- Email: [email protected]
Consensus Contours for CT Versus MRI in Image-Based Brachytherapy for Cervix Cancer to Generate an RTOG Atlas
S30
International Journal of Radiation Oncology Biology Physics
evaluable patients were excluded from further analysis. Toxicities were scored by CTC v 3.0. LAP07 final analysis will be done in March 2013. Results: Among the 442 included patients, 135 patients have been randomized in the CRT arm. Fifteen patients did not receive CRT. Five patients were not evaluable. RTQAs of the 115 patients were 24% PP, 50% MID and 16% MAD. MAD were mainly PTV coverage inhomogeneities (Dmin < 90% or Dmax > 110%), non-respect of required margins around the GTV, and a too high dose (V30 > 45%) received by the liver. Impact of RTQAs on survival and toxicity will be presented at the meeting. Conclusions: Quality assurance of complex radiation therapy was better than in previous large studies but needs improvement. Its impact on toxicity and outcome will be presented. Author Disclosure: F. Huguet: F. Honoraria; Roche. S. Racadot: None. D. Goldstein: None. N. Spry: None. J. Van Laethem: None. P. Van Houtte: None. B. Glimelius: None. M. Gubanski: None. F. Bonnetain: None. P. Hammel: F. Honoraria; Roche. G. Consultant; Roche.
and W. Bosch10; 1Brigham & Women’s Hospital/Dana-Farber Cancer Institute, Boston, MA, 2Medical College of Wisconsin, Milwaukee, WI, 3University of Pittsburgh Cancer Institute, Pittsburgh, PA, 4 London Health Sciences Centre and Western University, London, ON, Canada, 5University of North Carolina, Chapel Hill, NC, 6 Miller School of Medicine, Miami, FL, 7The Robert H. Lurie Comprehensive Cancer Center of Northwestern University, Chicago, IL, 8University of California, San Diego, CA, 9University of Utah Huntsman Cancer Hospital, Salt Lake City, UT, 10Washington University, St. Louis, MO
70 Trends in the Utilization of Brachytherapy in Cervical Cancer in the United States K. Han,1 M. Milosevic,1 A. Fyles,1 M. Pintilie,1 and A. Viswanathan2; 1 Princess Margaret Hospital, Toronto, ON, Canada, 2Brigham and Women’s Hospital, Boston, MA Purpose/Objective(s): To determine the trends in brachytherapy utilization in cervical cancer in the United States, and identify factors and survival benefit associated with brachytherapy treatment. Materials/Methods: Using the Surveillance, Epidemiology, and End Results (SEER) database, 7,359 patients with stage IB2-IVA cervical cancer treated with external beam radiation therapy (EBRT) between 1988 and 2009 were identified. Logistic regression analysis was performed on potential factors associated with brachytherapy use: age, marital status, race, ethnicity, metropolitan status, year of diagnosis, SEER region, histology, grade, and stage. Propensity-score matching was used to adjust for differences between patients who received brachytherapy and those who did not from 2000 onwards (after the National Cancer Institute alert recommending concurrent chemotherapy). Causespecific survival (CSS) and overall survival (OS) of matched patients were estimated using the Kaplan-Meier method, and compared using the log-rank test. Multivariate survival analysis was performed using the Cox proportional hazard model. Results: Sixty-three percent of the 7,359 women received brachytherapy in combination with EBRT and 37% received EBRT alone. The brachytherapy utilization rate has decreased from 83% in 1988 to 58% in 2009 (p < .001), with a sharp decline of 23% in 2003 to 43%. Factors associated with higher odds of brachytherapy use included younger age, married (versus single), earlier years of diagnosis, earlier stage and certain SEER regions. In the propensity-score matched cohort, brachytherapy treatment was associated with higher 5-year CSS (61.5% vs 48.7%, p < .001) and OS (54.8% vs 42.5%, p < .001). Brachytherapy treatment was independently associated with better CSS (HR 0.64; 95% CI, 0.57 to 0.71), and OS (HR 0.66; 95% CI, 0.60 to 0.74). Conclusions: This population-based analysis reveals a concerning decline in brachytherapy utilization, and significant geographic disparities in the delivery of brachytherapy in the U.S. Brachytherapy use is independently associated with significantly higher CSS and OS, and should be implemented in all feasible cases. Author Disclosure: K. Han: J. Funding Other; Fellowship salary support from the Canadian Association of Radiation Oncology-Elekta research fellowship and Canadian Institute of Health Research (CIHR) EIRR21 award. M. Milosevic: None. A. Fyles: None. M. Pintilie: None. A. Viswanathan: None.
71 Consensus Contours for CT Versus MRI in Image-Based Brachytherapy for Cervix Cancer to Generate an RTOG Atlas A.N. Viswanathan,1 B. Erickson,2 S. Beriwal,3 D.P. D’Souza,4 E.L. Jones,5 L. Portelance,6 W. Small,7 C.M. Yashar,8 D.K. Gaffney,9
Purpose/Objective(s): To generate a consensus RTOG atlas for CT and for MR image-based cervical cancer brachytherapy by analyzing variability in contouring. Materials/Methods: RTOG Gynecologic Cancer Steering Committee members received 3 cervical cancer cases to contour. Each case had a 3T MRI at diagnosis, an MRI and a CT performed at the time of brachytherapy within an hour of insertion, and clinical drawings based on the MRI at diagnosis and at the time of brachytherapy. Instructions mandated that CT contouring should be done first without viewing the MRI at the time of brachytherapy. For CT contouring, physicians drew a CTV-cervix that included the cervix and any notable parametrial extension at the time of brachytherapy, but not the entire parametrial region if not involved. The cervix contours started at the level of the applicator. MR contouring followed the GEC ESTRO guidelines for delineating an HR-CTV. The clinical cases were then analyzed for consistency and clarity of target delineation using an expectation maximization algorithm for simultaneous truth and performance level estimation (STAPLE), with kappa statistics as a measure of agreement between participants. The conformity index (CI), defined as the ratio between the common (mean) and encompassing (union) volume of a given pair of contours was calculated for each of the six data sets. Dice coefficients were generated to compare CT and MR contours of the same case. The Dice coefficient is the volume of the intersection divided by the average volume of the two structures. Results: A total of 23 physicians contoured as part of this protocol. Results are shown in the Table. The mean tumor volume was smaller on MR than CT (p < 0.05 for all 3 cases). Kappa estimates showed substantial agreement among physicians’ contours and were slightly higher for CT. Sensitivity and specificity were similar between CT and MR, indicating very little apparent difference in contours. Conformity index was slightly higher for CT compared to MR, indicating a higher level of agreement on CT. Dice coefficients of the 95% consensus volumes comparing CT to MR were 59% for case 1, 71% for case 2 and 88% for case 3. Conclusions: MRI contoured volumes are consistently smaller than CT volumes. However, CT has a higher level of agreement that may be due to the more distinct contrast visible on the images at the time of brachytherapy. A 95% consensus volume was generated for CT and for MR and separate online atlases will be created based on these results. Author Disclosure: A.N. Viswanathan: None. B. Erickson: None. S. Beriwal: None. D.P. D’Souza: None. E.L. Jones: None. L. Portelance: None. W. Small: None. C.M. Yashar: None. D.K. Gaffney: None. W. Bosch: None.
Oral Scientific Abstract 71; Table Structure Measure
Case 1 CT
Case 1 MR
Case 2 CT
Case 2 MR
Case 3 CT
Case 3 MR
Sensitivity Specificity Vol. Mean (cc) Vol. Std. Dev. (cc) STAPLE Vol. (cc) Kappa
0.684 0.977 33.70 10.10 39.16 0.65
0.729 0.978 16.20 6.26 17.50 0.64
0.652 0.998 55.16 16.10 77.63 0.71
0.720 0.985 39.77 11.93 45.21 0.67
0.749 0.983 59.36 15.45 69.20 0.70
0.660 0.996 44.54 15.54 59.68 0.66
International Journal of Radiation Oncology Biology Physics
evaluable patients were excluded from further analysis. Toxicities were scored by CTC v 3.0. LAP07 final analysis will be done in March 2013. Results: Among the 442 included patients, 135 patients have been randomized in the CRT arm. Fifteen patients did not receive CRT. Five patients were not evaluable. RTQAs of the 115 patients were 24% PP, 50% MID and 16% MAD. MAD were mainly PTV coverage inhomogeneities (Dmin < 90% or Dmax > 110%), non-respect of required margins around the GTV, and a too high dose (V30 > 45%) received by the liver. Impact of RTQAs on survival and toxicity will be presented at the meeting. Conclusions: Quality assurance of complex radiation therapy was better than in previous large studies but needs improvement. Its impact on toxicity and outcome will be presented. Author Disclosure: F. Huguet: F. Honoraria; Roche. S. Racadot: None. D. Goldstein: None. N. Spry: None. J. Van Laethem: None. P. Van Houtte: None. B. Glimelius: None. M. Gubanski: None. F. Bonnetain: None. P. Hammel: F. Honoraria; Roche. G. Consultant; Roche.
and W. Bosch10; 1Brigham & Women’s Hospital/Dana-Farber Cancer Institute, Boston, MA, 2Medical College of Wisconsin, Milwaukee, WI, 3University of Pittsburgh Cancer Institute, Pittsburgh, PA, 4 London Health Sciences Centre and Western University, London, ON, Canada, 5University of North Carolina, Chapel Hill, NC, 6 Miller School of Medicine, Miami, FL, 7The Robert H. Lurie Comprehensive Cancer Center of Northwestern University, Chicago, IL, 8University of California, San Diego, CA, 9University of Utah Huntsman Cancer Hospital, Salt Lake City, UT, 10Washington University, St. Louis, MO
70 Trends in the Utilization of Brachytherapy in Cervical Cancer in the United States K. Han,1 M. Milosevic,1 A. Fyles,1 M. Pintilie,1 and A. Viswanathan2; 1 Princess Margaret Hospital, Toronto, ON, Canada, 2Brigham and Women’s Hospital, Boston, MA Purpose/Objective(s): To determine the trends in brachytherapy utilization in cervical cancer in the United States, and identify factors and survival benefit associated with brachytherapy treatment. Materials/Methods: Using the Surveillance, Epidemiology, and End Results (SEER) database, 7,359 patients with stage IB2-IVA cervical cancer treated with external beam radiation therapy (EBRT) between 1988 and 2009 were identified. Logistic regression analysis was performed on potential factors associated with brachytherapy use: age, marital status, race, ethnicity, metropolitan status, year of diagnosis, SEER region, histology, grade, and stage. Propensity-score matching was used to adjust for differences between patients who received brachytherapy and those who did not from 2000 onwards (after the National Cancer Institute alert recommending concurrent chemotherapy). Causespecific survival (CSS) and overall survival (OS) of matched patients were estimated using the Kaplan-Meier method, and compared using the log-rank test. Multivariate survival analysis was performed using the Cox proportional hazard model. Results: Sixty-three percent of the 7,359 women received brachytherapy in combination with EBRT and 37% received EBRT alone. The brachytherapy utilization rate has decreased from 83% in 1988 to 58% in 2009 (p < .001), with a sharp decline of 23% in 2003 to 43%. Factors associated with higher odds of brachytherapy use included younger age, married (versus single), earlier years of diagnosis, earlier stage and certain SEER regions. In the propensity-score matched cohort, brachytherapy treatment was associated with higher 5-year CSS (61.5% vs 48.7%, p < .001) and OS (54.8% vs 42.5%, p < .001). Brachytherapy treatment was independently associated with better CSS (HR 0.64; 95% CI, 0.57 to 0.71), and OS (HR 0.66; 95% CI, 0.60 to 0.74). Conclusions: This population-based analysis reveals a concerning decline in brachytherapy utilization, and significant geographic disparities in the delivery of brachytherapy in the U.S. Brachytherapy use is independently associated with significantly higher CSS and OS, and should be implemented in all feasible cases. Author Disclosure: K. Han: J. Funding Other; Fellowship salary support from the Canadian Association of Radiation Oncology-Elekta research fellowship and Canadian Institute of Health Research (CIHR) EIRR21 award. M. Milosevic: None. A. Fyles: None. M. Pintilie: None. A. Viswanathan: None.
71 Consensus Contours for CT Versus MRI in Image-Based Brachytherapy for Cervix Cancer to Generate an RTOG Atlas A.N. Viswanathan,1 B. Erickson,2 S. Beriwal,3 D.P. D’Souza,4 E.L. Jones,5 L. Portelance,6 W. Small,7 C.M. Yashar,8 D.K. Gaffney,9
Purpose/Objective(s): To generate a consensus RTOG atlas for CT and for MR image-based cervical cancer brachytherapy by analyzing variability in contouring. Materials/Methods: RTOG Gynecologic Cancer Steering Committee members received 3 cervical cancer cases to contour. Each case had a 3T MRI at diagnosis, an MRI and a CT performed at the time of brachytherapy within an hour of insertion, and clinical drawings based on the MRI at diagnosis and at the time of brachytherapy. Instructions mandated that CT contouring should be done first without viewing the MRI at the time of brachytherapy. For CT contouring, physicians drew a CTV-cervix that included the cervix and any notable parametrial extension at the time of brachytherapy, but not the entire parametrial region if not involved. The cervix contours started at the level of the applicator. MR contouring followed the GEC ESTRO guidelines for delineating an HR-CTV. The clinical cases were then analyzed for consistency and clarity of target delineation using an expectation maximization algorithm for simultaneous truth and performance level estimation (STAPLE), with kappa statistics as a measure of agreement between participants. The conformity index (CI), defined as the ratio between the common (mean) and encompassing (union) volume of a given pair of contours was calculated for each of the six data sets. Dice coefficients were generated to compare CT and MR contours of the same case. The Dice coefficient is the volume of the intersection divided by the average volume of the two structures. Results: A total of 23 physicians contoured as part of this protocol. Results are shown in the Table. The mean tumor volume was smaller on MR than CT (p < 0.05 for all 3 cases). Kappa estimates showed substantial agreement among physicians’ contours and were slightly higher for CT. Sensitivity and specificity were similar between CT and MR, indicating very little apparent difference in contours. Conformity index was slightly higher for CT compared to MR, indicating a higher level of agreement on CT. Dice coefficients of the 95% consensus volumes comparing CT to MR were 59% for case 1, 71% for case 2 and 88% for case 3. Conclusions: MRI contoured volumes are consistently smaller than CT volumes. However, CT has a higher level of agreement that may be due to the more distinct contrast visible on the images at the time of brachytherapy. A 95% consensus volume was generated for CT and for MR and separate online atlases will be created based on these results. Author Disclosure: A.N. Viswanathan: None. B. Erickson: None. S. Beriwal: None. D.P. D’Souza: None. E.L. Jones: None. L. Portelance: None. W. Small: None. C.M. Yashar: None. D.K. Gaffney: None. W. Bosch: None.
Oral Scientific Abstract 71; Table Structure Measure
Case 1 CT
Case 1 MR
Case 2 CT
Case 2 MR
Case 3 CT
Case 3 MR
Sensitivity Specificity Vol. Mean (cc) Vol. Std. Dev. (cc) STAPLE Vol. (cc) Kappa
0.684 0.977 33.70 10.10 39.16 0.65
0.729 0.978 16.20 6.26 17.50 0.64
0.652 0.998 55.16 16.10 77.63 0.71
0.720 0.985 39.77 11.93 45.21 0.67
0.749 0.983 59.36 15.45 69.20 0.70
0.660 0.996 44.54 15.54 59.68 0.66