La ricerca e la terapia in adroterapia-2 R.Orecchia / P. Fossati
Dose (Gy) = energy (joule) / mass (kg) One degree of fever (from 37.5 to 38.5 ) > 4000 Gy RT small dose big damage
Photons : Dose Resposne
A uniform dose of photons Makes comparisons easy Maybe is the optimal distribution (under given conditions)
If I deliver a uniform dose of photons RT I care about: Prescribed dose If I deliver a Non uniform dose of photons RT I care about: voxel by voxel dose I never care (at least as a first approximation) Number of fields ; IMRT vs. 3D-CRT ; 5 MeV vs. 18 MeV Target shape; Target depth; Flattening filter
Two quite separate problems 1 delivering a optimal dose distribution (prescribing) 2 understanding each other (reporting)
Which is the optimal dose distribution? Uniform? BTV and dose painting? Shrinking fields?
Optimal dose distribution High cure probability Low risk of toxicity
When is uniform dose the optimal choice? For the patient: when there is uniform carcinogenic cells distribution within the CTV and uniform distribution of factors determining dose response (e.g. hypoxia) For the population when there is total ignorance about clonogens an hypoxia distributions
SHRINKING FIELDS / DOSE PAINTING GIVE MORE DOSE WHERE IT IS NEEDED
BAD DOSE NON-UNIFORMITY TECHNICALLY DRIVEN: BRACHYTHERAPY CYBERKNIFE GAMMAKNIFE ISODOSE PRESCRIPTION
PROSTATE HDR
Cold spots in the target are bad Hot spots in hypoxic areas with high density of clonogens are good Hot spots at random may be acceptable if they do not involve critical structures
Which are acceptable hot spots? Microsocopic Brachytherapy with seed imoplantations or catheter based HDR Central Cyberknife Gammknife Gyncological brachytherapy
Inacceptable hot spot: 100 Gy to left prostate lobe 50 Gy to right prostate lobe No cure A lot of toxicity
High LET, High RBE
Carbonio Ions high LET?(only where you need it) high LET lowlet
But you have to calculate it
Carbon ion treatment is not high LET RT, not even in the target KeV/μm Dose averaged LET Dose fisica NIRS, physical dose and dose averaged LET for 1 field Courtesy of Dr. Matsufuji
Protons uniform physical dose Not different form photons
Carbon ions More biological effect Whichever endpoint you choose
What if we give a uniform physical dose of carbon ions
It would be like prescribing 57 Gy to the right prostate lobe and 99 to the left prostate lobe
Beam line at Lanzhou, Institute of Modern Physics China
> 100 patients!!!
Most people (including IMP colleagues) agree that a lower dose is needed in areas where there are mainly slow particles with high LET and a higer dose is needed where there are mainly fast particles with low LET
How much less? How much more?
Passive systems (NIRS, HYOGO, GUNMA and IMP) You only need a shape of SOBP Distal and proximal planes receive different radiation quality but there is no relevant change perpendicular to the beam axis
active systems (GSI, HIT, NIRS, CNAO) Weight must be assigned to single monocromatic beamlets, each beamlet contributes to dose in many voxels A radiobiolgical mathematical model must be embedded in the TPS
All system used in the clinics are based on a very simple concept Less dose where there is more LET
All models do not refer to a given endpoint (10% survival of HSG cell lines) but unfortunately to a GyE
The worst possible reference radiationis X-RAy Non linear Higly sensitive to almost anything
Short outline of Kanai Model LEM I model LEM IV model MKM model
KANAI Model
Human salivary gland cell lines Maybe not so critical a choice
Measure several survival curves in several position along a monochromatic bragg peak Describe the single monochromatic braag peak and the mixed field as Linear quadratic Use Zaider Rossi formula Manufacture a less spiky ridge filter
From this point the story becomes complicated Why the red line? Why do we care about 80 KeV/μm?
Multiple endpoints vs singkle endpoint? HSG surviival or pig skin reddening or clinical toxicity 80 KeV/μm was the LET of fast neutron used at NIRS
More than 1000 patients from 1975 to 1984 Melanoma, NSCLC, H&N SCC, Gynecological cancer
1. Neutron at NIRS were used with RBE 3 2. Carbon with the same LET must have the same RBE 3. Pig skin reddening agrees with neutrons and not with HSG survival 4. you do not get curves out of pig skin reddening or clinicla experience Solution: Scale linearly the biological dose of HSG multipling by 1.5
We have measured that for HSG it is equivalent to 1.84 Gy but we Believe taht for the aptient it may be equivalent to 2.7 Gy
Even more complicated 1. Dose escalation trials have been carried out at NIRS escalating dose per fraction 2. SOBP shape has not been changed and RBE has been assumed to scale linearly
It is not straightforward to compare Kanai Model with anything else The dose distribution produce a uniform effect (at least for HSG cell lines) only if 2.7 GyE per fraction are prescribed
All clinical results from Japan (NIRS, Hyogo and Gunma) are based on Kanai Model
LEM I (Local Effect Model)
LEM I (Local Effect Model) The difference depends on microscopic pattern of dose deposition: Photons are like spanking, carbon like stabbing with a dagger
LEM IV (Local Effect Model)
Photons survival curves are used Survival means zero lethal events Probability of lethal events for a cell is derived from survival curves with Poisson statistics Lethal events are assumed uniformly spaced in the nucleus for photons
For carbon ions number of lethal events is integrated over the nucleus and local probability is derived from photons global curves Local dose is calculated based on the amorphous track There are some free parameters
The most critical free parameter is Transition dose from linear quadratic to linear Local doses can exceed 1000 Gy It is not possible to assume LQ relation between dose and survival, survival curves are linearized at a given dose
LEM model You can apply it to any mixed field of particles OK for spot scanning OK for inverse planning Predicts cell survival for complex beam arrangements You can change the reference cell line easily
All clinical results from Europe (GSI, HIT and CNAO) are based on LEM I Model with an idealized chordoma cell line as reference
LEM IV (Local Effect Model)
LEM I focuses on dose and survival to calculate local probability of lethal events LEM IV introduces a new concept : simple DSB vs. complex (clustered) DSB Better agreement with in vitro and in vivo study Never used with patients
Microdosimetric Kinetic Model (MKM)
The basic idea is not so much different from LEM Expected number of lethal event in a cell is obtained by summation of expected number of lethal events in a small domains Instead of integrating points over a volume a finite number of small domains is added. This allows to directly measure relvant radiation parameters down in the microscopic domain
MKM Once again a lot of dose clustered in a small volume is predict to create more damage Once again there is linear quadratic dependence The model is less of a black box respect to LEM as many of tis parameters can be derived form microdismetric measurments
Good fit of in vitro data Nasty mathematics
It is / will be used for NIRS spot scanning
Different way to prescribe carbon ion RT Everyone agree qualitatively but there is quantitative disagreement No one is right as there are many relevant endpoints and all are difficult to measure
BUT We risk not to understand each other Kanai vs. LEM one is a clinically relevant conversion The shape wil be different but we want to avoid systematic errors
10% difference is clinically relevant High dose (70.4 GyE jp ) 5yrs. 76.4% H&N Sarcoma at NIRS Low dose (64.0 or 57.6 GyE jp ) 5yrs. 21.4% LogRank p<0.0001
Can we compare physical dose? NO
Same model: comapre biological doses 2 fields big target 1 field big target 1 field, small target
Same DVH of RBE weighted dose Dosse GyE LEM
Different physical dose DVH
No easy way to compare carbon ion RT plans obtained with different models You cannot compare physical dose (not even with the same RBE model) You cannot compare RBE weigted dose
Possible solution: Compare physical dose after fixing reference conditions : 1. Same volumes 2. Same number of fields 3. Same field orientation
Water phantom 5 Cubic targets : (4, 6, 8, 10, 12 cm)
Same volumes in superficial position
Fields 1 single field 2 opposed fields 2 orthogonal fields
NIRS physical dose (4 GyE jp ) 6 cm SOBP
NIRS physical dose in 6 cm SOBP (4 GyE jp )
NIRS physical dose in 6 cm SOBP (4 GyE jp )
NIRS 4 GYE jp CNAO 4.05 GYE LEM
NIRS 4 GYE jp CNAO 4.05GYE LEM CNAO 4.2 GYE LEM
NIRS 4 GYE jp CNAO 4.3 GYE LEM
NIRS 4 GYE jp CNAO 4.4 GYE LEM
NIRS 4 GYE jp CNAO 4.5 GYE LEM
Physical dose in SOBP CNAO 4.5 GYE LEM CNAO 4.45 GYE LEM CNAO 4.05 GYE LEM NIRS 4 GYE jp
CNAO 4.45 GYE LEM Cost = Red area 2 Abs (red area) NIRS 4 GYE jp
Same size different prescriptrion doses 6 cm SOBP Cost (physical Gy) CNAO GyE LEM
Optimal dose is size dependant! NIRS 3.6 GyE JP Cost (physical Gy) CNAO GyE LEM
Sizes are weigthed according to real tumor size Number of pts. Number of pts. mm mm
Optimal dose depends on field arrangement 0,14 0,12 NIRS 3.6 GyE jp 0,1 Cost (physical Gy) 0,08 0,06 0,04 0,02 single field opposed fields 90 2 fields 0 4,1 4,15 4,2 4,25 4,3 4,35 4,4 CNAO GyE LEM
Cubes spheres or patients?
Tails of wasted SOBP were LEM is underdosing
Water phantom 5 spheric targets : (4, 6, 8, 10, 12 cm)
NIRS 3.6 GyE jp Cost (physical Gy) CNAO GyE LEM If we treat spheres we should give slightly lower doses
Final results
Head and neck NIRS 4 GyE x 16 frazioni = 64 GyE CNAO 4.45 GyE x 16 frazioni = 71,2 GyE Retroperitoneal sarcoma NIRS 4.4 GyE x 16 frazioni = 70,4 GyE CNAO 4.75 GyE x 16 frazioni = 76 GyE
Risultati finali
(2) Recalculating NIRS biological dose according to LEM We can recreate the same physical dose shape with CNAO spot scanning and then calculate the resulting biological dose, this would be the biological dose of NIRS plan calculated with LEM. Not possible with Syngo PT but with other non commercial softwares (MyLEM)
NIRS physical dose physical Gy mm
Monocromatic peaks (MyLEM) physical Gy mm
Fitted peaks to reproduce NIRS dose physical Gy mm
Recalculated biological dose of NIRS-fitted dose distribution according to LEM physical Gy GyE LEM mm
Perfect agreement between the two approaches? indication NIRS dose 1 field Min err quad Recalc biol dose Min err ab H&N 3,6 4,15 4,187276 4,2 H&N 4 4, 45 4,47313 4,5 ch ch skull base 3,8 4,35 4,331724 4,4 ch ch spine 4,2 4,6 4,640563 4,65 sarcoma body 4,4 4,75 4,778061 4,8 sarcoma H&N 4,4 4,7 4,747709 4,75
Indipendent calculation similar results
CONCLUSION Treatment plans for carbon ions should include enough information on local physical quantitties to enable anybody to recalculate RBE weighted dose according to their own radiobilogical model