MRI Based treatment planning for with focus on prostate cancer Xinglei Shen, MD Department of Radiation Oncology KUMC
Overview How magnetic resonance imaging works (very simple version) Indications for MRI in diagnosis and treatment planning MRI improves upon CT in prostate cancer Technique for incorporating MRI into prostate cancer treatment planning (If time allows) MRI in cervical brachytherapy
How MRI works (the very dummy version) Tissues are ultimately made of atoms (Hydrogen is by far most common) Because they spin, hydrogen atoms have tiny currents and thus tiny magnetic fields Putting atoms in a huge magnetic field (1 Tesla = 20,000 times earth s magnetic field) results in alignment of the atoms to the magnetic field and the nuclei in atoms to precess (spin) This generates a net magnetic field along direction of applied magnetic field (i.e. a human magnet)
How MRI works (the very dummy version) Hydrogen atoms precess at a specific frequency. Adding energy at the specific frequency to align hydrogen atoms Effect on hydrogen can be measured based on radiofrequency signal emitted by these atoms as they relax Relaxation characteristics after a energy pulse in different directions (longitudinal or perpendicular to main magnetic field) is affected by tissue composition T1 = time for longitudinal relaxation T2 = time for perpendicular relaxation
MRI Sequences Different tissue compositions have characteristic appearances on MRI sequences T1 or T2 weighted sequences show different properties of the tissue T1 (in line relaxation) related to energy transfer between molecules Free hydrogens in water and hydrogens bound to large molecules do not exchange energy well Bound water (i.e. attached to proteins) and hydrogens in fat exchange energy well T2 (transverse relaxation) related to interactions from nearby small magnetic fields from tissue More molecules in tissue = more interaction Free water has interactions than organized tissues
CT versus MRI
CT versus MRI
CT versus MRI
CT versus MRI
MRI vs CT in diagnostic imaging MRI Slower acquisition/sensitive to motion Cannot use with certain metal/pacemakers etc. Looks at multiple characteristics of tissues Better for soft tissue (i.e. tumor within soft tissue) CT Fast/less sensitive to motion No restrictions Looks at electron density only Better for density (i.e. calcification) Cannot use directly for RT Treatment Planning Can use directly for RT Treatment Planning
Use of MRI in Treatment planning Location CNS Head and neck Lung/thoracic Breast GI (esophagus, pancreas, gastric) Liver Spine Pelvis Role of MRI Important Not necessary (except for skull base) Not necessary Not necessary Not necessary Important Important Not necessary but nice to have
Take home point MRI demonstrates better soft tissue differentiation compared to CT MRI important where we treat gross tumor (GTV) which is surrounded by OAR with soft tissue of similar density
How does MRI help patients with pelvic cancers? Prostate cancer Staging (extracapsular invasion, seminal vesicle invasion, bladder neck invasion etc) Precise location of gland, urethra, bladder, rectum Cervical cancer Staging (parametrial invasion, local invasion etc) Brachytherapy boost tumor extent Sarcoma Delineating tumor from muscle
Current standards CT based volume definition is standard for pelvic cancers No requirement for MRI based treatment planning Prostate cancer MRI not in routine use Cervical cancer brachytherapy: GEC-ESTRO guidelines strongly recommends MRI based 3D-volume definition for cervical brachytherapy ABS (2011) guidelines also recommend 3D-volume definition with ultrasound, CT or MRI
Prostate contouring is not easy Visible Human Project Normal male cadaver CT at 3 mm slices Anatomy slices at 1 mm Six experience radiation oncologists asked to contour on CT Compare to true anatomy (tissue) Gao et al. Radiotherapy and Oncology. 2007. 85: 239-246
Problem with CT based volume definition Inter-observer variability 18.8% Contoured volume exceeded true prostate volume (30%) on average And did not include 9-21% of true prostate Gao et al. Radiotherapy and Oncology. 2007. 85: 239-246
Prostate MRI Impossible to see prostate cancer on CT MRI allows to distinguish: Minor extracapsular extention Minor seminal vesicle invasion Location of anterior fibrous stroma Location of GU diaphragm Bladder neck
MRI in prostate cancer
Common areas of mistakes on contouring (avoidable by MRI) McLaughlin et al. IJROBP 2010: 78: 369-378
How much does this matter? PTV expansion covers areas of inadequate CTV definition Over-treatment laterally, anteriorly do not necessarily have much consequence Superior bladder neck Inferior penile bulb/erectile tissues Posterior rectum
Ali et al. PRO. 2013: e1-9. MRI vs CT Target Delineation for PC Multi-institutional study of 155 consecutive patients CT vs CT + MRI Prostate volume smaller on MRI 43 cc vs 55.7 cc DVH parameters better Less acute GU toxicity with MRI
MRI vs CT Target Delineation for PC MRI planning had 21.5% smaller prostate volume (40.9 cc vs 52.1 cc) No difference in PSA outcome? Difference in late toxicity Sander et al. Acta oncologica. 2014: 53: 809-814.
Areas where MRI is especially helpful Rectal prostate interface Hip arthroplasty artifact Small margins (SBRT) Delineation of target and OAR structures within the prostate Avoid under-coverage of base Delineate SV from prostate (spare rectum) Focal therapy Urethra
Target delineation rectal interface
Hip prostheses
SBRT Selected RTOG 0938 prostate SBRT clinical trial guidelines: Margins: 3mm posteriorly/5 mm on sides Urethral dose < 107% hot spot
Avoid over contouring
Urethral sparing
Take home point MRI improves upon CT based planning Amount of benefit is not large in most cases and CT based planning is adequate MRI very helpful in select situations
Process for Incorporation of MRI in treatment planning for pelvic malignancy CT-simulation MRI simulation (future?) Contouring on CT/MRI Fusion/Registration Planning on CT MRI Other diagnostic studies (CT, PET)
How to incorporate MRI for PC Imaging characterstics: Hypodense on T2 Early enhancement and washout Restricted diffusion MR-Spect with decreased citrate, increased choline Typically use T2 images to fuse Fuse with axial and sagittal and coronal images
How I fuse MRI Auto-fusion software ALWAYS VERIFY Over-relies on bony anatomy Set appropriately small ROI Easy fusions: skull, spine, pelvis (regions limited by bony anatomy) Hard fusions: soft tissue (i.e. abdomen), altered anatomy (i.e. curve in spine) Contour on CT and match to appropriate structure on MRI Look at fusion in multiple planes Only need to make sure ROI fusion is good
Fusion using fixed match points (fiducials)
Iterative refinement May need to keep adjusting fusion between CT and MRI until fusion satisfactory Sometimes fusion simply does not work because of misalignment between scans or distortion but this is rare
Common pitfalls Changes in MRI based volume due to ADT Using autofusion (relies too much on bone) Not looking on multiple planes Not all MRI are the same need 3T multiparametric imaging for prostate cancer
Take home points For fusion to MRI: do not rely on bony landmarks and use iterative review
Future Directions for MRI in prostate cancer
Focal therapy Classically prostate cancer has always been treated to the whole prostate Treatment of only a portion of the prostate gland may reduce toxicity Potential to re-treat other parts of the gland Multiple modalities: HIFU, cryo, thermal ablation, brachytherapy etc.
Comparison of MRI with whole mount to detect prostate cancer mpmri: For Gleason > 6 or tumor size > 1 cm sensitivity = 72% Largest tumor detected in 80% Le et al. Eur Urology. 2015: 67: 569-76
Comparison of MRI with whole mount to detect prostate cancer Le et al. Eur Urology. 2015: 67: 569-76
Use of MRI in focal therapy MRI can identify target lesions (largest, most aggressive) MRI tends to underestimate size of lesions need 5-9mm margins Not ready for prime time yet!
MRI only treatment planning Challenges: Need to convert to electron density (already have in CT information) for dose calculation Geometric distortion (magnetic field, tissue susceptibility changes) in MRI Potential solutions: Automatic segmentation of bone and assign all other tissue water density (dose calc within 1-2%) External or internal reference markers Minimal distortion at isocenter 3 mm distortion at 9 cm from isocenter Kapanen et al. Magnetic Resonance in Medicine. 2013: 70 127-135
MRI only treatment planning Commercially available system!
Take away points MRI becoming more important in treatment planning in prostate cancer Not yet standard, but will become a greater part of treatment planning.
Thank you!
MRI brachytherapy in cervical cancer
Impact of image guidance in cervical brachytherapy: 2D vs 3D Prospective non-randomized French trial of 2D vs 3D brachytherapy in cervical cancer Better local control with 3D Less toxicity with 3D Charra-Brunaud et al. Radiotherapy and Oncology. 2012: 305-313
GEC-ESTRO Guidelines moving toward volume based implants High risk CTV = Whole cervix PLUS residual extra-cervical tumor at time of implant Large tumors difficult to see extra-cervical extension on CT American Brachytherapy society guidelines still allow point A dosing, but are moving towards 3D implants
MRI base brachytherapy ARRO Cervical cancer lecture
MRI based brachytherapy
MRI based brachytherapy
CT vs MRI cervical brachytherapy
Take away points MRI replacing point A dosing For most situations, CT is good enough in 3D treatment planning, but MRI is the best