A cohort study of childhood cancer incidence after

A cohort study of childhood cancer incidence after postnatal diagnostic X-ray exposure IMBEI, DKKR Dr von Hauner‘sche Kinderhospital Helmholz-München

Introduction Institut für Biometrie, Epidemiologie und Medizinische Informatik § Good epidemiologic knowledge on § Adults and children exposed to high doses of radiation § Atomic bomb victims § Patients following radiation radiotherapy § Adults exposed to low doses of radiation § NPW-workers, uranium mine workers, pilots… § Radon in homes § Adverse effects among children exposed to low level radiation? RICC Study 2

Introduction Source: German Federal Office for Radiation Protection RICC Study Institut für Biometrie, Epidemiologie und Medizinische Informatik 3

Background Institut für Biometrie, Epidemiologie und Medizinische Informatik Source: German Federal Office for Radiation Protection, Yearly Report 2005 RICC Study 4

Typical doses in paediatric radiology Institut für Biometrie, Epidemiologie und Medizinische Informatik Entrance surface dose (µGy) Age Examination 0 1 5 15 Abdomen AP Chest PA/AP Pelvis AP Skull LAT 110 60 170 / / 340 80 350 600 340 590 110 510 1250 580 2010 1300 / / Source: NRPB-W 14, 2000

Major studies of cancer risk following diagnostic radiation exposure in childhood and adolescence Institut für Biometrie, Epidemiologie und Medizinische Informatik Cohort/underlying condition Cancer site investigated Breast Tuberculosis, USA Tuberculosis, Canada Scoliosis, USA Diagnostic x-ray of broken bones, Canada Diagnostic x-rays, China Diagnostic x-rays, USA Cardiac catheterisation, Israel Cardiac catheterisation, Canada Leukaemia All sites X X X X

ERR of brain cancer by age at exposure among selected cohorts treated with radiotherapy Institut für Biometrie, Epidemiologie und Medizinische Informatik Dose ranges refer to doses to the brain. Source: Sadetzki & Mandelzweig, 2009

Estimated lifetime risk from a single dose of radiation as a function of age at exposure (BEIR 2006) Institut für Biometrie, Epidemiologie und Medizinische Informatik

Institut für Biometrie, Epidemiologie und Medizinische Informatik RICC: Radiation induced Cancer in Children: A cohort study of childhood cancer incidence after postnatal diagnostic X-ray exposure Objectives: To estimate precise radiation dose To estimate cancer risk (in children)

Material § § Institut für Biometrie, Epidemiologie und Medizinische Informatik Data collected at Dr. von Hauner Children’s Hospital - Munich National X-ray ordinance 1973 Protocol of all x-ray examinations since 1976 in databanks (as from 1998 RIS system) 270. 000 examinations until 2003 RICC Study 12

Available data Institut für Biometrie, Epidemiologie und Medizinische Informatik Patient history § Name, Gender, Date of birth, Date of examination, Height, Weight, Address Examination data § Projection radiography / Screening, Organ of interest, X-ray tube type, Radiation entry, Focus-skin distance (implicit), X-ray tube voltage, Total filtration (implicit), Exposure time / Duration of screening, m. As-Product / Dose area product Clinical data § Indication, Radiological diagnosis RICC Study 13

Base data Institut für Biometrie, Epidemiologie und Medizinische Informatik § Examination databases § MINDIUS I-III § Paper only § RIS / discharge letters 1976 – 1991 1992 - 1997 1998 - 2003 § Additional databases § X-ray machine details § Dosimeter details § Exposure modeling RICC Study 14

Cohort Institut für Biometrie, Epidemiologie und Medizinische Informatik Inclusion criteria § At least one diagnostic procedure performed at Dr. von Hauner’s Children’s Hospital § Age at “examination” ≤ 14. 5 years § Main residence in Germany § Time period 1976 -2003 § Cancer free at beginning (incl. first 6 months) RICC Study 16

Follow-Up Institut für Biometrie, Epidemiologie und Medizinische Informatik § Cancer: § Linkage of pseudonymised data with the German Childhood Cancer Registry § Observation period: 1980 -2006 § Person Years: § No individual follow-up ! § Calculating PY with age-specific mortality rates § Excluding children with “high” mortality risk RICC Study 17

Methods: Patient groups Institut für Biometrie, Epidemiologie und Medizinische Informatik § On the basis of clinical indications / diagnosis § Patients with elevated mortality risk § Syndromes with elevated cancer risk § Pre-term babies § On the basis of exposure records § Highly exposed (CT, contrast media, single dose > 5 m. Sv, cardiac patients) § RIS only patients (no specific diagnosis available!) RICC Study 18

Categories Institut für Biometrie, Epidemiologie und Medizinische Informatik Group „elevated mortality risk“ Disease Acute pericarditis Acute rheumatic fever Aneurysms Appendicitis Bronchopulmonary dysplasia Chronic rheumatic heart diseases Other coagulation defects Diverticular disease of intestine Endocarditis Inflammatory diseases of the central nervous system (Meningitis, Encephalitis) Epiglottitis Cardiac defect (complex defects, not isolated ASD or VSD) HIV-disease „elevated cancer Agranulocytosis and neutropenia risk“ Chromosome anomalies, not classified elsewhere Colitis ulcerosa Immunodeficiencies Crohn’s disease RICC Study ICD-10 -Codes I 30 I 00 -I 02 I 71 -I 72, Q 25. 4 K 35 -K 36 P 27 I 05 -I 09 D 68 K 57 K 60 K 62 I 38 G 00 -G 09 J 05. 1 J 37 Q 20 -Q 24 B 20 -B 24 D 70 Q 90 -Q 99 K 50 -K 51 D 80 -83 K 50 19

Methods: Dosimetry Institut für Biometrie, Epidemiologie und Medizinische Informatik Source: Michael Seidenbusch RICC Study 20

Radiation Exposure Institut für Biometrie, Epidemiologie und Medizinische Informatik § Organ dose was estimated § For risk analysis: § effective dose (=whole body dose) § Leukaemia: red bone marrow dose § Dose: continuous and categorical variable § Latency period: ½ year § Imputation for (some) missing values RICC Study 22

Statistical Analyses Plan (SAP) Institut für Biometrie, Epidemiologie und Medizinische Informatik § „External comparison“ § SIR (by sex): § All cancers, § Leukaemia, lymphocytic leukaemia, acute myeloid leukaemia, lymphoma, § CNS-tumours, other tumours) § Dose categories § Internal comparison: RR (multivariate) § Sensitivity analysis § Excluding specific subgroups § High exposed, elevated mortality risk § No. of examinations: 1, 2, 3+ § Cox Regression (time-dependent covariable) RICC Study 23

Excurse: The German Childhood Cancer Registry (GCCR) Institut für Biometrie, Epidemiologie und Medizinische Informatik § Begin of registration 1980 § Population base (children below 15) 13. 0 million § Number of reported cases (1980 -2008) 43. 014 § Completeness of registration ca. 95 % § Number of cases annually 1, 700 -1, 800 § Since 1991 inclusion of former GDR RICC Study 24

GCCR: Institut für Biometrie, Epidemiologie und Medizinische Informatik § Society for Paediatric Oncology and Haematology (GPOH) § 34 large treatment centres § treat 75% of all children § Each year nearly 70 hospitals report cases § since 1980 more than 130 reporting hospitals § 25 GPOH-clinical trials RICC Study 25

Power calculation for RICC Cancer endpoint Institut für Biometrie, Epidemiologie und Medizinische Informatik All cancers Expected Numbers 104 Leukaemia 36 Lymphocytic 30 Acute myeloid 5 CNS-tumours 22 Lymphoma 15 All others 31 RICC Study 29

RICC: Year of first examination RICC Study Institut für Biometrie, Epidemiologie und Medizinische Informatik 30

Results: Cohort Institut für Biometrie, Epidemiologie und Medizinische Informatik Highly exposed RICC Study 0, 3 3428 Premature children % 100 54 45 1 15 0, 4 279 All Patients Boys Girls unknown gender Patients with elevated mortality risk Syndromes with elevated cancer risk n 92957 50005 41432 1520 14174 398 5 31

Results: Cohort Institut für Biometrie, Epidemiologie und Medizinische Informatik All Patients Boys Girls unknown gender Patients with elevated mortality risk Syndromes with elevated cancer risk Premature children All 92957 50005 41432 1520 14174 398 % Cases 100 87 54 52 45 35 1 0 15 21 0, 4 0 279 RICC Study 0 3428 Highly exposed 0, 3 5 4 32

Results – Cohort Institut für Biometrie, Epidemiologie und Medizinische Informatik Age at inclusion 0 1 2 3 4 n 20546 9096 6945 6202 6387 % 22 10 7 7 7 5 -9 10 -14 24891 18890 27 20 RICC Study 33

Results – Cohort Institut für Biometrie, Epidemiologie und Medizinische Informatik Number of examinations (per patient) 1 2 3 4 5 6 7 8 9 10+ RICC Study All % 54605 59 17818 19 7515 8 4042 4 2341 3 1611 2 1128 1 737 1 561 1 2599 3 34

Results: incident cases Institut für Biometrie, Epidemiologie und Medizinische Informatik Gender ICCCDiagnosis Boys Girls 3 I-XII All cancers 52 35 Leukaemias, myeloproliferative diseases I 19 14 and myelodysplastic diseases Ia Lymphocytic leukaemia 15 9 Ib Acute myeloid leukaemia Ic-Ie Other leukaemias II Lymphoma III CNS-tumours IV-XII Other tumours RICC Study 2 2 11 7 15 3 2 2 3 16 36

Results: Cumulative exposure by status Institut für Biometrie, Epidemiologie und Medizinische Informatik 38 RICC Study 38

Results – Exposure per examination, by age Institut für Biometrie, Epidemiologie und Medizinische Informatik (whiskers show 10% and 90% percentiles) RICC Study 39

Results – Exposure per examination, by year Institut für Biometrie, Epidemiologie und Medizinische Informatik (whiskers show 10% and 90% percentiles) RICC Study 40

Results: Cumulative exposure (µSv) ATTENTION: µSv Institut für Biometrie, Epidemiologie und Medizinische Informatik n/N Median 92957 7. 0 3428 24. 0 RIS only 21319 3. 0 MUNDUS 71638 10. 0 14174 84. 0 elevated cancer risk 398 58. 0 premature children 279 51. 0 All patients Highly exposed elevated mortality risk RICC Study 43

External Comparison Institut für Biometrie, Epidemiologie und Medizinische Informatik O E SIR 95%-CI Boys 52 52. 8 0. 74 -1. 29 Girls 35 35. 2 0. 99 1. 00 87 88. 0 0. 79 -1. 22 33 30. 5 Lymphocytic leukaemia 24 24. 5 Acute myeloid leukaemia 5 4. 3 Lymphoma 13 13. 4 CNS-tumours 10 19. 3 Other tumours 31 24. 8 0. 99 1. 08 0. 98 1. 16 0. 97 0. 52 1. 25 Gender All cancers Leukaemia RICC Study 0. 69 -1. 38 0. 74 -1. 52 0. 63 -1. 45 0. 38 -2. 70 0. 52 -1. 66 0. 25 -0. 95 0. 85 -1. 77 44

Sensitivity Analysis Institut für Biometrie, Epidemiologie und Medizinische Informatik O E SIR 95%-CI No 58 56. 1 1. 03 0. 79 -1. 34 yes 21 16. 4 1. 28 0. 79 -1. 96 No 83 84. 4 0. 98 0. 78 -1. 22 Yes 4 3. 7 1. 09 0. 30 -2. 78 Elevated mortality risk Highly exposed RICC Study 45

SIR-Analysis: Effective dose Institut für Biometrie, Epidemiologie und Medizinische Informatik Trend test: p = 0, 32 Nov. 2009 RICC Study 47

RR-Analysis Institut für Biometrie, Epidemiologie und Medizinische Informatik Total elevated mortality risk excluded Nov. 2009 RICC Study Eff. Dosis (µSv) 01050+ All cancers RR* 1. 00 1. 02 1. 01 1. 00 1. 08 1. 04 95%-KI 0. 60 -1. 74 0. 60 -1. 71 0. 61 -1. 89 0. 55 -1. 96 49

RR-Analysis Total Institut für Biometrie, Epidemiologie und Medizinische Informatik Eff. Dosi s (µSv ) 0 - Leukaemia + Lymphoma RR* 95%-KI 1. 00 10 - 1. 00 0. 482. 07 50+ 1. 04 0. 512. 12 elevated mortality 01. 00 risk excluded 10 - 1. 13 0. 532. 44 50+Study 1. 24 0. 54 RICC 50

RR-Analysis Institut für Biometrie, Epidemiologie und Medizinische Informatik Eff. Solid Tumours Dosis (µSv) RR* 95%-CI Total 1. 00 10 - 1. 05 0. 49 -2. 27 50+ elevated mortality risk excluded 0 - 0. 98 0. 46 -2. 12 0 - 1. 00 10 - 1. 01 0. 44 -2. 34 50+ 0. 84 0. 31 -2. 25 RICC Study 51

Summary Institut für Biometrie, Epidemiologie und Medizinische Informatik § Cohort of 92957 children § Observed 1980 -2006; 7, 8 years § 726200 person years § 87 cancer cases § No increased incidence § No dose-effect relationship RICC Study 52

Strengths Institut für Biometrie, Epidemiologie und Medizinische Informatik + Unique patient collective + Large cohort + Prospective data acquisition + Good documentation + Excellent dosimetry + Cancer register with complete coverage, RICC Study 54

Strengths and weaknesses + Unique patient collective + Large cohort + Prospective data acquisition + Good documentation + Good dosimetry + Cancer register with complete coverage, extensive inquiries Institut für Biometrie, Epidemiologie und Medizinische Informatik - Small numbers of cases - Very low radiation doses - Underestimation of exposure - CT-exposures not yet quantified - Confounding by indication - No ascertainment of confounders - No cancer after age 15 RICC Study 55

Perspective Institut für Biometrie, Epidemiologie und Medizinische Informatik § Current cohort § Further follow-up possible § Nested case-control study on biological markers ? ? § Cohort of children with CT exposures § CT risks only roughly assessable up to now § Missing epidemiologic data § Pilot study is ongoing RICC Study 56

Institut für Biometrie, Epidemiologie und Medizinische Informatik Estimation Risk of CT Exposure

Estimated number of CT scans performed annually in the US Institut für Biometrie, Epidemiologie und Medizinische Informatik Source: Brenner et al. 2007

Special focus: CT scans in children Institut für Biometrie, Epidemiologie und Medizinische Informatik § Increase in CT use in children 1989: ~ 4% of all CT scans 1993: ~ 6% of all CT scans 2000: ~ 8 -11 of all scans

Estimated Organ Doses from Typical Single CT Scans of the Head Institut für Biometrie, Epidemiologie und Medizinische Informatik Source: Brenner et al. 2007

Estimated lifetime risk from a single dose of radiation as a function of age at exposure (BEIR 2006) Institut für Biometrie, Epidemiologie und Medizinische Informatik

Age-related effect of a single pediatric head CT scan on tumor occurrence and fatality Institut für Biometrie, Epidemiologie und Medizinische Informatik Age at exposure years 1 2 5 10 15 20 Lifetime risk of radiation induced cancer per 10 000 exposed children tumour fatality 220 150 120 80 50 40 70 60 50 33 20 15 Adjustment of table 4: Stein et al. 2008

CT-Cohort Study Institut für Biometrie, Epidemiologie und Medizinische Informatik Design: Similar to RICC Data sources: § Radiological and Neuro-radiology departments of several clinics (PACS and RIS) § Estimated size: 47 000 children exposed between 1990 and 2003 in Germany § Cohort planned: 5000 children § Feasibility study is ongoing § EU-consortium, BMBF funding Problems: § Data on indication, Confounding by indication § Dosimetry § Cancer after age 15!! 65

Acknowledgements Institut für Biometrie, Epidemiologie und Medizinische Informatik § Mainz: § IMBEI: § § § München: § Michael Seidenbuch § Karl Schneider § Dieter Regulla Gael Hammer Hajo Zeeb Doris Bardehle (!!) Susanne Seuchter § Data protection officer § § § § Irene Reinisch § GCCR § § Nov. 2009 IMBEI 12. 08. 2008 Claudia Bremensdorfer Irene Jung Claudia Spix Thomas Ziegler RICC Study Monika-Maria Deml Sieglinde Eberle Toni Galitzendorfer Sabine Heyn Renate Ritzer Ina Schneider 66

Publications Institut für Biometrie, Epidemiologie und Medizinische Informatik § Hammer GP, Seidenbusch MC, Schneider K, Regulla DF, Zeeb H, Spix C, Blettner M. A cohort study of childhood cancer incidence after postnatal diagnostic X-ray exposure. Radiat Res 2009; 171(4): 504 -512. § Seidenbusch MC, Regulla D, Schneider K. , Zur Strahlenexposition von Kindern in der pädiatrischen Radiologie. Teile 1 -6. Fortschr Röntgenstr 2008/2009 ; 180(5): 410 -422 ; 180(6): 522 -539 ; 180(12): 1061 -1081 ; 180(12): 1082 -1103 ; 181(5): 454 -471 ; 181(10): 945 -961. § Hammer GP, Seidenbusch MC, Schneider K, Regulla DF, Zeeb H, Spix C, Blettner M. Inzidenz von Kinderkrebs nach Röntgendiagnositik im Patientenkollektiv der Jahre 1976 -2003 einer Universitätsklinik. Fortschr Röntgenstr 2010, 182: 404 -414 RICC Study 67