New Evidence on Radiogenic Risks of CLL Where do we go from here?

New Evidence on Radiogenic Risks of CLL Where do we go from here? 22 nd Nuclear Medical Defense Conference May 10 th, 2017, Munich, Germany Lydia B. Zablotska Professor, Department of Epidemiology & Biostatistics University of California, San Francisco Outline Is chronic lymphocytic leukemia (CLL) radiogenic? - One of the biggest conundrums of radiation epidemiology Summary of findings from mortality and incidence studies Medically exposed patients Occupationally exposed workers 31 years after the Chernobyl accident - What do we know about the long-term risks of CLL? Epidemiological studies Extension studies: 1. Non-radiation risk factors study 2. Clinical follow-up of CLL cases study 3. Genetic studies of CLL cases Remaining knowledge gaps on radiation risks of CLL and future directions for research 2 1

Study of survivors of atomic bombings in Hiroshima and Nagasaki Life Span Study (LSS) N=86,572 (113,011 including those NIC) Less than 5 years after the bombings 1 : An excess of leukemia 20 years 2 : Significant increases in radiation risks of cataracts, leukemia and thyroid cancer; 5 CLL cases (Nagasaki) 30 years: A significant increase in solid cancers 50 years 3 : An unexpected increase was found in non-cancer diseases Solid cancers observed over 50 years: 10,127; due to radiation: 479 (4.7%) Leukemia observed over 50 years: Mortality: 296; due to radiation: 93 (31.4%), 55 years 4 : Incidence: 312; due to radiation: 91 (29.2%); 12 CLL cases 72 years: Leukemia forms only a small fraction of the accepted total radiation-related health detriment 3 1 Valentine 1951 and Heyssel et al. 1960; 2 Finch et al. 1969 3 Preston et al. 2004; 4 Hsu et al. 2013 CLL/SLL Descriptive Characteristics Aggregate category (CLL/SLL) because of shared clinicopathological features Dores et al. 2007: Incidence of CLL/SLL 90% higher M:F CLL 98% higher M:F SLL 67% higher M:F 2

CLL/SLL Descriptive Characteristics Aggregate category (CLL/SLL) because of shared clinicopathological features Dores et al. 2007: Incidence of CLL/SLL 90% higher M:F CLL 98% higher M:F SLL 67% higher M:F Incidence increases with age exponentially CLL/SLL Descriptive Characteristics Aggregate category (CLL/SLL) because of shared clinicopathological features Dores et al. 2007: Incidence of CLL/SLL 90% higher M:F CLL 98% higher M:F SLL 67% higher M:F Incidence increases with age exponentially Significant reporting delay for CLL 3

Radiation risk estimates from mortality studies of hematologic cancers UNSCEAR 2000 and 2006 Reports CLL excluded from the calculations of leukemia risks because there is little evidence that it is radiogenic. 7 Silver et al. 2007 Radiation risk estimates from mortality studies of hematologic cancers UNSCEAR 2000 and 2006 Reports CLL excluded from the calculations of leukemia risks because there is little evidence that it is radiogenic. 8 Silver et al. 2007 4

Radiation risk estimates from recent mortality studies of hematologic cancers Medical external radiation and brachytherapy for malignant conditions, non-malignant conditions, and diagnostic purposes With longer follow-up time (15 years), many studies have point estimates greater than 1.00 for irradiated patients Higher risks among Swedish and Danish Thorotrast patients (Travis et al. 2003) Statistical significance for doses at or above 100 mgy for CLL (but not for non-cll) among patients treated I-131 for hyperthyroidism in Sweden (Hall et a. 1992) but not for patients in the US (Ron et al. 1988) 11 million PY would be required to detect an RR of 1.63 for CLL comparing >100 msv to a referent category of 0 10 msv, with 80% power (alpha = 0.05). 9 Silver et al. 2007 Radiation risk estimates from recent incidence studies of hematologic cancers *significant at p<0.05 Study cohort Follow-Up Cohort ERR/Gy (95% CI) A-bomb survivors study (Hsu et al. 2013) U.K. radiation workers (Muirhead et al. 2009) Techa River cohort (Davis et al. 2015) Wismut uranium workers (Mohner et al. 2010) CLL NHL n=12 n=402 1950-2001 113,011 6 (0.3, 31)* 0.46 (-0.08, 1.29) n=128 n=305 1955-2001 174,541-0.12 (-1.42, 2.71) 1.28 (-0.38, 4.06) n=27-1953-2007 28,223 0.10 (<0, 1.20) - n=159-1953-1990 360,000 1.95 (-0.86, 4.99) - 10 5

Studies of Chernobyl Cleanup Workers Study cohort Chernobyl cleanup workers from Ukraine (Zablotska et al. 2013) Chernobyl cleanup workers from Belarus, Russia and Baltic countries (Kesminiene 2008) Chernobyl cleanup workers from Russia (Ivanov 2012) # Follow-Up Cohort Excess Relative Risk per gray (95% CI) CLL NHL n=65-1986-2006 110,645 2.58 - (0.02, 8.43)* n=21 n=20 1993-2000 ~146,000 4.7 28.1 (0.9, 243)* (<0, 76.1) 1986-2007 75,685 n=44 - -1.83 - (-3.33, 4.48) * Significant at the 0.05 level # Possible selection bias: Only 72% of the registered workers have personalized data on received external whole-body radiation doses, 77% of CLL cases. 11 Studies of Chernobyl Cleanup Workers Characteristic Belarus, Russia and Ukraine Russia Baltic countries (Zablotska et al. 2013) (Ivanov et al. 2012) (Kesminiene et al. 2008) Cases of leukemia 40 117 198 CLL cases (%) 21(53) 65 (56) 57 (29) Matched controls 165 739 NA Average cumulative dose, mgy NR Bone marrow - 91.3 Recorded 106 Percentage of workers with Doses below 50 mgy 78 66 31 Doses above 100 mgy 14 23 NR Estimates from the linear model (ERR/Gy and 95% CI), *- 90% CI Non-CLL leukemia 5.0 (-3.8, 57.0) 2.21 (0.05, 7.61) 0.44 (-1.68, 2,56)* CLL 4.7 (nd, 76.1) 2.58 (0.02, 8.43) -1.83 (-3.33, 4.48)* 12Official dose records (ODRs) of military liquidators overestimate EPR doses by a factor of 2.2 and should be adjusted prior to use in risk analyses (Chumak et al. 2015.) 6

NIOSH s Reclassification of CLL Richardson 2005: The epidemiologic evidence of radiation CLL associations is weak; however, given the limitations of the reviewed studies, these findings do not offer a persuasive basis for concluding that CLL is an exception to general principles of radiation carcinogenesis. In addition, there is a problem of logical inconsistency if the government continues to assert that CLL is nonradiogenic whereas SLL is radiogenic. November 2012: the NIOSH Office of Compensation Analysis and Support changed the eligibility for compensation for work-related diagnoses of CLL in U.S. nuclear weapons workers occupationally exposed to radiation making it a covered condition because of the high probability of causation by ionizing radiation, similar to all other types of leukemia. 13 Summary of results of radiationrelated risks of leukemia in Chernobyl cleanup workers from Ukraine 14 7

Chernobyl cleanup workers from Ukraine Epidemiological study Study population: Workers registered in the Ukrainian State Chornobyl Registry before 1992 Residents of Kyiv City or five oblasts 110,645 male clean-up workers (46% of all cleanup workers in Ukraine) Design: Nested case-control with incidence density sampling Matching (1:5) on year of birth and oblast of residence at the time of interview Proxies for dead cases and controls: Wives or next of kin for demographic portion of the dosimetric questionnaire Colleagues for the clean-up experience portion of the dosimetric questionnaire Case ascertainment All cases confirmed by the International Hematology Panel Phase I: cases of leukemia and 99 ancillary diagnoses from all medical institutions in the 6 study areas; cases diagnosed 1986-2000 and interviews conducted in 2002-2004 Phase II: cases identified through linkage with the Cancer Registry of Ukraine; cases diagnosed 2001-2006 and interviews conducted in 2006-2009 15 Romanenko et al. 2008 and Zablotska et al. 2013 RADRUE Dosimetry Realistic Analytical Dose Reconstruction with Uncertainty Estimates Basic idea: Dose = dose rate x (taking shielding into account) time 16 Kryuchkov et al. Health Phys 2009. 8

Time-and-motion method Database of exposure rates (time and location) Questionnaire: -what did you do? - when? and where? Radiation exposure Bone-marrow doses and uncertainties 17 A. Bouville 18 A. Bouville 9

19 Retrospective dose reconstruction The 30-km zone is split into five areas (base maps) Each base map subdivided into 10,000 cells Dose-rates extrapolated for each day of 1986-1990 for each cell Some cells have more spatial resolution of the radiation field than the areas with low exposure rate gradients Results: Annual doses 10,000 dose estimates based on random sampling from assumed distributions of model parameters Comparisons with other dose estimates: Official doses from the chornobyl state registry (<35% of the cohort) Doses of early liquidators from measured chromosomal aberrations (N=20) Doses for military liquidators from electron paramagnetic resonance (EPR) of tooth enamel (N=61) Kryuchkov et al. Health Phys 2009 Doses estimated by RADRUE 20 Chumak et al. 2015 10

21 Zablotska et al. Env Health Perspect 2013 Radiation-related risks of leukemia 22 Zablotska et al. Env Health Perspect 2013 11

56%! 23 Zablotska et al. Env Health Perspect 2013 24 Zablotska et al. Env Health Perspect 2013 12

Chernobyl cleanup workers from Ukraine Summary of findings Linear dose-response, relative risk at 1 Gy of 3.58 Similarity of effects for CLL and non-cll Possible modifying effect of age at exposure and time since exposure on risk of leukemia High average dose compared to other nuclear worker studies (mean=108 mgy bone marrow dose) Long follow-up (up to 20 years) 25 Zablotska et al. 2013 Chernobyl cleanup workers from Ukraine Unique Strength Ability to evaluate radiogenicity of chronic lymphocytic leukemia (CLL) Mortality vs. incidence Cancer Registry data vs. active case finding Under-ascertainment of indolent incident CLL in Cancer Registries 44% among males in Ukraine but 58% in the Chernobyl study (Zablotska et al. 2013) Active case finding vs. passive incidence collection nuclear worker studies High average dose compared to other nuclear worker studies (92 vs 14 msv) Up to 20 years of follow-up Generally high prevalence of CLL in Ukraine 26 13

Chernobyl cleanup workers from Ukraine Limitations Design limitations: Recall bias? Information bias? Comparability of data from direct respondents and proxies? Low statistical power to test interactions with dose Measurement errors in dosimetry: mostly random Monte-Carlo procedure to estimate 10,000 dose estimates for each subject Uncertainties associated with the responses given during the interviews? 27 Is this a true result? Could it be explained by exposures to other environmental factors at Chernobyl? Lifestyle factors? Could it be explained by hazardous occupational exposures of cleanup workers before or after the Chernobyl accident? Is the clinical course of the radiationassociated CLL in any way different from a typical CLL in the general population? Could it be due to genetic predisposition of Ukrainians to CLL? Is CLL in Ukrainian Chernobyl cleanup workers in any different from a typical CLL in the Western population? 1. Study of non-radiation risk factors 2. Clinical follow-up study 3. Comprehensive genomic characterization of samples from Chernobyl CLL cases and comparison with general population Ukrainian and Western CLL cases 28 14

1. Study of non-radiation risks factors Odds Ratios and 95%CIs for exposure to hazardous factors 67 Gudzenko et al. Env Research 2015 1. Study of non-radiation risks factors Odds ratios and 95% CIs for petrol exposure by subtypes of leukemia 30 Gudzenko et al. Env Research 2015 15

2. CLL Clinical Follow-up Study 5-year survival rate after CLL diagnosis US in 2004: < 65 years old: 83% 65 years and older: 68% Chernobyl cleanup workers in 2010: 48% 39% 31 CLL characteristics USA, Europe and Australia Chernobyl cleanup workers Proportion of all leukemia incident diagnoses: 40% 56% Age of diagnosis: Median: 70 years 57 years <65 years ~25% 84% 50 years ~6% 22% Study enrolled only males who were <60 years during Chernobyl cleanup work Chemotherapy ~50% with a community referral 86% base 32 Mulligan et al. in Advances in the Treatment of B-cell Chronic Lymphocytic Leukaemia., Keating MJ and Tam CS, Editors; 2012. 16

2. CLL Clinical Follow-up Study 33 Finch et al. 2016 2. CLL Clinical Follow-up Study 73 Finch et al. 2016 17

2. CLL Clinical Follow-up Study Factors Associated with Survival 35 Finch et al. 2016 2. CLL Clinical Follow-up Study Summary of findings Significant dose-related increase in the hazard of dying Radiation co-morbidities? or some other factors? - not clear at this time Younger age at first exposure to radiation during Chernobyl cleanup work associated with significantly shorter survival Median age at diagnosis=57 years compared to the median age at diagnosis in the US of 72 years Latent period was not associated with radiation dose, stage of disease, chemotherapy treatment or any other clinical characteristics. Smoking and higher frequency of visits to the doctor were significantly associated with a shorter latent period Screening bias in unlikely 27.8% never visited a doctor prior to diagnosis and 36.7% visited no more than once per two years 50% were diagnosed at Rai stage two or higher 36 18

Design 3. CLL Genetic Study B-cell CLL CLL cases from Ukrainian population CLL cases from previous large sequencing studies focusing on CLL (n=100) IR-exposed CLL cases - Chernobyl cleanup workers (n=19) Unexposed CLL cases (n=39) matched on age; males only 37 Ohja et al. 2017 submitted Approach Somatic mutations 3. CLL Genetic Study Targeted deep sequencing in 530 genes found predominantly mutated in various cancers (UCSF500 panel) Copy number alterations By aligning off- target reads from targeted sequencing hg19 reference genome - CNVkit software, GISTIC & CopywriteR Pathways analysis with recurrently mutated genes Mutation signature Analysis Telomere length (TL) Predominant pathways perturbed by acquired somatic lesions Go gene analysis Non-native matrix factorization method (NMF) Estimated using Tel-Seq algorithm 38 Ohja et al. 2017 submitted 19

Non-CLL vs. CLL Non-CLL Radiosensitivity and carcinogenicity of the immature cells of bone marrow (Law of Bergonie and Tribondeau) 1 The highest risk per unit of radiation dose among all radiation-induced cancers 2 The shortest latency period (2-5 years) 2,3 Those exposed at younger ages have higher risk 2,3 CLL B-cell radionsensitivity (NHL, Multiple Myeloma, CLL/SLL) Few reliable estimates, no pooled studies For a slow-growing malignancy, the latency period should be longer CLL is the disease of older age, need lifespan studies to determine risks of exposures at younger age 39 1 Bergonie J. and Tribondeau L. Comptes-Rendus des Séances de l'académie des Sciences 143 (1906). 2 UNSCEAR 2000 Report. Vol. II: Effects. 3 BEIR VII Phase II, National Research Council, 2005. Remaining Knowledge Gaps 1. Are increased CLL risks due to radiation? High probability Somatic mutations of POT1 and TL Need further studies 2. Due to Ukrainian genetics? NOT likely, need further studies Due to interaction and activation of previously dormant pathways? What are these pathways? Due to higher frequency of heritable genetic variants or the prevalence of other environmental risk factors? 3. Due to lifestyle factors? NO 4. Due to active screening? NO 5. Is radiation-related CLL more aggressive or in any way different from the typical CLL? Appears to be more aggressive 40 20

Future directions Epidemiological studies of incident CLL identified through active case finding Pooled analyses of Chernobyl cleanup workers/ radiation workers Modifying effects of time since exposure and age at exposure? Genetic studies Mutations in telomere-related genes may be critical to radiation-associated leukemogenesis The relationship between telomere maintenance, radiation exposure, and CLL prognosis merits further investigation Studies of of populations with individual-level doses and a large number of CLL cases InterLymph project with 12,000 CLL cases and GWAS studies of samples general population but no radiation doses from occupational or medical diagnostic radiation exposures 41 Collaborative Chernobyl Studies Ukraine Belarus NCI Columbia University/UCSF Research Center of Radiation Medicine, Kyiv, Ukraine Directors: D. Bazyka, A. Romanenko Head of Dosimetry Group: V. Chumak Head of Epidemiology Group: N. Gudzenko Head of Hematology Group: I. Dyagil Head of the DCC: Yu. Belyayev 21

Collaborative Chernobyl Studies Ukraine Belarus NCI Columbia University/UCSF UCSF Joseph Wiemels Juhi Ohja Paige Bracci Kyle Walsh NCI Mark P. Little Columbia University Stuart C. Finch Robert F. Reiss 43 22