Clinical relevance
Our institutional experience describes the early stages of SpaceOAR hydrogel implementation in high-dose per fraction SBRT setting, where the benefits of a well-placed spacer on rectum dosimetry quickly become clinically evident upon examination. Our objective was to develop a systematic method to quantify the perirectal space after hydrogel placement and identify a hydrogel placement metric to correlate perirectal space creation with rectum dosimetry. The findings showed increased perirectal space and optimal hydrogel placement have a positive impact on rectal dosimetry when we considered the θ*hydrogel volume metric. This metric was at least, if not more, predictive of rectum dosimetry than any perirectal space measurement. With longer follow up and greater sample size, this metric should have added utility of eventually tracking placement quality and operator experience.
Evidence of a learning curve in developing operative skillsets is well-established as seen in prostate brachytherapy implant quality that improves with experience up to a point [
16,
17], after which proficiency is maintained with a minimum annual caseload [
18]. The learning curve phenomenon for hydrogel placement was first reported by Pinkawa et al. in a study of 64 patients that showed improved lateral hydrogel symmetry, increased perirectal space, and better rectum dosimetry in the latter 32 patients compared with the first 32 patients [
19]. Such a learning-curve effect is minimized in a well-established training environment with appropriate mentorship and operator feedback [
20]. We expect that a well-designed hydrogel placement metric, such as we have described in this manuscript, would provide one such measure of feedback.
Quantifying perirectal space
While results from this study are in agreement with the general expectation that a well-placed hydrogel is important for rectum dosimetry, to our knowledge this is the first time the post-hydrogel perirectal space has been rigorously mapped. We showed that precise perirectal space measurements can be obtained independently from either the post-hydrogel T2 weighted MRI or CT simulation scans (Tables
2 and
3). Prior to hydrogel placement, the prostate apex and midgland lie close to the rectum on the pre hydrogel placement MRI (Table
3A). Following hydrogel placement, almost all regions of the prostate have increased separation from the rectum, with the greatest mean increase seen at the prostate apex and midgland (Table
3C). Several publications have described a wide range of post-hydrogel perirectal separation ranging from 0.6 to over 2 cm, but to-date few have described in detail the optimal location to obtain this perirectal measurement, or Δ measurement, resulting from hydrogel placement [
11,
14,
21‐
23].
From our perirectal space maps, we identified the perirectal distance measurement obtained posterior to the prostate at midgland, midline, or CTV center, as that which is most strongly correlated with rectum dosimetry (Table
4A). This is followed by the perirectal distance measurements 1 and 2 cm that are immediately inferior to it i.e. toward the prostate apex. As enlargement of the perirectal space at, or slightly inferior to, the prostate CTV center led to the greatest improvement in rectum dosimetry, this position represented the optimal location for hydrogel placement.
SpaceOAR characteristics and rectum dosimetry
Distribution of lateral hydrogel deviation in our patient cohort was nearly identical to that of earlier work by Fischer-Valuck et al. [
14] Half (
n = 10) of patients had symmetric gel placement (SYM1, lateral deviation < 1 cm) in all three axial slices, and 15% (
n = 3) had hydrogel in all three axial slices but with lateral deviation < 2 cm in only one axial slice (SYM2). Consistent with their conclusions, the rV95% and rD
max 1 cc were not significantly different between SYM1 and SYM2 hydrogels in our study (Fig.
2).
While the effect of lateral hydrogel deviation on rectum dosimetry is thus well-characterized, hydrogel distribution along the craniocaudal dimension has not been commented upon in existing literature. In our study minor hydrogel deviation along the craniocaudal axis had a more pronounced effect on rectum dosimetry than lateral deviation of a similar scale (~ 1 cm). The absence of hydrogel in the axial slice 1 cm inferior to midgland (
n = 4) correlated to an increase in rD
max 1 cc of 171 cGy (
p = 0.005, Fig.
2).
This is unsurprising as the native perirectal space is most limited inferomedially, averaging 1–2 mm in our analysis. In the vast majority of patients, such close proximity between the prostate and rectum begins inferiorly at the prostate apex and extends superiorly at least as far as the prostate midgland. Whereas previous analysis emphasized the importance of lateral hydrogel symmetry, we conclude that hydrogel deviation must be accounted for in both the lateral and cranio-caudal dimensions to accurately predict rectum dosimetry.
As a result, we defined θ to quantify hydrogel deviation from the optimal CTV center location in three-dimensional space rather than along a single axis. θ alone correlated moderately with rD
max 1 cc and rV95% and predicted only a minority of variance in these statistics (R
2 = 0.36 and 0.43, respectively; Table
4B) as it does not take into account the volume of hydrogel centered around the θ vertex.
Similarly, hydrogel volume alone was modestly correlated with rectum dosimetry (R2 rDmax 1 cc = 0.23; rV95%, 0.2). This is due to minimal enlargement of the perirectal space in the instance of suboptimal injection site (i.e. small θ). Indeed, hydrogel volume was not significant in two-variable regression of hydrogel volume and θ against rDmax 1 cc (pVOLUME = 0.06, pθ < 0.01, n = 20), but became significant only in the subset of patients with θ > 35° (pVOLUME = 0.04, pθ < 0.01, n = 16).
We thus inferred that an interaction exists between hydrogel volume and θ such that the correlation of each θ and hydrogel volume influenced rectum dosimetry. Based on this examination of data, we tested the product θ*hydrogel volume which showed strong ability to explain rectum dosimetry (R
2 rD
max 1 cc =0.64; rV95%, 0.6; Table
4B). A three-variable regression analysis of θ, hydrogel volume, and θ*hydrogel confirmed that θ*hydrogel was the only significant predictor of rV95 (R
2 = 0.694; F
-value 12.08;
p = 0.008). We thus conclude that the parameter θ*hydrogel volume quantifies perirectal space enlargement effect following hydrogel placement, and should be considered in evaluating hydrogel placement success.
Previous work has shown no statistically significant variation in rectum dose with CTV volume ranging up to 100 cc, and is consistent with our findings [
21].
Acute rectal toxicity and θ*hydrogel volume
Hydrogel spacer use in prostate SBRT was documented as early as 2013 by Alongi et al. and continues to represent a growing proportion of SpaceOAR utilization [
11,
24,
25]. Yet as of this study, the only published phase III randomized prostate hydrogel data were obtained in the conventionally fractionated setting [
22]. In their control arm without hydrogel, acute rectal toxicity ≥grade 2 benefit was not statistically significant, but any late toxicity was improved 4.5-fold at 3 years.
We expected that relative toxicity benefits attributed to hydrogel use would be at least as prominent in the high-dose per fraction SBRT setting as it is in conventionally-fractionated treatment. In addition, projected cost-benefit decision analysis suggests that the use of hydrogel spacer will be cost-effective for toxicity management in the long term for all forms of prostate radiotherapy, but particularly so for prostate SBRT [
26].
With the hydrogel spacer and prostate SBRT to a dose of 3625 cGy in five twice-weekly treatments, less than one third of patients developed grades 1–2 acute rectal toxicity either during or immediately after treatment, which resolved at the latest by the one-month post-treatment visit. It is noteworthy that these men had characteristics of rectum dosimetry, perirectal spacing, and θ*volume that we quantified as being in the less favorable half of our cohort (Fig.
4).
Likely as a result of small sample size, no statistically significant difference was observed in rDmax 1 cc, rV95%, perirectal distance at CTV center, and the metric θ*hydrogel volume, between men who developed acute rectal toxicity and men who did not. We suspect that the likelihood of statistically significant relationships between these parameters and symptomatic toxicity would emerge at the higher SBRT doses currently being evaluated in phase II clinical trials, with larger sample size, and longer follow-up.
We have nonetheless shown from our early hydrogel SBRT patient cohort a correlation between the metric θ*hydrogel volume and rectum dosimetry. We expect that the learning curve necessary to attain high-quality hydrogel placements will also become evident as measured by the θ*hydrogel volume metric as our experience with hydrogel placement increases.
The relatively wide range of hydrogel volumes as measured on post-hydrogel T2 weighted MRI bears further evaluation. Each SpaceOAR hydrogel is injected transperineally as a 10 cc suspension following dissection of Denonvillier’s fascia with saline solution. Potential explanations for small hydrogel volumes include suboptimal placement, hydrogel dispersion prior to polymerization and incidental withdrawal of hydrogel along with the needle following the procedure. Explanations for large hydrogel volumes include contouring software volume overinterpolation and overcontouring of hydrogel. The latter may occur as saline solution used in the fascial dissection and SpaceOAR hydrogel is indistinguishable on the MRI T2 image obtained after hydrogel placement. In our cohort, the earliest post-hydrogel MRI was obtained two days after hydrogel placement.