Concurrence between position of tumor estimated from the sterotactic platform and high

trityl amplitude tumor measurements

Interesting common (though not universal) concurrence of tumor and high EPR OX063 peak height from correlation with MRI Using Image Registration:• Registered T2 or T2 weighted MRI

defines tumor voxels in EPROI

MRI,

Therapeutic Puzzle

• Oxygen known to potentiate radiation therapy• TNF known to produce vascular destruction• One would think that combining TNF with

radiation would be a prescription for failure• BUT

• Using a genetically engineered adenvirus to locally produce TNF in response to radiation gives the following

Induction of TNFa expression IR-mediated cell killing

Effect of Gene Therapy + IR

Probe the physiology with DCE and EPRI oxygen images

• 9mm PC3 human prostate tumors in nude mice• Pre-treatment T2 and DCE MRI plus EPRI

Oxygen images• 10 Gy Radiation plus AdEGRTNFa or sham• 3 days post Rx T2 and DCE MRI plus EPRI

Oxygen images

Image pO2 Statistics• Control animals: Day 0 Day 3

– Mean pO2 55.1+/- 4 52.0+/-4.4– 10 torr hypoxic Fx 5.8+/-2.0 5.0+/-3.8

• Treated animals– Mean pO2 53.0+/-1.3 72.7+/-3.5– 10 torr hypoxic Fx 1.3 +/-0.3 0.2+/-0.2

• Significance of difference between Control and Treated on different days: – Day 0 pO2 p=0.8– Day 3 pO2 p=0.01 Students 2 tailed t test

Mouse 69: control, day 0 (top), day 3(bottom)

Mouse 73, treated: day 0 (top), day 3 (bottom)

How can this be?• Antivascular therapy is increasing the oxygenation: How

could this be• What if the antivascular therapy specifically targeted chaotic

tumor vessels and had less effect on the well functioning ones• Ad EGR TNFa is locally produced. Tumor vessels can’t get

rid of it but fully functional vessels can.• It’s a large molecule: on form 51KD, monomer 17KD• The TNF selectively eliminates chaotic tumor

vessels and allows better oxygenation from the functioning vessels.

EPRI Oxygen Images PredictRadiation Curability of Tumors

• MDAnderson FSa Fibrosarcomas in C3H mice• Radiation given to tumors of air breathing

animals + loosely and tightly clamped tumors• Doses given near the TCD50

• Images obtained prior to therapy in same state• Tumor voxels determined with either co-

registered MRI or stereotactic surface “touch”

EPRI Hypoxic Fraction Predicts Radiation Curability of Tumors

HF10: Fraction of tumor voxels w pO2<10 torrXRT: p=0.01 EPRI oxygen: p=0.023

Dose-HypoxiaBoundary

EPRI Hypoxic Fraction Predicts Radiation Curability of Tumors

• Bivariate logit model of response• Dose correlates with cure p=0.01• Fraction of voxels with pO2<10 torr correlates

with cure p=0.023• Slope of the boundary curve is approximately

1 Gy/%voxels less than 10 torr• Minimum cure dose (abscissa intercept):

25 Gy

Pulse: Another Approach to EPRI

• As indicated in the first lecture, information about the reality term, the G or friction term which is a determinant of linewidth equally affects the Rate of decay of the transverse magnetization,

• R2=1/T2. R2=dBg where g is the electron gyromagnetic ratio = 17.06 rad/sec/gauss

Deadtime Limits Pulse Detection• Magnetization created with a short high power pulse• To rotate magnetization from the direction of spin

sample magnetization preparation (z axis) to the transverse axis conventionally taken as the x axis, RF B1 in the y direction to precess <S> by p/2

• Bloch’s equation with T1, T2 ≈ ∞, d<S>/dt=gB1 or gB1Dt=p/2

• Broad band pulse covers frequency encoded sample (Dn large) => Dt small =>B1 large => high power pulse

• It takes time to dissipate the magnetization creation power to a low level for signal detection

Trityl Spin Probes Defeat Dead Time Limits for Oxygen Imaging

• Pulse Problem: • Magnetization may relax before the excitation

energy dissipates enough for detection • Solution: Spin probe with small relaxation

rate (R) or long relaxation time (T=1/R)• Trityls have long relaxation times ~ 5 ms• Dead time minimum ~ 0.5 ms so this works

Electron Spin Echo (ESE) Imaging

• Diminishes the dependence of image noise on dead time

• Pulse sequence π/2-τ-π spin echo imaging• t is the time between the p/2 and p pulses• Echo pulse is delayed t relative to the p pulse• Long relaxation times ~ 5 ms for trityls

Electron Spin Echo ExperimentMeasure magnetization voltage near here

Magnetization envelope decay time constant=R2 = 1/T2

2-D Spin Echo Image of 3 Vials with Different Oxygen Concentrations

3-D Spin Echo Images of Vials with 0 and 3% Oxygen for Different τ

Histogram of Decay Rates Across the Image for 0% and 2% (15torr)

Oxygen

T2 distribution (mG equivalent)

σ: 0.28 mG equivalent to ~ 0.5 torr

Mouse leg/tumor Sagittal

CW ESE CW ESE

Real Test of EPR Oxygen Imaging

• Treat with radiation to a single dose – TCD50

• Model 450 ml FSa fibrosarcomas, in legs of C3H mice

• Tumor oxygenation status with ESE EPROI

Do EPROI voxels with low pO2 predict radiation failure or success?

• HF10- Fraction of voxels pO2 <10 torr

Typical ESEOI slices from failed and controlled tumors

HF10: Fraction Tumor Voxels pO2< 10 torr

Tumor Control vs HF10

• HF10 < 9% Tumor control 18/20 or 90%

• HF10 > 9% Tumor control 7/19 or 37%

Failed

39 Animals/tumors: pO2 statistics of failures vs controlled tumors

Controlled Failed pStudents 2 tail t-test

Median pO2

(torr)30.9±1.4 27.8±1.8 0.181

HF10 7.0±1.7 11.7±1.7 0.041

Freedom from Failure HF10 > or < 9%

Wilcoxon Analysis p=0.0002

SBRT and IGRT enable safe delivery of high radiation doses/fraction (1-3Fx18-24Gy) to tumors residing in soft tissues, bone, lung and the liver

Single dose (≥24Gy) results in ~90% long-term local control of oligometastatic (M1) cancer regardless of the histological phenotype or the target organ of the metastatic lesion

Single large doses (>24 Gy) results in 80% disease free survival at 18 months for Stages I and II medically inoperable lung cancer, comparable with surgery.

The earliest visible response of tumors to high-dose radiation is an acute wave of endothelial apoptosis

Time after 15 Gy (hours)

Apo

ptot

ic T

umor

cel

ls

0

5

10

15

20

Apo

ptot

ic e

ndot

helia

l cel

ls

0

5

10

15

20

0 4 6 10 24

0.35%

0.2%17%

50%

Con

trol

17G

y

Human Ovarian CarcinomaMCA/129 Fibrosarcoma; 4hr after 15 Gy

Initiated by a plasma membrane, not DNA damage, response

Involves ASMase activation and ceramide generationInhibited in apoptosis-deficient cells (asmase-/-, Bax-/-)Does not effect DNA repair-deficient phenotypes (SCID,

atm-/-)

a-Ceramide a-ASMase

Con

trol

15 G

y30

sec

The Apoptosis Inhibits Repair of Radiation Damage

• Oxygen necessary for radiation damage repair• The vascular apoptosis creates rapid hypoxia• This sensitizes tumors to large doses of

radiation• THE EFFECT HAS ONLY BEEN SEEN IN

HISTOLOGIC SECTIONS, NOT IN VIVO

IS THERE A RAPID HYPOXIC RESPONSE IN VIVO?

• Electron Spin Echo allows oxygen images in 10 minutes

• Oxygen images can be obtained before a 20 Gy radiation dose

• Compared with the oxygenation after the dose• Voxels from the tumor provide a large

statistical sample.• Large variability of pO2 voxels in the tumor

pO2 slice before/after 20 Gy in ASMase +/+ C57 mouse

TumorT2 MRI before experiment.

Results: Δt in 50 min (~30 min) post 20 Gy

+/+: Significantly pO2 repair ↑ sensitivity

Results: Δt in 15 min post 20 Gy single image pre-post % change

+/+: Significantly pO2 repair ↑ sensitivity

Bigger: Rat tumor

4 cm Rabbit Tumor Key: local arterialtrityl perfusion

CW vs ESE

ESE vs CW

More Accurate R1 pO2 Images

• R1=1/T1: Relaxation rate of longitudinal magnetization– Self interaction exchanges energy within the spin

system• R2=1/T2: Relaxation rate of transverse

magnetization– Self interaction indistinguishable from interaction

with lattice or external spins

Try inversion recovery pulse sequence with electron spin echo detection (IRESE).

T1 EPR oxygen image

With 30 MHz BW, little spin diffusionin trityls

Concentration Dependence of OX063H Relaxation Rates

- R2e

- R1e

pO2 [torr]R

[mG]

0 10 20 30 40 50 60 70

10

15

20

25

30

35

40

45

50

0.5 mM in saline 370C

Average concentration observed in mice is ~0.3 mM but actual concentration may be ~ 1 mM with 0.3 mM uncertainty.

0.5 1 1.5 2 2.5 3 3.5 4 4.5 510

15

20

25

30

35

40

45

50

C [mM]

R [m

G]

0% O2 saline 370C

x 3x 5

- R2e

- R1e-R2ee

- R1e

R1e uncertainty ~ 0.5 mG~ 1 torr pO2 uncertainty

Orthogonal slices and histograms of: (left) T2 based ESE pO2 image (right) T1 based IRESE pO2 image of tumor bearing leg of C57 mouse.

Scaling up in volume• CW oxygen imaging of mouse

tumors:– Overmodulation – Rapid projection acquisition– Interpolation to recover

spectral fidelity• Implementation of an intermediate

size imaging system to– Explore rapid acquisition

while obtaining data from the older small imager

– Test proof of principle for the design of the larger magnet

– Explore scaling to larger objects

30 cm DHV Large Magnet

Achieves design homogeneity

Multidirectional access

Promise of EPRI Oxygen Images

Body part

Tumor

Hypoxic region

Use the DOSE vs HYPOXIC FRACTION diagramto modify the dose to the hypoxic region only

IMRT Dose “Painting” increasing dose as above based on EPRI oxygen map

Primary Tumor Dose Hypoxic Boost

Conclusions• CW 4D Images of phantoms can be obtained with ~1 mm

spatial resolution and ~1.5mG spectral resolution or ~ 3 torr in ~ 30 minutes

• Quantitative oxygen images spatially correspond to Oxylite oxymeter measurements.

• EPROI shows that anti-vascular agents attack ineffective tumor vessels -> improve oxygen delivery

• EPR Oxygen images + dose predict tumor cure• T1 based IRESE recover most of the oxygen accuracy at a

small expense (30%) in SNR and give a 1 torr resolution with a 1 mm pO2 resolution

• EPRI may enable radiation dose painting• Larger tumor capability promises human imaging

Chicago EPR Lab:

Howard HalpernMartyna ElasBoris EpelChad HaneyCharles PelizzariGene BarthVS SubramanianBrandi ButlerChanel Li

Chemistry:Viresh RawalT. ReddyTetsuo Iwama

MRI Lab:

Greg KarczmarJonathan RiverMarta Zamora

Trityl SynthesisNovosibirskInstitute forOrganic Chem.

Victor Tormyshev

http://www.geocities.com/boep777/intro.html