10/2/2015 tumor and normal-tissue responses p. 1 of 68 Illinois Institute of Technology Physics 561...

04/19/23 tumor and normal-tissue responses p. 1 of 68

Illinois Institute of Technology

Physics 561

Radiation Biophysics

Lecture 8:Fractionation; Carcinogenesis

Andrew Howard26 June 2014


Plans For This Class Dose fractionation

– Logic behind it– Mathematical models– Treatment regimens

Radiation Carcinogenesis– Animal studies– Cell-culture studies– Etiology of cancer– Dose-response relationships– Mutagenesis


And now for something mostly different

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


Fractionation 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


Power Law and Timing

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:lnD = a + blnn, i.e.D = ea+blnn = ea eblnn = ea elnnb

D = Qnb, where Q = ea.


Power-law treatments, continued Strandqvist: total time

of treatment T:D = UT1-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!

Strandqvist modelU=2 Gy1-p = 0.2


Normalized Standard Dose Ellis: tolerance dose D for

normal tissue is related to the number of fractions N and the overall treatment time in days, T:

D = T0.11N0.24

The value of is called the Normalized Standard Dose or NSD; it can be determined separately for each tissue and each treatment modality.

# of treatments


What are we really doing here?

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


Time of treatment and number of fractions

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.


Can we do better than this?

Explicit accounting for damage in terms of repairability:

– Sublethal– Potentially lethal– Nonrepairable

Model suggests that the limiting slope of lnS vs D as you fractionate a lot is determined by the single-hit (nonrepairable) mechanism


Effect of FractionationFig. 11.3: Repair capability;

limiting slope determined by fraction sizes < W

Sur

vivi

ng fr

actio

n

W X Y Radiation Dose

Limiting slope forfraction sizes < W orlow dose rates

Effective slope,fraction size X

Effective slope,fraction size Y

Limiting slopefor largedose fraction

0.1

0.01

0.001


Douglas & Fowler

Used mouse-foot skin reaction to fractionated doses: ≤ 64 fractions , constant overall time

For an isoeffect, the following equation held: n( + 2) = where n = # of fractions, = dose per fractionnote: I’m using where Alpen uses D, to reduce potential confusion with the overall dose.


Douglas-Fowler Assumptions

Repair occurs after single doses Biological outcome depends on surviving

fraction of critical clonogenic cells Every equal fraction will have same

biological effect


Survival fraction,Douglas&Fowler formulation

lnS = n(Fe/a) Note that a is not . 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


Douglas & Fowler Survival Fraction, Continued

Thus lnS = n(b/a + c2/a) cf. Standard LQ model, assuming constant

effect per fraction: lnS = -n() Defining E = -lnS, E/(n) = +

1/(n) = /E + /E plot vs Fe = 1/(n to get /E, /E. In practical situations we may not be able to

measure E directly


Fig. 11.4: finding /E, /E

/E = intercept = 1.75 Gy-1

/E = slope = 27 Gy-2

= ratio = 0.0648 Gy

1/to

tal D

ose

, Gy-1

Dose per fraction, Gy0

2

4

6

0.05 0.10 0.15


Applicability We don’t have to be using an LQ model to work

with the Douglas-Fowler formulation; we just need a nonzero slope of lnS vs. D at low dose.

Thus MTSH doesn’t work:With MTSH, S= 1 - (1 - exp(-D/D0))n

For n > 1,dS/dD = -n(1-exp(-D/D0))n-1 at D = 0, dS/dD = -n(1-e0)n-1= -n(0)n-1 = 0.

For n = 1, S = exp(-D/D0)dS/dD = (-1/D0)exp(-D/D0)at D = 0, dS/dD = (-1/D0)e-0 = -1/D0 ≠ 0.


Withers extension of Fe model

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)


Withers extension: results

Let’s pick a reference total dose Dref and a reference dose per fraction ref.

Then-lnSref = Nref(ref + ref

2),where Nref is the reference number of doses(Dref= Nrefref)

Then for a different total dose D and different dose per fraction , D = N ,-lnS= N (+ 2)


Withers result In order for the reference regimen to have the same

effect as the test regimen, S = Sref, or -lnS = -lnSref

ThereforeNref(ref + ref

2) = N( + 2), i.e. Nrefref + Nrefref

2 = N + 2

But Nrefref = Dref and N = D, so Nrefref

2 = Drefref and N2 = D Thus Dref(ref) = D()

D/Dref = (ref)/ () = ( + ref)/()


Withers plot (fig. 11.5)Comparison of three different Isoeffect curves,

depending on (with ref = 2 Gy):

Yellow: α/β=10 GyRed:α/β=3.33 GyBlue: α/β=1.66Gy


Homework for later in July

[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.


Homework for later in July, continued

(a) Assuming NSD=17Gy, calculate the tolerance dose associated with this regimen. Will we be able to deliver this treatment regimen without damage to the normal tissue?


Homework, concluded

(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.

Reminder about schedules

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 10:00am to 12:50pm in Stuart 107

I will do another makeup class on Friday 11 July from 9:00am to 11:50am 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.



Stochastic Effects

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.


Does cancer really work that way? Not entirely Fry (1976):

– 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

Is there a threshold at the population level?


Radiation-induced cancer

incidence: Cases per 105 people

Dose to population

Measured data

No threshold

Threshold model


Traditional View of Population Dose-Response Relationships

Notion is that there’s a nonzero slope at D=0, rather than a threshold:

Pro

babi

lity

of C

ance

r In

cide

nce

Dose

Background IncidenceNonzero slope at D=0


Radiation Carcinogenesis in Animals Earliest tool in understanding radiation-induced

cancer Consider mice with leukemia brought on by ionizing

radiation (fig. 12.1):

Inci

den

ce(%

of p

op

ula

tion)

Dose, Gray

1 2 3 4

Raw incidence

Corrected for mortality50


The Background Problem

(made-up data): Error bars make it impossible to

figure out which line is correct

Inci

denc

e, %

Dose, Gray

1

2

3

0.2 0.4 0.6


In fact, it’s worse!

Substantial error in the dose values too in many cases!

Inci

denc

e, %

Dose, Gray

1

2

3

0.2 0.4 0.6


Extrapolation to low dose 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:

Exc

ess

Inci

denc

e,%

of

popu

latio

n

Dose, Gray

2 4 6


Differential Sensitivity

Some individuals within a population are more susceptible than others

– To tumors– To other conditions

Why?– Defective DNA repair mechanisms– Problems in cell signaling– Lifestyle agents

(smoking, drinking, lack of exercise)– Genetic differences among individuals


How does differential sensitivity affect dose-response relationships?

Dose

Tu

mo

r in

cid

ence

Response in normal population

Responsesin sensitive populations

Additive

Supra-additive


Differential Exposure

Mean dose = 1 Gy Maximum dose = 10 Gy Minimum dose = 0 Gy Mode of dose distribution =

1.2 Gy

Fra

ctio

n r

ece

ivin

g g

iven

dos

e

Dose, Gy0 1


Upton’s Summary of the Animal Data

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


Upton, continued 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.


Events from transformationto mutated cells (fig. 12.2)

Many factors influence events up through malignancy

Radiation event:dose, dose rate, quality

Mutagenic eventsin cell

Killing orsterilizingof the cell

Nonproliferating

Oncogenes &Tumor Suppressor

Genes

Cells with oncogenic mutations RepairViral

Activation

Repair


Mutations through Malignancy Additional influences seen

Cells with oncogenic mutations

HormonesCell Cycle State

Proliferative stimuliOther mutations,

radiation,and/or chemicals

Malignancy withfull autonomy of growth

Neoplasia

Clonal selectionAltered immune state

Mitosis


Tumors: Definitions

Tumor: abnormal, de-differentiated cellular proliferation

– 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


Cancer: Prevalence and Significance

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


Inci

den

ce,

% o

f po

pul

atio

n

1.5 3.5 4.7

20

40

60

Observed

Adjusted for

Intercurrent mortality

Fig.12.1: myeloid leukemia in mice

Tumors and Radiation Stochastic late effects (cf. earlier in this lecture)

– Are these effects truly stochastic?– Even with cancer, there exists some dose-response effects in

the individual

Dose, Gy


Tumors and Radiation (Cont’d) Is there a threshold?

– Probably not (but is this a red herring?)– Not at the population level

Serious Inquiry: the ED01 experiment Brown & Hoel, Fundamental & Applied Toxicology 3: 458 (1983)

More about this in a few slides.

Po

pula

ti on

re

spo

nse

Dose


Upton’s rules (remember?) 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


How do Cancers Begin?: The Clonal Theory

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

Events from transformationto mutated cells (fig. 12.2)

Many factors influence events up through malignancy

Radiation event:dose, dose rate, quality

Mutagenic eventsin cell

Killing orsterilizingof the cell

Nonproliferating

Oncogenes &Tumor Suppressor

Genes

Cells with oncogenic mutations RepairViral

Activation

Repair


Mutations through Malignancy Additional influences seen

Cells with oncogenic mutations

HormonesCell Cycle State

Proliferative stimuliOther mutations,

radiation,and/or chemicals

Malignancy withfull autonomy of growth

Neoplasia

Clonal selectionAltered immune state

Mitosis


Modifying Factors

Immune system

Hormonal effects

Oncogenes

Oncogenic viruses

Environmental factors


How Cancers Develop: The Multistage Theory

Initiation– DNA damage– e.g. Intercalators

Promotion– 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


Potentiation of Effect of Radiation by Smoking

Inquiry into lung-cancer incidence among uranium miners and nearby office workers. Smokers and nonsmokers were surveyed.


How do we study radiation-induced carcinogenesis?

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


What Constitutes a Cancer?

Morphological change Cell immortality (escape from apoptosis) Tumorigenicity, i.e. spread of

undifferentiated cells


Oncogenes Genes that are activated or show enhanced

expression in tumors Limited data showing connection between

human radiation-induced tumors and activation of oncogenes


ED01 study

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:


ED01 study, continued

24000 mice in various exposure groups Endpoints and elements of study:

– Time to tumor incidence– Dose “fractionation” (but this is a chemical)

Sophisticated statistical analyses:– Initial analyses around 1981– Re-analysis a few years later

Compared various dose-response models


ED01 quantitation

Analyze tumor incidence according toP(t,d) = 1 - exp(-F(t,d)), wheret = time and d = dose.

P, the tumor incidence fraction, behaves like1-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!


Experimental Systems for Studying Rad-induced Tumors

We need these because we can’t deliberately do high-dose experiments on humans!

CHO cells– 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


CHO Cells (Continued) Key assay: resistance to contact inhibition

radiationor chemicals

No radiationor chemicals


Mouse Embryo Cells:

Experiment: growing total confluence Lose contact inhibition? Can induce tumors in syngeneic animals

Limitation in both systems:– Fibroblasts (mesenchymals)– Most human tumors are epithelial


Mutagenesis

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.


Is an Ames Test a Good Substitute for These Complex Systems?

No! 1,8-dinitropyrene is the

most mutagenic substance known in the Ames test; yet it is only weakly tumorigenic in rats.


Why might we care about dinitropyrene?

Most mutagenic substance known in Salmonella strain TA98: 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!


This is also a story about enzyme induction

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.


Animal Cell-Line Cancer Studies How similar are these rodent cell systems

(CHO, mouse) to human cells? Answer: Human cells:

– 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


More on humans vs. rodents

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


Radiation Carcinogenesisin Human Populations

Occupational: radiologists, miners, dial painters Medical exposures:

– 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)


Dose-Incidence in Cancer Studies We seek a relationship relating post-exposure

incidence ID to dose D and normal incidence In Model might be:

– Linear: ID = In + 1D

– Quadratic: ID = In + 2D2

– LQ: ID = In + 1D + 2D2

Corrected for loss of clonogenic potential: ID = (In + 1D + 2D2)exp(-1D+2D2)


10/2/2015 tumor and normal-tissue responses p. 1 of 68 Illinois Institute of Technology Physics 561...

Documents

Transcript of 10/2/2015 tumor and normal-tissue responses p. 1 of 68 Illinois Institute of Technology Physics 561...