Measurement of craniocaudal catheter displacement between fractions in computed tomography–based high dose rate brachytherapy of prostate cancer

The objective of the present work was to measure the craniocaudal displacement of catheters occurring between consecutive fractions of transrectal ultrasound (TRUS)‐guided high dose rate (HDR) prostate brachytherapy. Ten consecutive patients were treated with 2 fractions of 9.5‐Gy TRUS‐guided HDR brachytherapy, with dental putty being used for the fixation of catheters. For each patient, a computed tomography (CT) scan with 3‐mm slice thickness was acquired before each of the 2 fractions. Two different references were used to measure the catheter displacement between fractions: the ischial bone as a bony marker (BM) and the center of two gold markers (COGM) implanted in the prostate. Catheter displacement was calculated by multiplying the thickness of the CT slice by the difference in number of CT slices between the reference slice and the slice containing the tip of a catheter. The average magnitude of caudal catheter displacement was 2.7 mm (range: −6.0 mm to 13.5 mm) for the BM method and 5.4 mm (range: −3.75 mm to 18.0 mm) for the COGM method. The measurement data obtained from the BM and COGM methods verified that prostate movement and catheter displacement both occurred independently between fractions. The most anterior and medial two catheters (catheter positions 8 and 12) had the greatest tendency to be displaced in the caudal direction because they were located at the most distant position from the fulcrum, making them susceptible to rotation of the dental putty in the lateral plane because of the movement of the patients’ legs between fractions. In conclusion, the combination of the BM and COGM methods can demonstrate prostate and catheter movement relative to the BM between fractions. Our technique found a pattern of catheter displacement. Based on that finding, further improvement of our results may be possible by modification of our current technique. PACS number: 87.53.Jw

sparing critical organs adjacent to the target (1) . Another advancement made in HDR 20 brachytherapy is the development of the inverse planning software which allows the 21 optimization of dwell time distribution providing the desired dose distribution based on the 22 prescribed dose constraints (2)(3)(4) . Furthermore, functional imaging information from MR 23 spectroscopy can be used for treatment planning to better identify dominant intraprostatic 24 malignant lesions (5) . Despite the advantages mentioned above, dose uncertainties still remain 25 in the HDR brachytherapy for the prostate cancer. We recently addressed the dosimetric impact of prostate volume change due to the trauma caused by the insertion of catheters 1 together with the resolution of edema between fractions (6) and the dose uncertainty due to the 2 intrinsic characteristics of finite thickness of CT slice (7) . This translates into a discrepancy 3 between the source dwell positions observable on planning CT or MRI images and the actual 4 dwell positions during the dose delivery by the afterloader.

5
In this study, the cranio-caudal catheter displacement between fractions due to patient 6 and prostate motions was measured by acquiring CT scans before each fraction for 10 patients.

19
In our institute, HDR prostate brachytherapy boost is performed in two 9.5 Gy fractions after 20 45 Gy of external beam radiotherapy. During the procedure, a physician inserts Flexi-guide 21 catheters (Best Medical, Springfield, VA) into the prostate using a freehand transrectal 22 ultrasound (TRUS) guided technique. An in-house customized catheter fixation technique 23 using dental putty ( Fig. 1(a)

4
The first fraction is delivered in the early afternoon on the first day, while the second fraction 5 is delivered in the morning of the second day (6,8) . A single treatment plan is used for both 6 fractions. The treatment plans are obtained with our in-house anatomy based inverse planning 7 algorithm (IPSA, Inverse Planning based on Simulated Annealing), which optimizes the 8 dwell times once the dose constraints and the prescription are specified (2-4) . 9 10 B. General pattern of 16 catheters inserted into prostate 11 As seen in Fig. 1(b), in general 16 catheters were inserted into the prostate in four rows by 12 four columns. The labeling was from right first to left fourth column. The numbering scheme 13 of catheters is also shown in Fig. 1(b). The catheters fixed by the dental putty were not 14 parallel as in the conventional pre-fabricated template technique. They sometimes converged 15 or diverged to cover the entire target volume based on TRUS image. In Fig. 1(b) the midline 16 of the prostate was satisfactorily covered by the catheters on the second (position 5 to 8) and 17 third column (position 9 to 12) even though no catheters were located at the midline of 18 perineum in Fig. 1(a). the size of black dot was smaller than expected because the reconstruction volume for that CT 9 slice did not fully contain the tip of the catheter. Hence, the tip of the catheter was assumed to 10 be located between the current slice and the previous one (7) . For instance, if the size of black 11 dot shown on (i)th CT slice that contains the tip of a catheter is not as big as expected, the tip 12 of the catheter is considered to be located on the CT slice assigned half integer number such 13 as (i)-1/2. Therefore, the catheter depth calculated on (i)th CT slice was decreased by 1.  In this study, a positive displacement means the catheter has moved inferiorly from day 1 to 1 day 2 (caudal displacement), and a negative displacement means the catheter has moved 2 deeper into the patient superiorly (cranial displacement). The measurements were analyzed 3 using descriptive statistics for each patient and each catheter position. For the total of 160 4 catheters, the measurement was performed again by another observer to assess inter-observer 5 differences in the measured catheter displacement. 6 7 III. RESULTS

9
The measurement data are displayed in Fig. 2 with mean ± standard deviation value and 10 summarized in Table 1 for all 10 patients. All measurement data were changed into the 11 absolute values and replotted as Fig. 2(b). The average catheter displacement between the 12 first day and the second day was 4.1 mm (2.7 mm for BM and 5.5 mm for COGM 13 measurement, respectively). The range of measured displacements was -6.0 to 13.5 mm for 14 BM measurement and -3.8 to 18.0 mm for COGM measurement. The maximum catheter 15 displacement was observed in patient C for the BM measurement method and it was observed 16 in patient C and I for the COGM measurement method. For the converted measurement data 17 in the absolute values, the average catheter displacement was 3.4 mm for BM and 5.6 mm for 18 COGM measurement method, respectively. The catheter experiencing the maximum 19 displacement was stationed at the twelfth catheter position ( Fig. 1(b)). The measured catheter 20 displacements were greater when they were based on COGM measurement.

21
In Fig. 3, catheter displacements measured with both methods for 8 patients who had 16 22 catheters are represented with mean ± standard deviation value corresponding to their catheter 23 position. In addition, the measurement data were changed into absolute values and replotted 24 as Fig. 3(b). One can observe that catheter position 8 and 12 were most likely to have the 25 greatest displacement. These two catheters correspond to the two most anteriorly and medially located catheters ( Fig. 1(b)). Table 2 shows the statistics for the catheter 1 displacement depending on the catheter position.

2
In addition, the average ± standard deviation value in the difference of catheter 3 displacement measured by two different observers was 0.9 ± 0.9 mm with maximum of 4.5 4 mm (95% confidence interval: 0.8 -1.1 mm) for BA method and 1.0 ± 0.9 mm with 5 maximum of 5 mm (95% CI: 0.8 -1.1 mm) for COGM method, respectively. 6 7 IV. DISCUSSION 8 9 The average displacement (4.1 mm) between the first and the second fraction (on average, 10 19.5 hours difference) in this study is quite small compared with several reports (9)(10)(11)(12) (11) reported that the average template movement was 1 mm and the 20 catheter movement relative to the prostate was 9.7 mm, using 5 mm CT scan, between the 21 first and second fraction (over 18 -24 hours). Mullokandov and Gejerman (12) reported that 22 there was no displacement of catheters relative to the template and the mean consecutive 23 catheter displacement was 2, 8 and 10 mm for before the second, third and fourth fraction.

24
Because the time interval between fractions was 6 hours in their study, the displacement 25 before the fourth fraction (minimum 18 hours difference) in their measurement (10 mm) can be compared with our measurement before the second fraction (4.1 mm). The four fraction 1 HDR studies (9, 10, 12) in the literature showed a time dependent fashion of catheter 2 displacement between fractions. The maximum catheter displacement occurs up to ~ 12 hours 3 after the first fraction (20 mm before the second fraction (9) , 7.6 mm before the second 4 fraction (10) , and 6 mm before the third fraction (12) , respectively) and its magnitude is 5 subsequently decreased for the following fraction. 6 7 Our two measurement methods may contain some potential error. 8 1. Because of the thickness of the CT slices used, the lower limit of accuracy of our 9 measurement is 3 mm. Even if the half integer was assigned whenever the tip of 10 catheter was obscure on a CT slice, the possible maximum error between a reference 11 slice and the slice containing a catheter tip is 3 mm. 12 2. Artifacts generated from our gold seed markers. In general, a gold seed marker appears 13 over 2 or 3 CT slices because its dimension was 5 mm in length and 1 mm in 14 diameter. There are 2 possible ideal scenarios. First, a gold seed marker is seen as a 15 medium size of bright dot on 2 consecutive CT slices in which the position of a gold 16 seed is defined as the center of the 2 CT slices. Second, when it is seen over 3 CT 17 slices, a gold seed marker appears as a big bright dot on middle CT slice and a small 18 dot on the previous and next CT slices in which the position of gold seed was 19 defined as the middle CT slice. A gold seed marker is usually between two ideal 20 scenarios. Hence, the maximum error from artifact of a gold marker seed is 1.5 mm. 21 3. Gold seed migration in the COGM measurement method. In a literature (13) , the gold seed migration was measured by the inter-marker distance and its 96 percentile value 23 was less than 1.5 mm. In our study, the average inter-marker distance variation was 24 1.4 mm and the 95% percentile value was 1.9 mm. We believe that the migration of the center of two gold seed markers was much less than the actual movement of two 1 gold markers.

15
Second, we could sometimes see the individual movement of catheters relative to the putty 16 ( Fig. 5(a)). However, this event rarely happens based on a physician's visual inspection 17 before the second fraction. Consequently, catheters may be assumed to move together with 18 the putty to explain the average catheter displacement measured by either BM or COGM 19 method. Finally, in case we can ignore the movement of OARs, we may consider only two 20 movements (catheter and prostate movement) relative to the BM between fractions. If there is 21 no movement of prostate relative to the catheters, the catheter displacements measured by 22 either BM or COGM method (∆d BM or ∆d COGM ) should to be the same (∆d BM = ∆d COGM ).

23
Otherwise, four possible scenarios are shown in Fig. 4. In this study, the caudal catheter 24 displacements observed are similar to the scenario III (Fig. 4(d)) and IV (Fig. 4(e)). In 25 particular, in Fig. 2 the cases in which the average ∆d BM is greater than the average ∆d COGM 26 Deleted: 3

Deleted: 4
Deleted: comparing two CT scans obtained by different CT gantry angles ¶ 5 (for patient A and G) can be classified into the scenario IV in Fig. 4(e) while the remaining 8 1 cases in which the average ∆d COGM is greater than the average ∆d BM correspond to the 2 scenario III in Fig. 4(d), depending upon the prostate and catheter movement relative to the 3 BM. A recent study (13) on prostate position relative to the pelvic bony marker (BM) also 4 demonstrated significant interfractional movement of prostate relative to the pelvic BM for 5 external beam radiation therapy. Hence, we believe GOGM method is more accurate than BM 6 method in this study.

7
We found that the two most anterior and medial catheters (position 8 and 12 in Fig. 1(b)) 8 were more likely to have a large displacement (Fig. 3). The depth of catheter position 8 and 9 12 was the shallowest because the advancement of those catheters was blocked due to the 10 presence of bladder. Hence, these shallowest implanted catheters may be the most vulnerable 11 to displacement between fractions. Another reason for this large displacement of these 12 catheters may be due to the rotation of dental putty in the lateral plane (Fig. 5). The fulcrum 13 for the rotation of dental putty is located along the suture (Fig. 5(a)). Between fractions the 14 fulcrum can move either in the anterior (Fig. 5(b)) or posterior (Fig. 5(c) (Fig. 5(b)) of both right and left putties in Fig. 6(c). Accordingly, the catheter 23 positions at the posterior of the putty (1, 5, 9 and 13) show larger catheter displacement. As phenomenon can be observed at the catheter position 7 and 13 in Fig. 6(b) and 16 in Fig. 6(c), 1 deviated from the typical trend of catheter displacement due to the movement of fulcrum. In 2 this study, for catheter displacement scenarios the change of prostate volume between 3 fractions was ignored because its magnitude was insignificant. However, partial swelling or 4 shrinking of prostate between fractions may also cause an individual catheter displacement 5 between fractions. We believe that the large catheter movement depending on catheter 6 position can be avoided by giving more tension to the region by changing the placement of 7 the sutures. For instance, the suture can be done on the putty in the superior-inferior direction 8 instead of current lateral direction. Another remedy is the use of two lateral suture lines (one 9 at anterior portion of putty and the other at the posterior portion of putty) in place of one 10 suture line in the middle of putty. The additional suture on dental putty is a promising 11 approach to tightly fix putty on the perineum while additional suture to existing four corners 12 may not be appropriate for a conventional pre-fabricated rigid plastic template method.

14
In the literature (11,12) , the dose variation due to catheter displacement during fractions 15 was reported using axial CT images for treatment planning: median 9.7 mm of catheter 16 displacement reduced D90 (dose received by 90% of the target volume) by 40% (11) and 17 median 9 mm catheter displacement caused 35% of change of the dose to 90% of the prostate 18 volume (12) . In those studies, the dosimetric impact due to catheter displacement was 19 significant because the magnitude of catheter displacement is almost twice the spacing of the 20 consecutive dwell positions (5 mm). However, in this study, the dosimetric analysis between 21 fractions was not feasible due to the absence of contours for target and OARs on the second 22 day CT scan. The delineation of the target on CT slice has inter-observer and intra-observer 23 variation and can sometimes be overestimated by as much as 30% based on external beam 24 radiation therapy literature (14)(15)(16)(17)(18) . As the prostate movement is observed relative to the 25 catheter displacement, we can also imagine the movement of critical organs such as bladder and rectum (though urethra may move together with the prostate) between fractions. The 1 uncertainty due to delineating target and OAR on CT images can also make a contribution to 2 the dose variation between fractions. Therefore, the dosimetric impact due to the small 3 catheter displacement (~ 4 mm) between fractions in this study should be distinguished from 4 the uncertainty of organ contouring on CT images between fractions. In the future, we may be 5 able to investigate the dosimetric impact due to catheter displacement and all organ 6 movement between fractions by employing MRI images to contour the prostate and OARs 7 precisely. 8 9 To measure catheter displacement between fractions, 3 mm CT scan used in this study 10 was comparable to other studies using 3 mm CT scan 11 ; 2 and 5 mm CT scans 12 even though 11 the average catheter displacement is quite different, ~ 4 mm in our study versus ~ 10 mm for 12 others. The 3 mm CT slice thickness in this study may lead to measurement error in the same 13 range of measured catheter displacements. However, in this study the method of assigning 14 half integer to the CT slice containing obscured catheter tip, artifact of gold seed marker, or 15 inter-slice located bony marker can reduce the measurement error by half. The statistics of 16 inter-observer variability study showed the typical range of measurement error. Although 17 more than 3 mm error was observed for a few catheters, for most catheters the error was less