Quan%ta%ve Tools for Benefit/Risk Op%miza%on in Medical Imaging MIRD CommiAee Pamphlet

Transcription

1 Quan%ta%ve Tools for Benefit/Risk Op%miza%on in Medical Imaging MIRD CommiAee Pamphlet Wesley Bolch, PhD, PE, DABHP, FHPS, FAAPM Medical Physics Program Department of Biomedical Engineering University of Florida, Gainesville, FL Fall 2013 Mee%ng of the Florida AAPM / Florida HPS West Palm Beach MarrioA Hotel October 2013

2 Presenta%on Outline 1. Mo%va%on for imaging dose reduc%on and op%miza%on 2. Effec%ve dose for quan%fying the risk what is it and where does it come from? 3. Alterna%ves to the Effec%ve Dose organ specific cancer risks 4. Methods of quan%fying the benefits of medical imaging 5. Upcoming MIRD Pamphlet and Book on this topic

3 NCRP Report 160 Ionizing Radia%on Exposure of the US Popula%on Early 1980s 2006 ~15% medical ~48% medical

4 Smith- Bindman et al - JAMA 2010 (1996 to 2010)

5 Smith- Bindman et al - JAMA 2010 (1996 to 2010)

6 Smith- Bindman et al - JAMA 2010 (1996 to 2010)

7 Smith- Bindman et al - JAMA 2010 (1996 to 2010)

8 Poten%al Stochas%c Effects from CT RBM Dose from pediatric head CT Brain Dose from pediatric head CT

9 Demonstrated Determinis%c Effects from CT Cedars- Sinai Medical Center Overexposure of 206 stroke pa%ents undergoing brain perfusion studies Modifica%on of imaging protocols without understanding or considera%on of radia%on exposure Epila%on Mad River Community Hospital 2- year- old male scanned 151 %mes over 65 min Erythema shown and cataracts expected Doses es%mated to be from 2.8 to 11 Gy Life%me fatal cancer risk es%mated at 39%

10 Effec%ve Dose The effec%ve dose is a dosimetry quan%ty defined by the Interna%onal Commission on Radiological Protec%on (ICRP) and is widely used as a dose metric for stochas%c risk following ionizing radia%on exposure. It is widely used in medical imaging to compare and contrast radia%on risks associated with radiography, computed tomography, fluoroscopy, and nuclear medicine imaging. Ques%ons: What is the history of its development? What are the proper uses of this quan%ty? What are the alterna%ves when used improperly?

11 ICRP System of Radia%on Protec%on Three tenets of the ICRP system Jus%fica%on No prac%ce shall be adopted unless its introduc%on produces a net posi%ve benefit Very applicable to pa%ent exposures! Op%miza%on All exposures shall be kept as low as reasonably achievable, economic and social factors being taken into account Good use of effec%ve dose (or alterna%ves) in medical imaging Limita%on Doses to the individual shall not exceed the limits recommended for the appropriate circumstances by the ICRP Pa%ent exposures are not regulated defer to op%miza%on!

12 History of the Effec%ve Dose ICRP Publica%on 14 (1969) * Radiat Env Biophys 1975 ICRP Pub 26 (1977) ICRP Pub 60 (1991) *Radiosensi%vity and spa%al distribu%on of dose ICRP Pub 103 (2007)

13 History of the Effec%ve Dose ICRP Publica%on 14 (1969) Radiat Env Biophys 1975 ICRP Pub 26 (1977)

14 Absorbed Dose D= d ε /dm Dose Equivalent H=DQN ICRP Publica%on Absorbed Dose ( D ) and Dose Equivalent ( H ) where d ε is the mean energy imparted by ionizing radia%on to maaer of mass dm (unit Gy = J/kg ) where Q is the quality factor and N is the product of all other modifying factors (unit Sv) Q based upon RBE values for stochas%c effects in humans. Q photons = Q electrons = 1 and Q alphas = 20 func%on of L in water N could account for factors such as the absorbed dose rate or frac%ona%on scheme, but it was assigned a value of 1 in ICRP 26

15 ICRP Publica%on Effec%ve Dose Equivalent ( H E ) Under condi%ons of non- uniform irradia%on, where various organ and %ssues each receive a dose equivalent H T, the effec%ve dose equivalent is defined as the hypothe%cal, uniform, whole- body dose equivalent which, if delivered to the individual, would result in the same total detriment (fatal cancer risk / gene%c damage to offspring). Define R T - Life%me risk of fatal cancer or gene%c damage per unit dose equivalent to %ssue T (unit Sv - 1 ) These values of R T used in ICRP 26 were derived from our understanding of radia%on cancer risks as of the early 1970s. Note implicit assump%on is the linear no- threshold model (LNT)

16 ICRP Publica%on Effec%ve Dose Equivalent ( H E ) Consider a radiopharmaceu%cal that localized in the liver Components of risk are thus es%mated from both the self- dose to the liver: Risk Liver Cancer = H Liver (msv) R Liver ( msv 1 ) and the photon cross- dose to other organs: Risk Leukemia = H RBM R RBM Risk Kidneys = H Kidneys R Kidneys Risk Lungs = H Lungs R Lungs Risk Total = Risk Liver + Risk RBM + Risk Kidneys + Risk Lungs

17 ICRP Publica%on What then is the effec%ve dose equivalent H E? Hypothe%cal uniform exposure of the reference pa%ent yielding same total risk Real non- uniform exposure of the pa%ent This is the pa%ent s effec%ve dose equivalent H E

18 ICRP Publica%on Effec%ve Dose Equivalent ( H E ) ( Total Risk from uniform irradiation )=( Total Risk from nonuniform irradiation ) H TB ( i R T i )= H T 1 R T 1 + H T 2 R T 2 + H T 3 R T H T N R T N H TB ( i R T i )= ( i H T i R T i ) H E =H TB = i H T i ( R T i / i R T i ) = i H T i w T i = T H T w T H E = T w T H T

19 ICRP Publica%on Source of the Tissue Weigh%ng Factors w T Tissue Risk (msv - 1 ) Biological Effect wt Gonads 4 x 10-6 Risk to 1st two genera%ons 0.25 Breast 2.5 x 10-6 Cancer 0.15 Red Marrow 2 x 10-6 Leukemia 0.12 Lungs 2 x 10-6 Cancer 0.12 Thyroid 0.5 x 10-6 Cancer 0.03 Bone Endosteum 0.5 x 10-6 Cancer 0.03 Remainder 5 x 10-6 Cancer 0.30 Totals 16.5 x

20 History of the Effec%ve Dose ICRP Publica%on 14 (1969) Radiat Env Biophys 1975 ICRP Pub 26 (1977) ICRP Pub 60 (1991)

21 ICRP Publica%on Effec%ve Dose ( E ) Absorbed Dose D T,R is defined as the mean absorbed dose to %ssue T delivered by radia%on type R Equivalent Dose H T is defined as the product of the mean absorbed dose and a radia%on weigh%ng factor w R H T = R w R D T,R RBE alpha = ( D x rays / D alphas ) Same Biol w R Value of w R have been selected by the ICRP to be representa%ve of values of RBE of that radia%on for the induc%on of stochas%c effects at low absorbed doses. They replace the quality factor Q

22 ICRP Publica%on Radia%on Weigh%ng Factors w R Note These values of w R are appropriate for stochas%c effects only! Values RBE for alphas, for example, are only 5 for determinis%c effects

23 ICRP Publica%on Revised Tissue Factors w T ICRP Publica%on 26 Values of w T were based upon fatal cancer risks or severe gene%c damage ICRP Publica%on 60 Values of wt were based upon the concept of detriment which included: Risk of fatal cancer Allowance for years of life lost due to differences in latency periods Allowance for cancer induc%on for non- fatal cancers Allowance for the risk of severe hereditary disease

24 ICRP Publica%on Revised Tissue Factors w T

25 ICRP Publica%on Revised Tissue Factors w T

26 ICRP Publica%on Revised Tissue Factors w T

27 ICRP Publica%on Revised Tissue Factors w T

28 ICRP Publica%on Revised Tissue Factors w T Footnotes to ICRP 60 Table of w T values Age and sex averaging Defining the remainder %ssues

29 ICRP Publica%on Revised Tissue Factors w T Footnotes to ICRP 60 Table of w T values The splirng rule Note the splirng rule was dropped in ICRP 103

30 History of the Effec%ve Dose ICRP Publica%on 14 (1969) Radiat Env Biophys 1975 ICRP Pub 26 (1977) ICRP Pub 60 (1991) ICRP Pub 103 (2007)

31 ICRP Publica%on Revised Tissue Factors w T Fast forward from 1991 to the mid- 2000s Some 15 years of addi%onal follow up of the Bomb Survivors Epidemiological studies of other exposed popula%ons Cancer- specific advances treatment outcomes Time to once again update %ssue weigh%ng factors

32 ICRP Publica%on Organ / w T ICRP 26 ICRP 60 ICRP 103 Bladder Bone marrow Brain 0.01 Breast Colon Endosteum Esophagus Liver Lung Skin Salivary Glands 0.01 Stomach Thyroid Gonads Remainder

33 ICRP Publica%on Effec%ve Dose Belongs to the Reference Person Only!

34 MIRD Statements on H E and E

35 MIRD Statements on H E and E What effec%ve dose can be used for

36 MIRD Statements on H E and E What effec%ve dose cannot be used for Accordingly, the effec%ve dose for medical exposures cannot be assigned as an index of stochas%c risk to a single individual pa%ent (male or female), nor can it be assigned to male or female pa%ents of body morphometries significantly different from those of the ICRP reference individuals. These limita%ons stem from the fact that w T is both sex- and age- averaged. As a result, the sex- averaged value of w T for the breasts provides no informa%on on the risk of breast cancer in male pa%ents. Similarly, the age- averaged value of w T for the thyroid overemphasizes the risk of thyroid cancer in adult pa%ents and conversely underemphasizes that risk in children.

37 Alterna%ves to the Effec%ve Dose As presented previously in the MIRD Course on Dose Reduc%on in Pediatric Nuclear Medicine Age, gender, and organ specific cancer risk models are available from the documents such as the BEIR VII report and EPA Blue Book which require knowledge of the mean absorbed dose to different organs the very same values needed to es%mate the effec%ve dose. hap:// hap:// docs/bluebook/bbfinalversion.pdf

38 History of the BEIR Reports BEIR III (1980): General es%mates of radia%on risk at low doses BEIR V (1990): General es%mates of radia%on risk at low doses BEIR VII (2005): General es%mates of radia%on risk at low doses In the 15 years since the publica%on of the BEIR V report on low- LET radia%on, substan%al new informa%on on radia%on- induced cancer had become available from the Hiroshima and Nagasaki atomic bomb survivors (LSS Life Span Study), where slightly less than half of the original survivors were s%ll alive as of The commiaee evaluated nearly 13,000 incidences of cancer and approximately 10,000 cancer deaths in contrast to fewer than 6000 cancer deaths available to the BEIR V commiaee. Also, since comple%on of the 1990 report, addi%onal evidence had emerged from the LSS sugges%ng that other health effects, such as cardiovascular disease and stroke, could result from radia%on exposure.

39 BEIR VII Models The BEIR VII CommiAee used excess rela%ve risk (ERR) and excess absolute risk (EAR) to project radiogenic cancer risks to the U.S. popula%on for each cancer site. Defini%on Incidence Rate λ Incidence refers to new cases of disease occurring among previously unaffected individuals. It is calculated as the number of new cases of the disease occurring in the popula%on in a specified %me interval divided by the sum of observa%on %mes in that interval for all individuals who were disease free at the beginning of each %me interval.

40 Defini%ons Model Parameters BEIR VII Models AAained age of an individual a Age at exposure to radia%on Time since exposure Radia%on organ dose (in Sv) Sex of individual (1 female and 0 male) e t = a - e D s Study popula%on specific factors p

41 BEIR VII Models Excess Rela%ve Risk (ERR) Model λ I (a,e,d,s,p)= λ I 0 (a, s, p) [1+D ERR (a, e, s)] λ M (a,e,d,s,p)= λ M 0 (a, s, p) [1+D ERR (a, e, s)] where λ I (a,e,d,s,p) = projected cancer incidence rate in exposed persons λ M (a,e,d,s,p) = projected cancer mortality rate in exposed persons λ I 0 (a, s, p) = baseline cancer incidence rate by age and sex (i.e., rate for an unexposed popula%on) λ M 0 (a, s, p) = baseline cancer mortality rate by age and sex (i.e., rate for an unexposed popula%on) ERR (a, e, s) = excess rela%ve risk per unit dose (i.e., propor%onal increase in incidence rate)

42 BEIR VII Models Excess Absolute Risk (EAR) Model λ I (a,e,d,s,p)= λ I 0 (a, s, p)+ D EAR (a, e, s) λ M (a,e,d,s,p)= λ M 0 (a, s, p)+ D EAR (a, e, s) λ M 0 (a, s, p)/ λ I 0 (a, s, p) where EAR (a, e, s) = excess absolute incidence risk per unit dose (i.e., addi%onal contribu%on to the incidence rate) λ M 0 (a, s, p)/ λ I 0 (a, s, p) = incidence rates) lethality frac%on (ra%o of mortality/

43 Func%onal Form for ERR and EAR BEIR VII Models ERR (a, e, s) or EAR (a, e, s) = β s exp (γ e ) (a/60) η where β s is the sex- dependent model parameter in units of (Sv - 1 ) for the rela%ve risk model and in (per 10,000 person- Sv) for the absolute risk model γ and η are model constants showing dependences on e and a, respec%vely e = min(e, 30) 30/10

44 BEIR VII Models Model Parameter Values

45 BEIR VII Models Age Dependence for Cancer Incidence

46 BEIR VII Models Age Dependence for Cancer Mortality

47 BEIR VII Models

48 BEIR VII Models The parameter life%me aaributable risk or LAR approximates the probability that a given radia%on exposure at age e will result in a premature cancer incidence or death at age a. This probably can be thought of as weighted sums of the age- specific excess probabili%es of radia%on- induced cancer incidence or death given by the func%on M(D, e, a). The dependence on sex is implied. LAR(D,e)= e+l 110 M(D,e,a) S(a)/S(e) da where M(D,e,a) = risk at aaained age a from radia%on dose D at age e (based on either ERR or EAR models and can be either for cancer incidence or cancer mortality) S(a) = probability of surviving to age a S(e) = probability of surviving to age e L = minimum latency period (2 years for leukemia and 5 years for solids cancers)

49 BEIR VII Models Baseline rates for Cancer Incidence (blue) and Cancer Mortality (green)

50 Equa%ons for LAR Calcula%ons BEIR VII Models Cancer Incidence M I (D,e,a) = ERR I (D,e,a) λ I 0 (a) ERR Model M I (D,e,a) = EAR I (D,e,a) EAR Model Cancer Mortality M M (D,e,a) = ERR I (D,e,a) λ M 0 (a) ERR Model M M (D,e,a) = EAR I (D,e,a) λ M 0 (a)/ λ I 0 (a) EAR Model

51 BEIR VII Models Ques%on Which model to use then in providing final es%mates of LAR those based upon rela%ve or absolute risk model? Answer BEIR VII commiaee adopted geometric averaging of LAR values LAR= ( LAR ERR ) w ( LAR EAR ) 1 w Thyroid cancer: w = 1.0 (100%:0% weigh%ng) Breast cancer: w = 0.0 (0%:100% weigh%ng) Lung cancer: w = 0.3 (30%:70% weigh%ng) All other cancers: w = 0.7 (70%:30% weigh%ng) Note EPA has chosen to use a straight arithme%c weigh%ng

52 BEIR VII Models LAR Tables for Cancer Incidence

53 BEIR VII Models LAR Tables for Cancer Mortality

54 BEIR VII Models Summary BEIR VII provides LAR incidence and mortality es%mates for all solid cancers, for leukemia, and following specific cancer sites: Stomach Colon Liver Lung Bladder Thyroid Prostate Breast (Female) Uterus Ovary Other Sites (opera%onally equivalent to remainder in the effec%ve dose)

55 2011 EPA Blue Book Updates to BEIR VII Models In addi%on to the sites men%oned in BEIR VII, EPA es%mates LARs for the following addi%onal sites: - Kidney - Bone Endosteum - Skin fatal cancers only Different Breast Cancer mortality model is used (complex) - Accounts for long survival %mes between incidence and death Different Thyroid model is used (similar to NCRP Report 159) - Explicitly accounts for the dependence of ERR on both age- at- exposure and %me- since- exposure

56 MIRD CommiAee Ac%vi%es In a forthcoming MIRD Pamphlet, LAR es%mates for addi%onal sites were developed in collabora%on with David Pawel, biosta%s%cian at EPA - Oral cavity - Pancreas - Gallbladder - Central Nervous System (CNS) - Esophagus - Rectum

57 Example Thyroid Cancer Incidence For thyroid cancer, the modest discon%nui%es evident in LAR at ages 5, 10, and 15 are an ar%fact of the categoriza%on used for age- at- exposure in the thyroid risk model.

58 Example Breast Cancer Incidence

59 Example Benefit to Risk Ra%o in NM?

60 Example Benefit to Risk Ra%o in NM?

61 Data Review by Pat Zanzonico (2010 SNM) Conventional pre-op work-up Thoracotomy: 81% (78 / 97) Thoracotomy futile: 41% (39 / 78) Conventional pre-op work-up Thoracotomy: 65% (60 / 92) w/ PET Thoracotomy futile: 21% (19 / 60) Surgery (Sx)-related mortality: 6.5% w/ PET Avoided futile Sx: 20% Van Tinteren et al. Lancet 359: 1388, 2002 Extrapolation New lung cancers in US (2006): 174,470 /yr Conventional pre-op work-up Futile-Sx deaths: 3,766 /yr Conventional pre-op work-up Futile-Sx deaths: 1,547 /yr + PET Gross benefit of pre-op PET - Lives saved w/ PET: 2,219 /yr 18 FDG ED / 10 mci: 7 msv Excess cancer deaths: 61 /yr Net benefit of pre-op PET - Lives saved w/ PET: 2,158 /yr B/R Ratio - 36

62 Alterna%ves to the Effec%ve Dose Thus, as a metric for op%miza%on of benefit (e.g., image quality) and risk (cancer induc%on) in medical imaging, direct conversion of mean organ dose to cancer risk is increasingly used as an alterna%ve to the effec%ve dose. One advantage over the effec%ve dose is that one can then compare that risk to other risks (or benefits) in the medical treatment of pa%ents.

63 MIRD Pamphlet Outline 1.Introduc%on ongoing efforts to minimize radia%on exposure in imaging 2.Stochas%c effects from ionizing radia%on 3.ICRP dosimetry quan%%es related to stochas%c risk 4.Quan%fica%on of stochas%c risk in medical imaging 5.Quan%fica%on of medical benefit in medical imaging 6.Examples of risks and benefits associated with medical imaging of pa%ents 7.Conclusions

64 Forthcoming Book by Hendee & Dauer

65 Forthcoming Book by Hendee & Dauer Introduc*on Part I. Jus%fica%on in Medical Imaging Part II. Radia%on Dose In Medical Imaging Defining, Measuring and Assessing Part III. Radia%on Risks in Medical Imaging Part IV. Op%miza%on and Dose Reduc%on in Medical Imaging Part V. Shared Decision Making in Medical Imaging Part VI. Medical Imaging Safety Approaches Conclusions

66 This concludes my presenta%on I would be happy to entertain any ques%ons!