Illinois Institute of Technology Physics 561 Radiation Biophysics

Illinois Institute of Technology Physics 561 Radiation Biophysics Lecture 8: Fractionation; Carcinogenesis Andrew Howard 26 June 2014 3/6/2021 tumor and normal-tissue responses p. 1 of 68

Plans For This Class u Dose fractionation – – – u Radiation Carcinogenesis – – – 3/6/2021 Logic behind it Mathematical models Treatment regimens Animal studies Cell-culture studies Etiology of cancer Dose-response relationships Mutagenesis tumor and normal-tissue responses p. 2 of 68

And now for something mostly different u u u We’ll move away from nonstochastic late effects to a discussion of treatment modalities for tumors. Here we’re focusing on the fact that tumors are somewhat DNA repair-deficient and therefore respond differently to ionizing radiation than healthy tissues do We’ll look, in particular, at dose fractionation as a component of treatment planning for cancer 3/6/2021 tumor and normal-tissue responses p. 3 of 68

Fractionation u u u Radiotherapy can’t wait for research: people need answers now Even in the 1930’s and 40’s it was recognized that there was an advantage in treating tumors to fractionate the dose, i. e. if the total dose you wanted to deliver was 5 Gy, you got a better therapeutic ratio if you delivered it in several small doses rather than all at once. We’ll now explore some quantitative models of the relationship between damage and fractionation 3/6/2021 tumor and normal-tissue responses p. 4 of 68

Power Law and Timing u u Witte: measured dose D required to reach the threshold for skin erythema as a function of dose rate or number of fractions n: Power law: ln. D = a + blnn, i. e. b a+blnn a lnn D=e =e e D = Qnb, where Q = ea. 3/6/2021 tumor and normal-tissue responses p. 5 of 68

Power-law treatments, continued u u Strandqvist: total time of treatment T: D = UT 1 -p; 1 -p for skin was about 0. 2. Cohen: 1 -p is tissue specific (0. 30 normal, 0. 22 for carcinomas); this enables radiotherapy! 3/6/2021 Dose, Gy 4 3. 5 3 Strandqvist model U=2 Gy 1 -p = 0. 2 2. 5 2 1. 5 1 0 5 tumor and normal-tissue responses 10 15 20 25 p. 6 of 68

Normalized Standard Dose u u u Ellis: tolerance dose D for normal tissue is related to the number of fractions N and the overall treatment time in days, T: D = r. T 0. 11 N 0. 24 The value of r is called the Normalized Standard Dose or NSD; it can be determined separately for each tissue and each treatment modality. 3/6/2021 Tolerated Dose as function of # of treatments 9 8. 5 8 7. 5 7 6. 5 6 5. 5 5 4 # of treatments 0 tumor and normal-tissue responses 50 100 150 p. 7 of 68

What are we really doing here? u u u This is curve-fitting in its most unapologetic form. As far as I know there is no attempt to attach physical meanings to the exponent (1 -p) in the Strandqvist model. Nor is there a reason to think there’s anything physically significant about the 0. 11 and 0. 24 exponents in the Ellis formulation 3/6/2021 tumor and normal-tissue responses p. 8 of 68

Time of treatment and number of fractions u u 3/6/2021 Clearly time and number of fractions are (anti-)correlated variables BUT this approach can be helpful in treatment planning, at least within the range of conditions for which the formulas are valid. tumor and normal-tissue responses p. 9 of 68

Can we do better than this? u Explicit accounting for damage in terms of repairability: – – – u Sublethal Potentially lethal Nonrepairable Model suggests that the limiting slope of ln. S vs D as you fractionate a lot is determined by the single-hit (nonrepairable) mechanism 3/6/2021 tumor and normal-tissue responses p. 10 of 68

Effect of Fractionation Fig. 11. 3: Repair capability; limiting slope determined by fraction sizes < W Surviving fraction WX Radiation Dose Y Limiting slope for fraction sizes < W or low dose rates 0. 1 Effective slope, fraction size X 0. 01 Limiting slope for large dose fraction 0. 001 3/6/2021 Effective slope, fraction size Y tumor and normal-tissue responses p. 11 of 68

Douglas & Fowler u u Used mouse-foot skin reaction to fractionated doses: ≤ 64 fractions , constant overall time For an isoeffect, the following equation held: n(a + b 2) = g where n = # of fractions, = dose per fraction note: I’m using where Alpen uses D, to reduce potential confusion with the overall dose. 3/6/2021 tumor and normal-tissue responses p. 12 of 68

Douglas-Fowler Assumptions u u u Repair occurs after single doses Biological outcome depends on surviving fraction of critical clonogenic cells Every equal fraction will have same biological effect 3/6/2021 tumor and normal-tissue responses p. 13 of 68

Survival fraction, Douglas&Fowler formulation u u u ln. S = n(Fe/a) Note that a is not a. For an appropriate choice of a, Fe = 1/(n ) Single-dose cell survival is S = exp[(Fe/a) ] So we do an isoeffect plot of Fe vs. : Fe = b + c 3/6/2021 tumor and normal-tissue responses p. 14 of 68

Douglas & Fowler Survival Fraction, Continued u u u Thus ln. S = n(b /a + c 2/a) cf. Standard LQ model, assuming constant effect per fraction: ln. S = -n(a + b 2) Defining E = -ln. S, E/(n ) = a + b 1/(n ) = a/E + b /E plot vs Fe = 1/(n ) to get a/E, b/E. In practical situations we may not be able to measure E directly 3/6/2021 tumor and normal-tissue responses p. 15 of 68

Fig. 11. 4: finding a/E, b/E u 1/total Dose, Gy-1 6 u 4 2 0 3/6/2021 u /E = intercept = 1. 75 Gy-1 /E = slope = 27 Gy-2 / = ratio = 0. 0648 Gy 0. 05 0. 10 Dose per fraction, Gy tumor and normal-tissue responses 0. 15 p. 16 of 68

Applicability u u We don’t have to be using an LQ model to work with the Douglas-Fowler formulation; we just need a nonzero slope of ln. S vs. D at low dose. Thus MTSH doesn’t work: With MTSH, S= 1 - (1 - exp(-D/D 0))n For n > 1, d. S/d. D = -n(1 -exp(-D/D 0))n-1 at D = 0, d. S/d. D = -n(1 -e 0)n-1= -n(0)n-1 = 0. For n = 1, S = exp(-D/D 0) d. S/d. D = (-1/D 0)exp(-D/D 0) at D = 0, d. S/d. D = (-1/D 0)e-0 = -1/D 0 ≠ 0. 3/6/2021 tumor and normal-tissue responses p. 17 of 68

Withers extension of Fe model u u Define flexure dose as the dose per fraction below which no further protection is provided by interfraction repair. It turns out the flexure dose is a multiple of / (units are correct: / is in Gy) 3/6/2021 tumor and normal-tissue responses p. 18 of 68

Withers extension: results u u u Let’s pick a reference total dose Dref and a reference dose per fraction ref. Then -ln. Sref = Nref( ref + ref 2), where Nref is the reference number of doses (Dref= Nref ref) Then for a different total dose D and different dose per fraction , D = N , -ln. S= N ( + 2) 3/6/2021 tumor and normal-tissue responses p. 19 of 68

Withers result u u u In order for the reference regimen to have the same effect as the test regimen, S = Sref, or -ln. S = -ln. Sref Therefore Nref( ref + ref 2) = N( + 2), i. e. Nref + Nref 2 = N + N 2 But Nref = Dref and N = D, so Nref 2 = Dref and N 2 = D Thus Dref( + ref) = D( + ) D/Dref = ( + ref)/ ( + ) = ( / + ref)/( / + ) 3/6/2021 tumor and normal-tissue responses p. 20 of 68

Withers plot (fig. 11. 5) Comparison of three different Isoeffect curves, depending on / (with ref = 2 Gy): Yellow: α/β=10 Gy Red: α/β=3. 33 Gy Blue: α/β=1. 66 Gy 3/6/2021 tumor and normal-tissue responses p. 21 of 68

Homework for later in July u u [This is a variation on problem 1 of chapter 11 in the book. I don't understand the wording of Alpen's problem, so I made up my own version] Suppose that the Ellis power law equation (11. 2) is valid in a particular tissue. A typical tumor dosing regimen consists of twenty treatments over four weeks using weekdays only, i. e. 26 days from the first Monday through the last Friday. Thus if the total dose delivered is 60 Gy, we deliver 3 Gy in each of the 20 treatments. 3/6/2021 tumor and normal-tissue responses p. 22 of 68

Homework for later in July, continued u (a) Assuming NSD=17 Gy, calculate the tolerance dose associated with this regimen. Will we be able to deliver this treatment regimen without damage to the normal tissue? 3/6/2021 tumor and normal-tissue responses p. 23 of 68

Homework, concluded u (b) If we wish to shorten the treatment time to three weeks (19 days from the first Monday to the last Friday) we will have to deliver larger doses per day, e. g. 60/19 = 3. 16 Gy/day if we include weekends. If we allow more than one dose delivery per day we can reduce the dose delivered in each treatment back to lower levels, though (1. 052 Gy/treatment). Calculate the number of doses we will have to deliver over the 19 -day period if we wish to ensure that the full 60 Gy will be tolerated. Determine the dose per treatment. 3/6/2021 tumor and normal-tissue responses p. 24 of 68

Reminder about schedules u u I will be out of the country from Wednesday 2 July through Wednesday 9 July, so there will be no class on Thursday 3 July or Tuesday 8 July. I will do a makeup class on Monday 30 June from 10: 00 am to 12: 50 pm in Stuart 107 I will do another makeup class on Friday 11 July from 9: 00 am to 11: 50 am in Stuart 213 If the students who normally attend the live section cannot attend one or both of those makeups, they are welcome to view the videos instead. 3/6/2021 tumor and normal-tissue responses p. 25 of 68

Stochastic Effects u u These are defined as effects for which the percentage of the population affected by the exposure is dependent on dose BUT the severity of the [medical] condition in an individual is independent of dose. 3/6/2021 tumor and normal-tissue responses p. 26 of 68

Does cancer really work that way? u u Not entirely Fry (1976): – – u Harderian gland tumors seldom invasive after low doses of low LET radiation More invasivity and metastasis after higher doses of low LET radiation Ullrich & Storer (1979): maybe there’s a threshold dose 3/6/2021 tumor and normal-tissue responses p. 27 of 68

Is there a threshold at the population level? Radiationinduced cancer incidence: Cases per 105 people Measured data No threshold Threshold model Dose to population 3/6/2021 tumor and normal-tissue responses p. 28 of 68

Probability of Cancer Incidence Traditional View of Population Dose. Response Relationships u Notion is that there’s a nonzero slope at D=0, rather than a threshold: Nonzero slope at D=0 3/6/2021 Background Incidence Dose tumor and normal-tissue responses p. 29 of 68

Radiation Carcinogenesis in Animals u Earliest tool in understanding radiation-induced cancer Consider mice with leukemia brought on by ionizing radiation (fig. 12. 1): Incidence (% of population) u rtality o m r o f d Correcte 50 Raw incidence 1 2 3 4 Dose, Gray 3/6/2021 tumor and normal-tissue responses p. 30 of 68

The Background Problem u Incidence, % 3 u (made-up data): Error bars make it impossible to figure out which line is correct 2 1 0. 2 0. 4 0. 6 Dose, Gray 3/6/2021 tumor and normal-tissue responses p. 31 of 68

In fact, it’s worse! u Incidence, % 3 Substantial error in the dose values too in many cases! 2 1 0. 2 0. 4 0. 6 Dose, Gray 3/6/2021 tumor and normal-tissue responses p. 32 of 68

Extrapolation to low dose Excess Incidence, % of population u The only reliable experimental measurements are made at doses much higher than the levels for which we want to set regulatory limits. Therefore we extrapolate, somehow: 4 2 6 Dose, Gray 3/6/2021 tumor and normal-tissue responses p. 33 of 68

Differential Sensitivity u Some individuals within a population are more susceptible than others – – u To tumors To other conditions Why? – – 3/6/2021 Defective DNA repair mechanisms Problems in cell signaling Lifestyle agents (smoking, drinking, lack of exercise) Genetic differences among individuals tumor and normal-tissue responses p. 34 of 68

How does differential sensitivity affect dose-response relationships? ive dit d a a pr Tumor incidence Su Responses in sensitive populations e Additiv n opulatio p l a m r o n nse in Respo Dose 3/6/2021 tumor and normal-tissue responses p. 35 of 68

Differential Exposure Fraction receiving given dose u u 0 1 3/6/2021 Mean dose = 1 Gy Maximum dose = 10 Gy Minimum dose = 0 Gy Mode of dose distribution = 1. 2 Gy Dose, Gy tumor and normal-tissue responses p. 36 of 68

Upton’s Summary of the Animal Data u u u Neoplasms of almost any type can be induced by irradiation of a suitable animal in a suitable way. Not every type of neoplasm is increased in frequency by irradiation of animals of one strain. Carcinogenic effects are interconnected through a variety of mechanisms. Some mechanisms involve direct effects on the tumor-forming cells; others don’t. High-LET radiation produces dose-dependent rather than dose-rate-dependent effects 3/6/2021 tumor and normal-tissue responses p. 37 of 68

Upton, continued u u u Development of tumors is multicausal and multistage; effects of radiation may be modified by other agents. Low to intermediate doses produce no tumors unless promoted by other agents. At high doses the effect is suppressed by sterilization of potentially transformed cells; this causes saturation. Time distribution of appearance of tumors varies with type of tumor, genetics and age, conditions of irradiation. Dose-response curves vary significantly. 3/6/2021 tumor and normal-tissue responses p. 38 of 68

Events from transformation to mutated cells (fig. 12. 2) u Many factors influence events up through malignancy Radiation event: dose, dose rate, quality Repair Mutagenic events in cell Viral Activation 3/6/2021 Oncogenes & Tumor Suppressor Genes Nonproliferating Killing or sterilizing of the cell Cells with oncogenic mutations tumor and normal-tissue responses Repair p. 39 of 68

Mutations through Malignancy u Additional influences seen Cells with oncogenic mutations Hormones Cell Cycle State Proliferative stimuli Mitosis Neoplasia Clonal selection Altered immune state Other mutations, radiation, and/or chemicals Malignancy with full autonomy of growth 3/6/2021 tumor and normal-tissue responses p. 40 of 68

Tumors: Definitions u Tumor: abnormal, de-differentiated cellular proliferation – – u u Benign: small mass reaches a certain size and then stops growing Malignant: those capable of uncontrolled growth & metastasis Cancer: a malignant tumor Carcinogen: a chemical or physical agent that increases the likelihood of cancer 3/6/2021 tumor and normal-tissue responses p. 41 of 68

Cancer: Prevalence and Significance u u 550, 000 cancer deaths per year in the US 20 -40% caused by environmental and workplace agents Others caused by smoking, diet, and natural causes Teasing apart these statistics is tricky: – – Probability of any individual getting cancer under a particular set of circumstances is small Multistage model makes multiple causes likely 3/6/2021 tumor and normal-tissue responses p. 42 of 68

Tumors and Radiation u Stochastic late effects (cf. earlier in this lecture) – Incidence, % of population – Are these effects truly stochastic? Even with cancer, there exists some dose-response effects in the individual 60 40 Fig. 12. 1: myeloid leukemia in mice for d e t s u j ality t r Ad o m t rren Intercu d Observe 20 3/6/2021 4. 7 1. 5 Dose, Gy 3. 5 tumor and normal-tissue responses p. 43 of 68

Tumors and Radiation (Cont’d) u Is there a threshold? – – u Serious Inquiry: the ED 01 experiment Brown & Hoel, Fundamental & Applied Toxicology 3: 458 (1983) More about this in a few slides. Population response u Probably not (but is this a red herring? ) Not at the population level 3/6/2021 Dose tumor and normal-tissue responses p. 44 of 68

Upton’s rules (remember? ) u u u u Irradiation can produce almost any kind of neoplasm if we do it right Not every type of neoplasm has its incidence increased by irradiation of animals of any one species or strain Carcinogenic effects depend on a variety of mechanisms Some effects are direct, some are indirect Incidence rises more steeply with dose for high-LET radiation than for low-LET radiation Irradiation interacts with other causative agents Promotion may depend on other agents 3/6/2021 tumor and normal-tissue responses p. 45 of 68

How do Cancers Begin? : The Clonal Theory u u u In general, mutational events in a single cell are sufficient to begin the process of tumorigenesis Often several mutations must arise in order for cancer to be a likely outcome Generally the mutation must be in one of the 50 or so genes that control cell replication and differentiation The mutagenic events are never enough to guarantee development of cancer (but that still leaves open the possibility that radiation could cause cancer all by itself, if it can act as a promoter too …) Mutations must be followed by promotional events, which stimulate uncontrolled cell division 3/6/2021 tumor and normal-tissue responses p. 46 of 68

Events from transformation to mutated cells (fig. 12. 2) u Many factors influence events up through malignancy Radiation event: dose, dose rate, quality Repair Mutagenic events in cell Viral Activation 3/6/2021 Oncogenes & Tumor Suppressor Genes Nonproliferating Killing or sterilizing of the cell Cells with oncogenic mutations tumor and normal-tissue responses Repair p. 47 of 68

Mutations through Malignancy u Additional influences seen Cells with oncogenic mutations Hormones Cell Cycle State Proliferative stimuli Mitosis Neoplasia Clonal selection Altered immune state Other mutations, radiation, and/or chemicals Malignancy with full autonomy of growth 3/6/2021 tumor and normal-tissue responses p. 48 of 68

Modifying Factors 3/6/2021 u Immune system u Hormonal effects u Oncogenes u Oncogenic viruses u Environmental factors tumor and normal-tissue responses p. 49 of 68

How Cancers Develop: The Multistage Theory u Initiation – – u Promotion – – – u DNA damage e. g. Intercalators Generally not mutational Involves changes in control systems, e. g. arachidonic acid cascade Tumors are present and capable of metastasis but haven’t necessarily metastisized Progression – Development of metastatic tumors 3/6/2021 tumor and normal-tissue responses p. 50 of 68

Potentiation of Effect of Radiation by Smoking u Inquiry into lung-cancer incidence among uranium miners and nearby office workers. Smokers and nonsmokers were surveyed. 3/6/2021 tumor and normal-tissue responses p. 51 of 68

How do we study radiation-induced carcinogenesis? u u u Induction and progress of cancer in experimental animals Transformation of cells grown in tissue culture Human epidemiological studies – – – Accidental exposures: Radium-dial workers, Chernobyl victims, foot fluoroscopes Medicinal exposures Atomic bomb victims 3/6/2021 tumor and normal-tissue responses p. 52 of 68

What Constitutes a Cancer? u u u 3/6/2021 Morphological change Cell immortality (escape from apoptosis) Tumorigenicity, i. e. spread of undifferentiated cells tumor and normal-tissue responses p. 53 of 68

Oncogenes u u Genes that are activated or show enhanced expression in tumors Limited data showing connection between human radiation-induced tumors and activation of oncogenes 3/6/2021 tumor and normal-tissue responses p. 54 of 68

ED 01 study u u We mentioned this a bit earlier Study run by scientists at the National Institute of Environmental Health Sciences in Research Triangle Park, North Carolina BALB-C mice analyzed for liver tumors Test compound was 2 -acetylaminofluorene, a known carcinogen in rodents: 3/6/2021 tumor and normal-tissue responses p. 55 of 68

ED 01 study, continued u u 24000 mice in various exposure groups Endpoints and elements of study: – – u Sophisticated statistical analyses: – – u Time to tumor incidence Dose “fractionation” (but this is a chemical) Initial analyses around 1981 Re-analysis a few years later Compared various dose-response models 3/6/2021 tumor and normal-tissue responses p. 56 of 68

ED 01 quantitation u u u Analyze tumor incidence according to P(t, d) = 1 - exp(-F(t, d)), where t = time and d = dose. P, the tumor incidence fraction, behaves like 1 -S in our survival curve studies Some analyses suggest that 2 -AAF is primarily a promotor, not an initiator – – So it isn’t a great model for what radiation does… But it still illustrates the importance of careful statistics! 3/6/2021 tumor and normal-tissue responses p. 57 of 68

Experimental Systems for Studying Rad-induced Tumors u u We need these because we can’t deliberately do high-dose experiments on humans! CHO cells – – u Chinese Hamster Ovary Good for looking at early effects--Initiation Difficult to model the promotional events. Transformation results in loss of contact inhibition Mouse embryo fibroblasts – – Immortalized Still display contact inhibition 3/6/2021 tumor and normal-tissue responses p. 58 of 68

CHO Cells (Continued) u Key assay: resistance to contact inhibition radiation or chemicals No radiation or chemicals 3/6/2021 tumor and normal-tissue responses p. 59 of 68

Mouse Embryo Cells: u Experiment: growing total confluence Lose contact inhibition? Can induce tumors in syngeneic animals u Limitation in both systems: u u – – Fibroblasts (mesenchymals) Most human tumors are epithelial 3/6/2021 tumor and normal-tissue responses p. 60 of 68

Mutagenesis u u u Many chemicals, as well as radiation, can be shown to cause mutations. It is therefore logical to test for mutagenicity as a first-stage inquiry into the likelihood that a compound or a radiation treatment might be carcinogenic Standard mutagenic test: The Ames test (developed by Bruce Ames), in which Salmonella cells are exposed to a chemical and mutation rates in the cells are measured. 3/6/2021 tumor and normal-tissue responses p. 61 of 68

Is an Ames Test a Good Substitute for These Complex Systems? u u No! 1, 8 -dinitropyrene is the most mutagenic substance known in the Ames test; yet it is only weakly tumorigenic in rats. 3/6/2021 tumor and normal-tissue responses p. 62 of 68

Why might we care about dinitropyrene? u u u Most mutagenic substance known in Salmonella strain TA 98: 72900 revertants/nanomole Nitroarenes like this one were found to be present in used toner, i. e. , combustion waste from Xerox toner When this appeared, Xerox chemists reformulated their toner to drastically reduce the nitroarene content in the used toner. Mermelstein (1981) Mutation Research 89: 187 -196. Löfroth et al(1980) Science 209: 1037 -1039 and Mermelstein et al (1980) Science 209: 1039 -1043. So: all’s well that ends well! 3/6/2021 tumor and normal-tissue responses p. 63 of 68

This is also a story about enzyme induction u u u Nitroarenes like dinitropyrene and other polynuclear aromatic hydrocarbons, (e. g. benzo(a)pyrene) are known to be inducers of enzyme activities Some of these enzyme activities actually activate toxicants rather than detoxifying them Most of the activity of these enzymes will detoxify; But if 1% makes things worse, we want to understand that 1% activation So we found that pretreatment with these compounds could induce subsequent binding of other compounds to mouse DNA: Howard et al (1986), Biochem. Pharm. 35: 2129 -2134. 3/6/2021 tumor and normal-tissue responses p. 64 of 68

Animal Cell-Line Cancer Studies u u How similar are these rodent cell systems (CHO, mouse) to human cells? Answer: Human cells: – – – 3/6/2021 Are more resistant to spontaneous immortalization Tend to give more nearly log-linear responses to dose Radical scavengers and cold don’t protect as much: That suggests that direct mechanisms prevail in humans and indirect mechanisms are more important in rodents tumor and normal-tissue responses p. 65 of 68

More on humans vs. rodents u u u 3/6/2021 High-LET studies indicate that repair is less effective in humans than in humans Is the timescale a factor in that? Humans live a lot longer than rodents. Promotion can be studied in animal cells, along with initiation tumor and normal-tissue responses p. 66 of 68

Radiation Carcinogenesis in Human Populations u u Occupational: radiologists, miners, dial painters Medical exposures: – – – u u Ankylosing spondylitis Nonmalignant disease in pelvis and breast Multiple fluoroscopies to chest (e. g. in TB patients) Infants & children with enlarged thymus and ringworm Children exposed in utero to diagnostic X-rays Nuclear accidents and weapon detonations Environmental background (see last chapter) 3/6/2021 tumor and normal-tissue responses p. 67 of 68

Dose-Incidence in Cancer Studies u u We seek a relationship relating post-exposure incidence ID to dose D and normal incidence In Model might be: – – – u Linear: ID = In + 1 D Quadratic: ID = In + 2 D 2 LQ: ID = In + 1 D + 2 D 2 Corrected for loss of clonogenic potential: ID = (In + 1 D + 2 D 2)exp(- 1 D+ 2 D 2) 3/6/2021 tumor and normal-tissue responses p. 68 of 68