Let s speak the same language: Standardization in Terminology

Transcription

1 Faculty Disclosure Let s speak the same language: Standardization in Terminology Michael Torrens Gamma Knife Department, Hygeia Hospital, Athens, Greece

2 Why is standardization necessary? Neurosurgical stereotactic radiosurgery has evolved over years and may be said to have led radiotherapy in a new direction. Developed empirically and with its own terminology, parallel to radiotherapy and parallel to the recommendations of the International Commission on Radiation Units and Measurements (ICRU) whose mission is to develop and promulgate internationally accepted recommendations on radiation related quantities and units, terminology, measurement procedures, and reference data for the safe and efficient application of ionizing radiation to medical diagnosis and therapy, radiation science and technology, and radiation protection of individuals and populations and parallel to the QUANTEC (Quantitative Analysis of Normal Tissue Effects in the Clinic) report in respect of quality and safety. The ICRU Small Field committee is working on a new report provisionally entitled Prescribing, Recording, and Reporting Stereotactic Treatments with Small Photon Beams (Report Committee 23)

3 Why is standardization necessary? Great variation in target and organs at risk (OAR) contouring. Some units do not delineate the target at all. Radiosurgery has assumed a target fixed in space, this is not always true. Possible need for planning target volume (PTV), especially in non-frame based SRS. Actual dose received by the target is not consistently recorded and prescription dose is not the actual dose. Absorbed dose D 95% should be quoted for comparison. What is the impact of plan quality on clinical results? Dose reporting currently is inconsistent. Need to move toward a dose volume histogram (DVH)-based approach Dose limits to OAR should be based on DVH data

4 Variations in volume delineation 1. Interobserver variability. AVM nidus across observers can have positional shifts of 2.8 ± 2.6mm (Borden 2000) Target volumes have standard deviations of 11%-105% across 20 observers (Sandstrom 2011) Target volume variation extending between 35% and 402% (Mitine 2006) 2. Image sets used for contouring T2 images are 12% smaller than T1 (Woo et al. 2010) Tumour volume may vary with differences in timing between contrast administration and image acquisition (Tao et al. 2009). 3. Image co-registration. Some studies report co-registration at a sub millimeter level (Chang 2003, Ken 2007) Others find inconsistencies up to 4mm or more (Cohen 1995, Tonetti 2012) 4. Different algorithms used in SRS planning systems, have been shown to result in volume differences, e.g. 3.6% to 22% in a phantom and 1.0%-10.2% in a patient (Ma 2012).

5 The target may not be where the MRI shows it to be There are various types of magnetic distortion, machine induced, patient induced and contrast induced. In one example the apparent position of the target depends on the direction (polarity) of the MRI read gradient. The vector of the total geometric distortion exhibits a directional dependence with respect to the frequency encoding axis and read gradient polarity (direction) selected during MRI acquisition. The selection of the y-axis for frequency encoding during MRI acquisitions results in accuracy degradation along the z-axis which for one specific MR scanner examined was found to be of the order of 0.5 mm.

6 Geometric uncertainty and underdosage results associated with the forward series in a patient with 19 metastases. (A) The geometric uncertainty vectors are presented as scaled arrows. (B) Target underdosage results were derived from target coverage of the prescription dose in the average images (the center and the diameter of each sphere coincides with the center and diameter of a target metastasis, whereas its color denotes the percentage of underdosage according to the presented color bar). P Karaiskos, A Moutsatsos, E Pappas, E Georgiou, A Roussakis, M Torrens, I Seimenis. A simple and efficient methodology to improve geometric accuracy in Gamma Knife radiation surgery: Implementation in multiple brain metastases. Int J Radiation Oncol Biol Phys 2014 Oct 13;90(5):

7 Quote the imaging and contouring parameters 1. The details of acquisition (e.g. type of scanner, field strength, MR sequence and slice thickness, read gradient polarity etc.). 2. Details of contrast administration: type, dose, injection timing 3. The contouring method and images used for target/oar delineation (adherence to recommendations of LGKS contouring consensus). In the future it may be appropriate to make standard recommendations about the ideal imaging parameters.

8 Why is standardization necessary? Great variation in target and organs at risk (OAR) contouring. Some units do not delineate the target at all. Radiosurgery has assumed a target fixed in space, this is not always true. Possible need for planning target volume (PTV), especially in non-frame based SRS. Actual dose received by the target is not consistently recorded and prescription dose is not the actual dose. Absorbed dose D 95% should be quoted for comparison. What is the impact of plan quality on clinical results? Dose reporting currently is inconsistent. Need to move toward a dose volume histogram (DVH)-based approach Dose limits to OAR should be based on DVH data

9 Reliability of frame fixation? The results of evaluating 170 system tests over 5 years showed that the mean displacement vector of the complete Gamma Knife system (including MRI imaging) was 0.48 mm ± 0.23mm (Mack et al. 2002). Most results for linac-based stereotactic radiation therapy are less accurate (Chang et al. 2003). Weight and deformity. Arn and Carlsson (2014) showed that a 5 kg load produced only a 0.02 mm deformation in an adult sized head phantom, but care should be taken with insulated Leksell posts where the deformity is greater and for longer screws (in children) which allowed a deformity of up to 0.44 mm. Two studies (Bednarz et al 1999, Massager et al. 2011) using image coregistration, have evaluated the difference between the frame coordinates and the position of a resulting stereotactically directed lesion, a procedure that evaluates the end to end error of the treatment process; both studies report a root mean square (RMS) error of 0.9 mm. Respiration etc. Kartaria et al.(2013) using cone beam CT measurements of target position have shown a RMS interfraction shift of 0.16 mm. Tonetti et al. (2012) showed movement of up to 4 mm of targets at the C1 level with change of neck position despite fixation of the skull by frame.

10 Reliability of other fixation? Using the Elekta EXTEND system, a relocatable frame based on a dental impression and bite block, the mean 3D intrafraction motion was found to be ± 0.3 mm (Ruschin 2010 et al., Schlesinger et al. 2012). Masi et al. (2008) report that variabilities for bite block and mask (2.9 ± 1.3 mm) were smaller than those for thermoplastic mask alone (3.2 ± 1.5 mm). In another study the typical radial overall uncertainty for positioning and fixation for stereotactic radiosurgery using a mask alone has been recorded as 1.80 ± 0.60 mm (Karger et al. 2001). For the Elekta ICON mask based immobilisation - end-to-end locational accuracy 0.2mm for single and multi-targets target treatment, target movement during treatment 0.43mm. Chung C et al. Clinical Evaluation of a Novel Thermoplastic Mask System with Intrafraction Motion Monitoring using IR Tracking and Cone-beam CT for Gamma Knife Radiosurgery. ASTRO, San Francisco, September 14-17, 2014 As the indications and techniques evolve beyond frame fixed cranial targets, certain procedural modifications will be necessary including the routine quotation of the mean displacement vector of the radiosurgery system being used and the incorporation of PTV when necessary. If PTV is required is it radiosurgery?

11 Why is standardization necessary? Great variation in target and organs at risk (OAR) contouring. Some units do not delineate the target at all. Radiosurgery has assumed a target fixed in space, this is not always true. Possible need for planning target volume (PTV), especially in non-frame based SRS. Actual dose received by the target is not consistently recorded and prescription dose is not the actual dose. Absorbed dose D 95% should be quoted for comparison. What is the impact of plan quality on clinical results? Dose reporting currently is inconsistent. Need to move toward a dose volume histogram (DVH)-based approach Dose limits to OAR should be based on DVH data

12 Recurrent Meningioma GTV 17.7 cm 3 ( )

13 Recurrent Meningioma GTV 17.7 cm 3 ( ) Two plans: Conformity index PCI 0.77 in both (courtesy Chryssa Paraskevopoulou) STEEP DOSE GRADIENT UNDERDOSING

14 Recurrent Meningioma GTV 17.7 cm 3 ( ) Two plans: Conformity index PCI 0.77 in both (courtesy Chryssa Paraskevopoulou) Plan 1 Prescription dose 13.5 Gy Coverage 95% D Gy Optic max dose 8.4 Gy Plan 2 Prescription dose 13.5 Gy Coverage 82% D Gy Optic max dose 8.4 Gy

15 Quote the dosage parameters 1. Dose Computation: The algorithm used to express the dose delivered (e.g. TMR 10, TMR Classic or Convolution in GammaPlan ) must be quoted since the dose may vary depending on the algorithm used. 2. Prescription Dose: rather than Planning, Planned and Peripheral or Marginal dose. Though this may not be the best basis for prescription see communication Thursday 19 th at 10am Significance of dosimetry in treatment of meningiomas 3. Absorbed Dose: a standard absorbed dose value such as D 95% should be recorded for all treatments as recommended by ICRU.

16 Why is standardization necessary? Great variation in target and organs at risk (OAR) contouring. Some units do not delineate the target at all. Radiosurgery has assumed a target fixed in space, this is not always true. Possible need for planning target volume (PTV), especially in non-frame based SRS. Actual dose received by the target is not consistently recorded and prescription dose is not the actual dose. Absorbed dose D 95% should be quoted for comparison. What is the impact of plan quality on clinical results? Dose reporting currently is inconsistent. Need to move toward a dose volume histogram (DVH)-based approach Dose limits to OAR should be based on DVH data

17 Variation in results versus conformity The basic principle of radiosurgery is to produce accurately delivered conformal therapy. However target response may be inversely related to conformity indexes because: 1. Conformity is better and outcome/response is worse in larger targets. 2. Less conformal plans include local microscopic disease in malignant cases and so produce better results. Good conformity has been shown to be related to a reduced incidence of complications in AVM s and vestibular schwannomas by Freidman et al (2003) and Liscak et al. (2000) but the number of studies showing this correlation are small. Petti et al. (2011) suggest that the range of conformity indices represented in most studies may not be wide enough to show a statistical correlation between improved conformity and decreased toxicity. There is no simple relationship between conformity and outcome but that does not dictate that planning quality indexes should be disregarded, rather reassessed more critically and with a standardized approach.

18 Why is standardization necessary? Great variation in target and organs at risk (OAR) contouring. Some units do not delineate the target at all. Radiosurgery has assumed a target fixed in space, this is not always true. Possible need for planning target volume (PTV), especially in non-frame based SRS. Actual dose received by the target is not consistently recorded and prescription dose is not the actual dose. Absorbed dose D 95% should be quoted for comparison. What is the impact of plan quality on clinical results? Dose reporting currently is inconsistent. Need to move toward a dose volume histogram (DVH)-based approach Dose limits to OAR should be based on DVH data

19 Maximum/minimum dose Maximum and minimum point-based dose reporting can be misleading due to the nature of the calculation and dose grid. There are two recognized alternatives to report a more representative value using the DVH: 1. The dose over (e.g.) a 1mm 3 region. 2. The absorbed dose e.g. D 2% which conforms to the recommendations of IAEA and ICRU. As a percentage, this value is dependent on the total volume of the target or structure. The standardization report therefore favors the former definition. The ICRU may change their recommendation on our advice.

20 Why is standardization necessary? Great variation in target and organs at risk (OAR) contouring. Some units do not delineate the target at all. Radiosurgery has assumed a target fixed in space, this is not always true. Possible need for planning target volume (PTV), especially in non-frame based SRS. Actual dose received by the target is not consistently recorded and prescription dose is not the actual dose. Absorbed dose D 95% should be quoted for comparison. What is the impact of plan quality on clinical results? Dose reporting currently is inconsistent. Need to move toward a dose volume histogram (DVH)-based approach Dose limits to OAR should be based on DVH data

21 Recurrent Meningioma GTV 17.7 cm 3 ( ) Two plans: Conformity index PCI 0.77 in both (courtesy Chryssa Paraskevopoulou) Plan 1 - steep dose gradient Optic max dose 8.4 Gy Optic mean dose 5.6 Gy Optic 8 Gy volume 0.3mm 3 Optic 6 Gy volume 69.0mm 3 Optic max 1mm Gy Plan 2 - underdosing near OAR Optic max dose 8.4 Gy Optic mean dose 6.8 Gy Optic 8 Gy volume 2.9mm 3 Optic 6 Gy volume 180.0mm 3 Optic max 1mm Gy

22 Is standardization necessary? Great variation in target and organs at risk (OAR) contouring. Some units do not delineate the target at all. Radiosurgery has assumed a target fixed in space, this is not always true. Possible need for planning target volume (PTV), especially in non-frame based SRS. YES Actual dose received by the target is not consistently recorded and prescription dose is not the actual dose. Absorbed dose D 95% should be quoted for comparison. What is the impact of plan quality on clinical results? Dose reporting currently is inconsistent. Need to move toward a dose volume histogram (DVH)-based approach Dose limits to OAR should be based on DVH data

23 The Standardization of Terminology in Stereotactic Radiosurgery: First Report from the Standardization Committee of the International Leksell Gamma Knife Society MICHAEL TORRENS, CAROLINE CHUNG, HYUN-TAI CHUNG, PATRICK HANSSENS, DAVID JAFFRAY, ANDRAS KEMENY, DAVID LARSON, MARC LEVIVIER, CHRISTER LINDQUIST, BODO LIPPITZ, JOSEF NOVOTNY, IAN PADDICK, DHEERENDRA PRASAD, CHUNG PING YU.

24 Scope of first report 1. Introduction 2. Objectives of standardization report 3. Definition of Stereotactic Radiosurgery 4. Target delineation and volume measurement 5. Dose prescription and measurement (target) 6. Dose definition in areas of risk

25 Progress of standardization proposal committee discussion at ISRS Paris white paper discussed and voted at LGKS Sydney with 86 % approval subject to revision internet vote on revised paper by all society with 92 % approval liaison with and influence on ICRU accepted for publication in LGKS 2014 supplement J Neurosurg (Suppl 2) 121:2-15, 2014

26 Highlights of first report Difficult to agree a definition of radiosurgery. The need for all to use the same terminology, for example Prescription Dose - not peripheral, planning or marginal dose. Incorporation of terminology in the Gamma Knife Registry. Need for PTV in relation to various uncertainties especially in non-frame based SRS. Prescription dose is not the actual dose. Absorbed dose D 95% should be quoted as well for comparison. To consolidate the concept of a Treated Target Volume, important for planning quality assurance. More studies needed of the relation between planning quality assurance and outcome/complications. When considering Accepted Tolerance Dose to Organs at Risk (OAR) to use the dose to a specific volume (and not % of volume as in ICRU).

27 Consensus statement Revision Implementation LGKS 2016 Involvement: ISRS ICRU Audit Problems: Legitimacy Single machine consensus

28 Conformity with first report at LGKS 2016 Review of submitted abstracts (Benign Tumours) 17 % correct terminology 66% no relevant terms used in abstract 17% use of Margin Dose traditional in USA but: Prescription Dose This report recommends the use of the term Prescription in relation to Dose and Isodose, rather than Planning, Planned and Peripheral or Marginal. In view of the fact that the question of what is the most appropriate Prescription dose (e.g. D95%) should be under review, perhaps Marginal can be accepted pro tem until the prescription method is finally agreed?

29 Where do we go from here? Second report contouring consensus for OAR s is in progress and hopefully will be ready for publication in the 2016 LGKS supplement. When consensus is obtained for contouring OAR s then studies will be needed to define the Accepted Tolerance Doses of each structure. Third report - target definition is possibly the largest source of error in radiosurgery. A contouring consensus is needed also for targets. Further coordination with other machine specific societies, with ISRS and ICRU to ensure agreement across technologies.

30 THANK YOU