1. Clinical decisions in the optimization process I. Emphasis on tumor control issues Avi Eisbruch University of Michigan

2. Standard radiotherapy Limited ability to control dose distribution across the treated volume Large body of past experience that dictates standard care regarding doses prescribed to tumors, dose homogeneity, and maximal doses allowed to critical organs.

3. Optimized intensity modulated radiotherapy Vastly improved ability to control dose distributions. How should we “optimize” dose distributions? –The limiting factor is our knowledge of what is clinically desired.

4. Optic nerve damage Following definitive, standard RT of paranasal sinus cancer: 21/78 (27%) patients had unilateral blindness and 4 (5%) had bilateral blindness. –Katz, Mendenhall, et al. Head Neck 2002.

5. Martel et al, IJROBP 1997 3D conformal RT of paranasal sinus cancer

6. Optic nerve damage following conformal-3D RT of paranasal sinus cancer 2 cases of blindness out of 20 patients treated definitively and were free of disease >2 years. Parameters for normal tissue complication probabilities (NTCP) were derived from the dosimetric and the clinical data. Martel et al, 1997

7. IMRT of paranasal sinus cancer: sparing the optic nerves “ralatively easy case” Tsien, Eisbruch, et al, IJROBP 2003

8. Paranasal sinus cancer: sparing the optic nerves Tsien, Eisbruch, et al, IJROBP 2003

9. Paranasal sinus cancer: sparing the optic nerves This was more difficult ! Tsien, Eisbruch, et al, IJROBP 2003

10. Paranasal sinus cancer: sparing the optic nerves Tsien, Eisbruch, et al, IJROBP 2003

11. Sparing the optic nerves by IMRT While IMRT provides highly conformal dose distributions, trade-offs between organ sparing and target under-dose do exist. An intelligent choice between these trade- offs cannot be made unless the parameters of tumor control and complication probability models are known with a higher certainty.

12. Tumor control vs. Dose The most elusive issue due to the large heterogeneity in the density of tumor stem cells and their sensitivity to radiation. The doses we prescribe in order to kill tumors are dictated by the presumed sensitivity of the adjacent normal tissue, rather than by the needs to kill the tumor.

13. A recent example Locally advanced cancer of the nasopharynx Most of the planning treatment volume for the gross, observed disease was encompassed by the desired dose (70 Gy) The parts of the tumor adjacent to the brain stem received less than the desired dose to reduce the risk of a severe complication.

22. Assessment of the sites of recurrences relative to the dose delivered

23. CT at time of Recurrence recurrence volume

24. Transfer of Recurrence Volume Planning CT recurrence volume location 50 Gy 60 Gy

25. Recurrence Definitions % recurrence volume receiving minimal intended dose (70 Gy to GTV, 60 to resection bed or high-risk CTV, 50 to low-risk CTV): In field  95% Marginal20 - 95% Outside field  20%

26. 0 20 40 60 80 Mean Dose Dose (Gy) Mean Dose to Recurrence Volumes Recurrence

27. 0 20 40 60 80 Dose (Gy) Minimum Dose to Recurrence Volumes Recurrence Minimum Dose

28. Rouvière, Anatomy of the Human Lymphatic System, Edwards Bros., 1938

29. Som et al., Arch. Otolaryngol. Head Neck Surg.1999 Neck Node Levels

30. Update: Pattern of failure 160 patients treated with IMRT since 1994 24 local-regional failures 20 in-field All 4 marginal failures can be explained by the clinical decisions that had been made regarding the targets.

31. Most failures are in-field Should we escalate the dose? Would dose inhomogeneities (“hot spots”) within the targets help improve tumor control? –Fowler, Deasy: modeling: Over-dosage in part of the target may compensate for volumes of underdosage within the target (as long as these volumes are not too large)

32. Xia et al, IJROBP 2000 Over-dosing the GTV

33. In addition to the nominal dose… “Hot volumes” in a highly nonuniform plan are delivered at a higher dose/fraction than the prescribed dose. The biologically equivalent dose in the hot volume is higher than the DVH implies –80 Gy over 35 treatments: 2.3 Gy/fraction, NTD=85 Gy.

34. Where are the “hot spots” located

35. Vineberg et al.

36. Tumor Control Probability vs. Treatment Delivery Time JZ Wang et al, IJROBP 2003

37. Should limiting treatment time be incorporated into the treatment optimization process?

38. Interim conclusions The pattern of recurrence in our and other series suggests that dose escalation to all or parts of the targets at highest risk of failure may improve tumor control. This is supported by models of tumor control probability (TCP) Clinical validation is still lacking.