Radiobiological Foundations of Stereotactic Body Radiation ......2016/10/14 · • Stereotactic Body Radiation Therapy (SBRT) and Stereotactic Ablative Radiotherapy (SABR) T

© University of Washington Department of Radiation Oncology

Robert D. Stewart, Ph.D.Associate Professor of Radiation OncologyUniversity of Washington School of MedicineDepartment of Radiation Oncology1959 NE Pacific StreetSeattle, WA 98195-6043206-598-7951 office206-598-6218 [email protected]

Presented at the Hands-on SBRT Workshop (Day 1)Date and Time: Friday October 14, 2016, 9:15 to 10:15 amLocation: Seattle, WA

Radiobiological Foundations of Stereotactic Body Radiation Therapy (SBRT)

(SAM #2)

© University of Washington Department of Radiation Oncology Slide 2

Acknowledgement and a Disclaimer

Many thanks to Stan Benedict for providing information and guidance on the activities of the AAPM Working Group on Biological Effects of Hypofractionated Radiotherapy/SBRT (WGSBRT)

Thanks also to the MMTI and UW faculty Kristi Hendrickson and Juergen Meyer for organizing the workshop and the opportunity to speak

No conflicts of interest to disclose but…


Learning Objectives

To gain an appreciation for differences in cell death mechanisms of action in vitro and in vivo

To understand the clinical significance of deviations from a linear-quadratic (LQ) dose-response model

To know the sources of normal tissue tolerances for hypofractionated radiation therapy• Stereotactic Body Radiation Therapy (SBRT) and Stereotactic Ablative Radiotherapy (SABR)

To become familiar with the WGSBRT publications• AAPM Working Group on Biological Effects of Hypofractionated Radiotherapy/SBRT• https://www.aapm.org/org/structure/default.asp?committee_code=WGSBRT


Outline

Radiobiology at the Molecular, Cellular and Tissue-levels• Induction and repair of initial DNA damage, modes and kinetics of cell death• Relevance to clinical endpoints of concern• (Expected) implications of the Four R’s of Radiobiology as applied to SBRT

Linear-Quadratic (LQ) Cell Survival Model• Concepts and mechanisms of action in vitro• Limitations and caveats of the LQ for in vivo (tumor and organ at risk) radiobiology

Practical Application of the LQ• Isoeffect calculations (iso-TCP, iso-NTCP, iso-BED, …)• Equivalent tumor prescriptions• Equivalent tolerance dose estimates

Relevant Literature


106 s

Physics → Chemistry → Biology → Clinic

Radiation

Ionization Excitation

DNA damage

10-6 s

10-18 to 10-10 s

ChemicalRepair10-3 s

Enzymatic Repair(BER, NER, NHEJ, …)

CorrectRepair

102 s

Incorrect or Incomplete Repair

Cell Death

103 s 105 s

Small- and large-scale mutations(point mutations and chromosomal aberrations)

104 s 105 s

104 s

1 Gy ~ 1 in 106

O2 fixation

Neoplastic Transformation

Somaticcells

ClonalExpansion

107 s2nd Cancer

108 s

HeritableEffects

Germline105 s

Loss of Function and Tissue Remodeling

Angiogenesis and Vasculogenesis

Self renewal and Differentiation

Inflamatory and Immunological Responses

Early Effects(erythema, …)

108 sLate Effects(fibrosis, …)

Absorbed Dose

Chronic hypoxia(> 1-2 h)

Chronic hypoxia

(> 4-10 h?)

Acute hypoxia


Mechanisms and Classification of Initial DNA Damage

Types of Elementary DNA Lesion Direct Effect Indirect Effect Classification of Clusters of DNA Lesions


Elementary Types of Damage

Individual nucleotides that become damaged are termed a DNA lesion• Abasic or AP (apurinic/apyrimidinic) sites = base loss• Base damage (A, T, G or C)• Strand breaks (damage to sugar or phosphate), usually accompanied by base loss

Strand break

Base damage

abasic site


Direct Ionization of the DNA (“direct effect”)

1.8 to 2.3 nm

+ =


Ionization of H2O close to the DNA (“Indirect Effect”)

+2 2+H O H O eγ −→ +

+ +2 2 3H O + H O OH H O→ ⋅ +

ionized water

hydroxl radical

0.5

μm

1 ps 1 ns 0.1 μs 1 μs

10 μs

Spatial distributions of ·OH in liquid water. Red dot indicates location of a 1 μm segment of a 24 MeV 4He2+ ion (26 keV/μm) directed into the image

Image adapted from Muroya Y, Plante I, Azzam EI, Meesungnoen J, Katsumura Y, Jay-Gerin JP. High-LET ion radiolysis of water: visualization of the formation and evolution of ion tracks and relevance to the radiation-induced bystander effect. Radiat Res. 165(4), 485-491 (2006).


Clusters of DNA Lesions

One of the more unique characteristics of ionizing radiation is its ability to produce several DNA lesions within one or two turns of the DNA, i.e., a cluster of DNA lesions*

* “Clusters of DNA lesions” are also referred to in the literature as locally multiply damaged sites (LMDS) or multiply damaged sites (MDS)

+ =


Double Strand Break (DSB)

A DSB is a cluster that contains at least two strand breaks on opposing strands within ~ 10 bp of each other

Complex DSB (11 lesions)

Simple DSB (2 lesions)

Number of lesions forming a cluster is a measure of “complexity”

Undamaged DNA segment

BaseSugar-phosphate backbone

Strand breaks formed by radiation are chemically reactive (“sticky”)

Red square ( ) denotes base damage


Non-DSB Clusters

Base damage (5 lesions)

Simple SSB (1 lesion) Complex SSB (6 lesions)

SSB (single strand break) denotes the family of all types of cluster other than the DSB that contain at least one strand break

Clusters that do not contain any strand breaks are referred to as “base damage”


DSB Induction ∝ Dose

6 to 9 DSB Gy-1 Gbp-1

0.2 keV/μm

23 keV/μm

Frankenberg D, Brede HJ, Schrewe UJ, Steinmetz C, Frankenberg-Schwager M, Kasten G, Pralle E. Induction of DNA double-strand breaks by 1H and 4He ions in primary human skin fibroblasts in the LET range of 8 to 124 keV/microm. Radiat Res. 151(5), 540-549 (1999).

human skin fibroblasts

DSB are directly or indirectly formed through the action of individual tracks + spatially and temporally correlated δ-rays

Not formed by 2 or more independent tracks up to at least a few hundred Gy –would be a linear-quadratic function of dose if DSB created by > 2 tracks.


Reproductive Cell Death – Modes and Kinetics

Uni- and bi-nucleated giant cells

Several distinct modes contribute to reproductive cell death…

Cell death kinetics often take hours to days or even weeks and months after ionizing radiation to fully manifest…

Image adapted from Lauber K, Ernst A, Orth M, Herrmann M, Belka C. Dying cell clearance and its impact on the outcome of tumor radiotherapy. Front Oncol. 2012 Sep 11;2:116. doi: 10.3389/fonc.2012.00116.


Necrosis

Intracellular contents released in uncontrolled fashion (membrane ruptures)

Chu K, Leonhardt EA, Trinh M, Prieur-Carrillo G, Lindqvist J, Albright N, Ling CC, Dewey WC. Computerized video time-lapse (CVTL) analysis of cell death kinetics in human bladder carcinoma cells (EJ30) X-irradiated in different phases of the cell cycle. Radiat Res. 158(6):667-77 (2002).

200X phase-contrast images of EJ30 human bladder carcinoma cell undergoing necrosis 45-48 h after exposure to 6 Gy of 220 kVp x-rays


Apoptosis (HCT116 colorectal carcinoma)

Chu K, Teele N, Dewey MW, Albright N, Dewey WC. Computerized video time lapse study of cell cycle delay and arrest, mitotic catastrophe, apoptosis and clonogenic survival in irradiated 14-3-3sigma and CDKN1A (p21) knockout cell lines. Radiat Res.162(3):270-86 (2004).


Uni-nucleated Giant Cells (arrested cell)

200X phase-contrast images of EJ30 human bladder carcinoma cell after exposure to 6 Gy of 220 kVp x-rays



Binucleated giant cells

Cell completes mitosis but progeny fuse back together

200X phase-contrast images of EJ30 human bladder carcinoma cell after exposure to 6 Gy of 220 kVp x-rays



Ultimate Fate of Giant Cells

Prieur-Carrillo G, Chu K, Lindqvist J, Dewey WC. Computerized video time-lapse (CVTL) analysis of the fate of giant cells produced by X-irradiating EJ30 human bladder carcinoma cells. RadiatRes. 159(6):705-712 (2003). Observed 203 giant EJ30 human

bladder carcinoma cells after 6 Gyof x-rays

83 died, 95 divided at least once, 25 remained intact and alive


What’s the connection between DNA Damage and Cell Survival?

Are all DSB lethal? What about SSB? Base Damage? Breakage and reunion therapy, a.k.a., what happens when pairs of

DSB interact?


Are all DSB Lethal?

After 1 Gy dose of low LET radiation, a typical human cell sustains 45 + 10 DSB Gy-1 cell-1. If DSB are always lethal, the fraction of cells that will survive a 2 Gy dose is

For comparison, many published studies indicate a surviving fraction of 0.1 (repair compromised) to 0.9 (repair proficient) cells after a 2 Gy dose of radiation. The DSB and cell survival measurements imply that

< 2% of initial DSB formedin a cell are lethal

Only those cells that do not sustain a radiation-induced DSB survive (Poisson distribution of DSB among irradiated cells)

∴ cells must be really good at repairing/rejoining DSB!

See also the classic review: DT Goodhead. Initial events in the cellular effects of ionizing radiations: clustered damage in DNA. IJRB 65(1): 7-17 (1994).


Repairing a DNA Double Strand Break (DSB)

Cells have three key mechanisms to repair DSBs:• Nonhomologous End Joining (NHEJ) is a

fast, relatively accurate process which is available throughout the cell cycle;

• Homologous Recombination (HR) is much more accurate but is slower and depends on the availability of a sister chromatid, meaning it is only available in S and early G2 phases of the cell cycle;

• Alternative- or Backup-End Joining is a fall-back process which is typically extremely slow and error-prone.

Slide courtesy S. McMahon


Time-Lapse Video of (colon carcinoma) HCT116 p21 -/- cells

Chu K, Teele N, Dewey MW, Albright N, Dewey WC. Computerized video time lapse study of cell cycle delay and arrest, mitotic catastrophe, apoptosis and clonogenic survival in irradiated 14-3-3sigma and CDKN1A (p21) knockout cell lines. Radiat Res. 162(3):270-86 (2004).


Breakage and Reunion Theory

DSB

DSBDSBBreak-ends

Break-ends associated with one DSB incorrectly rejoined to break-end associated with a different DSB

Proximity Effects: pairs of DSB formed in close spatial and temporal proximity are more likely to rejoin incorrectly than pairs of DSB separated in time and\or space (→ dose rate and LET effects)


Lethal and Non-Lethal Aberrations

CentromereAcentric fragment

Dicentric

Symmetric (reciprocal) translocation

Dicentrics and acentric fragments are usually lethal in the reproductive sense because segregation of chromosomes at mitosis is disturbed. Correct DSB rejoining and symmetric (reciprocal) translocations are usually consistent with continued cell division (cell survival)

Correct rejoiningCentric ring and acentric fragment

Hlatky L, Sachs RK, Vazquez M, Cornforth MN. Radiation-induced chromosome aberrations: insights gained from biophysical modeling. Bioessays 2002, 24(8):714-23.


( ){ }2( ) expS D D Dα β= − +

Conceptual Basis for the Linear-Quadratic (LQ) Survival Model

D αβ

=

Half of cell killing due to one-track mechanism and half due to two-track mechanism

EJ Hall and AJ Giaccia, Radiobiology for the Radiologist, 6th Edition, Lippincott Williams & Wilkins (2006)

≅ lethal aberrationor point mutation

Probability (fraction) of cells that do notacquire a lethal point mutation or chromosome aberration


1:1 relationship between lethal aberrations and cell survival

Source: Cornforth and Bedford, Rad. Res., 111, p 385-405 (1987). See also Figure 3.4 in Hall (p. 37)

AC 1522 normal human fibroblasts irradiated by x-rays

S = e-Y

S = fraction that surviveY = average number of lethal

aberrations per cell


Mitotic Death, Apoptosis, the LQ and Fraction Size Effects

Apoptosis Dominant

{ }( ) expS D Dα≅ − (α/β) → ∞

( ){ }2( ) expS D D Dα β= − +

Mitotic Death Dominant (apoptosis absent)

(α/β) ~ dose range< 10-14 Gy

insensitive to fraction size

Shift from apoptosis to mitotic death associated with (a) reduced or no p53 activity and (b) increased duration of radiation-induced G2 delay. Duration of G2 phase delay tends to increase with increasing (a) dose and (b) particle LET. p53 often inactive or mutated in many cancers (less apoptosis) but “wild-type” in normal tissue (more apoptosis) – potential argument favoring hypofractionation over conventional fractionation!

Image: Hall and Giaccia, Radiobiology for the Radiologist, 6th Edition Lippincott, Williams & Wilkins (2006)


What about SSB? Base Damage?

After 1 Gy dose of low LET radiation, a typical human cell sustains 1000 + 200 SSB Gy-1 cell-1. Using the same logic as for DSB, < 0.1% of initial SSB are involved in cell killing, i.e., cells are even better at repairing SSB than DSB. Same argument applies to base damage.

Base damage (5 lesions)

Simple SSB (1 lesion) Complex SSB (6 lesions)

Small-scale (point) mutations are far easier for a cell to tolerate than larger-scale chromosome aberrations!


Linear Quadratic (LQ) Survival Model

( )2( ) expS D D GDα β= − −

One-Track Damage

Inter-Track Damage(pairs of DSB interact)

G relates to the cell’s capacity for DSB repair and is influenced by the temporal pattern of radiation delivery

Absorbed Dose (Gy)0 5 10 15 20 25 30 35

Surv

ivin

g Fr

actio

n

10-4

10-3

10-2

10-1

100

0.12 Gy h-1

0.5 Gy h-1

45 Gy h-1

When the protraction factor Gis included in the LQ model, good fits are possible for small and large doses

LQ with G correction for fraction size and dose rate effects has 3 cell- and tissue specific adjustable parameters (α, α/β and τ).

Measured data from Stackhouse M.A. and Bedford J.S. Radiat. Res. 136, 250-254 (1993)


Dose Rate Effects in IMRT – Relevant to SBRT?

( ){ }2( ) expS D D GDα β= − +

βGD2 term much more significant for SBRT than IMRT because

(a) dose per fraction “large” compared to α/β

(b) dose fraction delivery times comparable to half-time for sub-lethal damage repair (τ)


Clinical Relevance of In Vitro Radiobiology – Good News

Little or no evidence in literature that fundamental molecular mechanisms are different in vitro and in vivo.• Radiation creates about the same numbers and types of DSB in vitro

and in vivo – linear function of dose. Damage to other cellular constituents (membranes, proteins, …) as well as SSB and other types of DNA damage also a linear function of dose up to at least a few hundred Gy (low and high LET radiations)

• Same DNA repair (BER, NER, HR, NHEJ, …) and tolerance (lesion bypass, …) mechanisms convert a subset of the initial DNA damage into small-scale (“point”) mutations and larger-scale chromosome aberrations

• Chromosome aberrations that are lethal in vitro are almost certain to also be lethal in vivo. But… some types of non-lethal DNA damage in vitro may become (initiate) controlled or uncontrolled reproductive cell death in vivo (and vice versa) because interactions of the damaged cell with the extracellular and tissue environment, immune surveillance, …

Why lethal in vitroand not in vivo??

Dicentric chromosome aberrations + acentric fragments


Clinical Relevance of In Vitro Radiobiology – Good News (?)

Little or no evidence in literature that the initiating (molecular) events and the cell death modes (apoptosis, necrosis, mitotic catastrophe, …) are absent or operate in very different ways in vitro and in vivo• Above statement does not mean the absolute frequency and kinetics of the cell death process are the same in

vitro and in vivo – only that the fundamental molecular and cellular mechanisms are conserved (present and operating) in vitro and in vivo

• Nor does it mean that there are not additional modes and mechanisms relevant to tumor biology and/or the responses of normal tissues

( ){ }( ){ } { }

2

2

( ) exp

exp factors and mechanismsN A S

S D D GD

D GD in vivo

α β

α α α β

= − +

= − + + + + ×

Mostly lethal chromosome aberrations

Several cell death modes initiated by SSB, DSB, ..., oxidative damage to membranes or other cellular constituents, energy deposition in the extracellular matrix, …

Abscopal effects and …


Conventional RT and SBRT – Four R’s of Radiobiology

Less Repair (TCP ↑ NTCP ↑) Less Repopulation (TCP ↑ NTCP ∼) Less Reoxygenation (TCP↓ NTCP ∼) More Redistribution in Cell Cycle

(TCP ↓ NTCP ∼)

SBRT compared to Conventional Fractionation. For large tumor dose per fraction, expect

DNA repair most significant of Four R’s for SBRT

( ){ }2( ) expS D D GDα β= − +

{ }( ) expS D Dα≅ −

Tumor

OAR (if d << α/β)


Biology Beyond the Cell – in vitro vs in vivo radiobiology

In vivo always dealing with complex interactions between extracellular environment, multiple cell types and system-wide physiological and immune system biology…

In vitro usually dealing with a relative homogeneous population of dividing or non-dividing cells in an artificial (cell culture) environment.

Multi-cellular spheroids and 2D or 3D tissue models are intermediate between in vitro cell cultures and in vivo models


Clinical Relevance of In Vitro Radiobiology – Bad News (?)

Kinetics of tumor and normal-cell birth/death are not static during a course of fractionated radiation therapy• Differs among patients, tumor site and type, fractionation schedule, total dose, low or high LET

radiation, …• Re-oxygenation, tumor and tissue-level

repopulation and recovery (e.g., cell migration from adjacent regions of tissue), heterogeneity in the characteristics of an evolving (genetically unstable) tumor cell population, …


Fraction Size (Gy)0 5 10 15 20 25 30

Sur

vivi

ng F

ract

ion

10-6

10-5

10-4

10-3

10-2

10-1

100

SF for a Heterogeneous Cell Population

α = 0.15 Gy-1 α/β = 25 Gy

Total Dose (Gy)0 5 10 15 20 25 30 35 40 45 50

Surv

ivin

g Fr

actio

n

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

100

Resistant (5%)

Sensitive (95%)α = 0.05 Gy-1

α/β = 1.5 Gy

Composite

α = 0.33 Gy-1

α/β = 93 Gy

LQ fit to composite α = 0.27 Gy-1

α/β = 53.5 Gy

LQ fit to composite

5 Gy per fraction


Local Tumor Control Probability (TCP)

Total Treatment Dose (Gy)60 65 70 75 80

Tum

or C

ontro

l Pro

babi

lity

0.0

0.2

0.4

0.6

0.8

1.0

Total Treatment Dose (Gy)60 65 70 75 80

Tum

or C

ontro

l Pro

babi

lity

0.0

0.2

0.4

0.6

0.8

1.0 Fraction: 1.7 – 2.3 Gy(35 fractions) (+5% uncertainty)

α = 0.15 Gy-1, α/β = 3.1 Gy

α = 0.15 Gy-1, α/β = 3.1 Gy

12% uncertainty in dose

Even small levels of uncertainty in the biological parameters (α and α/β) have a significant impact on our ability to predict TCP


Take Away…

LQ cell survival does have some mechanistic underpinnings when applied to a homogeneous cell population

But… application of the LQ to a heterogeneous cell population has fundamental conceptual and practical challenges unrelated to the hotly debated accuracy of the LQ for large doses per fraction!

Fraction Size (Gy)0 5 10 15 20 25 30

Sur

vivi

ng F

ract

ion

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

100

LQ withG = 1

22( 1) /ln 2 /

xG e x xx t tλ τ

−= + −≡ Δ = Δ

LQ with

“linear-quadratic-linear” dose response


Radiation Therapy and the Immune Response (“Abscopal Effects”)

Apoptosis less effective stimulating anti-tumor immune response

Uncontrolled cell death modes (mitotic catastrophe, necrosis, …) generate pro-inflammatory cytokines that tend to enhance anti-tumor immunity

Role of senescent cells is ?

Image adapted from Lauber K, Ernst A, Orth M, Herrmann M, Belka C. Dying cell clearance and its impact on the outcome of tumor radiotherapy. Front Oncol. 2012 Sep 11;2:116. doi: 10.3389/fonc.2012.00116.


SBRT and Anti-Tumor Immune Responses

Image adapted from Derer A, Deloch L, Rubner Y, Fietkau R, Frey B, Gaipl US. Radio-Immunotherapy-Induced Immunogenic Cancer Cells as Basis for Induction of Systemic Anti-Tumor Immune Responses - Pre-Clinical Evidence and Ongoing Clinical Applications. Front Immunol. 2015 Oct 8;6:505.

Large SBRT-like (~ 10-15 Gy/fx) of low LET radiation and smaller doses of high LET radiations may be especially effective a shifting cell killing from non-immunogenic cell death modes (apoptosis) towards immunogenic cell death modes (mitotic catastrophe or necrosis).

With IR(large dose/fx)

apoptosis

necrosis

Enhanced anti-tumor activity

No IR(or small dose per fx)

necrosisapoptosis

Tumor cells evolve to evade the immune system


Apoptosis and the p53 gene (low LET)

In mouse embryonic fibroblasts, loss of the p53 (“tumor suppressor”) gene substantially decreases sensitivity to ionizing radiation

p53 null (-/-) = immune stimulating mitotic death (negligible apoptosis)

p53 wild-type (+/+) = lots of apoptosis

Hardenbergh PH, Hahnfeldt P, Hlatky L, Takemoto C, Shimamura A, McGill G, Fung CY, Bodis S, Fisher DE. Distinct mathematical behavior of apoptotic versus non-apoptotic tumor cell death. Int J Radiat Oncol Biol Phys. 43(3), 601-605 (1999).

p53 activity reduced or inactive (mutated) in many cancers (less apoptosis) but “wild-type” p53 activity in normal tissue (more apoptosis)


Apoptosis Induced by High LET Radiations

p53 null (-/-)

p53 wild-type (+/+)

p53 wild-type (+/+)

p53 null (-/-)

200 keV x-ray(1.7 keV/μm)

Fe ion(200 keV/μm)

If treat with high LET, p53 status of tumor may have little impact on the numbers of dead or dying cells.

With low LET radiations (MV x-rays and protons), p53 status of tumor may be very significant.

Takahashi A, Matsumoto H, Furusawa Y, Ohnishi K, Ishioka N, Ohnishi T. Apoptosis induced by high-LET radiations is not affected by cellular p53 gene status. Int J Radiat Biol. 2005 Aug;81(8):581-6.

Human non-small lung carcinoma cells (H1299) irradiated in vitro


Optimal Fraction Size?

To increase the therapeutic ratio, want to increase anti-tumor immune response and suppress inflammatory/and immune stimulating signaling in relevant OAR

Optimal fraction size and total dose might depend on whether or not apoptosis dominant or (largely) absent• Optimal fraction size may be very different for radiation sensitive

(apoptosis dominant) and radiation resistant tumors (apoptosis absent)

Apoptosis Dominant

{ }( ) expS D Dα≅ − ( ){ }2( ) expS D D GDα β= − +

Apoptosis Absent

Resistant Tumors(some OAR?)

Radio-responsive Tumors (?)(most OAR?)


Clinical Relevance of In Vitro Radiobiology – Bad News (?)

Cancers almost always treated with a combination of two or more modalities• Potential to have synergistic or antagonistic treatment interactions – clinical outcomes not just about cell

killing by ionizing radiation!

Image adapted from Gaip US, Multhoff G, Scheithauer H, Lauber K, Hehlgans S, Frey B, Rödel F. Kill and spread the word: stimulation of antitumor immune responses in the context of radiotherapy. Immunotherapy. 2014;6(5):597-610.

For all of these reasons (and more), the usefulness of the LQ model and related concepts for guiding SBRT and SABR is hotly debated and controversial topic…


Use of the LQ for Practical Clinical Guidance

Seems a bit hopeless to have any faith in the usefulness of the LQ, no? Fortunately, there are ways to work around the substantial seeming

conceptual and practical limitations of the LQ Method 1

• Treat the LQ and related metrics, such as the biologically equivalent dose (BED), as a “good enough” empirical or quasi-empirical metric for the numbers of dead or dying cells in an organ, tissue or tumor target

• Fit the combined LQ + TCP (or NTCP) model to clinical data to determine relevant parameters (e.g., α/β) • Use the model to guide and inform clinical decision making with appropriate awareness of the limits of such

models arising from uncertainties associated with the determination of biological parameters

Method 2• Re-cast the clinical problem as iso-effect calculation• Derive estimates of biological parameters (α/β) from an analysis of clinical outcomes – or – from our practical,

somewhat subjective, experience treating patients


Example of Method 1

Brown JM, Carlson DJ, Brenner DJ. The tumor radiobiology of SRS and SBRT: are more than the 5 Rs involved? Int J Radiat Oncol Biol Phys. 88(2):254-62 (2014)

ln 1 1/ /

S GD dBED D Dα α β α β

≡ − = + = +

Biologically Equivalent Dose (BED)

because G ≅ 1/n and d = D/n

Analysis of TCP for stage I non-small cell lung cancer at > 2 year survival.

No compelling evidence that LQ dose-response model does not suffice to explain fraction-size effects in SBRT and 3D-CRT treatments. Occam's razor applies…

Correction for DSBrepair capacity


Method 2: Iso-Effect Calculations (iso-TCP = iso-survival = iso-BED)

( ) ( )2 2

( ) ( )

exp expR

R R

S D S D

D GD D GDα β α β

=

− − = − −

Reference Treatment = Alternate Treatment Seeking the total treatment dose D that produces the same cell survival level as a reference treatment that delivers total dose DR.

Take logarithm, apply quadratic formula and rearrange terms

4 4( / ) 1 1 1 ( / ) 1 12 ( / ) ( / ) 2 ( / )

R R R

R

n nn n

D D BEDDn

α β α βα β α β α β

= − + + + = − + +

Equivalent (physical) dose D determined by a single biological parameter (α/β) and one known, physical parameter n (number of fractions).

This formula converts BED (a non-physical dose) back into a true physical dose that can directly serve as a tumor prescription and/or dose-volume constraint for treatment planning.


Why are iso-effect calculations useful?

Eliminate (at least) one adjustable, highly uncertain biological parameter (α) from the modeling process

Standard of care clinical experience used to offset limitations of the LQ and our incomplete understanding of the relevant biology

4( / ) 1 12 ( / )

RBEDnn

D α βα β

= − + +

Uncertainty in D mainly arises from uncertainties associated with α/β.

New (alternate) Treatmentn = desired number fractions

Reference Treatment(“clinical experience”)

1( / )

RR R

dBED Dα β

= +

Current Standard of CareTotal Dose(new treatment)


Equivalent Fractionation Schedules

Number of Fractions, n5 10 15 20 25 30 35 40 45 50

Frac

tion

Siz

e (G

y)

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0


Frac

tion

Siz

e (G

y)

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0

α/β = 1.5 Gy α/β = 1.5 Gy

44 x 1.8 Gy

DR = 79.2 Gy (44 × 1.8 Gy) DR = 58.8 Gy (20 × 2.94 Gy)

Use any point along isoeffect line for the reference treatment

20 x 2.94 Gy

Stanford Cyberkife

1 4( / ) 1 12 ( / )

RD BEn

Ddn

α βα β

= = − + +


Effects Intra-Tumor and Inter-Patient Heterogeneity


Frac

tion

Siz

e (G

y)

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0


Frac

tion

Siz

e (G

y)

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0

44 x 1.8 Gy

20 x 3 Gy

DR = 79.2 Gy (“clinical experience”) DR = 60 Gy (“clinical experience”)

95% CI

10,000 values for α/β sampled from a uniform pdf (range 1 to 10 Gy)


Equivalent OAR Tolerance Dose (TD)

4( ) ( / ) 1 1 12 ( / ) ( / )

R Rn TD TDTD nn

α βα β α β

= − + + +

Treat TDR and (α/β) as adjustable parameters and perform a non-linear regression analysis of published or in-house treatment planning constraints.

Number of Fractions0 5 10 15 20 25 30 35 40 45 50

Eso

phag

us T

D (G

y)0

5

10

15

20

25

30

35

40

45

50

55

60

65TDavg

TD(< 33-50 cc) TD(< 5-15 cc)TD (< 0.5 cc)

(α/β) = 2.92 GyTDR = 15.35 Gy (< 0.5 cc), TDR = 10.81 Gy (entire OAR)

Practical way to leverage current and rapidly developing clinical experience without the need to collect clinical data for many years…


TDR for Trachea and Spine


Trac

hea

TD (G

y)

0

5

10

15

20

25

30

35

40

45

50

55

60

65

TD(< 4 cc)TD (< 0.5 cc)

(α/β)R = 2.09 GyTDR = 11.86 Gy (< 0.5 cc and < 10% of 5-6 mm vol)TDR (1.2 cc) = 10.81 Gy

(α/β)R = 20.4 GyMaximum TDR = 22.6 Gy (< 0.5 cc) Average TDR = 11.55 Gy (entire OAR)



TD fo

r Che

st a

nd R

ibs

(Gy)

0

5

10

15

20

25

30

35

40

45

50

55

60

65TD (< 0.5 cc)Emami et al. 1991

Stomach, Chest Wall and Ribs


Sto

mac

h TD

(Gy)

0

5

10

15

20

25

30

35

40

45

50

55

60

65TDavgTD(< 5-15 cc)TD (< 0.5 cc)Emami et al. 1991

(α/β)R = 3.76 Gy. Solid black line: TDR(< 0.5 cc) = 15.53 Gy. Black dashed line: TDR(< 5-15 cc) = 12.8 Gy, Black dotted line: TDR(mean) = 8.03 Gy. (α/β)R = 19.32 Gy and TDR(< 0.5 cc) = 24.0 Gy


Sources of OAR Tolerance Dose Parameters

Emami et al. 1991• Emami B, Lyman J, Brown A, Coia L, Goitein M, Munzenrider JE, Shank B, Solin LJ, Wesson M. Tolerance of

normal tissue to therapeutic irradiation. Int J Radiat Oncol Biol Phys. 21(1):109-22, May 1991• Burman C, Kutcher GJ, Emami B, Goitein M. Fitting of normal tissue tolerance data to an analytic function. Int

J Radiat Oncol Biol Phys. 21(1), 123-135, May 1991.

QUANTEC 2010 (special supplemental issue – 27 papers)• Int J Radiat Oncol Biol Phys. Volume 76(3), March 2010

AAPM Working Group on SBRT (WGSBRT) (special issue – 12 papers)• Seminars in Radiation Oncology Volume 26(2), April 2016

Start with above... Use cited papers to drill down on additional study details


Parting Thoughts …

Many decades of basic, translational and clinical studies suggest that TCP and NTCP modeling, especially in the form of iso-effect calculations, provide useful biological guidance when evidence from randomized clinical trials is lacking.

Conceptual framework underpinning TCP and NTCP modeling is (implicitly or explicitly) based on a working hypothesis in which the spatial and temporal pattern of dead, dying and damaged cells, regardless of the underlying mechanisms of action (DNA damage and repair, immune response, …), largely determines clinical outcome

Is this a reasonable working hypothesis? Or are we in for a major surprise(s) as we move into an era of individualized and complex multi-modality cancer treatments?


Supplemental Slides

Repair of non-DSB clusters by Base Excision Repair (BER)• Delayed formation of “enzymatic” DSB after irradiation• Mutagenic potential of complex SSB and complex base damage clusters

Additional Details and Examples of the Use of the LQ model with the dose protraction factor G to correct for dose and dose rate effects

RBE Models relevant to Particle Therapy


Base Excision Repair (BER) of Individual and non-DSB Clusters

Four Major Steps• (1) Damage recognition and removal, (2) Strand incision and end trimming, (3) Nucleotide insertion

using opposed strand as a template for base selection (A, T, G or C), (4) Ligation (“seal the nick”)

Short Patch BER• Removes and replaces individual DNA lesions (strand breaks, abasic sites, base damage)• Active in dividing and non-dividing cells• Very accurate (error frequency < 10-5 to 10-6 per lesion)

Long Patch BER• Removes and replaces > 2 or more nucleotides/DNA lesions (strand breaks, abasic sites, base damage)• Mostly active in dividing cells• More error prone, especially if opposed (template) DNA strand has one or more lesions


Long patch BER of a cluster (> 2 lesions)

DNA polymerase δ/ε remove and resynthesize nucleotides in the 5′ to 3′ direction. Opposing strand used as template to insert new nucleotide

A-T or G-C Error rate ~ 10-6

If patch extends past other lesions (5′→3′ direction), a single round (“cycle”) of BER may remove multiple lesions

Repair is complete when a DNA ligase seals the nick.

base pair rules


Base Substitution (point mutation)

Initiate BER of lesion with equal probability (?)

Error rate 10-6

(undamaged template)

5′ 3′

5′3′

Error rate ¾ (?) (damaged template)

Randomly insert A, G, T or C=

5′ 3′

5′3′

base substitution(point mutation)


Conversion of SSB into DSB

Equal chance of selecting either lesion

If base damage selected…

Glycosylase generates an abasic site

AP endonuclease generates a DSB“enzymatic DSB”


Enzymatic DSB formed by Ionizing Radiation (60Co γ-rays)

Incubation time (minute)10-1 100 101 102 103

Unr

ejoi

ned

DSB

(%)

0

10

20

30

40

50

60

70

80

90

100

110

120

SP or LP BER (with enzymatic DSB)LP BER (no enzymatic DSB)SP BER (no enzymatic DSB)Gulston et al. (2004)

Gulston M, de Lara C, Jenner T, Davis E, O'Neill P. Processing of clustered DNA damage generates additional double-strand breaks in mammalian cells post-irradiation. Nucleic Acids Res. 32(4):1602-9 (2004).

Filled symbols – delayed (post-irradiation) DSB (“enzymatic DSB”) formed in mutant hamster cells (xrs-5) deficient in NHEJ DSB repair.

Semenenko VA, Stewart RD. Monte carlo simulation of base and nucleotide excision repair of clustered DNA damage sites. II. Comparisons of model predictions to measured data. Radiat Res. 164(2):194-201 (2005).

Monte Carlo BER model used to determine numbers of non-DSB clusters converted to enzymatic DSB.


Absorbed dose (Gy)0 5 10 15 20

Surv

ivin

g fr

actio

n

10-3

10-2

10-1

100

76.8 Gy h-1

0.6 Gy h-1

Absorbed dose (Gy)0 10 20 30

Surv

ivin

g fr

actio

n

10-5

10-4

10-3

10-2

10-1

100

81 Gy h-1

0.24 Gy h-1

0.71 Gy h-1

Additional Examples of LQ Fits (with G)

RT112 human bladder carcinoma

AG1522 normal human fibroblast

Measured data from Ruiz de Almodovar JM, Bush C, Peacock JH, Steel GG, Whitaker SJ, McMillan TJ. Radiat. Res. 138, S93-S96 (1994)

Absorbed dose (Gy)0 10 20 30 40 50 60 70 80

Surv

ivin

g fr

actio

n

10-5

10-4

10-3

10-2

10-1

100

55 Gy h-1

0.06 Gy h-1

C3H 10T1/2

Measured data from Wells R.L. and Bedford J.S. Radiat. Res.94(1), 105-134 (1983).

Measured data from Hall EJ, Marchese MJ, Astor MB, and Morse T. Int. J. Radiat. Onc. Biol. Phys. 12, 655-659 (1986).


ln 2λτ

= { }2

2 ( ) ( )exp ( )t

G dt D t dt D t t tD

λ∞

−∞ −∞

′ ′ ′= − −

Conceptual Basis for G (protraction factor)?

Instantaneous absorbed dose rate (e.g., Gy/h) at time t

Absorbed dose (Gy)Probability per unit time sub-lethal damage (= DSB) is rejoined

Half-time for sub-lethal damage (i.e., DSB) repair

( ){ }2( ) expS D D GDα β= − +

Can determine G for any temporal pattern of radiation delivery from the above integral.

Sachs RK, Hahnfeld P, Brenner DJ. The link between low-LET dose-response relations and the underlying kinetics of damage production/repair/misrepair. Int J Radiat Biol. 1997 Oct;72(4):351-74.


Irradiation time (h)10-3 10-2 10-1 100 101 102 103 104

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

τ = 10 h

τ = 0.1 h

1 fraction

g

fraction delivery time

( ){ }( ){ }

2( ) exp

exp , where /

S D D GD

D dD d D n

α β

α β

= − +

≅ − + =

Protraction Factor – n daily fractions

g is always between 0 (large delivery time compared toτ ) and 1 (short delivery times compared toτ )

Dose d (fraction size) delivered during time interval Δt (fraction delivery time)

22( 1) /ln 2 /

xg e x xx t tλ τ

−= + −≡ Δ = Δ

1gGn n

= ≅

For a series of n daily fractions

(assumes repair complete between fractions)

if Δt << τ

24 hours


Trends in RBEDSB with proton LET (same overall trends other ions)

DSB are only major category of initial DNA damage that increases with increasing particle LET

LET (keV/μm)0 5 10 15 20 25 30 35 40 45 50 55 60 65

RB

E

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

2.2

2.4

2.6

2.8

DSB

SSB

Base Damage

Lines: Monte Carlo Damage Simulation* (MCDS)

Filled Yellow Symbols: Track Structure Simulation (Nikjoo et al. 1997, 2001, 2002)

Filled Red Symbols: Track Structure Simulation (Friedland et al. 2003)

* Semenenko and Stewart 2004, 2006, Stewart et al. 2011


A Mechanism-Inspired RBE Model

Combines a kinetic repair-misrepair model very similar to the earlier RMR (Tobias 1985), LPL (Curtis 1986), and MK (Hawkins 1996, 1998, 2003) models with an independently tested Monte Carlo Damage Simulation (MCDS) for initial DSB induction (Stewart et al. 2004, 2006, 2011, 2015).


Connection between RBEDSB and the RBE for cell survival?

In the MCDS+RMF system of models, the RBE for cell survival in the limit of low dose (RBELD) and high dose (RBEHD) is related to RBEDSB by

Intra-track DSB-DSB interactionsUnrepaired and

Mis-repaired DSB

HD DSBRBE RBEγ

ββ

≡ =

Adjust αγ and (α/β)γ to fit cell survival data for reference radiation. Then, use MCDS to compute RBEDSB as a function of particle type and energy.

1( / )

2 F DSBLD DSB

z RBERBE RBEγ γ

αα α β

≡ = +

Inter-track DSB-DSB interactions


Dashed lines: MCDS+RMF model prediction for higher LET radiations –based on MCDS estimate of RBEDSB for 4He2+ ions and

Human Kidney T1 Cells (aerobic)

Measured data from Barendsen circa 1960-1966

Absorbed Dose (Gy)0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Surv

ivin

g Fr

actio

n

10-4

10-3

10-2

10-1

100 200/250 kVp x-rays (1.3-1.5 keV/μm)

Absorbed Dose (Gy)0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Surv

ivin

g Fr

actio

n

10-4

10-3

10-2

10-1

100 200/250 kVp x-rays (1.3-1.5 keV/μm)28.6 MeV 4He2+ (24.9 keV/μm)25 MeV 4He2+ (26.3 keV/μm)

Absorbed Dose (Gy)0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Surv

ivin

g Fr

actio

n

10-4

10-3

10-2

10-1

100 200/250 kVp x-rays (1.3-1.5 keV/μm)28.6 MeV 4He2+ (24.9 keV/μm)25 MeV 4He2+ (26.3 keV/μm)5.2 MeV 4He2+ (89 keV/μm)

Solid line: LQ fit to survival data for x-rays [αγ = 0.266 Gy-1, (α/β)γ = 10.1 Gy].

( 2/ )( / )

( / ) 1 F DSB

DSB

z RBERBE

γ

γ

α βα β

α β

= ⋅ +

21( / )

F DSBDSB

z RBERBEγ

γαα β

α

= ⋅ +

2 , 5 FLETz d m

dμ

ρ≅ =

Absorbed Dose (Gy)0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Sur

vivi

ng F

ract

ion

10-4

10-3

10-2

10-1

100 14.9 2H+ (5.7 keV/μm)8.3 MeV 4He2+ (62.7 keV/μm)4.0 MeV 4He2+ (109 keV/μm)3.1 MeV 4He2+ (131 keV/μm)


Human Kidney T1 Cells (anoxic)

Absorbed Dose (Gy)0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Sur

vivi

ng F

ract

ion

10-3

10-2

10-1

100

250 kVp x-rays (1.3 keV/μm)25 MeV 4He2+ (26.3 keV/μm)4.0 MeV 4He2+ (109 keV/μm)2.5 MeV 4He2+ (148 keV/μm)

Absorbed Dose (Gy)0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Sur

vivi

ng F

ract

ion

10-3

10-2

10-1

10014.9 MeV 2H+ (5.7 keV/μm)8.3 MeV 4He2+ (62.7 keV/μm)5.1 MeV 4He2+ (90.3 keV/μm)

RMF model predictions (dashed lines) based on reference radiation αγ and (α/β)γ values for aerobic conditionsand MCDS estimates of RBEDSB for anoxic conditions (i.e., no fitting to measured data in these figures)


Power of a Mechanism-Inspired Biophysical Model

LQ analysis (15 × 2 = 30 adjustable parameters)

MCDS+RMF analysis (2 adjustable parameters, RBEDSB from first principle MCDS simulations)

Models and measurements in about equally good agreement given the number of adjustable parameters!

Absorbed Dose (Gy)0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Surv

ivin

g Fr

actio

n

10-4

10-3

10-2

10-1

100 200/250 kVp x-rays (1.3-1.5 keV/μm)28.6 MeV 4He2+ (24.9 keV/μm)25 MeV 4He2+ (26.3 keV/μm)5.2 MeV 4He2+ (89 keV/μm)

Documents

Radiobiological Foundations of Stereotactic Body Radiation ......2016/10/14 · • Stereotactic Body Radiation Therapy (SBRT) and Stereotactic Ablative Radiotherapy (SABR) T