Radiobiology Behind Dose Fractionation
Radiobiology Behind Dose Fractionation
Radiobiology Behind Dose Fractionation
Bill McBride Dept. Radiation Oncology David Geffen School Medicine UCLA, Los Angeles, Ca. wmcbride@mednet.ucla.edu
WMcB2009
www.radbiol.ucla.edu
Objectives
To understand the mathematical bases behind survival curves Know the linear quadratic model formulation Understand how the isoeffect curves for fractionated radiation vary with tissue and how to use the LQ model to change dose with dose per fraction Understand the 4Rs of radiobiology as they relate to clinical fractionated regimens and the sources of heterogeneity that impact the concept of equal effect per fraction Know the major clinical trials on altered fractionation and their outcome Recognize the importance of dose heterogeneity in modern treatment planning
WMcB2009
www.radbiol.ucla.edu
The number of non-homogeneous treatment plans (IMRT) and extreme hypofractionated treatments are increasing. Do existing models cope?
WMcB2009
www.radbiol.ucla.edu
In theory, knowing relevant radiobiological parameters one day may predict the response for
Dose given in a single or a small number of fractions
SBRT, SRS, SRT, HDR or LDR brachytherapy, protons, cyberknife, gammaknife
Combination therapies with chemo- or biological agents Different RT options when tailored by molecular and imaging theragnostics
If you know the molecular profile and tumor phenotype, can you predict the best delivery method?
WMcB2009
www.radbiol.ucla.edu
www.radbiol.ucla.edu
In general, history has shown repeatedly that single high doses of radiation do not allow a therapeutic differential between tumor and critical normal tissues. Dose fractionation does.
SBRT/SRS often aims at TISSUE ABLATION
WMcB2009
www.radbiol.ucla.edu
WMcB2009
www.radbiol.ucla.edu
P survival (when x = 0) 100 targets 100 hits m=1 e-1=0.368 100 targets 200 hits m=2 e-2=0.137 100 targets 300 hits m=3 e-3=0.05
N.B. Lethal hits in DNA are not really randomly distributed, e.g. condensed chromatin is more sensitive, but it is a reasonable approximation
WMcB2009
www.radbiol.ucla.edu
This Gives a Survival Curve Based on a Model where one hit will eliminate a single target
1.0
0.37
S.F.
0.1
S.F.= e-1 = 0.37 This is the mean lethal dose D0 D10 = 2.3 xD0 In general, S.F. = e-D/D0 or LnS.F. = -D/D0 or S.F. = e-aD , i.e. D0 = 1/a
Where a is the slope of the curve and D0 the reciprocal of the slope
0.01
0.001
D0 D10 DOSE Gy
How many logs of cells would be killed by 23 Gy if D0 = 1 Gy?
WMcB2009
www.radbiol.ucla.edu
WMcB2009
www.radbiol.ucla.edu
Puck and Marcus, J.E.M.103, 563, 1956 First in vitro mammalian survival curve
0.1
0.01
0.001
dose
WMcB2009
www.radbiol.ucla.edu
S.F.
0.1
S.F.=e-D/1D0[1-(1-e-D/nD0)n]
Single hit Extrapolation Number Accumulated damage
0.01
nD 0
0.001
WMcB2009
www.radbiol.ucla.edu
S.F.
Multi-fraction survival curves can be considered linear if sublethal damage is repaired between fractions
they have an extrapolation number (n) = 1.0
5 fractions .1 3 fractions
S.F. = e-D/eD0
If S.F. after 2Gy = 0.5, eD0 = 2.9Gy; eD10 = 6.7Gy and 30 fractions of 2 Gy (60Gy) would reduce survival by (0.5)30 = almost 9 logs (or 60/6.7) If a 1cm tumor had 109 clonogenic cells, there would be an average of 1 clonogen per tumor and cure rate would be about 37%
Single dose
.01
0 4 8 12 16 Dose (Gy) 20 24
WMcB2009
www.radbiol.ucla.edu
1.0
aD bD2
Cell kill is the result of single lethal hits plus accumulated damage from 2 independent sublethal events S.F. = e-aD Single lethal hits
S.F. 0.1
0.01
0.001
a/b in Gy
DOSE Gy
a/b is dose at which death due to single lethal lesions = death due to accumulation of sublethal lesions i.e. aD = bD2 and D = a/b in Gy
WMcB2009
www.radbiol.ucla.edu
Over 90% of radiation oncologists use the LQ model: it is simple and has a microdosimetric underpinning a/b is large (> 6 Gy) when survival curve is almost exponential and small (1-4 Gy) when shoulder is wide the a/b value quantifies the sensitivity of a tissue/tumor to fractionated radiation. But: Both a and b vary with the cell cycle. At high doses, S phase and hypoxic cells become more important. The a/b ratio varies depending upon whether a cell is quiescent or proliferative The LQ model best describes data in the range of 1 6Gy and should not be used outside this range
WMcB2009
www.radbiol.ucla.edu
The slope of an isoeffect curve changes with size of dose per fraction depending on tissue type Acute responding tissues have flatter curves than do late responding tissues a/b measures the sensitivity of tumor or tissue to fractionation i.e. it predicts how total
dose for a given effect will change when you change the size of dose fraction
Douglas and Fowler Rad Res 66:401, 1976 Showed and easy way to arrive at an a/b ratio
Slope = b
WMcB2009
www.radbiol.ucla.edu
S.F.
S.F.
.1
.1
.01
a/b is high (>6Gy) when survival curve is almost exponential and low (1-4Gy) when shoulder is wide
8 12 Dose (Gy)
16
12 16 Dose (Gy)
20
www.radbiol.ucla.edu
WMcB2009
www.radbiol.ucla.edu
Breast Cancer a/b = 4.0Gy (1.0-7.8) Breast appearance a/b = 3.6Gy; induration a/b = 3.1Gy
If fractionation sensitivity of a cancer is similar to dose-limiting healthy tissues, it may be possible to give fewer, larger fractions without compromising effectiveness or safety WMcB2009
www.radbiol.ucla.edu
So, for late responding tissue, what total dose in 1.5Gy fractions is equivalent to 66Gy in 2Gy fractions? Dnew (1.5+2) = 66 (2 + 2) Dnew = 75.4Gy
NB: Small differences in a/b for late responding tissues can make a big difference in estimated D!
WMcB2009
www.radbiol.ucla.edu
E = nd(a + bd)
E/a = nd(1+d/a/b)
Total dose
Relative Effectiveness
Normalized total dose2Gy = BED/RE = BED/1.2 for a/b of 10Gy = BED/1.67 for a/b of 3Gy
WMcB2009
35 x 2Gy = B.E.D.of 84Gy10 and 117Gy3 NOTE: 3 x 15Gy = B.E.D.of 113Gy10 and 270Gy3
www.radbiol.ucla.edu
80 70
2.0Gy/fx
a/b=30Gy; 4Gy/fx
60
D new
50 40 30 20
a/b=3Gy; 4Gy/fx
20
30
40
50
60
70
80
D old
WMcB2009
www.radbiol.ucla.edu
Prescribed Dose: 25 fractions of 2Gy = 50Gy Hot spot: 110% Physical dose: 55Gy Biological dose: 60.5Gy
Double Trouble
WMcB2009
www.radbiol.ucla.edu
WMcB2009
www.radbiol.ucla.edu
HT29 cells
N.B. Survival curves may deviate from L.Q. at low and high dose!!!!
Certain cell lines, and tissues, are hypersensitive at low doses of 0.050.2Gy. The survival curve then plateaus over 0.05-1Gy Not seen for all cell lines or tissues, but has been reported in skin, kidney and lung At high dose, the model probably does not fit data well because D2 dominates the equation
WMcB2009
www.radbiol.ucla.edu
WMcB2009
www.radbiol.ucla.edu
Repopulation
Redistribution Repair
WMcB2009
www.radbiol.ucla.edu
Repopulation
spares acute responding normal tissue, no effect on late effects, danger of tumor repopulation
Reoxygenation
increases tumor damage, no effect in normal tissues
WMcB2009
www.radbiol.ucla.edu
Repair
Repair between fractions should be complete - N.B. we are dealing with tissue recovery rather than DNA repair Correction for incomplete repair is possible (Thames) In general, time between fractions for most tissues should be >6 hours Some tissues, such as CNS, recover slowly making b.i.d. treatment inadvisable Bentzen - Radiother Oncol 53, 219, 1999 CHART analysis HNC showed that late morbidity was less than would be expected assuming complete recovery between fractions Is the T1/2 for recovery for late responding normal tissues 2.54.5hrs?
WMcB2009
www.radbiol.ucla.edu
WMcB2009
www.radbiol.ucla.edu
local control
no local control
4 weeks to start of accelerated repopulation. Thereafter T1/2 of 4 days = loss of 0.6Gy per day
Withers, H.R., Taylor, J.M.G., and Maciejewski, B. Acta Oncologica 27:131, 1988
WMcB2009
oxic Dose
Physical Dose Need to know more about the importance of dose-volume constraints
WMcB2009
www.radbiol.ucla.edu
Heterogeneity within and between between tumors in dose-response characteristics, often resulting in large error bars for a/b values In spite of this, the outcome of clinical studies of altered fractionation generally fit the models, within the constraints of the clinical doses used
WMcB2009
www.radbiol.ucla.edu
Altered Fractionation
or
How to optimally distribute dose over time
WMcB2009
www.radbiol.ucla.edu
Players
Total dose (D) Dose per fraction (d) Interval between fractions (t) Overall treatment time (T) Tumor type Acute reacting normal tissues Late reacting normal tissues
WMcB2009
www.radbiol.ucla.edu
Definitions
Conventional fractionation
Daily doses (d) of 1.8 to 2 Gy Dose per week of 9 to 10 Gy Total dose (D) of 40 to 70 Gy
Hyperfractionation
The number of fractions (N) is increased T is kept the same Dose per fraction (d) less than 1.8 Gy Two fractions per day (t)
WMcB2009
www.radbiol.ucla.edu
Definitions
Accelerated fractionation
Shorter overall treatment time Dose per fraction of 1.8 to 2 Gy More than 10 Gy per week Rationale: Overcome accelerated tumor repopulation
Hypofractionation
Dose per fraction (d) higher than 2.2 Gy Reduced total number of fractions (N) Rationale: Tumor has low a/b ratio and there is no therapeutic advantage to be gained with respect to late complications
WMcB2009
www.radbiol.ucla.edu
TCP or NTC
Complication-free cure
TCP or NTC
Accelerated Fractionation
Hyperfractionation
Dose
www.radbiol.ucla.edu
WMcB2009
Conventional
70 Gy - 35 fx - 7 wks
Hyperfractionated
81.6 Gy - 68 fx - 7 wks
Moderately accelerated
72 Gy - 42 fx - 6 wks
WMcB2009
www.radbiol.ucla.edu
Hyperfractionated
Barcelona (586), Brazil (112), RTOG 90-03 (1113), EORTC 22791 (356), Toronto (331)
Very accelerated
CHART (918), Vancouver (82), TROG 91-01 (350),GORTEC 94-02 (268)
Moderately accelerated
RTOG 90-03 (1113), DAHANCA (1485), EORTC 22851 (512) CAIR (100), Warsaw (395)
Other
EORTC 22811 (348), RTOG 79-13 (210)
www.radbiol.ucla.edu
Years
Years
WMcB2009
www.radbiol.ucla.edu
54 Gy - 36 fx - 12 days
Loco-regional control
Survival
conventional CHART
conventional CHART
WMcB2009
www.radbiol.ucla.edu
CHART: Morbidity
Dische 1997
54 Gy - 36 fx - 12 days
P = 0.04
Moderate/severe subcutaneous fibrosis and oedema Mucosal ulceration and deep necrosis
P = 0.04
P = 0.009
Laryngeal oedema
Moderate/severe dysphagia
WMcB2009
www.radbiol.ucla.edu
Moderately Accelerated
DAHANCA 6: only glottic, (N = 694) DAHANCA 7: all other sites, + nimorazole (N = 791)
Overgaard 2000
66-68 Gy - 33-34 fx - 6 wks Actuarial 5-year rates Local control DAHANCA 6 DAHANCA 7 Nodal control DAHANCA 6 + 7 Disease-specific survival DAHANCA 6 + 7 Overall survival Late effects (edema, fibrosis)
73% 56%
87% 65% n.s. n.s. .
WMcB2009
www.radbiol.ucla.edu
Moderately Accelerated
CAIR: 7-day-continuous accelerated irradiation (N = 100)
Skladowski 2000
CAIR
Probability
CONTROL
log-rank p=0.00001
Follow-up (months)
WMcB2009
www.radbiol.ucla.edu
RTOG 90-03, Phase III comparison of fractionation schedules in Stage III and IV SCC of oral cavity, oropharynx, larynx, hypopharynx (N = 1113)
Fu 2000
Conventional
70 Gy - 35 fx - 7 wks
Hyperfractionated
81.6 Gy - 68 fx - 7 wks
www.radbiol.ucla.edu
WMcB2009
www.radbiol.ucla.edu
WMcB2009
www.radbiol.ucla.edu
Fu 2000
Maximum toxicity Conventional Hyperfract Concom Acc + per patient boost split Grade 1 15% 4% 4% 7% Grade 2 57% 39% 36% 41% Grade 3 35% 54% 58% 49% Grade 4 0% 1% 1% 2%
Late
Maximum toxicity per patient Grade 1 Grade 2 Grade 3 Grade 4 Grade 5 Conventional 11% 50% 19% 8% 1% Hyperfract 8% 56% 19% 9% 0% Concom Acc + boost split 7% 16% 44% 50% 29% 20% 8% 7% 1% 1%
WMcB2009
www.radbiol.ucla.edu
Toxicity of RT in HNSCC
Acute effects in accelerated or hyperfractionated RT Author Regimen Grade 3-4 mucositis Cont Exp 49% 67% 50% 67% 43% 73% 25% 46% 25% 41% 25% 42% 56%
WMcB2009
Horiot (n=356) HF Horiot (n=512) Acc fx + split Dische (n=918) CHART Fu (n=536) Acc fx(CB) Fu (n=542) Acc fx + split Fu (n=507) HF Skladowski (n=99) Acc fx 26%
www.radbiol.ucla.edu
Survival benefit: 3.4% (36% 39% at 5 years, p = 0.003) Loco-regional control benefit: 7% (46.5% 53% at 5 years, p < 0.0001)
WMcB2009
www.radbiol.ucla.edu
WMcB2009
www.radbiol.ucla.edu
www.radbiol.ucla.edu
www.radbiol.ucla.edu
Questions:
The Radiobiology Behind Dose Fractionation
WMcB2009
www.radbiol.ucla.edu
Modeling of radiation responses are based on 1. Random events occurring in cell nuclei 2. Random events in space as defined by the Poisson distribution 3. A Gaussian distribution 4. Logarithmic dose response curves
WMcB2009
www.radbiol.ucla.edu
D0 is 1. Is a measure of the shoulder of a survival curve 2. Is the mean lethal dose of the linear portion of the dose-response curve 3. Represents the slope of the log linear survival curve 4. Is constant at all levels of radiation effect
WMcB2009
www.radbiol.ucla.edu
Dq is 1. A measure of the inverse of the terminal slope of the survival curve 2. A measure of the inverse of the initial slope of the survival curve 3. A measure of the shoulder of the survival curve 4. A measure of the intercept of the terminal portion of the survival curve on the y axis
WMcB2009
www.radbiol.ucla.edu
If Dq for a survival curve is 2Gy, what dose is equivalent to a single dose of 6Gy given in 2 fractions, assuming complete repair and no repopulation between fractions. 1. 4 Gy 2. 6 Gy 3. 8 Gy 4. 10 Gy
WMcB2009
www.radbiol.ucla.edu
A whole body dose of 7 Gy of x-rays would produce severe, potentially lethal hematologic toxicity. Assuming that the Do of the hematopoietic stem cells is 1 Gy and that these cells have a negligible capacity to repair sublethal radiation damage, what is the surviving fraction of these stem cells after this dose of radiation? 1. 0.0001 2. 0.001 3. 0.025 4. 0.067 5. 0.1167
WMcB2009
www.radbiol.ucla.edu
If 90% of a tumor is removed by surgery, what does this likely represent in term of radiation dose given in 2 Gy fractions? 1. 1-2 Gy 2. 3-4 Gy 3. 6-7 Gy 4. 9-12 Gy 5. 20-30 Gy
WMcB2009
www.radbiol.ucla.edu
What is true for the a/b ratio 1. It is unitless 2. It is a measure of the shoulder of the survival curve 3. It measures the sensitivity of a tissue to changes in size of dose fractions 4. It is the ratio where the number of nonrepairable lesions equals that for repairable lesions
WMcB2009
www.radbiol.ucla.edu
The alpha component in the linear quadratic formula for as radiation survival curve represents 1. Unrepairable DNA double strand breaks 2. Lethal single track events 3. Multiply damaged sites in DNA 4. Damage that can not be altered by hypoxia
WMcB2009
www.radbiol.ucla.edu
Which parameter is most relevant for standard clinical regimens in RT 1. The a/b ratio 2. Do 3. Alpha 4. Beta 5. The extrapolation number
WMcB2009
www.radbiol.ucla.edu
If cells have a Do of 2 Gy, assuming no shoulder, what dose is required to kill 95% of the cells? 1. 6 Gy 2. 12 Gy 3. 18 Gy 4. 24 Gy 5. 30 Gy
WMcB2009
www.radbiol.ucla.edu
The extrapolation number N for a multifraction survival curve, allowing complete repair between fractions and no repopulation is 1. 1 2. < 1 3. >1 4. Dependent on the size of the dose per fraction
WMcB2009
www.radbiol.ucla.edu
The extrapolation number N for a single dose neutron survival curve is 1. 1 2. < 1 3. >1 4. Dependent on the size of the dose per fraction
WMcB2009
www.radbiol.ucla.edu
The extrapolation number N for a low dose rate survival curve is 1. 1 2. < 1 3. >1 4. Dependent on the size of the dose per fraction
WMcB2009
www.radbiol.ucla.edu
The inverse of the slope of a multifraction survival curve (effDo) is generally within the range 1. 1.0-1.5 Gy 2. 1.5-2.5 Gy 3. 2.5-5.0 Gy 4. 5.0-10.0 Gy
WMcB2009
www.radbiol.ucla.edu
If the effDo for a multifraction survival curve is 3.5 Gy, what dose would cure 37% of a series of 1cm diameter tumors (109 clonogens). 1. 56 Gy 2. 64 Gy 3. 72 Gy 4. 80 Gy
WMcB2009
www.radbiol.ucla.edu
If the effDo for a multifraction survival curve is 3.5 Gy, what dose would cure 69% of a series of 1cm diameter tumors (109 clonogens). 1. 56 Gy 2. 64 Gy 3. 72 Gy 4. 80 Gy
WMcB2009
www.radbiol.ucla.edu
If a tumor has an effective Do of 3.5 Gy,what is the S.F. after 70 Gy? 1. 2 x 10-11 2. 2 x 10-9 3. 2 x 10-7 4. 2 x 10-5 5. 2 x 10-3
WMcB2009
www.radbiol.ucla.edu
If 16 x 2 Gy fractions reduce survival by 10-4, what dose would be needed to reduce survival to 10-10? 1. 50 Gy 2. 60 Gy 3. 64 Gy 4. 70 Gy 5. 80 Gy
WMcB2009
www.radbiol.ucla.edu
If 16 x 2 Gy fractions reduce survival by 10-4, what is the effective D0? 1. 2.0 Gy 2. 2.3 Gy 3. 3.0 Gy 4. 3.5 Gy 5. 3.8 Gy
WMcB2009
www.radbiol.ucla.edu
WMcB2009
www.radbiol.ucla.edu
Which of the following human tumors Is thought to have an a/b ratio of 1-2 Gy 1. Oropharyngeal Ca 2. Prostate Ca 3. Glioblastoma 4. Colorectal Ca
WMcB2009
www.radbiol.ucla.edu
The TD5/5 for a certain tissue irradiated at 2 Gy/fraction is 60 Gy whereas at 4 Gy/fraction it is 40 Gy. Assuming that the linear quadratic equation, -lnSF= N (aD + bD2), accurately represents cell survival for this tissue, what is the value of a/b? 1. 1 Gy 2. 2 Gy 3. 4 Gy 4. 10 Gy 5. 20 Gy
WMcB2009
www.radbiol.ucla.edu
It is decided to treat a patient with hypofractionation at 3 Gy/fraction instead of the conventional schedule of 60 Gy in 2 Gy fractions. What total dose should be delivered in order for the risk of late normal-tissue damage to remain unchanged according to the linear-quadratic model with a/b for late damage = 3 Gy? 1. 40 Gy 2. 48 Gy 3. 50 Gy 4. 55.4 Gy 5. 75 Gy
WMcB2009
www.radbiol.ucla.edu
A standard treatment for HNSCC tumors is 70 Gy delivered at 2 Gy/fraction. Hyperfractionation is being attempted with a fraction size of 1.2 Gy. What total treatment dose should be used to maintain the same complication rate for the late responding normal tissues. Assume full repair of sublethal damage between fractions and an a/b of 3 Gy. 1. 42 Gy 2. 58 Gy 3. 70 Gy 4. 83 Gy 5. 117 Gy
WMcB2009
www.radbiol.ucla.edu
A standard treatment for HNSCC tumors is 70 Gy delivered at 2 Gy/fraction. Hyperfractionation is being attempted with a fraction size of 1.2 Gy. What total treatment dose should be used to maintain the same complication rate for the late responding normal tissues. Assuming no proliferation and complete repair between fractions, an a/b of 3 Gy for late responding tissue and 12 Gy for tumor, what would be the therapeutic gain. 1. 6% 2. 12% 3. 18% 4. 24%
WMcB2009
www.radbiol.ucla.edu
Which of the following sites is the least suitable for b.i.d. treatment 1. Head and neck 2. Brain 3. Lung 4. Prostate
WMcB2009
www.radbiol.ucla.edu
The rationale behind accelerated fractionation is 1. To spare late responding normal tissue 2. To combat encourage tumor reoxygenation 3. To exploit redistribution in tumors 4. To combat accelerated repopulation in tumors
WMcB2009
www.radbiol.ucla.edu
The CHART regimen for HNSCC of 54Gy in 36 fractions over 12 days compared with 66 Gy in 33 fractions in 6.5 weeks, overall showed 1. Superior locoregional control, no increase in overall survival, increased late effects 2. Superior locoregional control that translated into an increase in overall survival, no change in late effects 3. No change in locoregional control and overall survival, decreased late effects 4. Superior locoregional control, no increase in overall survival, increased acute effects
WMcB2009
www.radbiol.ucla.edu
DAHANCA 6 and 7 clinical trials with 6668Gy given in 6 compared to 7 weeks 1. Was a hyperfractionation trial 2. Treated 6 days a week 3. Showed no increase in local control 4. Showed no increase in disease-specific survival
WMcB2009
www.radbiol.ucla.edu
RTOG 90-03, which compared hyperfractionation, accelerated fractionation with a split, and accelerated fractionation with a boost showed 1. Hyperfractionation to be superior in terms of loco-regional control and late effects 2. Accelerated fractionation with a split to be equivalent to hyperfractionation in terms of locoregional control 3. There to be no advantage to altered fractionation 4. Accelerated fractionation to be superior to hyperfractionation
WMcB2009
www.radbiol.ucla.edu
Answers
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. NA 2 2 4 1 5 2 4 2 4 4 3 2 1 3 2 1 2 3 3 1 3 1 2 5 1 3 1 3 3
WMcB2009
www.radbiol.ucla.edu