Individualising prescription dose to lung tumours based on NTCP

Transcription

1 Agenzia Provinciale per la Protonterapia Trento Individualising prescription dose to lung tumours based on NTCP M. Schwarz

2 RT for the lung: where are we? Typical radical doses < 70 Gy (Very) disappointing local control CRT has technical limitation Use of functional imaging for target delineation still at the early stages

3 Rationale for RT dose escalation 100 Local control (%) Vijayakumar Martel Arriagada Dose (Gy)

4 AvL-NKI Phase I/II dose escalation study using 3D CRT Establish maximum tolerated dose (MTD) using 3D CRT Grade 3 RP chosen as primary endpoint Dose escalation scheme based on risk estimate of RP Belderbos J. et al. IJROBP 2006

5

6 Delivered dose by patient subgroup Patients grouped by rmld Belderbos et al, IJROBP 2006

7 Benefits of dose escalation Belderbos et al, IJROBP 2006 Kong et al, IJROBP 2005

8 What do we need for a safe dose escalation? Consistency between treatment planning and delivery Geometry Anatomy representation Target delineation Dosimetry Robust dose response models for the OARs Irradiation techniques allowing to escalate the dose

9 Dose algorithm The need of accurate dose engine for lung treatmement plans should not be a matter of discussion these days. Dose (% of central axis) TPS1 TPS2 Film d (from central axis, cm) M. Engelsman, R&O 2001

10 De Jaeger et al, R&O 2003 Dose algorithm

11 Influence of dose calculation on NTCP parameters De Jaeger, R&O 2003

12 Dose response relations A fit is available based on our lung clinical data and dose grids recomputed with a convolution algorithm (De Jaeger et al., R&O 2003) t = (MLD-TD 50 )/m TD 50 These are NTD corrected doses! TD 50 =29.2 Gy n=1 m=0.45

13 Mean Lung Dose and RT pneumonitis Kong et al. Semin Oncol 2005

14 Dose escalation/imrt for NSCLC Engelsman et al (IJROBP 2001): iso-ntcp dose escalation with heterogeneous dose distributions in the target volume. Grills et al (IJROBP 2004): IMRT to spare OAR (S&S). Dose heterogeneity in the PTV seemed inevitable. Murshed et al (IJROBP 2004): IMRT to spare OAR (dmlc). Increase of lung tissues irradiated at doses < 5Gy.

15 Aim of the planning study Separately assess the benefits of dose heterogeneity in the target volume and IMRT for the purpose of dose escalation Analyse the properties of IMRT dose distributions in the Organs At Risk when compared to conventional CRT treatments Schwarz et al. IJROBP 2005

16 Patients Data Datasets of 10 pts treated in the phase I/II trial 8 patients chosen according to risk group 2 patients with the oesophagus as dose limiting organ CRT replanned with a convolution dose calculation algorithm (Pinnacle) IMRT planned with Hyperion (MC) and recalculated in Pinnacle

17 Es (III) Es (III) IV IV III III II II I I Group PTV Vol. (cm 3 ) 52 T+N 32 T+N 70 T+N 74 T+N 127 T 123 T+N 57 T 43 T 8 T 6 T GTV Vol. (cm 3 ) T2N 2M0 T2N 2M0 T2N 2M0 T2N 2M0 T3N 0M0 T2N 3M0 T2N 0M0 T2N 0M0 T1 N0 M0 T2N 0M0 TNM Patient ID Patients characteristics

18 Which technique should be used to escalate the dose? CRT1: Stdev(D PTV ) δ 3% D prescr CRT2: Dmin(PTV) CRT2 ε Dmin(PTV) CRT1 IMRT1: Stdev(D PTV ) δ 3% D prescr IMRT2: Dmin(PTV) IMRT2 ε Dmin(PTV) IMRT1 Prescribed dose = mean PTV dose For each technique the prescribed dose was the maximum PTV mean dose that did not violate the OAR constraints (In silico phase I trial)

19 Tolerances for the OARs Lung: MLD δ 16 Gy, NTD corrected ( / =3 Gy) 17% risk of grade II and 2% of grade III complications Oesophagus: EUD < 74 Gy (n=0.06), NTD corrected ( / =3 Gy) Spinal cord: Dmax δ 50 Gy, NTD corrected ( / =2 Gy) Heart: Usual DVH points for CRT (V 40 < 100%, V 50 < 66%, V 66 < 33%) do not prevent unnecessary heart irradiation. DVH points chosen individually for each patient to achieve dose conformity.

20 IMRT optimisation Dose tolerances directly translated into the cost function (including / =3 Gy) Hyperion performs by design constrained optimisation of PTV dose Accurate dose calculation also during optimisation Two step optimisation, including segment shapes and weights: More efficient delivery in terms of MU Less interplay effects

21

22

23 How to control the dose in the lung? V20

24 Local sigmoidal dose effect

25 Results

26 CRT Plans CRT1: For group I and II (patients 1 to 4) Gy could be achieved with low MLD (10-12 Gy) and few beams (δ 4). For patient 5 to 10: The achievable doses were significantly lower (down to 54 Gy) The lung was always the dose limiting organ CRT2: average dose escalation of 6%

27 IMRT Plans Either IMRT1 or IMRT2 could achieve a dose of at least 85 Gy in all except one cases. For more than half of the patients the esophagus was the dose limiting organ. The average number of segments per beam was 8 for IMRT1 and 10 for IMRT2 The typical MU ratio between IMRT and CRT was equal to 2

28 Benefit of CRT2, IMRT1 and IMRT2 - Dmin Increase of minimum dose in the PTV(Gy) CRT2 IMRT1 IMRT Patient Schwarz et al. IJROBP 2005

29 IMRT vs CRT Dmin = 65 Gy TCP 20 % Dmin = 74 Gy TCP 39 % CRT IMRT

30 IMRT and dose heterogeneity in the target 75 Gy TCP 39 % 75 Gy 84 Gy TCP 57 % IMRT and dose homogeneity IMRT and dose heterogeneity

31 Dose distribution in the PTV

32 Dose distribution in the lung MLD = 16 Gy for all three techniques V5-V20 nearly identical

33 Did we stretch the dose response model? What was the effect of optimizing / corrected dose values?

34 Esophagus P6 P7 P8 P9 P10 CRT Lung Lung Lung Lung Lung IMRThom Esoph. Lung Lung Esoph. Lung IMRTinhom Esoph. Lung+ Esoph. Esoph. Esoph. Esoph. Different dose tolerances are being used at different institutions (e.g. V60 vs. EUD 0.06 ) Different volumetric response for late and acute toxicity (EUD 0.06 vs. EUD 0.7 ) These differences might lead to different optimisation problems.

35 Esophagus

36 Conclusions NTCP-based IMRT optimisation is an effective method for dose customisation. IMRT ± dose heterogeneity in the PTV allows to escalate the dose in NSCLC pts with large/concave tumor volumes. The IMRT dose distributions were subsequently proved safe w.r.t geometrical uncertainties (Schwarz et al, IJROBP 2006) More data needed on: dose response for the esophagus

37 What if we used protons? HT vs IMPT - IsoTCP plan

38 p+ IsoNTCP plan Lung_HT Lung_IMPT

39 p+ IsoNTCP plan-2

40

41 Dose volume relationships Pelvis + Head&Neck Marco Schwarz Sara Broggi (HSR Milano)

42 PELVIS Radiother. Oncol QUANTEC, IJROBP 2010

43 Organs at risk Rectum Bladder Small Bowel Penile bulb Bone marrow (hematologic toxicity)

44 RECTUM Late bleeding: Dose volume constraints High dose region : larger convergence. DVH constraints for V70 andv75 predictive of a very low incidence of late bleeding (Dose 70-80Gy, 1.8-2Gy/fr) V50 < 55% V60 < 40% V65< 30 % V70< 20-25% V75< 5%

45 RECTUM Late bleeding: NTCP models

46 QUANTEC parameters: n=0.09, m=0.13, TD50=76.9Gy

47 RECTUM late bleeding: DVH + clinical risk factors [ Fiorino et al, 2009 ] [ Fellin et al, 2009 ]

48 RECTUM: Acute rectal toxicity - Severe acute toxicity => interruption of treatment with a potentially detrimental effects - Evidence that acute damage may play a significant role in late toxicity [ Fiorino et al, 2009 ] Dose-volume effect: Dmean and DVH constraints in Gy dose region: most predictive parameters for acute side effects

49 BLADDER: dose volume effects Evidence of a dose effect for whole bladder irradiation D5= 65 Gy / D50 = 80 Gy In irradiation of pelvic tumors (prostate, rectum, gynecological cancer) the bladder is only partial irradiated at the prescribed dose Quantec, 2010 Large variations in bladder shape during treatment (variable filling) => lack of knowledge regarding dose-volume modeling of bladder toxicity Bladder

50 Serial behaviour for late mild-severe toxicity [ Cheung 2007] Mixed serial-parallel behaviour for chronic urinary moderate/severe toxicity [ Harsolia 2007] A small fraction of bladder (few cc) receiving more than Gy is highly predictive of late GU toxicity (clinically confirmed by a large increase of moderate/severe late GU toxicity with the escalation to high doses) Pre- treatment GU complaints, TURP, TURPT, presence of acute toxicity are factors probably involved in conditioning urinary morbidity

51 we know very little about bladder volume effect Bladder HT Bladder IMPT Volume (%) Dose(Gy)

52 BOWEL: dose volume effects Common knowledge that the irradiation of large volumes of the bowel to doses around Gy (1.8/2 Gy/day) during whole pelvis irradiation with conventional RT is associated with moderate /severe acute toxicity; the probability an the severity of these effects increases with field width and with dose/fraction Evidence of a large volume effect, but few quantitative studies Increasing interest for pelvic nodes irradiation IMRT as a valid tool to spare bowel Critical dependence on the definition of bowel/ intestinal cavity /bowel loops

53 BOWEL definition Bowel is a mobile structure => blurring of the evidence for a dose/volume relationship with toxicity The expansion required to cover all possible locations of intestine in 90% of patients during RT has been estimated equal 3 cm around the bowel. [ Hysing, RO 2006] Intestinal cavity (IC) a robust contour with respect to bowel motion [Sanguineti et al. RO 2008]

54 Quantec recommendation V15 <120 cc (bowel loops contours ) V45 < 195 cc ( intestinal cavity)

55 Grade 2-3 acute bowel toxicity: Average DVHs 800 TOMO: n = 58 IMRT: n = 26 4-fields box.: n = 91 Volume (cc) Dose (Gy) TOMO: 1/58 (1.7%) IMRT: 2/26 (7.7 %) 3/84 (3.6 %) 4-fields box: 19/ 91 (20.9%)

56 INTESTINAL CAVITY OUT PTV V40 vs Tox Probability V40 < 172 cc 172 cc < V40 < 296 cc V40 > 296 cc Tox probability 0,6 0,55 0,5 0,45 0,4 0,35 0,3 0,25 0,2 0,15 0,1 0, V40 (cc) IMRT PATIENTS!!

57 Summary: BOWEL dose-volume relationships Evidence of a large volume effect for WPRT. Quantitative dose-volume relationship recently reported for intestinal cavity/loops Using IMRT with an avoidance approach combined to DVH constraints on intestinal cavity/loops dramatically reduces upper gastro-intestinal toxicity (both incidence and severity!)

58 QUANTEC, IJROBP 2010 HEAD&NECK

59 HEAD-NECK: Organs at risk Salivar glands (parotids, submandibular) Larynx /Pharynx (related muscles structures) Mucosae Spinal cord Tyroid / Esophagous Mandible / TMJ Internal hear (cochlea) Optical Nerves/ Chiasma Brain /Brain stem

60 QUANTEC recommendation - HN region Dose limits EQD2/BED/ NTCP recommendations Probability curves Brain Yes Predictors for 5 and 10% are given in BED and EQD2. (α/β =3Gy) Incidence as function of BED Optical nerves Chiasm Yes - - Brain stem Yes Total dose vs fraction - dose curves for EQD2 using α/β =3.3, 2.5, 2.1Gy - Spinal cord Yes EQD2.; α/β= 3Gy Probability as function of EQD2 Cochlea Yes - - Salivary glands Yes - Tox severity vs mean dose; TD50(50% function loss) vs followp up months Larynx/Pharynx Yes - Probability as function of mean dose

61 PAROTID GLANDS: dose volume effects Stimulated salivary production is largely (60-70% of the total) derived from the parotid glands. Unstimulated salivary production is due primarily to the submandibular and sublingual glands. Outcomes: Reduced salivary function, xerostomia, alterations in speech and taste, nutritional problems Evaluation of outcome: - Subjective: assessment based on patient s symptoms by questionnaires, interviews - Objective: measurements of salivary gland flow rate (stimulated/unstimulated) ; imaging end-points (scintigraphic activity)

62 PAROTID GLANDS: dose volume effects Large volume effect dependency (parallel [QUANTEC, 2010 ]

63 PAROTID GLANDS : entity and speed of recovery Unstimulated saliva flow rate 142 pts. (3DCRT & IMRT) 266 parotid glands Saliva flow rate: 1, 3, 6, 12, 18, 24 months after RT Measurements for each single parotid gland; stimulated and unstimulated flow rate Stimulated saliva flow rate Saliva flow reduction within 1-3 months after RT; then gradually recovery Dmean < Gy => almost complete recovery at months [Li, IJROBP 2007]

64 PAROTID GLANDS: NTCP models Large volume effect Parotid function recovery Parotid flow reduction: : TD50 ~30-45 Gy [Quantec, IJROBP 2010] TD50(Gy) m n Emami (1991) No 3D Eisbruch(1999) 88pts 28.4 ( ) 0.18 ( ) 1(fixed) Roesink(2001) 180pts; Flowratio<25% 6 weeks:31 (26-35) 6 months:35 (30-40) 1 year:39 (34-44) 0.54 ( ) 0.46 ( ) 0.45 ( ) 1(fixed) Roesink(2004) 96pts; SEF<45% 6 weeks:29 (25-34) 1 year:43 (37-51) 0.73 ( ) 0.53 ( ) 1(fixed) Munter(2007) 75pts; SEF<50% 3 months 36.4 ±15.9 (3DCRT) 35 ±3.5 (IMRT) 2.2 ±1.8 (3DCRT) 6.5 ±2.1 (IMRT)

65 PAROTID GLANDS.. recent update [Dijkema, IJROBP 2010] 222 pts. 384 parotids (Michigan:157; Utrecht:227) Stimulated flow rates 1y after RT NTCP end point: reduction <25% baseline Confirmation of NTCP parameters (n=1)

66 Submandibular glands Dmean > 40 Gy => reduction of submandibular gland-stimulated salivary function [Murdoch-Kinch 2008] Parotid and Submandibular glands: entity and speed of recovery [Strigari, IJROBP 2010] 63 pts. (IMRT) Salivary flow rate : 3, 6, 12, 18, 24 months after RT Dmean parotids and submandibular glands Saliva flow reduction within 1-3 months after RT; then gradually recovery Dmean < Gy => almost complete recovery at 12 months

67 Endpoints: LARYNX /PHARYNX : dose volume effects Laryngeal edema, due to inflammation and lymphatic disruption. Progressive edema and associated fibrosis can lead to long-term problems with phonation and swallowing Dysphagia/Aspiration Evaluation of outcome: - Larynx edema: flexible fiberoptic examination - Dysphagia : videofluorography / esophagography to visualize phases of swalllowing - Vocal function /Dysphagia : can be objectively assessed using several instruments. Subjective assessments can be made with validated patient-focused questionnaires to assess various combinations of voice, eating, speech and social function

68 LARYNX : dose volume effects QUANTEC recommendation Edema /Vocal dysfunction V50 < 27% Dmean < 44 Gy Dmax < Gy (if possible, according to the tumour extent) [QUANTEC, 2010 ]

69 LARYNX edema: NTCP models [Rancati, IJROBP 2009] - 48 pts - Videofluoroscopy - End Point: G2 Larynx edema - Fit: LKB (LEUD) and logistic model (LOGEUD) TD50 m n LKB model (LEUD) 47.3 ± 2.1 Gy 0.23 ± ± 0.6 TD50 k n Logit model (LOGEUD) 46.7 ± 2.1 Gy 7.2 ± ± 0.8 TD50 = 46.7 Gy K = 7.2 n fixed= 1 TD50 = 47 Gy k= 7.33 Reduction of risk of G2-G3 larynx edema EUD < Gy

70 PHARYNX (Dysphagia) : dose volume constraints [QUANTEC, 2010 ] Swallowing: complex process that involves voluntary and involuntary movements through several cranial nerves and muscles. => Several possible anatomic structures whose dose-volume parameters could have the major effect with dysphagia ( pharyngeal constrictors muscles, glottic/supraglottic larynx,pharynx,.)

71 Dysphagia: dose volume constraints 36 pts; Stage III-IV Oropharynx / Nasopharynx Endpoint: Swallowing problems videofluoroscopy ( before RT vs 3 months after RT) 3 OARs: pharyngeal constrictors (PC); glottic e supraglottic larynx (GSL), esophagous [Feng, IJROBP 2007] GSL PC Mean dose, V50-V65 to pharyngeal constrictors muscles (PC) and supraglottic larynx (SL) correlated. QUANTEC recommendation V65 (PC) < 50% (more predictive) Dysphagia/ Aspiration Reduction as low as possible V60 / V50 pharyngeal constrictors and larynx

72 Dysphagia: NTCP models 81 pts Follow-up: 18 months - 50% NTCP for Dmean Gy to supraglottic larynx - For Dmean < 50 Gy, NTCP<20% [Quantec, IJROBP 2010] Significant correlation between Dmean to SCM and late dysphagia Probability of late dysphagia increases around 19% every 10Gy [Levendag, Radiother.Oncol. 2007]

73 [Narayan, IJROBP 2008] MUCOSAE : dose volume effects Threshold effect for local toxicity ( Dose > 32 Gy). Duration of mucositis depends on absorbed dose (>3 weeeks for D>39 Gy) Local effect..severity of mucositis depends on the irradiated volume! Few quantitative data [Werbrouck et al.,

74 Avoiding unnecessary irradiation of PC/larynx/esophagous.outside PTV!!! Conformal avoidance approach 70 Gy Gy 3

75 Parotid glands Summary: HN dose-volume relationships Consensus for Xerostomia => parotid as a parallel organ: Mean dose, V20-40 best predictors Faster recovery of the damage with IMRT Larynx, Pharynx, Oral mucosa Increasing evidence of dose-volume effects for dysphagia (especially sup. constrictor, larynx) and for PEG insertion; NTCP model available for laryngeal edema Mucositis: first quantitative data Dose-volume effect for Oral mucosa and constrictors in predicting PEG risk/swallowing problems Conformal avoidance approach with IMRT may reduce toxicities (even without quantitative dose-volume relationship)