A Phase I Trial of Highly Conformal, Hypofractionated Post-Prostatectomy Radiotherapy

Patients were treated with three increasingly abbreviated courses of radiotherapy as specified in Table 1. Each dose level was selected as the isoeffective dose schedule to a standard, conventionally fractionated prescription (68.4Gy in 38 fractions) at its respective treatment course length as determined by the linear-quadratic model with an ⍺/β of 3Gy representing normal tissues. The initial course length of 20 fractions was selected as prior studies demonstrated the tolerability of a 26 fraction course of treatment [17]. The clinical target volume (CTV) was defined by the RTOG consensus definition [18]. A PTV margin of 7 mm was used concentrically with allowance for minimization to 5 mm posteriorly to meet dose constraints if necessary. The dose was prescribed to the 100% isodose line using intensity modulated radiation therapy (IMRT). The maximum heterogeneity allowed was a minimum of 95% and a maximum of 115% of the prescription dose within the PTV. Organs at risk (OARs) were contoured. Dose constraints were calculated as isoeffective to previously, widely established values for normal tissues using the linear-quadratic model and an ⍺/β ratio of 3 as detailed in Supplementary Table 1. Treatments were delivered with daily fractionation, 5 days per week, using daily cone beam CT image guidance with physician approval of alignment prior to each fraction.

Androgen deprivation therapy (ADT) was allowed at the discretion of the treating physician. When clinically appropriate, 6 months of ADT was given with a depot GnRH antagonist with or without a run-in period of a nonsteroidal antiandrogen.

Subjects were followed for 24 months after completion of radiotherapy. Patients with biochemical progression during follow-up were removed from study. Adverse events (AEs) were classified using Common Terminology Criteria for Adverse Events version 4.0 (CTCAE v4.0) and were scored weekly during treatment and at each follow-up visit. After the completion of treatment, patients were followed every 3 months while enrolled on study and outcome measures were assessed including PSA, testosterone, and AEs. AEs were classified as acute if occurring from the start of treatment to 90 days after treatment and late if occurring > 90 days after treatment. Patient reported QoL was assessed with the Expanded Prostate Cancer Index Composite Short Form (EPIC-26), Sexual Health Inventory for Men (SHIM), American Urologic Association-Symptom Index (AUA-SI), Patient-Reported Outcomes Measurement Information System (PROMIS) Psychosocial Illness Impact-Positive (PII), PROMIS Anxiety (PROMIS-A), PROMIS Depression (PROMIS-D), and Decision Regret Scale (DRS). All patient reported QoL instruments were collected at baseline and at the 6-, 12-, 18-, and 24-month timepoints.

The primary objective of this study was to determine the maximally tolerated hypofractionated dose schedule in the post-prostatectomy setting. Acute dose limiting toxicity (DLT) was defined as a grade 3 or higher in-field, physician-assessed toxicity which did not resolve to grade 2 or less within 4 days and occurred during the treatment course or within three weeks after completion of treatment. As an additional safety measure, a late DLT constraint was also included in the trial. A late DLT was defined as a grade 4 or higher gastrointestinal (GI) or genitourinary (GU) toxicity occurring after the DLT assessment period (more than 3 weeks after completion of therapy). Both acute and late DLTs were considered in regard to escalating dose levels.

Three patients were allocated sequentially to each dose level. If none of the three patients experienced a DLT during the assessment period, escalation to the next dose level was initiated. If one of three participants experienced a DLT at any dose level, three additional patients would be enrolled at that dose level. If two participants experienced a DLT at any dose level, that DL would be considered the maximally tolerated hypofractionated dose schedule. If no DLT was observed at the most hypofractionated dose level, that level was considered the maximally tolerated hypofractionated dose schedule.

Secondary objectives were an evaluation of the safety and tolerability of the investigational treatment, the rate of biochemical control (defined as a PSA < 0.1) at 12- and 24-months post-treatment, and patient-reported outcomes (PROs) measured with the SHIM, AUA-SI, and EPIC-26 instruments. Exploratory objectives included additional psychometric testing with the PROMIS-PII, PROMIS-A, PROMIS-D, and DRS.

Changes to PROs from baseline to each follow-up visit were tested by the Wilcoxon signed-rank test. Changes with respect to ADT use and castrate level of testosterone (< 50 ng/mL) for overall sexual function of EPIC and overall SHIM score were compared by Kruskal rank test. In evaluating these measures, the minimally important difference (MID), or threshold beyond which changes in PRO measurements are considered clinically relevant, was defined based on previously used values for the AUA-SI [19], EPIC [20], PROMIS-A [21], PROMIS-D [22], and DRS [23]. As previous work has shown the MID to be dependent on the baseline SHIM score, the MID was variably defined as 2 for mild (score 0-10), as 5 for moderate (score 11-16), and 7 for severe erectile dysfunction (score 17-25) [24]. P-values < 0.05 were considered statistically significant. As these analyses were exploratory in nature, no adjustment for multiple comparisons was made.

From January 29, 2018 to August 9, 2019, a total of 15 patients were enrolled. All participants were followed until 24 months or until removal from protocol after biochemical failure. The median follow-up was 24 months. The median age was 69.9 years (range: 46.8-76.0 years). The majority patients were treated in the salvage setting for biochemical persistence or recurrence (n=14/15). Twelve of the 15 patients (80%) had margin negative resections at time of prostatectomy, and most patients (60%) had NCCN high-risk disease defined by pathologic staging. The remainder of the key patient characteristics are summarized in Table 2.

The maximal tolerated hypofractionated dose schedule was found to be 44.2 Gy in 10 daily fractions over two weeks (DL3) on which 9 of the 15 patients were treated. No acute or late DLTs were encountered at any dose level. Adverse events possibly, probably, or definitely related to radiation are summarized in Table 3. Overall, the most common acute adverse events were grade 1 GU toxicity and grade 2 GI toxicity each occurring in 60% of the patient population. The most common acute grade 2 GI toxicity was proctitis, occurring in 7 patients. A single patient encountered acute grade 3 proctitis that resolved to grade 2 within 3 days with medical management and did not meet the pre-specified criterion for a DLT due to its duration. No other acute grade 3 toxicity was observed. Two late grade 3 toxicities occurred. These consisted of one episode of noninfective cystitis that resolved after 11 weeks in one patient on DL2 and an increase of baseline post-operative incontinence from grade 2 to grade 3 which developed 6 months after treatment and was ongoing at 24 months of follow-up in one patient on DL3. No grade 4 or 5 adverse events were observed.

One notable toxicity occurred in a patient treated on DL3 after removal from the protocol for biochemical progression, which occurred 14 months after the completion of treatment. This patient experienced intermittent, gross hematuria adjudicated to likely be treatment-related. After infectious causes were excluded, imaging revealed left ureteral and renal pelvis dilation. Cystoscopy with biopsies of the prostatic urethra and bladder trigone revealed necroinflammatory debris from the former and reactive appearing urothelium from the latter. There was no evidence of prostate or bladder cancer. Unilateral ureteral stenosis was addressed with a combination of balloon dilation and temporary stent placement. As this event occurred after removal from the protocol, it is not reported in Table 3.

Of the 15 patients in this study, a single patient met the pre-specified definition for biochemical failure (PSA > 0.1, confirmed on subsequent measurement) by 12 months and a second patient met these criteria for biochemical failure by 24 months. This corresponded to a 12-month rate of biochemical control of 93% and a 24-month rate of 86%. Of the two biochemical failure events, the first exhibited no PSA response to treatment with a PSA trajectory which continued to rise on all post-treatment measurements. The second biochemical failure occurred after an initial PSA response at 22 months post-treatment in a patient who also received ADT. While initial restaging at time of biochemical failure did not detect any recurrent disease in either case, PSMA PET/CTs obtained at PSA values of > 1.0 ng/mL did eventually document an out-of-field pattern of failure and did provide radiographic verification of in-field disease control.

There were two key findings from our QoL analysis. First, there was no significant change in GI QoL from baseline at any timepoint during the study period. This finding is in contrast to the results from a previous phase I/II trial from Wages et al. [27], which investigated three similar hypofractionated PORT regimens delivered in 20, 15, and 10 fractions and found a statistically significant decrease at the 1-year timepoint (92.08 vs. 85.80, p = 0.026) in EPIC GI QoL. While the mean difference from baseline to 1-year was very similar between our trial and this previous trial (-5.36 vs. -6.28 points), a difference in the mean change from baseline did emerge at 2 years between the two studies (-16.07 vs. -4.49 points). As the treatment volume in the prior study was the same as that employed in the current study, and our rectal planning constraints were slightly more permissive than those in the prior study, the differences in these findings are likely explained by patient-related or technical factors, such as subtle differences in treatment planning or delivery. Future larger follow-up studies are needed to further profile the impact of hypofractionated PORT on GI QoL.

The second key finding was that there is a substantial increase in patient reported GU symptom burden as indicated by higher scores on the AUA-SI and corresponding worse GU QoL as measured by changes in the AUA-SI QoL score and EPIC-26 overall GU score. In a previous, detailed, prospective study of PORT delivered via standard fractionation, similar decrements in GU QoL as measured by the EPIC-26 inventory were not seen in long term follow-up [28]. Further, of note, this finding also differs from that of the previous trial examining a hypofractionated dose schedule [27]. In the prior study[27], the phase II endpoint was the change in the EPIC-26 QoL subdomain scores at 1 year from baseline. In that study, no significant change in this measure of GU QoL was seen at this timepoint. Our trial corroborated these this finding up to 1 year follow-up, showing no significant differences in GU QoL at 6 or 12 months. However, a noticeable decrement on GU QoL became apparent with later follow-up. At the 18- and 24-month timepoints, a statistically significant and clinically meaningful decrement in GU QoL emerged emphasizing the necessity of long-term follow-up to uncover the full QoL burden of hypofractionated PORT schedules on patient reported QoL.

Hypofractionated PORT may also increase the objective rates of GU toxicity. Overall, 2 of the 15 patients (13.3%) enrolled on our trial experienced a late grade 3 GU toxicity. These consisted of one patient with self-resolving cystitis and a second with an exacerbation of baseline grade 2 incontinence. This rate of late toxicity is higher than that expected for conventional fractionation; contemporary phase III trials such as SAKK 09/10 [25] and RTOG 0534 [29] report rates of late grade 3+ GU toxicity within the range of 5.3% to 7.9% in patients undergoing post-operative, prostate bed-directed, salvage radiotherapy. The preliminary results of a recent phase III non-inferiority trial, NRG GU003, designed to compare the 2-year EPIC-26 QoL scores of a more protracted hypofractionated regimen than in our trial (62.5Gy in 25 fractions) and a conventionally fractionated regimen (66.6Gy in 37 fractions) also showed a possible signal of grade 3 adverse events in the hypofractionated arm with a 4.7% rate of grade 3 cystitis as compared to 0% in the conventionally fractionated arm [12]. The aforementioned trial by Wages et al. reported a rate of late grade 3 GU toxicity of 9.4% at 24 months, but were unable to find any statistically significant dosimetric predictors of grade 2+ or grade 3+ GU toxicity on exploratory analysis [27]. Additionally, 4 of the 26 patients (15.4%) in a separate phase I five-fraction dose escalation trial encountered late grade 3 GU toxicity [30] with exploratory analysis revealing a candidate pool of dosimetric predictors of late GU toxicity [31]. These findings from our study and the prior available reports underscore the need for further study of predictors of late GU toxicity in patients who receive PORT. Given that serious GI toxicity appears to be rare, a modification in the dosimetric objectives for treatment planning to prioritize sparing of bladder dose at the expense of higher integral doses to GI structures may produce a superior therapeutic index.

Similarly, our findings may argue for a modification of treatment volumes. Current recommendations for the definition of a post-operative clinical target volume (CTV) are based primarily on the principle of encompassing the entirety of the surgical field and the inclusion of the most frequent sites of local failure [18, 32-35]. Another implicit criterion used in the definition of the planning target volume (PTV) is the feasibility of delivering a therapeutic dose to a certain treatment volume without undue toxicity. While this is common in routine clinical practice, formal evidence of this directive is provided by the specification of anisotropic PTV margins used in several clinical trials which recommend more conservative posterior margins [25, 36] despite the lack of evidence that setup uncertainty or internal organ motion is lower in magnitude posteriorly at the rectal interface [37, 38]. Furthermore, in other reports of hypofractionated PORT there has been some heterogeneity in PTV margin definitions from as low as 2mm [30] or 3mm [39] to as high as 7mm [27] despite the use of daily image guidance in all cases.

Based on the observation that hypofractionated PORT appears to have a more substantial impact on late grade 3+ GU toxicity as compared to GI toxicity as well as higher rates of GU toxicity as compared to prior conventionally fractionated treatments [12, 25], the risk-benefit ratio of the inclusion of various post-prostatectomy anatomic zones in the treatment volume is likely different than in conventionally fractionated radiotherapy. Current CTV recommendations have been reported to inadequately provide coverage of the area posterior to the vesicourethral anastomosis (VUA) [40] and in the posterolateral aspect of the treatment field at the anorectal interface [41]. Based on the possibility that serious late GI toxicity appears less frequent than corresponding rates of serious late GU toxicity when delivering treatment with a hypofractionated as compared to a standard dose schedule, expansion the posterior aspect of the treatment volume to ensure adequate coverage of the VUA and posterolateral anorectal recess may improve oncologic control without substantial cost in terms of GI toxicity.

A final implication of these findings is that hypofractionated PORT may be most appropriate in a select patient population in whom rational field reduction strategies may be employed. One such group of these patients may be those in whom recurrent or residual disease can be radiographically detected within the prostate bed. In such a population, treatment volume reduction to the area of visible recurrence may be a viable strategy to control disease while limiting the volume of normal tissue exposure and consequent toxicity of treatment. A recent multi-institutional report of patients undergoing 5-fraction PORT to a limited CTV defined as gross disease with a 2mm margin supports this conclusion [42]. This study reported a favorable rate of biochemical remission in patients who did not receive ADT despite otherwise unfavorable baseline characteristics such as a mean pre-salvage PSA of 2.3 ng/mL. These results are somewhat more favorable than older, small, retrospective series reporting on the efficacy of focal brachytherapy in the salvage setting [43, 44] likely owing to the sensitivity of PET imaging to detect local recurrences earlier in their natural history in the contemporary reports.