Int. J. Radiation Oncology Biol. Phys., Vol. 53, No. 3, pp. 766 –774, 2002 Copyright © 2002 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/02/$–see front matter
PII S0360-3016(02)02752-9
PHYSICS CONTRIBUTION
IMPROVING THE CONSISTENCY IN CERVICAL ESOPHAGEAL TARGET VOLUME DEFINITION BY SPECIAL TRAINING PATRICIA TAI, M.B.,* JAKE VAN DYK, M.SC.,* JERRY BATTISTA, PH.D.,* EDWARD YU, PH.D., M.D.,* LARRY STITT, M.SC.,* JON TONITA, M.SC.,† OLUSEGUN AGBOOLA, M.D.,‡ JAMES BRIERLEY, M.D.,§ RASHID DAR, M.B.,* CHRISTOPHER LEIGHTON, M.D.,㛳 SHAWN MALONE, M.D.,‡ BARBARA STRANG, M.D.,¶ PAULINE TRUONG, M.D.,* GREGORY VIDETIC, M.D.,* C. SHUN WONG, M.D.,§ REBECCA WONG, M.B.,# AND YOUSSEF YOUSSEF, M.D.** *Department of Oncology, London Regional Cancer Center, London, Ontario, Canada; †Epidemiology Department, Allan Blair Cancer Center, Regina, Saskatchewan, Canada; ‡Department of Radiation Oncology, Ottawa Regional Cancer Center, Ottawa, Ontario, Canada; § Department of Radiation Oncology, Princess Margaret Hospital, Toronto, Ontario, Canada; 㛳Department of Radiation Oncology, Windsor Regional Cancer Center, Windsor, Ontario, Canada; ¶Department of Radiation Oncology, Hamilton Regional Cancer Center, Hamilton, Ontario, Canada; #Department of Radiation Oncology, Toronto-Bayview Regional Cancer Center, Toronto, Ontario, Canada; **Department of Radiation Oncology, Kingston Regional Cancer Center, Kingston, Ontario, Canada Purpose: Three-dimensional conformal radiation therapy requires the precise definition of the target volume. Its potential benefits could be offset by the inconsistency in target definition by radiation oncologists. In a previous survey of radiation oncologists, a large degree of variation in target volume definition of cervical esophageal cancer was noted for the boost phase of radiotherapy. The present study evaluated whether special training could improve the consistency in target volume definitions. Methods and Materials: A pre-training survey was performed to establish baseline values. This was followed by a special one-on-one training session on treatment planning based on the RTOG 94-05 protocol to 12 radiation oncologists. Target volumes were redrawn immediately and at 1–2 months later. Post-training vs. pre-training target volumes were compared. Results: There was less variability in the longitudinal positions of the target volumes post-training compared to pre-training (p < 0.05 in 5 of 6 comparisons). One case had more variability due to the lack of a visible gross tumor on CT scans. Transverse contours of target volumes did not show any significant difference pre- or post-training. Conclusion: For cervical esophageal cancer, this study suggests that special training on protocol guidelines may improve consistency in target volume definition. Explicit protocol directions are required for situations where the gross tumor is not easily visible on CT scans. This may be particularly important for multicenter clinical trials, to reduce the occurrences of protocol violations. © 2002 Elsevier Science Inc. Esophageal cancer, Target volume definition, Training, Quality assurance, Three-dimensional (3D) radiation treatment planning.
INTRODUCTION
volume definition of cervical esophageal tumor among 48 radiation oncologists (1). The data were obtained by measuring the transverse diameter and length of volumes hand-drawn by the responding radiation oncologists on the computerized tomography (CT) scan films using a fine-tip pen. For the sites of brain, bladder, and prostate cancers, other investigators also found a great variation in target volume delineation (2– 4). The hypothesis of this study is
Accurate and reproducible target volume definition is a prerequisite for three-dimensional conformal radiation therapy (3D-CRT). The use of 3D-CRT reduces the dose to the normal tissues and may allow dose escalation to the tumor, thus possibly improving the clinical outcome. This approach requires precise definition of the target volume. In 1996, our team completed a trans-Canada survey that revealed a large degree of variation of target
Ontario, Division of Research and Education. Acknowledgment—The authors thank Barbara Barons for her secretarial assistance and Matthew Schmid for his assistance with graphs. Received Jul 24, 2001, and in revised form Jan 10, 2002. Accepted for publication Jan 16, 2002.
Reprint requests to: Patricia Tai, Radiation Oncology Department, Allan Blair Cancer Center, 4101 Dewdney Avenue, Regina, SK S4T 7T1 Canada. Tel: (306) 766-2206; Fax: (306) 766-2845; E-mail:
[email protected] Presented in part at the first UK Radiation Oncology Conference (UKRO 1), 23–25 April 2001, York, UK. Supported by Research Fellowship Award from Cancer Care 766
Target volume definition
● P. TAI et al.
767
Table 1. Elements of the training process 1. 2. 3. 4. 5.
ICRU 50 definitions Summary of the variation in the 1996 study Design of the present study Normal anatomy of esophagus Definition of boost target volumes for concurrent chemotherapy and radiation, based on RTOG 94-05 protocol; Longitudinal Transverse
CTVII GTV ⫹ 1 cm PTVII GTV ⫹ 2 cm Note: Nonuniform margins are permitted to avoid critical structures. 6. Demonstration films of a sample case 7. Question-and-answer period
GTV ⫹ 1 cm GTV ⫹ 2 cm
Abbreviations: GTV ⫽ gross tumor volume; CTVII ⫽ boost phase clinical target volume; PTVII ⫽ boost phase planning target volume.
that variation in treatment practice can be minimized by clear presentation of protocols through training sessions. A special training session was arranged and the variation of target volume definition by radiation oncologists before and after the training was evaluated to see if the objective of reducing the variation can be achieved. This is relevant to clinical practice as well as for quality assurance of multicenter trials. METHODS AND MATERIALS The details of the study methodology were described in our previous report (1). Briefly, a questionnaire was developed using three cervical esophageal cancer cases to obtain target volume information as determined from CT scans. All 62 qualified radiation oncologists who treat esophageal cancers in Canada were eligible to participate in the first study in 1996. One of the three cases, each representing a different stage of cervical esophageal cancer, was randomly selected and mailed to the 58 radiation oncologists who had agreed to participate. Case A was an advanced carcinoma that closely abutted the aorta. Case B was an early cancer such that, after biopsy, no gross tumor was seen on CT imaging. Case C was intermediate in tumor bulk; gross tumor was seen in the esophageal wall. These cases with different disease stages were included as a representation of different situations encountered clinically. Radiation oncologists were asked to complete a questionnaire regarding treatment techniques and to outline “boost” target volumes on CT scans using the International Commission on Radiation Units and Measurements Report 50 (ICRU 50) definitions (5): gross tumor volume (GTV), boost phase clinical target volume (CTVII), and boost phase planning target volume (PTVII). The mailing kit consisted of a questionnaire, detailed instructions, the ICRU target volume definitions, patient history, physical findings, full radiology and endoscopy reports, barium swallow films, magnified planning CT scan films, a fine-point pen, and self-addressed return envelope. Planning CT scans were selected at 1.2-, 1.0-, 1.0-cm slice intervals, for Case A, B, and C respectively, instead of providing all slices at closer intervals to
allow radiation oncologists to finish within a reasonable time. In 1998, special training was given to 12 radiation oncologists in six Southern Ontario centers. These radiation oncologists had participated in the drawing of target volumes in 1996 and were a subset of the Trans-Canada survey (1). They were chosen due to geographic proximity from the London Regional Cancer Center, where the trainer was located. To examine whether these 12 Ontario radiation oncologists were representative of all of the radiation oncologists in the 1996 study, the variability of the target volume definition for physicians from Southern Ontario was compared with those from the rest of Canada using the 1996 survey data. There was no statistically significant difference in the target volume definition variability between Southern Ontario radiation oncologists and other radiation oncologists surveyed in 1996 (p ⬎ 0.35, 0.24, 0.22 respectively for Case A, B, and C, unpaired t test) (6). A training package was developed to aid consistency in boost target volume definition, based on the Radiation Therapy Oncology Group (RTOG) 94-05 protocol recommendations (Table 1). RTOG 94-05 is a concurrent chemoradiation trial to compare 50.4 Gy/28 fractions/5.5 weeks with and without a cone down boost of 14.4 Gy/8 fractions/1.5 weeks. Our present study consisted of three steps: Step 1: Pre-training Radiation oncologists were asked to define target volumes before training to establish baseline values. The mailing package was the same as in the previous study. Radiation oncologists were given the same case as was assigned randomly in 1996, to compare consistency and to assess the impact of training. Step 2: Training Special training in target volume definition was given. One of the investigators (P.T.) met with the participating radiation oncologists in person and provided a 10-min standardized training session (Table 1). The concepts of GTV, CTVII, and PTVII according to ICRU 50 were reviewed. The CTVII is not intended to cover supracla-
768
I. J. Radiation Oncology
● Biology ● Physics
vicular lymph nodes. Periesophageal lymph nodes are potentially at risk and can be included with acceptable morbidity. PTVII includes the CTVII with a margin for geometric variation in setup or movement in the boost phase. Illustrative schematic diagrams were shown to the participants, as in our previous publication (1). A summary of the variation encountered in the 1996 study was given. The ratio of the 95 to 5 percentiles of GTV, CTVII, PTVII diameters range from 1.5 to 2.6. The ratio of the 95 to 5 percentiles of GTV, CTVII, PTVII lengths range from 1.9 to 5.0. They were reminded that the purpose of the present study was to assess the differences in target volumes drawn before, immediately after, and at 1 to 2 months following the training. Those given Case B were informed that there was no gross tumor seen on CT images after a generous biopsy. They were given a detailed anatomic diagram from a standard radiotherapy textbook (7) that showed the relationships of vertebral body levels to other anatomic structures. The vertebral level is an easy landmark for localization in barium swallow and scout view of CT scan. Relationships to aorta, tracheal bifurcation, and diaphragm were also in the same figure. The RTOG 94-05 protocol with required margins was shown: CTVII ⫽ GTV ⫹ 1 cm margin, PTVII ⫽ GTV ⫹ 2 cm margin all around (Table 1). Radiation oncologists were asked to use this specific instruction for target volume definition of the study cases as opposed to the general ICRU 50 definition. Finally, a completed planning CT scan film with boost target volumes of a sample case was shown to the oncologists. The sample case was intentionally different from any of the 3 cases chosen for the study. At the end of the session, a question-and-answer period was provided for further clarification on the tutorial. The radiation oncologists were asked to draw the target volumes on the planning CT scans immediately after this training session. Step 3: One to two months post-training The participating radiation oncologists were asked to redefine target volumes by mail as in step 1 (pre-training) of the study. In this last definition of target volumes, the radiation oncologists were provided with instructions, a summary of the protocol, and a questionnaire in the mail package. They were also asked for comments on radiotherapy planning techniques in the questionnaire. The data of this study were analyzed similarly to those in our previous study (1). First, a comparison between the 1996 and 1998 pre-training results was performed to assess the potential impact of new information and the clinical experience over the elapsed time period. We also made a comparison between the pre- and post-training responses to evaluate the effectiveness of the training. Redrawing the film at 1–2 months after training was intended to see if there was a decay in the learning process and whether the oncologists would revert back to their previous planning habits. For analysis, the superior and inferior slice positions of
Volume 53, Number 3, 2002
the target volumes in the planning CT scan were used. The actual slice position, not the length or variability, was compared by statistical tests. Length was not used as a parameter for comparison because it is less sensitive than the slice positions. While two persons may draw exactly the same length, the positions of these lengths may be different longitudinally. Paired t test was used to calculate the p values for the comparison of individual GTV, CTVII, PTVII results among the different steps. Then the GTV, CTVII, PTVII were pooled together to show if there was any overall difference between the different steps. The t test was used as the distribution of measurements was normally distributed (6).
RESULTS In Tables 2– 4, the ID numbers are not consecutive because the order follows the same number the radiation oncologist got assigned in 1996. ID 1 for Case A is not the same doctor as ID 1 for Case B. The data for 12 radiation oncologists are presented, 4 for each case. One radiation oncologist withdrew from the study after the pre-training step due to personal reasons. The remaining 11 radiation oncologists completed the study and were compliant with respect to instructions. Tables 2– 4 list the data of the superior and inferior CT scan slices on which GTV, CTVII, PTVII were drawn by the individual radiation oncologists. Figure 1 is a graphic representation of the data in Tables 2– 4. Interobserver variations were expressed as standard deviations in Table 5 where an improvement in consistency is shown by a smaller standard deviation and less spread of the target volumes in Fig. 1. Due to the small number of radiation oncologists in this pilot study, we do not define a presumed limit for variability before the training. Intraobserver variations were not measured because the primary question of the study was to improve consistency of target volume definition among radiation oncologists. For Case A (Table 2 and Fig. 1), by 1–2 months posttraining, the oncologists approach conformity on the definition of PTVII longitudinal positions. When the GTV, CTVII, and PTVII were combined as a group for analysis, the pre-training vs. immediate post-training results showed a statistically significant difference in the inferior slice position (p ⫽ 0.05). This means training imparts a statistically significant difference in target volume defined. The longitudinal positions of the target volumes of Case B in Table 3 also show a statistically significant difference when comparing the results of different steps of the study, before and after the training. Actual data for the superior and inferior slices of Case C are presented in Table 4: when the GTV, CTVII, and PTVII are combined as a group for analysis, the pre-training vs. immediate post-training results showed a statistically significant difference in both the superior and inferior slice positions (p ⬍ 0.05). For Case C, the special training
Target volume definition
● P. TAI et al.
769
Table 2. The superior and inferior extent of GTV, CTVII, and PTVII according to the planning CT scan slices for Case A defined by individual radiation oncologists I. Data CT scan slice numbers* Volume
ID†
1996
Pre-training
Immediately post-training
1–2 months post-training
GTV
1 6 9 11 1 6 9 11 1 6 9 11
18–36 21–36 21–33 6–51 9–48 12–48 9–45 6–51 9–48 9–51 9–45 6–51
21–36 15–42 21–33 15–42 9–48 15–42 9–45 15–42 9–48 15–42 9–45 15–42
21–36 15–42 21–33 18–33 18–39 12–45 18–36 15–36 15–42 9–48 15–39 12–39
21–36 21–36 20–35 21–36 18–39 18–39 18–36 18–39 15–42 15–42 15–36 15–42
CTVII
PTVII
II. Comparisons p value Volumes GTV CTVII PTVII GTV, CTVII, PTVII
Comparison
Superior slice position
Inferior slice position
pre- vs. immed post-training immed vs. 1–2 m post-training pre- vs. immed post-training immed vs. 1–2 m post-training pre- vs. immed post-training immed vs. 1–2 m post-training pre- vs. immed post-training‡ immed vs. 1–2 m post-training
0.39 0.30 0.31 0.22 0.82 0.22 0.24‡ 0.02
0.39 0.90 0.16 0.72 0.49 0.50 0.05‡ 0.44
* CT slice spacing was 0.4 cm, only CT slices 3, 6, 9, 12, 15, etc., separated 1.2 cm apart, were given to participating radiation oncologists. † ID ⫽ each radiation oncologist has a confidential identification number. ‡ Two pre- vs. immed post-training comparisons. Abbreviation: immed ⫽ immediate.
conferred uniformity to the longitudinal positions with a smaller standard deviation for the PTVII (Table 5). Data in Tables 2– 4 showed that between the immediate post-training and pre-training target volumes, the variability of the superior and inferior slice positions was less in 5 of the 6 pooled comparisons of GTV, CTVII, PTVII (p ⬍ 0.05), marked by ‡ in the tables. The AP and lateral extents of delineated target volume have been simplified to be expressed as the diameter, because the transverse target volume is almost circular most of the time. Details of the discussion are presented in our previous publication (1). The transverse contours, or diameters, of target volumes did not show a trend toward improved consistency in all three cases and the standard deviations were small (Table 5). One aim of the study was to assess protocol compliance. Table 5 shows that the means of diameters and lengths of CTVII are generally 2 cm bigger than GTV and similarly, PTVII are generally 2 cm bigger
than CTVII. So the radiation oncologists did comply to the margin recommended in Table 1.
DISCUSSION Two aspects of this study are worthy of note: (a) the multidisciplinary study team consisted of clinicians, physicists, and statisticians to develop the training protocol, presentation, and analysis tools; (b) respondents had different years of experience in radiation oncology and represented a spectrum to reflect the reality of clinical practice. Any differences between 1996 and 1998 pre-training results reflect intraobserver variation and possibly the impact of intervening clinical experience, participation in learning activities, or clinical trials during this time interval. Any differences between the 1998 results of pre- and immediately post-training results are likely due to intraobserver variation and teaching. Any differences between the
770
I. J. Radiation Oncology
● Biology ● Physics
Volume 53, Number 3, 2002
Table 3. The superior and inferior extent of GTV, CTVII, and PTVII according to the planning CT scan slices for Case B defined by individual radiation oncologists I. Data CT scan slice numbers* Volume
ID†
1996
Pre-training
Immediately post-training
1–2 months post-training
1 10 13 17 1 10 13 17
n/a 19–31 25–33 11–23 18–35 nd 21–35 11–23 16–37
n/a 13–37 21–33 29–35 9–21 nd 19–35 27–37 7–23
n/a 9–23 25–33 19–27 13–17 9–25 23–35 17–29 11–19
n/a 13–43 23–31 19–27 17–21 nd 21–33 17–29 15–23
GTV CTVII
PTVII
II. Comparisons p value Volumes GTV CTVII PTVII GTV, CTVII, PTVII
Comparison
Superior slice position
Inferior slice position
pre- vs. immed post-training immed vs. 1–2 m post-training pre- vs. immed post-training immed vs. 1–2 m post-training pre- vs. immed post-training immed vs. 1–2 m post-training pre- vs. immed post-training‡ immed vs. 1–2 m post-training
n/a n/a 0.69 0.39 0.90 0.74 0.04‡ 0.35
n/a n/a 0.12 0.35 0.23 0.74 0.02‡ 0.67
* CT slice was spacing 0.5 cm, only CT slices 1, 3, 5, etc. separated 1 cm apart, were given to participating radiation oncologist. † ID ⫽ each radiation oncologist has a confidential identification number. ‡ Two pre- vs. immed post-training comparisons. Abbreviations: n/a ⫽ Case B does not have any gross tumor visible in the CT scan; nd ⫽ not drawn since the dosimetrist will draw PTVII later, based on the CTVII; immed-immediate.
results of immediately after training and 1–2 months later are likely due to intraobserver variation and decay in retention of training. It is possible that reinforcement by a second exposure to the protocol, and redrawing the same case for the fourth time, the respondents became familiar with the case, instructions and films, although there is no clear evidence of this in the data. If interobserver variation increases with time, then training needs to be reinforced. Our data show that at 1–2 months post-training, PTVII of Case A had a smaller standard deviation in Table 5 whereas Case C had the same standard deviation. Hence, the uniformity for PTVII was maintained at 1–2 months after training, with the aid of a summarized handout of the training session. Case B had more variability despite training than the other two cases, due to the lack of a visible gross tumor on CT scans. Cases A and C had an easily defined target volume. To summarize Fig. 1 and Table 5, (a) the training managed to reduce the variation of the longitudinal positions of GTV, CTVII, and PTVII for all cases, as reflected by the standard deviations for these three parameters in Table 5;
(b) the improvement is shown in the longitudinal positions of the target volume only; and (c) generally the effect of the training is maintained after 1–2 months post-training. Currently we do not have any plans to repeat the study for further comparison. Importance of training Major multi-institutional clinical studies still suffer from major to minor protocol deviations. Participating oncologists are often given lengthy written protocols. Despite the use of diagrams, these protocols are sometimes hard to follow. Clinical trials with complicated treatment planning are not likely to be implemented by individual centers. Treatment complexity may also deter accrual of patients if the protocol is not clear to clinicians or to patients. The feedback method currently used in most trials is to check the radiation records and films after completion of treatment. This feedback process is slow. Sometimes protocol violations can result in the patient being ineligible for data analysis. In addition, if each center has an initial long learning curve, accrual can be slow. Improvements in the
Target volume definition
● P. TAI et al.
771
Table 4. The superior and inferior extent of GTV, CTVII, and PTVII according to the planning CT scan slices for Case C defined by individual radiation oncologists I. Data CT scan slice numbers* Volume
ID†
1996
Pre-training
Immediately post-training
1–2 months post-training
GTV
8 17 18 20 8 17 18 20 8 17 18 20
10 8–12 8–14 8–13 9–16 6–17 6–17 7–13 9–16 5–18 6–18 7–13
4–14 8–12 8–14 8–13 3–14 7–13 6–16 7–15 3–14 6–15 5–17 7–15
6–12 8–12 8–12 nd 6–14 7–13 7–13 nd 4–14 6–14 6–14 nd
7–8 8–12 8–12 nd 5–14 7–13 6–14 nd 4–14 6–14 5–15 nd
CTVII
PTVII
II. Comparisons p value Volumes GTV CTVII PTVII GTV, CTVII, PTVII
Comparison
Superior slice position
Inferior slice position
pre- vs. immed post-training immed vs. 1–2 m post-training pre- vs. immed post-training immed vs. 1–2 m post-training pre- vs. immed post-training immed vs. 1–2 m post-training pre- vs. immed post-training‡ immed vs. 1–2 m post-training
0.42 0.42 0.27 0.18 0.18 0.42 0.04‡ 0.35
0.18 0.42 0.42 0.42 0.27 0.42 0.02‡ 0.67
* CT slices 1, 2, 3, 4, etc., separated 1 cm apart, were given to participating radiation oncologists. ID ⫽ each radiation oncologist has a confidential identification number. ‡ Two pre vs. immed post-training comparisons. Abbreviations: nd ⫽ cannot draw due to personal reasons; immed ⫽ immediate. †
protocol write-up are imperative (8). Thus, we assessed the use of special training as a means to improve protocol compliance. Training is important in rare clinical sites. Cervical esophageal cancer is an example. In Canada, the total number of esophageal cancers was 1081, based on the Canadian Cancer Registry in 1996. The cervical and upper thoracic esophageal subsites constituted 12% of all referred esophageal cancer at the London Regional Cancer. The reason for combining cervical and upper thoracic esophageal subsites is because the lesion can extend to involve a long length and it can be difficult to tell if it originates from cervical or upper thoracic esophagus. We must bear in mind that some esophageal cancer patients were never referred to a Cancer Center but treated by surgery as the sole primary modality. Since only 62 radiation oncologists treated the esophageal site in Canada, on average each radiation oncologist treats less than 2 cases per year (1,081 ⫻ 0.12/62) for this subsite of esophagus. Case A has a tumor abutting on the aorta. The volume
of the aorta and mediastinum that needs to be treated remains controversial. For the situation where the radiation oncologists were not given specific instructions for target volume definition when the GTV is not clearly visible on CT scans, the results showed a wider variation as demonstrated by the results of Case B. The teaching package gave general anatomic information only. Providing details of the position of cricopharyngeus muscle in the barium swallow and CT scans and instructions on how to draw the PTV when there is no visible gross tumor volume might help to improve target volume definition for this situation. When such instructions were not given, a wide variation in response was observed. Case C had the clearest GTV and showed a definite reduction of variation by training. One limitation of this pilot study is the small sample size. Based on the variation of the target volume longitudinal positions observed in this pilot study, for Case A 16 radiation oncologists are required to detect a 1-cm difference at each end of the planning target volume (power ⫽ 0.8, ␣ ⬍
772
I. J. Radiation Oncology
● Biology ● Physics
Volume 53, Number 3, 2002
Fig. 1. Graphic representation of the distribution of the gross tumor volume (GTV), clinical target volume (CTVII), planning target volume (PTVII) for boost phase of radiotherapy for the three cases at different time periods: 1996; 1998 step 1 (pre-training); 1998 step 2 (immediately after training); 1998 step 3 (1 to 2 months after training). Individual line within each time period corresponds to each response from the radiation oncologist. For Case B, there is no GTV because there is no gross tumor visible in the CT scan. Some data for PTVII were missing because the dosimetrist will draw PTVII later, based on the CTVII. One radiation oncologist did not complete Case C owing to personal reasons.
Target volume definition
● P. TAI et al.
773
Table 5. Diameters and lengths of Case A, B, and C over time, showing interobserver variations with standard deviations 1996 Volumes Case A GTV Diameter (cm) Length (cm) CTVII Diameter (cm) Length (cm) PTVII Diameter (cm) Length (cm) Case B GTV Diameter (cm) Length (cm) CTVII Diameter (cm) Length (cm) PTVII Diameter (cm) Length (cm) Case C GTV Diameter (cm) Length (cm) CTVII Diameter (cm) Length (cm) PTVII Diameter (cm) Length (cm)
Pre-training
Immediately posttraining
1–2 months posttraining
mean
(SD)
mean
(SD)
mean
(SD)
mean
(SD)
2.22 9.30
(0.49) (5.98)
2.51 8.10
(0.21) (3.15)
2.33 6.90
(0.53) (2.66)
2.39 6.00
(0.13) (0)
4.70 15.60
(0.84) (1.70)
4.79 12.90
(0.48) (2.48)
4.24 9.10
(0.46) (2.82)
4.50 8.05
(0.58) (0.70)
6.23 16.20
(0.68) (1.54)
6.56 13.15
(0.79) (2.72)
6.13 11.30
(0.57) (2.98)
6.33 10.10
(0.83) (1.40)
n/a n/a
n/a n/a
n/a n/a
n/a n/a
1.40 6.13
(1.61) (1.84)
1.90 6.75
(1.68) (3.77)
1.36 4.25
(1.93) (2.06)
1.92 6.25
(1.42) (5.91)
3.68 7.83
(0.01) (2.36)
4.99 7.00
(0.33) (1.73)
* 6.00
(1.63)
4.57 5.33
(0.60) (1.15)
1.57 5.00
(0.11) (1.00)
1.85 6.00
(0.14) (2.83)
1.77 4.67
(0.19) (1.15)
1.16 3.00
(1.00) (1.73)
3.02 9.50
(0.68) (1.91)
3.53 8.75
(0.79) (2.22)
3.81 6.67
(0.31) (1.15)
3.44 7.67
(0.59) (1.53)
4.88 10.50
(0.99) (3.11)
5.14 10.00
(1.48) (1.83)
5.79 8.67
(0.73) (1.15)
5.48 9.33
(1.55) (1.15)
Abbreviations: SD ⫽ standard deviation; GTV ⫽ gross tumor volume; n/a ⫽ Case B does not have any gross tumor visible in the CT scan. * Exact diameter not drawn in the common representative slice, although it was indicated that PTVII should cover this slice.
0.05, two-sided test). For a difference of 1.5 cm, seven radiation oncologists are required. Similarly for Case C, six and three radiation oncologists are required to detect a difference of 1 and 1.5 cm respectively in each end of the PTVII. This study is important for the development of clinical protocols, especially as we move into an era of 3D-CRT and dose escalation studies. It can help to improve precision in radiotherapy in parallel with other efforts to achieve accurate dosimetry (9). Type of training The ICRU 50 was first introduced in 1993. It had not gained widespread use in Canadian clinics by 1998. After leaving the residency program, much of the postgraduate learning is through reading of current literature, which by itself may not be very effective. The occasional conferences are usually in the form of brief capsule lectures, often lacking details, and do not always contain clear teaching objectives. Interactive learning by personal presentations combined with discussion has been found to be the most effective learning method (10 –12).
In this study, we have assessed the use of one-on-one teaching and actual patient films as audiovisual aids in an attempt to improve consistency. Other possible methods of improving protocol compliance include prompt feedback if there is protocol violation, more involvement in protocol development, demonstration of different outcomes to participants, and interactive teaching sessions via personal or video conferences. We have not studied these approaches on protocol compliance. A larger study may consider incorporating these strategies. Frequent reminders may improve consistency over a longer period of time. Whatever presentation method is selected, it should be concise, since clinicians are busy and have limited attention span for lengthy protocol descriptions. CONCLUSION This study aimed to improve consistency in target volume definition for cervical esophageal tumor by the use of specialized personal training. It shows that more detailed presentation of protocol guidelines may improve consistency, especially in the longitudinal tumor dimension.
774
I. J. Radiation Oncology
● Biology ● Physics
This study may have an impact on quality assurance and consensus guideline development, and provide information on appropriate training of oncologists for target volume definition used in clinical trials. Further clinical studies may be required for this site and other tumor sites to determine
Volume 53, Number 3, 2002
the degree of tolerance to variation in target volume definition for 3D-CRT. The survey methodology described here may be an important tool in this regard. Protocols for improving target volumes can be developed using this study as a model and adjusted for different sites.
REFERENCES 1. Tai P, Van Dyk J, Yu E, et al. Variability of target volume delineation in cervical esophageal cancer. Int J Radiat Oncol Biol Phys 1998;42:277–288. 2. Leunens G, Menten J, Weltens C, et al. Quality assessment of medical decision making in radiation oncology: Variability in target volume delineation for brain tumors. Radiother Oncol 1993;29:169 –175. 3. Logue JP, Sharrock CL, Cowan RA, et al. Clinical variability of target volume description in conformal radiotherapy planning. Int J Radiat Oncol Biol Phys 1998;41:929 –931. 4. Valicenti RK, Sweet JW, Hauck WW, et al. Variation of clinical target volume definition in three-dimensional conformal radiation therapy for prostate cancer. Int J Radiat Oncol Biol Phys 1999;44:931–935. 5. International Commission on Radiation Units, and Measurements. ICRU Report 50: Prescribing, recording, and reporting photon beam therapy. Bethesda, MD: ICRU; 1993. p. 3–16. 6. Armitage P, Berry G. Statistical inference. In: Armitage P,
7. 8. 9. 10. 11. 12.
Berry G, editors. Statistical methods in medical research. 3rd ed. Oxford: Blackwell Science Ltd; 1994. p. 106 –114. Coia L. The esophagus. In: Cox JD, editor. Moss’ radiation oncology. Rationale, technique, results. 7th ed. St. Louis: Mosby; 1994. p. 406. Bolla M, Bartelink H, Garavaglia G, et al. EORTC guidelines for writing protocols for clinical trials of radiotherapy. Radiother Oncol 1995;36:1– 8. Dutreix A. When and how can we improve precision in radiotherapy? Radiother Oncol 1984;2:275–292. Harden RM. Ten questions to ask when planning a course or curriculum. Med Edu 1986;20:356 –365. Sibbald RG, Tipping J. Rounds and maintenance of certification program. Ann RCPSC 2000;33:69 –72. Davis D, O’Brien MAT, Freemantle N, et al. Impact of formal continuing medical education: Do conferences, workshops, rounds, and other traditional continuing education activities change physician behavior or health care outcomes? JAMA 1999;282:867– 874.
