Validation of the bladder neck as an important organ at risk in prostate seed brachytherapy based on D2cc: A single-institution, retrospective review

Patient and treatment characteristics

We conducted a retrospective review of 209 consecutive patients treated with LDR monotherapy between and March, 2015 and September, 2018. All patients had biopsy-confirmed prostate cancer, and had magnetic resonance imaging (MRI) as part of their staging, unless contraindicated. Full baseline details are displayed in Table 1.

Study design

We recorded clinical, dosimetric, and toxicity data for 209 consecutive patients treated with definitive monotherapy for prostate cancer. This included at least 100 patients who were treated both before and after we began routinely contouring the BN intra-operatively.

Our primary interest was whether rates of grade ≥ 2 AUT and LUT changed after we began contouring the BN, and aiming to keep the dose to this novel OAR as low as achievable.

Secondly, we examined the constraint which Hathout et al. [9] had identified as a predictor for AUT and LUT by comparing rates of grade 2+ AUT and LUT in patients whose BN D_2cc was > or ≤ 50% prescription dose. The study was approved by the hospital’s research ethics committee.

Brachytherapy technique

Three different brachytherapists performed the procedure during the study period, but the majority were undertaken by one clinician. The patients were under anesthesia, either general or, less frequently, spinal. They were catheterized for the duration of the procedure and received anti-microbial cover at the time of anesthetic induction only. Insertion of transperineal interstitial needles to the prostate gland was aided by a perineal template, and performed under image-guidance via axial and sagittal TRUS images.

The prostate, rectum, and urethra were contoured on axial ultrasound slices and transferred to the brachytherapy planning system. The urethra was contoured with assistance of trans-urethral aerated gel. The method for contouring the BN was described in detail in Hathout’s paper [9] and, in short, it is a 5 mm isometric expansion around the Foley catheter, as defined by the aerated gel, distal to the catheter balloon, proximal to the prostatic urethra, and expanded to a minimum volume of 2 cc.

Iodine-125 seeds were used for all patients, and the prescription dose was 160 Gy to the prostate without a margin. Needles were placed with reference to template coordinates. Loose seeds were placed peripherally via a Mick applicator, and real-time dosimetry via a nomogram-based technique was evaluated using VariSeed planning system to guide placement of the central needles and seeds. Constraints for treatment included a volume percentage of the prostate receiving 100% of prescription dose (V₁₀₀) > 95%, prostate D₉₀ > 100%, and rectum V₁₀₀ < 1 cc. We considered a D_2cc ≤ 50% to be an optimal constraint for the BN based on Hathout’s study.

After completion of the seed implant, the catheter was removed and the patient was transferred to the recovery room. Patients were given an α-blocker in recovery and discharged later the same day with a three-month prescription for the same medication.

Outcome measurement

Patients completed an international prostate symptom score (IPSS) questionnaire at their initial consultation before brachytherapy. Follow-up appointments were scheduled at 1, 3, 6, and 12 months after the procedure, and either 6 monthly or annually thereafter. At each outpatient visit, clinical information was recorded regarding patients’ subjective recording of symptoms reflecting toxicity.

For the purpose of this study, medical records were reviewed and urinary toxicities were graded according to the Common Terminology Criteria for Adverse Events (CTCAE) v. 5.0. In a minority of cases, clinicians had assigned CTCAE scores at the time of review. Otherwise, these were graded by the author reviewing the medical records. The CTCAE contains multiple domains under the category of ‘Renal and Urinary Disorders’, and this was applied as a guide. The worst score in any of these domains was used to assign a grade of urinary toxicity for each clinic visit. For example, a need for medical management for urinary frequency was categorized as grade 2. Need for tamsulosin was also considered grade 2. According to CTCAE, minor toxicities are generally grade 1, whereas moderate toxicities are generally grade 2. Although retrospective, the medical records typically provided sufficient information to make this evaluation. IPSS questionnaires were not routinely issued to patients at follow-up visits. This practice has changed since these patients were treated, but post-therapy IPSS scores are only available for a small minority of patients.

AUT was defined as urinary toxicity reported within 3 months of the procedure, and LUT was urinary toxicity reported beyond 3 months.

Statistical analysis

We applied medians and interquartile ranges for continuous variables, and frequencies and percentages for categorical variables. A comparison of AUT and LUT before and after we began contouring the BN was performed using a χ² test. Rates of AUT and LUT, and pre- and post-therapy IPSS scores, for those with BN D_2cc of > and ≤ 50% were also analyzed with χ², or Fisher’s exact test when data were sparse. A significance level of 0.05 was applied in all cases.

A relationship with other potential predictive factors for urinary toxicity was assessed by independent-samples Mann-Whitney U test.

Values before and after institutional contouring of the BN

Table 2 demonstrates outcomes before (group 1) and after (group 2) institutional contouring of the BN commenced. Grade 2-3 AUT reduced from 15/101 (15%) to 9/104 (8.7%), p = 0.245. LUT reduced from 32/100 (32%) to 18/100 (18%), χ² = 4.5, p = 0.034. All except 3 of these toxicities were grade 2 in severity, and all grade 3 toxicities occurred in patients who were treated without BN contouring.

Five patients experienced urinary retention, of whom three were treated without BN contouring. A Fisher’s exact test indicated no statistically significant association between BN contouring and acute or late urinary retention, p = 0.674 (n = 209); although it should be noted that the number with urinary retention was very small.

5/209 patients (n = 5; missing = 9) were taking alpha blockers or other pharmacotherapy for lower urinary tract symptoms (LUTS) at baseline. 42/101 (42%) patients treated without BN contouring and 33/108 (31%) of those treated with a contoured BN had a new requirement for such medications extending 3 months or more after the procedure. The difference was not statistically significant (p = 0.243).

Toxicity rates for those with D_2cc > and ≤ 50%

Baseline characteristics for the patients treated after we began contouring the BN are displayed in Table 3.

Of these 108 patients, 67 (66%) had values > 50% of prescription dose, and 34 (34%) had values ≤ 50%. Data were missing for 7 patients.

No episodes of grade 3+ toxicity were seen in this population. Grade 2 AUT rates were 4/63 (6%) for those with D_2cc > 50% and 5/34 (15%) with D_2cc ≤ 50% of prescription dose, p = 0.270. Grade 2 LUT was reported in 11/62 (18%) patients with D_2cc > 50% and 5/32 (15.6%) with D_2cc ≤ 50% of prescription dose, p = 1.0 (Table 4).

Association of other variables with toxicity

A Mann-Whitney U test revealed a statistically significant association between IPSS pre-implant score and grade 2-3 LUT. Those with grade 2-3 LUT had a higher score (median, 7) than those who did not develop grade 2-3 LUT (median, 4) (n = 188, p = 0.001). Of the 5 patients with grade 3 LUT, scores ranged from 3 to 22. No significant association was observed between urethra D₃₀, prostate D₉₀, or gland volume, and AUT or LUT.

Possible explanations for our findings and difference from Hathout et al. [9]

The groups treated before and after we began contouring the bladder neck appeared to have similar clinical characteristics. The higher toxicity rates observed in patients treated without the BN contoured may reflect increasing operator experience, hot spots in the region of the urethra, or may simply be the result of chance due to the small numbers involved. In addition, our prescription dose of 160 Gy was higher than Hathout’s, which may have influenced why BN D_2cc was not predictive in our cohort.

The vast majority of implants were performed by a single operator, and the outcomes may improve over time as a result of increased experience. This may not necessarily be reflected in DVCs seen at the time of implantation, but similar findings have previously been reported regarding declining rates of urinary retention as operators gain experience [14]. It may be that the slight change in technique introduced by BN contouring may have had some influence on outcomes aside from dosimetric changes.

An advantage of the generating the urethral expansion intra-operatively is that it helps to guide seed placement. It allows for visualization of the urethra when implanting seeds in the sagittal view, and thus acts as a prompt as to the proximity of the urethra. As a result, performing the procedure with the BN contoured may reduce the likelihood of placing seeds in close proximity to the urethra. This may reduce the risk of significant hot spots in its vicinity, which may not be reflected by the BN D_2cc. Significantly elevated doses in close proximity to the urethra have previously been shown to correlate with rates of urinary toxicity [15], and it is possible that the BN has similar properties.

We recognize that there may be a difference in ultrasound-based contouring with the patient in the dorsal lithotomy position, as opposed to at CT where the patient is in the supine position. There is incomplete drainage of urine against gravity at this time of the procedure, whereas the catheter can drain more freely in the supine position at the time of CT. It is possible that this would result in a slightly different contour. However, it seems unlikely that any such difference would have been so great as to fully explain the differences between our results and those of Hathout’s [9].

Strengths

To our knowledge, this is the first report of outcomes based on intra-operative ultrasound-based contouring of the BN and its usage as an OAR. The patient cohort was treated sequentially, and all patients treated with monotherapy over this timeframe were included. This is an actual population of patients undergoing brachytherapy in a tertiary referral center.

Limitations

It is difficult to retrospectively grade toxicity and this is perhaps the greatest weakness of the present study. We were reliant on individual recording in clinical practice in the absence of express guidance on how toxicity should be recorded. For the purpose of this study, two authors were involved in the retrospective grading of toxicity rates but, given the limitations, our absolute toxicity rates could not be considered externally valid. Toxicities for all patients within the population were assessed in the same way, so comparisons within our population sub-groups can be considered valid. Consistent prospective recording according to a set standard, including CTCAE, or a patient-reported outcome, such as the IPSS, would have been a more reliable metric. However, these were not available for this population.

In addition, the absolute number of patients involved was reasonably small, which reduced our power to detect a difference between the groups with higher or lower BN D_2cc. Many referrals came from outside institutions, which impacted the availability of results for LUT. Follow-up time was also longer in the group treated without their BN contoured. This had little impact on the results for AUT, but may have contributed to underpowering results for LUT. However, the majority of patients who developed LUT in our population had done so before 12 months, and the group treated after we began contouring the BN had a median follow-up of 15 months. Therefore, it seems unlikely that a shorter duration of follow-up could fully explain the significantly lower rates of LUT observed in that group.

Implications for practice

These data were collected retrospectively and lend limited support to the prior findings by Hathout et al. [9], but the overall literature in this area is conflicting (Table 5). A prospective phase 2 trial (n = 61) examining the impact of BN dose in the context of MRI-guided high-dose-rate prostate brachytherapy also demonstrated no obvious link between a high BN dose and grade 2 AUT or above [16]. BN doses within the highest quartile of the population were observed in 3 of 4 patients with urinary retention in that trial. The small sample size is a limitation of that study, and it is hard to know how applicable those results are to ours given the different dose-rate. On the other hand, a cohort study evaluating predictors of urinary toxicity in 159 patients treated with caesium-131 prostate brachytherapy identified the BN maximum dose as the only consistent dosimetric predictor of toxicity [17].

GEC-ESTRO [7, 8] and the American Brachytherapy Society [6] have recommended constraints for the urethra when performing this procedure. Associations between other factors, including baseline IPSS, larger prostate size, greater number of implanted needles, use of neoadjuvant androgen-deprivation therapy, and a higher prostate D₉₀ with higher rates of urinary toxicity have been reported [10, 18-20]. Dose to the lower bladder and trigone have also been linked to rates of urinary toxicity using external beam radiotherapy [21].

Looking at other potentially predictive factors in our population, we noted greater rates of toxicity in those with higher baseline IPSS values, but no association with other variables. We did not identify any specific peri-implant constraints which were associated with lower rates of urinary toxicity.

There is no obvious detriment to contouring the BN and striving to limit the dose it receives, especially in those with high baseline IPSS. MRI guidance may help with more accurate delineation [16]. Additional prospective studies examining toxicity rates would further help to evaluate the BN as an OAR during prostate seed implantation.