Between August 23, 1994 and October 30, 2000, 1084 cases were registered on this study. All follow-up information on patients submitted and data entered at RTOG headquarters through October 2007 is included in this analysis. Pretreatment patient characteristics and median follow-up at time of this analysis are listed in . The median follow-up for patients treated in level I exceeds 12 years and for level V it exceeds 6 years. The mean ICRU reference doses delivered to patients in each patient grouping and prescription dose level for the current study are listed in . Because minimum dose prescriptions to the PTV (CTV on level III) were used, the ICRU reference doses are approximately 4% greater than the minimum PTV dose. These are actual doses delivered based upon final dosimetry review by the RTOG 3D QA Center.
The protocol allowed treating physicians to use a range 5 to 10 mm for the CTV to PTV margins. Institutions were encouraged to systematically evaluate their setup error and internal organ motion for their patient population, localization and immobilization methods. For dose level I, II, III, IV, and V the average PTV margins were 6.96mm, 7.00mm, 6.92mm, 7.49mm and 7.76mm, respectively. The margins used in the latter dose levels were statistically significantly larger than the margin used in the earlier arms of the study (p<0.05). For this reason we examined both PTV margin and the volume of the high dose PTV (PTV-HD) in the multivariate analyses examining toxicity by treatment arm.
The primary endpoint of this study is the incidence of RTOG late grade 3 or worse GI or GU toxicity. In this report the incidence of GI and GU toxicities are reported collectively and separately to allow comparison with other published series. The incidences of late grade 3 or worse GI or GU toxicity in group 1 were 3%, 4%, 6%, 7% and 9% and those for group 2 were 6%, 2%, 6%, 9%, and 12% at dose levels I, II, III, IV and V, respectively. Group 3 patients who were treated with dose level II showed 6% of late grade 3 or worse GI/GU toxicity. The incidences of grade 3 or worse GI toxicity are 1%, 0%, 1%, 3% and 4% for group 1 and 3%, 0%, 3%, 2% and 7% for group 2 at dose levels I through V. The incidences of grade 3 or worse GU toxicity are 1%, 4%, 5%, 3% and 5% for group 1 and 3%, 2%, 3%, 6% and 6% for group 2.
The rates of late grade 3 toxicities are not significantly large enough to be modeled. The incidences of RTOG late grade 2 or worse GI, GU, and GI or GU toxicity were analyzed instead. For levels I to III (1.8Gy fraction size) the incidence of grade 2 or worse GI or GU toxicity ranges from 21 to 31% in groups 1 and 2 and 44% for group 3 treated to dose level II. For dose levels IV and V (2.0 Gy fraction size), the incidence of grade 2 or worse GI or GU toxicity ranged from 32% to 44% for groups 1 and 2. The incidences of grade 2 or worse GI toxicity are 9%, 7%, 11%, 10% and 25% for group 1 and 13%, 9%, 14%, 16% and 26% for group 2 at dose levels I through V. The incidences of grade 2 or worse GU toxicity are 24%, 22%, 18%, 29% and 23% for group 1 and 19%, 16%, 21%, 21% and 28% for group 2. and show the cumulative incidences of time to late grade 2 or worse GI, GI, and GI or GU toxicity for disease groups 1 and 2, respectively. There is a significantly higher incidence of grade 2 or worse GI toxicity for patients receiving 78Gy (p=0.0001 for group 1 and p=0.0063 for group 2). There is no similar increase in grade 2 or worse GU toxicity (p=0.2067 for group 1 and p=0.2021 for group 2). As for grade 2 or worse GI or GU toxicity, there is a significantly higher incidence of toxicity for the 78 Gy level (p=0.0005 for group 1 and p=0.0016 for group 2).
Figure 1a. Disease group 1, Time to Late GI Grade 2+ (The cumulative incidence method is used to estimate the time to late GI grade 2+ failure rate and the Gray's test is used to test the difference between the levels in this univariate analysis.)
Figure 2a. Disease group 2, Time to Late GI Grade 2+ (The cumulative incidence method is used to estimate the time to late GI grade 2+ failure rate and the Gray's test is used to test the difference between the levels in this univariate analysis.)
Patients treated with dose level V have a higher probability of grade 2 or worse GI toxicity but no significant increase in grade 2 or worse GU toxicity compared to levels I, II, III and IV in group 1 and levels II, III, and IV in group 2 patients. In the disease group 1, patients who are treated with dose level V have higher probability of grade 2 or worse late GU or GI toxicity compared to dose level I, II, and III (). Also in the disease group 2, patients who are treated with dose level V have higher probability of grade 2 or worse late GU or GI toxicity compared to dose level II, III, and IV. These differences are primarily related to GI toxicity.
Univariate Cox Proportional Hazard Model-Time to Late GI/GU Grade 2+ Toxicity in Disease Group
We analyzed these results in a multivariate analysis which adjusted for age, induction hormone, PTVHD volume () or 3D margin between GTV and PTVHD(). The results remained the same in groups 1 and 2. While in some instances the magnitude of the margins or the size of the PTV-HD volume may have contributed to an increase in GI or GU toxicity, the effect of the 78Gy in 2.0Gy day fractions remained a significant factor for toxicity.
Multivariate Cox Proportional Hazards Models1 within Groups- Time to Late GI/GU Grade 2+ Toxicity
Multivariate Cox Proportional Hazards Models1 within Groups -Time to Late GI/GU Grade 2+ Toxicity
In group 3, we did not compare the two dose levels (I vs. II) since there are only 4 patients in dose level I. In a univariate Cox proportional hazard model, patients in group 3 did experience a higher rate of grade 2 or worse GI toxicity than patients in groups 1 or 2 (HR 3.87, p=0.0017). There were no significant covariates associated with late grade 2 or worse GI or GU toxicity for group 3 patients.