Julian Rosenman, MD, PhDUniversity of North Carolina

IMRT REVEALED!

(Or, the Second Law of Thermodynamics Strikes Again)

The First Law :“You can’t win”

The Second Law :“You break even only if you get to absolute zero”

The Third Law :“You can never get to absolute zero”

The Laws of Thermodynamics

What we will cover in this talk:

What we will NOT cover in this talk:1) 3-D conformal therapy2) Optimization engines3) Forward versus inverse planning4) Especially forward versus inverse

planning

1) What does IMRT actually do?2) What good is it?, with a

digression on lung cancer

What does IMRT (wedges!) do in this case?

Wouldn’t this do just as well?

H C

The concept behind IMRT is to exploit the tremendous redundancy in possible beam profiles that lead to the same (or nearly the same) tumor dose distribution, but deliver very different non-target doses.

!!!

OK, so what kind of non-target dose distributions are possible?

To help us answer this question I want to define the “integral dose” of a treatment as:

Then Goitein’s theorem states that the integral dose is approximately constant,

irrespective of beam arrangement!

where volume (V) = tumor + patient.(let’s call the patient dose “extraneous dose”)

dose dV,ID = Ú

Integral dose = 100 ay

To show how this works: Consider these AP PA beams

a100

100

100

y

x

a

a

y

x

100

50

50

5050ID = 50a (x+y)

Compared to 100ay

∴∆ = 50a (x-y)

With four fields the answer is a little different

a

a

So going from two fields to four reduces the dose around the tumor, but actually increases the extraneous dose.

What does IMRT do to integral dose?

A decrease in dose here... Means an increase in dose

here and here...

Can we sweep dose entirely out of the body?

Yes, by shortening the beam path

Even better, can we sweep dose into the tumor?

Again yes, but only under certain circumstances:

e = 8

e = 6

e = 5

Does brachytherapy decrease extraneous dose?

*

What is this factor of 2 about?a

rDa2

x2

D(πa2 2r-4πa3/3) =2πa2D (r-2a/3)*

{ar

4πx2 dx = 4πa2D (r-a)Ú

{ar

{ar

{ar

{ar

{ar

{ar

{ar

{ar

So, if implants improve on the tumor/patient dose ratio it has to be mu!

D = e-ur/r2

(Sorry about that)

So, how can the ability to redistribute dose be useful clinically?

A clinical reason to redistribute dose

Physical modulators, similar to the wedges, are a good way to modulate the beam

100 110 120 130908070

% Dose

30

20

10

0

% v

olum

e of

tum

or

2 modulators

1 modulator

2 wedges

1 wedge

Differential dose volume histograms

Using IMRT to make doses homogeneous is fairly straight forward. A more interesting

use is to reduce dose to certain organs...

An example is in the treatment of prostate cancer

And we can often get what we want

0

20

40

60

80

100

Perc

ent

PSA R

elap

se-F

ree

Sur

viva

l

0 12 24 36 48 60 72 84 96 108

Months

75.6 Gy (193)81 Gy (65)

64.8 - 70.2 (134)

UnfavorableT1-3PSA >10; Gleason =7

p = 0.05

p = 0.00121%

62%

43%

Is it worth it?Yes, if you believe this data.

Because IMRT can clinically protect the rectum

0 12 24 36 48 60 72 84 96

Months

5

10

15

20

81 Gy 3D-CRT (61)

81 Gy IMRT (189)

p< 0.001

Perc

ent

of G

rade

=2

Rect

al B

leed

ing

IMRT for prostate cancer treatment is probably best done with segments, static or dynamic.

There are consequences of pushing dose around, however:

1. Because of the GoiteinTheorem at least one normal tissue DVH must cross. (The Second Law strikes!)

2. You may push the dose to an unlabled volume and make it “disappear”

“Sweeping dirt under the rug” is a dangerous business!

Anterior brainstem/CordDose (%)

Right and left laterals

10090

80706050403020

100

Volume%

Parotid Gland

0 20 40 60 80 100 120

10090

80706050403020

100

Volume%

Dose (%)

6-field non-coplanarRight and left laterals

6-field non-coplanar

A potentially dangerous example: parotid sparing in head and neck cancer treatment.

0 20 40 60 80 100 120

QuickTime™ and aBMP decompressor

are needed to see this picture.

And then there is lung cancer

Is “good” radiation therapy better than

“bad” radiation therapy in the treatment of

NSCLC?

Redrawn from Perez et alRTOG 73-01

0 1 2 3 4 5

100

75

50

25

0

TREATMENT TOTAL ALIVE4000 split 101 74000 continuous 102 55000 continuous 90 16000 continuous 80 4

Surv

ival

Years

I think we act as if it is not!

“microscopic” < 10-3 cm3 5000 cGy excellentlarynx 10-2 cm3 6000 cGy excellentprostate ~50 cm3 7000-7500 cGy fair to goodcervix 102 cm3 7500-8500 cGy fair to goodNSCLC 103 cm3 6000 cGy lousy

How can 6000 cGy be right?

Tumor Typical tumor size Typical dose Control rate

Study

RTOG 7301SWOG 7628RTOG 8311CALGB 8433EORTC 8844QARC

# Patients

316140832155332

Major Errors

39 (12%)44 (31%)50 (6%)34 (23%)50 (15%)

(40%)

Minor Errors

n/an/a

183 (22%)n/a

56 (17%)

Did it matter?

noyesyes

probablyprobably

yes

In the 2-D era there were substantial targeting errors.

To improve we needed “a basic new design!”

3-D treatment planning

The “3-D paradigm”

In 1996 at UNC we decided to dose escalate the treatment of Stage III NSCLC.

Putting these ideas together lead to theUNC Study 9603 for

PS 0-1, Stage IIIA and IIIB NSCLC patients

Induction

carboplatin, AUC 6paclitaxel 225 mg/m

cycles start days 1 and 22

Concurrent CT/RT (day 43)carboplatin (AUC 2/wk)paclitaxel 45 mg/m /wk

Radiation doses escalated from:60 66 70 74 Gy

Survival of all 62 patients1.0

0.8

0.6

0.4

0.2

0.00 6048362412

1 year survival: 71%2 year survival: 50%3 year survival: 38%4 year survival: 29%

Median Survival 24 monthsMedian follow-up of survivors 43 months

• 2 cycles of carboplatin, paclitaxel, CPT-11• Concurrent carbo/paclitaxel & radiation• Radiation dose: 78 82 86 90

• Currently at 8600 cGy

Follow on—LCCC 2001

Can we use IMRT in the lung?

QuickTime™ and aBMP decompressor

are needed to see this picture.

The “usual” way to deal with motion

1) The “Gaters”, both passive and aggressive

2) The “Trackers”, both movers and shakers.

The “bounding box” is known as the PTVNewer approaches to getting rid of the PTV concept with lung and other motion include:

But what if we just did this?

And, of course, this is what happens

to the dose

Which we know how to

change to this

Modulate

I would like to call this approach “4-D” treatment planning.

In 4-D planning IMRT has a pivotal role; it is not just an add on. This was unexpected, I think.

Where h(x,t) is the time dependent warping of one image to the next

DVHstatic (d) = Volume {x e GTV: d(x) > d}

DVHdynamic (d) = Volume {x e GTV: d(h(x,t))dt > d }

The 4-D approach leads us to a new definition of the DVH

Úa

r

1. Some protection of normal tissues such asa. esophagusb.spinal cordc. heartd.possibly even liver, in some cases

2.Compensation for tissue/air interfaces

IMRT in the lung, enabled by this 4-D approach, may also lead to other benefits:

Are we under-dosing small amounts of tumor at the interface?

We can find out with Monte Carlo dose calculation at 1 mm resolution.

Ratio of Doses to Centers of Synthetic Tumors, 15 MV

0.75

0.8

0.85

0.9

0.95

1

1 1.5 2 2.5 3 3.5 4 4.5

Margin (cm)“Conventional”Monte Carlo

Preliminary results at 2 mm resolution.

I think modulators, not MLC IMRT may be needed for lung cancer

I think modulators, not MLC IMRT may be needed for lung cancer

“2-D treatment planning is fine as long as your radiation ports are not more than 1 cm high, or your patient is a coffee can”

A paraphrase of Allan Lichter’s comments circa 1988

“3-D treatment planning is fine as long as your beams are not on for more than a second, or your patient agrees not to breath”

Julian Rosenman, 2002

What? Now 4-D IMRT? This is too confusing!

“As you all know, the astrophysicist Steven Hawking has Lou Gehrig’s Disease. But what isn’t generally known is the because of a mix-up at the hospital, Lou Gehrig actually had Hodgkin’s Disease, Hodgkin had Parkinson’s Disease, and Parkinson had Alzheimer’s Disease. Alzheimer couldn’t remember what disease he had!”

-Paraphrased from George Carlin