1. Clinical decisions in the optimization process II. Emphasis on the avoidance of normal tissue complications Avi Eisbruch University of Michigan

2. Xerostomia following standard RT The salivary glands are very sensitive to radiation Standard radiation for head and neck cancer typically delivers a high dose to the salivary glands, causing permanent decline in saliva output The resulting xerostomia (mouth dryness) is the most frequent complaint of long term survivors.

4. Nasopharynx cancer, T3N2c GTV CTV Parotid

5. Measuring Parotid gland Salivary flow

6. Early decline of the salivary flow rates following RT Eisbruch et al, IJROBP 1996

7. Single Parotid cDVHs Volume (%) < 25% saliva flow at 12 months Dose (Gy)

8. Salivary flow (ml/min) Stimulated Salivary Flow Contralateral Parotid Bilateral neck RT Unilateral neck RT

9. Xerostomia summary scores Median and 25 th and 75 th percentiles 0 20 40 60 80 Pre-RT 1 month3 months6 months12 months18 months24 months Quality of Life Score Unilateral RT group Bilateral RT group Time after RT 100 75th percentile 25th percentile

10. Xerostomia following IMRT of nasopharyngeal cancer (RTOG grading) Sultanem et al, IJROBP 2000

11. QOL Each of the four domains of the QOL instrument (Eating, Communication, Pain, Emotion) was correlated with the mild/moderate xerostomia following IMRT. These correlations suggests that the efforts to reduce xerostomia through IMRT may improve general QOL.

12. All is well in the xerostomia front (once IMRT is employed) Is it?

13. All is well in the xerostomia front (once IMRT is employed) Problem No. 1: we cannot spare the submandibular salivary glands when treating the neck bilaterally.

14. v II Submandibular glands CTVs (Need to expand to yield PTVs)

15. Measuring Submandibular/sublingual glands salivary flow

16. Salivary flow (ml/min) Stimulated Salivary Flow Submandibular Gland Bilateral neck RT Unilateral neck RT

17. Limits in controlling dose distributions Would proton modulated RT improve this aspect? –Protons have a dosimetric advantage over photons: It is possible to determine the depth of the energy delivered in tissue

18. All is well in the xerostomia front once IMRT is employed. Problem No. 2: Weak correlation between the amount of saliva spared and patients’ subjective symptoms.

19. Correlation between saliva output from the major salivary glands and xerostomia scores: P=0.02; r=0.3

20. Weak correlation between the amount of parotid saliva spared and patients’ subjective symptoms. Could the output of the minor salivary glands be important, too?

22. Multivariate model for post-RT xerostomia scores Variable p-value Baseline Scores<0.001 Time 0.003 Major Salivary Gland Mean Dose 0.009 Oral Cavity Mean Dose0.002 Oral Cavity Mean Dose0.002

23. Unexpected clinical issues Careful monitoring of clinical outcome may reveal many issues that were not apparent beforehand. There may be a very large number of such issues that will need to be accounted for in optimizing therapy.

24. All is well in the xerostomia front once IMRT is employed. Problem No. 3: Uncertainties in the relationships between dose, volume, and reduced salivary output.

25. % parotid receiving > 15 Gy 0 25 50 75 100 125 150 175 200 0102030405060708090100 Stimulated Unstimulated % saliva flow rate % Parotid Saliva Flow Rate 12 Months Post-RT Relative to pre-RT flow

26. % parotid receiving > 30 Gy 0 25 50 75 100 125 150 175 200 0102030405060708090100 Stimulated Unstimulated % saliva flow rate % Parotid Saliva Flow Rate 12 Months Post-RT Relative to pre-RT flow

27. Stimulated Unstimulated % saliva flow rate % parotid receiving > 45 Gy 0 25 50 75 100 125 150 175 200 0102030405060708090100 % Parotid Saliva Flow Rate 12 Months Post-RT Relative to pre-RT flow

28. % Saliva Flow vs. Mean Dose Relative to Pre-RT Flow (1 & 3 months after RT) unstimulatedstimulated Mean Dose

29. % Saliva Flow vs. Mean Dose Relative to Pre-RT Flow (6 & 12 months after RT) unstimulatedstimulated Mean Dose

30. Dose, volume, and effect relationships Each of the partial volume dose thersholds, as well as the mean dose threshold, described these relationships well. In a multivariate analysis, the mean dose was found to be the best descriptor.

31. Vineberg, Eisbruch, et al. IJROBP 2002

32. Complication Probability Curves

33. Dose/response relationships Author RT Technique MeasuringMethodFittingModel Mean Dose reducing saliva to <25% RoesinkStandardSelectiveparotid First order 39 Gy Chao IMRT and standardWholemouthExponential 32 Gy EisbruchIMRTSelectiveparotidThreshold 26 Gy MunterIMRTScintigraphy 26-30 Gy MaesConformalScintigraphy 20 Gy

34. Dose/response relationships What are the reasons for the wide spread of the reported mean doses causing significant salivary flow decline?

35. #1: Different techniques --- different dose distributions ---- different relationships between the mean dose and partial volumes receiving any dose. What are the reasons for the wide spread of dose/response results?

36. Coopes et al, Groningen, The Netherlands Measuring rat parotid salivary flows following irradiation of different parts of the glands

37. Spatial dose distributions Irradiating different regions of the rat parotid gland yields different dose/volume/effect relationships (Coopes et al, ESTRO 2003). The spatial dose distribution within the gland is important DVH-related metrics alone are not enough.

38. In addition to dose... Clinical factors affecting xerostomia following parotid-sparing RT –Dehydration in the malnourished patient –Drugs (diuretics, antihistamines, antidepressants) Eisbruch et al, IJROBP 2001

39. In addition to dose... Physical/ statistical models alone cannot explain all the variation in the salivary output. There is (still) a lot of uncertainty –how much of the parotid glands should we spared, if their sparing requires some trade- off with target coverage?

40. IMRT: Lower total dose to a critical organ means also lower dose/fraction. The biologically equivalent dose delivered to the critical organ is lower than the nominal dose. For the same total maximal organ dose, IMRT is safer than standard or 3D RT.

41. Biologically equivalent doses: dose/fraction <2 Gy (30 fractions) Wu et al, IJROBP 2003

42. IMRT: Lower total dose to a critical organ means also lower dose/fraction. Dose/response data from standard or 3D RT, where full dose/fraction (1.8-2.0 Gy) is delivered to the critical organ, may not be relevant to IMRT. We need much more data correlating dose/response in tumor and in normal tissue before we can use IMRT optimally.

43. Illustration of the current process of research in using IMRT to solve a clinical issue

44. Acute and late mucosal/pharyngeal toxicity Pharyngeal toxicity is the main barrier for winning the battle with head and neck cancer (K. Robbins, Editorial, IJROBP 2002).

45. Stricture Aspiration Videofluoroscopy after chemo-RT

46. Aspiration Videofluoroscopy After chemo-RT

47. RT concurrent with gemcitabine for nonresectable HN cancer High rate of tumor control High rate of pharyngeal toxicity, aspirations, and pneumonia The therapeutic index is not satisfactory Eisbruch et al, JCO 2001

48. Improve the theraputic index by physical means IMRT must solve this problem!

49. 65-73 Gy 58-64 Gy 51-57 Gy PTV1 PTV2 PTV3

50. Retrospective comparisons of pharyngeal toxicity following intensive chemo-RT: Standard RT vs. IMRT Mittal et al (ASTRO 2001): IMRT is better Milano et al (ASCO 2003), Garden et al (ASTRO 2003): No difference.

51. Can we use IMRT specifically to reduce dysphagia/aspiration following intensive chemo-RT? Which anatomical structures are affected? –36 different muscles and 10 nerves participate in swallowing and airway protection

52. Which anatomical structures are affected? Videofluoroscopy: Functional abnormalities following chemo-RT –Reduced contraction of the pharyngeal constrictors –Reduced movement of the epiglottis to protect the airway –Reduced coordination between pharyngeal contraction, closure of the larynx, and opening of the upper esophageal sphincter

53. Which structures are affected? CT: Anatomical abnormalities following chemo-RT Pre therapy 3 months post therapy Pharyngeal constrictors Epiglottis Pharyngeal constrictors Epiglottis

54. The pharyngeal constrictors: Affected by chemo- RT both functionally and anatomically

55. Include the non-involved pharyngeal constrictors in the IMRT cost function “Standard IMRT”“Dysphagia/aspiration-specific IMRT”

56. Include the non-involved pharyngeal constrictors and larynx (glottic & epiglottic) in the IMRT cost function “Standard IMRT” “Dysphagia/aspiration-specific IMRT”

57. Reducing the dose to the dysphagia-related structures Will it help the patient swallow better? How much one needs o reduce the dose to gain a clinical benefit? What is the important dosimetric parameter? –Mean dose to the structure? –Maximal dose? –Volume receiving a specified dose?

58. Optimization using biologic cost functions vs. Dose-volume Biologic cost functions use models describing the presumed effects of the non-uniform dose distribution – Equivalent Uniform Dose (EUD) –Normal Tissue Complication Probability (NTCP) –Tumor Control Probability (TCP)

59. PTV60 Ipsilateral Parotid Vineberg et al

60. Achieving homogeneous target dose distributions is feasible with dose-based or biological-based optimization if enough flexibility is provided to the optimizer. The same tradeoffs are made between target and normal tissue, regardless of the type of cost function. Biologic cost functions

61. Once the parameters of the biologic cost function, derived from clinical data, are known with some confidence, they will be superior to dose/volume cost functions for optimization. Biologic cost functions

62. Conclusions Data regarding the relationships between clinical endpoints like tumor control or toxicity risks and the distribution of radiation doses and volumes, are essential for an intelligent decision making or model building for optimization.

63. Conclusions Much of these data do not yet exist Some of these data will never be available due to their complex nature and the inability to express them in physical/statistical terms.