and TBI, because each treatment type included not enough survivors and/or AEs. Also, the effect of missing CRT doses was largest in the TBI group, and a small number of survivors was classified to 2 treatment types; consequently, it is ambiguous to which of the treatment types the AEs in involved survivors could be attributed. Finally, we must be aware that the ORs obtained from the logistic regression analyses cannot be interpreted as relative risks (RRs) because of the high prevalence of AEs (>10%) in the cohort; in fact the ORs overestimated RRs. Because we did not transform the ORs to population-averaged RRs, it is not possible to compare the ORs in this study with the RRs reported in our previous study (1). In conclusion, our study confirms that long-term childhood cancer survivors who have received CRT are at high risk for the development of AEs and that these AEs are more often severe, life-threatening, or disabling in comparison with survivors who have not received CRT. Moreover, analyzing dose-effect relationships with the cumulative EQD2 instead of total physical dose
Late adverse events after cranial radiation therapy
775
connects the knowledge from radiation therapy and radiobiology with the clinical experience.
References 1. Geenen MM, Cardous-Ubbink MC, Kremer LC, et al. Medical assessment of adverse health outcomes in long-term survivors of childhood cancer. JAMA 2007;297(24):2705-2715. 2. Oeffinger KC, Mertens AC, Sklar CA, et al. Chronic health conditions in adult survivors of childhood cancer. N Engl J Med 2006;355: 1572-1582. 3. Pui CH, Campana D, Pei D, et al. Treating childhood acute lymphoblastic leukemia without cranial irradiation. N Engl J Med 2009;360: 2730-2741. 4. Armstrong GT, Liu Q, Yasui Y, et al. Long-term outcomes among adult survivors of childhood central nervous system malignancies in the Childhood Cancer Survivor Study. J Natl Cancer Inst 2009;101: 946-958. 5. Zeltzer LK, Recklitis C, Buchbinder D, et al. Psychological status in childhood cancer survivors: a report from the Childhood Cancer Survivor Study. J Clin Oncol 2009;27:2396-2404. 6. Hua C, Bass JK, Khan R, et al. Hearing loss after radiotherapy for pediatric brain tumors: effect of cochlear dose. Int J Radiat Oncol Biol Phys 2008;72:892-899. 7. Sklar CA, Constine LS. Chronic neuroendocrinological sequelae of radiation therapy. Int J Radiat Oncol Biol Phys 1995;31:1113-1121. 8. Gurney JG, Ness KK, Sibley SD, et al. Metabolic syndrome and growth hormone deficiency in adult survivors of childhood acute lymphoblastic leukemia. Cancer 2006;107:1303-1312. 9. Byrne J, Fears TR, Mills JL, et al. Fertility of long-term male survivors of acute lymphoblastic leukemia diagnosed during childhood. Pediatr Blood Cancer 2004;42:364-372. 10. Banerjee J, Paakko E, Harila M, et al. Radiation-induced meningiomas: a shadow in the success story of childhood leukemia. Neuro Oncol 2009;11:543-549. 11. Taylor AJ, Little MP, Winter DL, et al. Population-based risks of CNS tumors in survivors of childhood cancer: the British Childhood Cancer Survivor Study. J Clin Oncol 2010;28:5287-5293. 12. Joiner MC, Bentzen SM. Fractionation: the linear-quadratic approach. In: Joiner MC, van der Kogel A, editors. Basic Clinical Radiobiology. London: Hodder Arnold,; 2009. 13. Cardous-Ubbink MC, Heinen RC, Bakker PJ, et al. Risk of second malignancies in long-term survivors of childhood cancer. Eur J Cancer 2007;43(2):351-362. 14. National Cancer Institute. CTEP.CTCAE v3.0; http://ctep.cancer.gov/ reporting/CTC.html. 2003. Publish Date: August 9, 2006. 15. van Dijk IW, Oldenburger F, Cardous-Ubbink MC, et al. Evaluation of late adverse events in long-term Wilms’ tumor survivors. Int J Radiat Oncol Biol Phys 2010;78(2):370-378. 16. Schmiegelow M, Lassen S, Poulsen HS, et al. Cranial radiotherapy of childhood brain tumours: growth hormone deficiency and its relation to the biological effective dose of irradiation in a large population based study. Clin Endocrinol (Oxf) 2000;53:191-197.
Radiation Oncology biology
physics
www.redjournal.org
Clinical Investigation: Pediatric Cancer
Dose-Effect Relationships for Adverse Events After Cranial Radiation Therapy in Long-term Childhood Cancer Survivors Irma W.E.M. van Dijk, MSc,* Mathilde C. Cardous-Ubbink, PhD,y Helena J.H. van der Pal, MD, PhD,y,z Richard C. Heinen, MSc,z Flora E. van Leeuwen, PhD,x Foppe Oldenburger, MD,* Rob M. van Os, MSc,* Ce´cile M. Ronckers, PhD,jj Antoinette Y.N. Schoutenevan Meeteren, MD, PhD,z Huib N. Caron, MD, PhD,y,z Caro C.E. Koning, MD, PhD,* and Leontien C.M. Kremer, MD, PhDy,z Departments of *Radiation Oncology and yMedical Oncology, Academic Medical Center; zDepartment of Pediatric Oncology, Emma Children’s Hospital/Academic Medical Center; xDepartment of Epidemiology, Netherlands Cancer Institute, Amsterdam; and jjDutch Childhood Oncology Group, Long-term Effects after Childhood Cancer, The Hague, The Netherlands Received Apr 24, 2012, and in revised form Jul 12, 2012. Accepted for publication Jul 13, 2012
Summary Adverse events (AEs) occur in the majority of long-term childhood cancer survivors, especially after cranial radiation therapy (CRT). In this retrospective study we converted CRT doses into the equivalent dose in 2-Gy fractions (EQD2) to account for fractionation dose. Using the EQD2 in multivariable models enabled us to assess dose-effect relationships for the prevalence and severity of AEs and for specific categories of AEs. Our analyses confirm that CRT
Purpose: To evaluate the prevalence and severity of clinical adverse events (AEs) and treatment-related risk factors in childhood cancer survivors treated with cranial radiation therapy (CRT), with the aim of assessing dose-effect relationships. Methods and Materials: The retrospective study cohort consisted of 1362 Dutch childhood cancer survivors, of whom 285 were treated with CRT delivered as brain irradiation (BI), as part of craniospinal irradiation (CSI), and as total body irradiation (TBI). Individual CRT doses were converted into the equivalent dose in 2-Gy fractions (EQD2). Survivors had received their diagnoses between 1966 and 1996 and survived at least 5 years after diagnosis. A complete inventory of Common Terminology Criteria for Adverse Events grade 3.0 AEs was available from our hospital-based late-effect follow-up program. We used multivariable logistic and Cox regression analyses to examine the EQD2 in relation to the prevalence and severity of AEs, correcting for sex, age at diagnosis, follow-up time, and the treatment-related risk factors surgery and chemotherapy. Results: There was a high prevalence of AEs in the CRT group; over 80% of survivors had more than 1 AE, and almost half had at least 5 AEs, both representing significant increases in number of AEs compared with survivors not treated with CRT. Additionally, the proportion of severe, life-threatening, or disabling AEs was significantly higher in the CRT group. The most frequent AEs were alopecia and cognitive, endocrine, metabolic, and neurologic events. Using the EQD2, we found significant dose-effect relationships for these and other AEs. Conclusion: Our results confirm that CRT increases the prevalence and severity of AEs in childhood cancer survivors. Furthermore, analyzing dose-effect relationships with the cumulative EQD2
Reprint requests to: Irma W. E. M. van Dijk, MSc, Department of Radiation Oncology, Academic Medical Center (AMC), University of Amsterdam, Postbox 22660, 1100 DD Amsterdam, The Netherlands, Tel: þ31-20-5666823, Fax: þ31-20-6091278; E-mail: [email protected] Int J Radiation Oncol Biol Phys, Vol. 85, No. 3, pp. 768e775, 2013 0360-3016/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ijrobp.2012.07.008
Supported by the KiKa Foundation, The Netherlands. Conflict of interest: none. Supplementary material for this article can be www.redjournal.org.
found
at
Volume 85 Number 3 2013
increases the risk for the development of AEs.
Late adverse events after cranial radiation therapy
769
instead of total physical dose connects the knowledge from radiation therapy and radiobiology with the clinical experience. Ó 2013 Elsevier Inc.
Introduction Owing to better treatment results, long-term childhood cancer survival has improved enormously. With the enhanced survival, however, the incidence of late tumor and treatment-related effects becomes more evident. Late effects occur in the majority of childhood cancer survivors, especially in those treated with cranial radiation therapy (CRT) (1, 2). For a long time, prophylactic brain irradiation (BI) has been part of therapy directed to the central nervous system (CNS) in patients with leukemia. Nowadays, prophylactic BI has been replaced by intrathecal and systemic chemotherapy (3). Nevertheless, patients with persisting or recurrent CNS disease still receive BI. Besides chemotherapy, CRT also is a major nonsurgical treatment method for high-grade CNS tumors; craniospinal irradiation (CSI) is indicated for medulloblastoma, and patients who receive total body irradiation (TBI) are also exposed to CRT. Late effects after CRT include neurocognitive and psychosocial outcomes, hearing loss, endocrine and metabolic disorders, fertility problems, and secondary tumors (4-11). The association between CRT and late effects is often analyzed by the use of dichotomous variables (ie, radiation therapy [RT] yes/no), dose categories, or total dose. However, fractionation studies have shown that late effects also depend on fractionation dose (12). Because treatment protocols have changed considerably over the years and fractionation doses vary within and between patients, the fractionation dose should also be considered in the evaluation of late effects. Our previous studies showed that the risk of secondary tumors (1, 13) was elevated in survivors treated with RT, and that RT involving the head-and-neck region increased the risk of several selected adverse events (AEs) (1). In the present study, we evaluated the prevalence and severity of AEs and treatment-related risk factors among survivors treated with CRT. Assessment of dose-effect relationships using the equivalent dose in 2-Gy fractions (EQD2) enabled us to more precisely examine the effect of CRT dose on specific categories of AEs.
Methods and Materials Patients The study population, previously described by Geenen et al (1), consisted of 1362 childhood cancer survivors diagnosed with a primary cancer at age of 0 to18 years between January 1, 1966, and January 1, 1996, in the Emma Children’s Hospital/Academic Medical Center (EKZ/AMC, Amsterdam, The Netherlands). They survived at least 5 years after diagnosis. Information on survivors’ characteristics, primary cancer diagnosis, and treatment was extracted from the EKZ/AMC Childhood Cancer Registry. From this registry, established in 1966, we identified 285 survivors who received CRT. Detailed CRT treatment data, including total dose, fractionation dose, and radiation schedule, were retrieved from their individual RT files. Since the establishment of the Outpatient Clinic for Late Effects of Childhood Cancer (Polikliniek Late Effecten
Kindertumoren; PLEK, Emma Children’s Hospital) in 1996, all living EKZ/AMC survivors were invited to this clinic for riskstratified, predefined screening and for the care of AEs. All clinically relevant AEs were extracted from the PLEK database and defined and graded as described in detail in our previous study (1) using the Common Terminology Criteria for Adverse Events (CTCAEv3.0) (14). This scoring system for both acute and late effects differentiates AEs as mild (grade 1), moderate (grade 2), severe (grade 3), and life-threatening or disabling (grade 4). Grade 5 refers to AE-related deaths.
Converting CRT dose into EQD2 CRT was defined as external beam radiation exposure of the brain resulting from these treatment types: brain irradiation (BI), craniospinal irradiation (CSI), and total body irradiation (TBI). All CRT doses including boost doses, delivered for the primary cancer or for metastatic or recurrent disease, were converted into the equivalent dose in 2-Gy fractions (EQD2) (15). The EQD2 was calculated with the formula based on the linear quadratic model: EQD2 Z D*(d þ a/b)/(2 þ a/b), in which the total dose D is the number of fractions multiplied by the fractionation dose d. The a/b ratio for late-responding brain tissue was estimated at 2 Gy (12). We calculated both the cumulative EQD2 and the EQD2 per treatment type (ie, EQD2_BI, EQD2_CSI, and EQD2_TBI). In case a survivor had undergone more than 1 treatment type, the cumulative EQD2 represents the total dose of all treatment types. For RT to other sites of the body, including all RT other than CRT, no EQD2 was calculated.
Statistics We described survivors’ characteristics and the prevalence and severity of AEs, comparing survivors who had received CRT with those who had not received CRT. Treatment-related risk factors for the prevalence and severity of AEs were evaluated with multivariable logistic regression models and reported as odds ratios (ORs). In model 1 we crudely analyzed CRT (yes/no) as a risk factor for the prevalence and severity of AEs and for specific AEs. In model 2 the cumulative EQD2 for CRT was used to assess dose-effect relationships for the prevalence and severity of AEs, and for the subset of specific AEs that showed a statistically significantly increased risk (P<.05) after CRT in model 1. In a third model, we attempted to asses dose-effect relationships per treatment type. Additionally, we performed stratified analyses for brain tumor survivors and survivors of other cancers to examine the role of primary cancer type. Besides CRT (y/n) or EQD2, all models included sex, age at diagnosis (years), follow-up time since diagnosis (years), and these treatment-related risk factors: surgery (y/n), chemotherapy (y/n), and RT other sites (y/n). We evaluated the assumptions for linearity of the EQD2; no violations of linearity assumptions were observed. Because the time-to-event for secondary tumors was known, we used Kaplan-Meier analyses to estimate the risk of both malignant and benign secondary tumors in the head-and-neck area 20 years after diagnosis. Additionally, Cox models were used to evaluate treatment-related risk factors for secondary tumors. All
770
International Journal of Radiation Oncology Biology Physics
van Dijk et al.
Table 1
Characteristics of the 1966-1996 EKZ/AMC cohort of 5 years childhood cancer survivors Number of survivors, (NZ1362) Characteristic
Sex M F Age at diagnosis childhood cancer, y Median (range) 0-5 >5-10 >10-15 >15 Primary cancer diagnosis Leukemia Lymphoma Kidney/Wilms tumor Bone tumor Soft tissue sarcoma Neuroblastoma Other* Brain/central nervous system Astrocytoma Ependymoma/medulloblastoma Intracranial germ cell tumor Primitive neuroectodermal tumor Sarcoma Calendar y of diagnosis 1966-1975 1976-1985 1986-1996 Overall treatment category Surgery only (S) Chemotherapy (CT) only, with/without S Radiation therapy (RT) only, with/without S CT þ RT with/without S first treatment, no recurrencey CT þ RT with/without S including recurrence treatmentz 131 I-MIBG with/without CT and/or RT and/or S Other Type of chemotherapy Anthracyclines with/without other chemotherapy Alkylating agents with/without other chemotherapy Anthracyclines and alkylating agents, with/without other chemotherapy Other chemotherapy only Specific agent Cisplatin Methotrexate Follow-up time, y Median (range) 5-15 >15-25 >25 Attained age at end of follow-up time, y Median (range) 5-15 >15-25 >25-35 >35
CRT, nZ285 (%)
No CRT, nZ1077 (%)
156 (54.7) 129 (45.3)
589 (54.7) 488 (45.3)
6.3 113 101 61 10
5.8 483 277 248 69
(0.0-17.6) (39.6) (35.4) (21.4) (3.5)
(0.0-17.8) (44.8) (25.7) (23.0) (6.4)
151 (53.0) 42 (14.7) 0 0 3 (1.1) 1 (0.4) 0 88 (30.9) 19 (6.7) 61 (21.4) 4 (1.4) 3 (1.1) 1 (0.4)
184 217 189 116 148 84 118 21 16 1
(17.1) (20.1) (17.5) (10.8) (13.7) (7.8) (11.0) (1.9) (1.5) (0.1) 0 4 (0.4) 0
49 (17.2) 161 (56.5) 75 (26.3)
163 (15.1) 383 (35.6) 531 (49.3)
0 0 41 (14.4) 151 (53.0) 93 (32.6) 0 0
102 650 49 175 85 15 1
(9.5) (60.4) (4.5) (16.2) (7.9) (1.4) (0.1)
33 (11.6) 61 (21.4) 77 (27.0)
77 (7.1) 199 (18.5) 375 (34.8)
73 (25.6)
155 (25.2)
2 (0.7) 177 (62.1)
75 (7.0) 285 (26.5)
19.6 102 119 64
(5.0-36.1) (35.8) (41.8) (22.5)
16.6 446 440 191
(5.0-38.1) (41.4) (40.9) (17.7)
26.3 36 92 121 36
(5.9-44.0) (12.6) (32.3) (42.5) (12.6)
24.0 169 421 355 132
(5.2-48.9) (15.7) (39.1) (33.0) (12.3)
(continued on next page)
Volume 85 Number 3 2013
Late adverse events after cranial radiation therapy
771
Table 1 (continued ) Number of survivors, (NZ1362) Characteristic
CRT, nZ285 (%)
Vital status at end of follow-up Alive Dead
No CRT, nZ1077 (%)
230 (80.7) 55 (19.3)
1011 (93.9) 66 (6.1)
Abbreviations: CRT Z cranial radiation therapy; EKZ/AMC Z Emma Children’s Hospital/Academic Medical Center; 131I-MIBG Z iodine-131-metaiodobenzylguanidine. * Including 43 gonadal germ cell tumors, 21 retinoblastomas, 13 thyroid carcinomas, 12 hepatoblastomas, 11 malignant histiocytoses, and 18 miscellaneous tumors. y Also treated with RT to other sites: mediastinum with/without abdomen (nZ10). z Also treated with RT to other sites: mediastinum with/without abdomen (nZ3), orbit (nZ4), testicles (nZ3), neck (nZ2), breast (nZ1), and brachytherapy of the mastoid (nZ1).
meta-iodobenzylguanidine [131I-MIBG]). The remaining 753 survivors had not received RT. We obtained detailed treatment information including fractionation doses for 95.8% (273/285) of survivors. This enabled us to calculate the EQD2_BI for 183 of 189 survivors who had received BI, the EQD2_CSI for 82 of 83 CSI survivors, and the EQD2_TBI for 23 of 29 TBI survivors. Sixteen of the 189 BI survivors additionally received CSI or TBI for a recurrence (Table 2).
data were analyzed with SPSS version 18.0 (SPSS, Chicago, IL) and the statistical program R version 2.13.0 (http://www.Rproject.org).
Results Table 1 shows the characteristics of the study cohort. The median follow-up time for all survivors was 17.0 years. The majority of the CRT group consisted of leukemia survivors, who were younger at diagnosis than were brain tumor survivors in the same group (median ages 4.7 and 8.3 years, respectively). Besides CRT, 24 survivors (8.4%) received RT to other sites of the body. Of the survivors not treated with CRT, 324 (30.1%) received RT to other sites of the body, including orbital RT, nasopharyngeal RT, brachytherapy, and treatment with radioactive iodine (iodine-131-
Table 2
Prevalence and severity of AEs Figure E1 presents an overview of all AEs. The most frequently scored AEs in the CRT group were cognitive disorders (14.4%), endocrine disorders (12.2%), neurologic disorders (9.2%), and
Cranial radiation therapy in the 1966-1996 EKZ/AMC cohort of 5 years childhood cancer survivors by primary tumor type. CRT, nZ285 (%) CRT dose; EQD2 (Gy) (median, range)
Treatment group Brain irradiation Whole brain Partial brain Whole brain with boost Unknown Craniospinal irradiation With boost Without boost Unknown Total body irradiation (TBI) TBI Unknown
Brain tumor, nZ88 (30.9) 31 (35.2) 0 21 (23.9) (51.3, 43.2-94.6) 6 (3.0) (50.2, 34.2-55.2) 4 (4.5) 57 (64.8) 53 (60.2) (49.3, 37.0-53.6) 3 (3.4) (42.9, 42.9-42.9) 1 (1.1) 0 0
Other cancer, nZ197 (69.1) 158* (80.2) 144 (73.1) (24.8, 13.6-48.3) 7 (3.6) (30.0, 9.6-54.1) 3 (1.5) (43.5, 22.3-47.9) 4 (2.0) 26 (13.2) 0 26 (13.2) (21.0, 13.1-45.6) 0 29 23 (11.8) 16.4 (0.8-14.3) 6 (3.0)
Abbreviations: CRT Z cranial radiation therapy; EKZ/AMC Z Emma Children’s Hospital/Academic Medical Center; EQD2 Z equivalent dose in 2-Gy fractions. * Includes survivors also treated with craniospinal irradiation (nZ11) or total body irradiation (nZ5).
772
International Journal of Radiation Oncology Biology Physics
van Dijk et al.
alopecia (9.0%). Table E1 shows that 6.3% of the survivors in the CRT group had no AEs, vs 23.3% in the non-CRT group. Of all survivors with known AEs, 81.2% of the CRT group and 53.4% of the non-CRT group had more than 1 AE. Furthermore, 46.3% of the CRT group had at least 5 events, vs 19.2% of the non-CRT group. Additionally, at least 1 severe, life-threatening, or disabling event was present in 45.1% of the CRT group, vs 34.8% of the non-CRT group. The proportion of AE-related deaths was twice as high in the CRT group as in the non-CRT group.
Treatment-related risk factors for AEs Figure E2, based on multivariable logistic regression models, shows the exponentially increasing risk for the development of any AE and of severity of AEs with increasing CRT doses in the whole cohort. Table 3 shows the results of the multivariable logistic regression analyses for treatment-related risk factors for the prevalence and severity of AEs, stratified by primary cancer type (brain tumor vs other cancers). CRT (y/n) significantly increased the risk of any AE in brain tumor survivors and in survivors of other cancers. CRT was also a significant risk factor for the severity of AEs (ie, for mild to moderate AEs vs no AEs, and for severe, lifethreatening, or disabling events and AE-related deaths vs no AEs [model 1]). Replacing CRT (y/n) by the cumulative EQD2 showed significantly increased risks of 1.07 Gy1 and 1.06 Gy1 for any AE in respectively brain tumor survivors and in survivors of other cancer types (model 2). The risk estimates were generally similar for different definitions of severity. We also tried to assess doseeffect relationships per treatment type using the EQD2_BI, EQD2_CSI, and EQD2_TBI, but each treatment group contained too few survivors and/or events (model 3, not shown).
Treatment-related risk factors for selected AEs In the whole cohort, CRT (y/n) significantly increased the risk of a dozen of outcome categories (results not shown). Dose-effect
relationships for these outcome categories were further analyzed with the cumulative EQD2 and stratified by primary cancer typedthat is, for brain tumor survivors and survivors of other childhood cancers separately. Figure 1 illustrates that the risk of selected AEs increases exponentially with increasing EQD2 among brain tumor survivors and survivors of other cancers, with different ORs in both groups. The analyses (Table E2) (model 2) showed significant dose-effect relationships for alopecia and for endocrine and cognitive events in both survivor groups. Survivors of childhood cancers other than brain cancer also had significantly increased risks of fatigue, headache, hearing loss, male fertility problems, short stature, and metabolic and neurologic events. The risk of male fertility problems and social-emotional problems was slightly higher in brain tumor survivors. The risk estimates for EQD2_BI, EQD2_CSI, and EQD2_TBI in model 3 (not shown) were generally similar to those of the cumulative EQD2 in model 2. However, for several selected AEs, it was not feasible to assess dose-effect relationships by treatment type because of the low number of survivors and/or AEs in each treatment group.
Secondary tumors in the head-and-neck area Of the secondary tumors in the CRT group, 29 occurred as solid tumors in the head-and-neck area (Fig. E1), counting 5 thyroid neoplasms, 6 basal cell carcinomas of the skin, 6 CNS tumors, 8 meningiomas, and 4 miscellaneous. The meningiomas occurred in survivors treated for leukemia (nZ6; median EQD2_BI Z 20.9 Gy), lymphoma (nZ1; EQD2_TBI Z 20.0 Gy), and medulloblastoma (nZ1; EQD2_CSI Z 45.9 Gy) after a median latency time of 17.8 years (range, 8.6-23.8 years). Figure 2 shows the Kaplan-Meier curves for secondary tumors in the head-and-neck area (including meningiomas) and for meningiomas separately. In survivors who had received CRT, the estimated cumulative risk for secondary tumors in the head-and-neck area 20 years after diagnosis was 10.2% (95% CI, 5.9%-14.5%), and 3.5% (95% CI,
Table 3 Treatment-related risk factors for severity of adverse events in the 1966-1996 EKZ/AMC cohort of 5 years childhood cancer survivors (NZ1362) stratified by primary cancer type Any adverse event by severity OR (95% CI) Grades 1-5 vs no AE Risk factor Model 1* CRT (y/n) RT-other-sites (y/n) Model 2* CRT; EQD2y RT-other-sites (y/n)
Grades 1-2 vs no AE
Grades 3-5 vs no AE
Brain tumor
Other cancers
Brain tumor
Other cancers
Brain tumor
Other cancers
30.7 (2.26-416) NAx
3.99 (2.26-7.06) 7.69 (4.42-13.4)
163 (1.25-2.14$104) NAx
2.99 (1.64-5.48) 5.99 (3.32-10.8)
27.8 (1.84-420) NAx
6.37 (3.33-12.2) 10.5 (5.91-18.8)
1.07 (1.01-1.13) NAx
1.06 (1.03-1.08) 7.61 (4.37-13.2)
1.11 (1.00-1.24)z NAx
1.04 (1.01-1.07) 5.93 (3.29-10.7)
1.07 (1.01-1.13) NAx
1.07 (1.04-1.10) 10.3 (5.79-18.4)
Abbreviations: AE Z adverse event; CI Z confidence interval; CRT Z cranial radiation therapy; EKZ/AMC Z Emma Children’s Hospital/Academic Medical Center; EQD2 Z equivalent dose in 2-Gy fractions; NA Z not applicable; OR Z odds ratio; RT Z radiation therapy. * Each model represents a multivariable logistic regression analysis including the variables listed for that particular model and corrected for sex, age at diagnosis, follow-up time, and these treatment-related risk factors: surgery and chemotherapy. y OR for EQD2 represents the increase in risk per Gray. Brain tumor survivors (nZ109): model 1 excluded 8 survivors with lack of information on AEs, model 2 excluded 12 survivors with lack of information on AEs and/or EQD2. Survivors of other cancers (nZ1253): model 1 excluded 70 survivors with lack of information on AEs, model 2 excluded 74 survivors with lack of information on AEs and/or EQD2. z P<.05. x An analysis for brain tumor survivors was not feasible because of the exceptionally low numbers in most of the cells.
Volume 85 Number 3 2013
Late adverse events after cranial radiation therapy
773
Fig. 1. Dose-effect relationships for selected adverse events (AEs) in the 1966-1996 EKZ/AMC cohort of 5 years childhood cancer survivors (nZ1362) stratified by primary cancer type. ORs result from multivariable logistic regression models, corrected for sex, age at diagnosis, follow-up time, surgery, chemotherapy, and radiation therapy to other sites of the body. ORs represent the increase in risk per Gray. Brain tumor survivors (nZ109); logistic regression excluded 12 survivors with lack of information on AEs and/or EQD2. Survivors of other cancers (nZ1253); logistic regression excluded 74 survivors with lack of information on AEs and/or EQD2. Male fertility; male brain tumor survivors (nZ66); logistic regression excluded 5 survivors with lack of information on AEs and/or EQD2, male survivors of other cancers (nZ679); logistic regression excluded 44 survivors with lack of information on AEs and/or EQD2. Abbreviations: OR Z odds ratio; CI Z confidence interval; EQD2 Z equivalent dose in 2-Gy fractions; SD Z standard deviation.
774
International Journal of Radiation Oncology Biology Physics
van Dijk et al.
Fig. 2. Kaplan-Meier curves for secondary tumors in the headand-neck area (including meningioma), and for meningioma separately. Abbreviation: CRT Z cranial radiation therapy. 0.8%-6.2%) for meningiomas. Cox regression analyses (Table 4) showed that both CRT and RT to other sites increased the risk of secondary tumors in the head-and-neck area (HREQD2, 1.04 Gy1 [CI, 1.02-1.05]; HRRTothersites, 2.83 [CI, 1.55-5.15]). CRT was the only significant risk factor for meningiomas (HREQD2, 1.04 Gy1 [CI, 1.01-1.08]).
Discussion Our results show a high prevalence of AEs in long-term childhood cancer survivors treated with CRT. Over 80% of survivors in the
CRT group had more than 1 AE, and almost half had at least 5 AEs, both representing significant increases in the number of AEs compared with survivors not treated with CRT. AEs after CRT were also significantly more often severe, life-threatening, or disabling. In previous studies (1, 13) we investigated the role of radiation therapy without assessing dose-effect relationships. In this study we particularly focused on CRT, with the aim of examining late effects in relation to CRT dose. We demonstrated that CRT increased the risk of alopecia; endocrine disorders; hearing loss; fatigue; male fertility problems; metabolic, neurologic, and ophthalmologic events; headache; social-emotional, and cognitive problems; short stature; and secondary tumors in the head-andneck area. Moreover, using the cumulative EQD2, we assessed significant dose-effect relationships for these selected AEs. Unexpectedly, the analyses stratified on primary cancer type showed slightly higher ORs for dose-effect relationships in survivors treated for other cancers than in survivors treated for brain tumor. A possible explanation may be that most survivors of other cancers were leukemia survivors who had received their diagnoses at younger ages than brain tumor survivors. Second, we did not include irradiated volume in our analyses, and brain tumor survivors received high doses in relatively small fields. Third, it seems that the brain tumor itself and/or brain surgery levels off the effect of radiation dose. Meningiomas occurred after a median interval of almost 18 years, which is similar to median latency times as reported in other studies (10, 11). Factors that increase the risk of meningioma include younger age at time of treatment, longer follow-up time, and higher CRT dose. In our cohort, meningiomas mainly occurred in leukemia survivors, who generally received lower CRT doses than did brain tumor survivors, but who were treated at younger ages and had longer follow-up times.
Table 4 Cox regression analyses for treatment-related risk factors of secondary tumors in the 1966-1996 EKZ/AMC cohort of 5 years childhood cancer survivors (NZ1362). Secondary tumor, HR (95% CI) Risk factor Model 1* Sex (F/M) Age at diagnosis Surgery (y/n) Chemotherapy (y/n) Surgery and/or chemotherapy (y/n) CRT (y/n) RT-other-sites (y/n) Model 2* Sex (F/M) Age at diagnosis Surgery (y/n) Chemotherapy (y/n) Surgery and/or chemotherapy (y/n) CRT; EQD2y RT-other-sites (y/n)
Head-and-neck area (incl. meningioma)
Meningioma
1.38 0.97 0.81 0.63
(0.80-2.35) (0.91-1.03) (0.42-1.54) (0.31-1.28) NA 5.08 (2.66-9.72) 3.07 (1.68-5.63)
0.30 (0.06-1.45) 1.00 (0.84-1.18) NA NA 1.34 (0.25-7.15) 34.6 (3.48-344) 1.71 (0.32-9.25)
1.45 0.96 0.52 0.72
0.33 (0.07-1.59) 0.98 (0.84-1.15) NA NA 0.49 (0.10-2.44) 1.04 (1.01-1.08) 0.98 (0.18-5.43)
(0.84-2.49) (0.90-1.02) (0.29-0.94) (0.35-1.47) NA 1.04 (1.02-1.05) 2.83 (1.55-5.15)
Abbreviations: CRT Z cranial radiation therapy; EKZ/AMC Z Emma Children’s Hospital/Academic Medical Center; EQD2 Z equivalent dose in 2-Gy fractions; HR Z hazard ratio; NA Z not applicable; RT Z radiation therapy. * Each model represents a multivariable Cox regression analysis including all variables listed for that particular model. y HR for EQD2 represents the increase in risk per Gray. Model 1 excluded 78 survivors with lack of information on adverse events; model 2 excluded 86 survivors with lack of information on adverse events and/or EQD2.
Volume 85 Number 3 2013 We consider the use of EQD2 to be the major strength of our study in which one of the aims was to assess the relationship between CRT dose and AEs. Obtaining detailed information about CRT treatment schedules enabled us to calculate the EQD2 for almost 96% of survivors who had received CRT. The EQD2 is useful in assessing dose-effect relationships because it includes both total dose and fractionation dose. As shown in fractionation studies, the latter is also important in determining late effects (12). The EQD2 accounts for the difference in biologic effect (ie, the biologic effect of 1 times 8 Gy differs from the biologic effect of 4 times 2 Gy, even though the total physical dose is equal). Most studies reporting on late effects after RT or CRT do not include fractionation dose (4-11), although the biologic effective dose has been analyzed in relation to late effects for similar reasons as we used the EQD2 (16). However, we prefer the EQD2 because it is easier to relate to everyday clinical practice, inasmuch as 2 Gy is a commonly used fractionation dose. Furthermore, we used the continuous cumulative EQD2, not categoric dose variables, so as not to lose power in our analyses. Other strengths of our study include the completeness of the cohort and the near complete inventory of CTCAE3.0 graded AEs from our hospital-based late-effect follow-up program. Our study also has some limitations. We evaluated a broad spectrum of selected AEs in a cohort that was heterogeneous with respect to primary tumor diagnosis and treatment modalities. Consequently, for the current study it was not feasible to address the role of irradiated volume. Focusing on a particular outcome would allow for reconstruction of dose-volume histograms in addition to the assessment of dose-effect relationships. Furthermore, we could not distinguish risk estimates for the EQD2 of the treatment types BI, CSI, and TBI, because each treatment type included not enough survivors and/or AEs. Also, the effect of missing CRT doses was largest in the TBI group, and a small number of survivors was classified to 2 treatment types; consequently, it is ambiguous to which of the treatment types the AEs in involved survivors could be attributed. Finally, we must be aware that the ORs obtained from the logistic regression analyses cannot be interpreted as relative risks (RRs) because of the high prevalence of AEs (>10%) in the cohort; in fact the ORs overestimated RRs. Because we did not transform the ORs to population-averaged RRs, it is not possible to compare the ORs in this study with the RRs reported in our previous study (1). In conclusion, our study confirms that long-term childhood cancer survivors who have received CRT are at high risk for the development of AEs and that these AEs are more often severe, life-threatening, or disabling in comparison with survivors who have not received CRT. Moreover, analyzing dose-effect relationships with the cumulative EQD2 instead of total physical dose
Late adverse events after cranial radiation therapy
775
connects the knowledge from radiation therapy and radiobiology with the clinical experience.
References 1. Geenen MM, Cardous-Ubbink MC, Kremer LC, et al. Medical assessment of adverse health outcomes in long-term survivors of childhood cancer. JAMA 2007;297(24):2705-2715. 2. Oeffinger KC, Mertens AC, Sklar CA, et al. Chronic health conditions in adult survivors of childhood cancer. N Engl J Med 2006;355: 1572-1582. 3. Pui CH, Campana D, Pei D, et al. Treating childhood acute lymphoblastic leukemia without cranial irradiation. N Engl J Med 2009;360: 2730-2741. 4. Armstrong GT, Liu Q, Yasui Y, et al. Long-term outcomes among adult survivors of childhood central nervous system malignancies in the Childhood Cancer Survivor Study. J Natl Cancer Inst 2009;101: 946-958. 5. Zeltzer LK, Recklitis C, Buchbinder D, et al. Psychological status in childhood cancer survivors: a report from the Childhood Cancer Survivor Study. J Clin Oncol 2009;27:2396-2404. 6. Hua C, Bass JK, Khan R, et al. Hearing loss after radiotherapy for pediatric brain tumors: effect of cochlear dose. Int J Radiat Oncol Biol Phys 2008;72:892-899. 7. Sklar CA, Constine LS. Chronic neuroendocrinological sequelae of radiation therapy. Int J Radiat Oncol Biol Phys 1995;31:1113-1121. 8. Gurney JG, Ness KK, Sibley SD, et al. Metabolic syndrome and growth hormone deficiency in adult survivors of childhood acute lymphoblastic leukemia. Cancer 2006;107:1303-1312. 9. Byrne J, Fears TR, Mills JL, et al. Fertility of long-term male survivors of acute lymphoblastic leukemia diagnosed during childhood. Pediatr Blood Cancer 2004;42:364-372. 10. Banerjee J, Paakko E, Harila M, et al. Radiation-induced meningiomas: a shadow in the success story of childhood leukemia. Neuro Oncol 2009;11:543-549. 11. Taylor AJ, Little MP, Winter DL, et al. Population-based risks of CNS tumors in survivors of childhood cancer: the British Childhood Cancer Survivor Study. J Clin Oncol 2010;28:5287-5293. 12. Joiner MC, Bentzen SM. Fractionation: the linear-quadratic approach. In: Joiner MC, van der Kogel A, editors. Basic Clinical Radiobiology. London: Hodder Arnold,; 2009. 13. Cardous-Ubbink MC, Heinen RC, Bakker PJ, et al. Risk of second malignancies in long-term survivors of childhood cancer. Eur J Cancer 2007;43(2):351-362. 14. National Cancer Institute. CTEP.CTCAE v3.0; http://ctep.cancer.gov/ reporting/CTC.html. 2003. Publish Date: August 9, 2006. 15. van Dijk IW, Oldenburger F, Cardous-Ubbink MC, et al. Evaluation of late adverse events in long-term Wilms’ tumor survivors. Int J Radiat Oncol Biol Phys 2010;78(2):370-378. 16. Schmiegelow M, Lassen S, Poulsen HS, et al. Cranial radiotherapy of childhood brain tumours: growth hormone deficiency and its relation to the biological effective dose of irradiation in a large population based study. Clin Endocrinol (Oxf) 2000;53:191-197.