Tumor Control Probability Modeling and Systematic Review of the Literature of Stereotactic Body Radiotherapy for Prostate Cancer

Abstract

BACKGROUND

Dose escalation improves localized prostate cancer disease control and moderately-hypofractionationed external beam radiation is non-inferior to conventional fractionation. The evolving treatment approach of ultra-hypofractionation with stereotactic body radiation therapy (SBRT) allows possible further “biological-dose escalation” and shortened treatment time.

METHODS

The American Association of Physicists in Medicine Working Group on Biological Effects of Hypofractionated Radiotherapy/SBRT (WGSBRT) included a subgroup to study the prostate tumor control probability (TCP) with SBRT. We performed a systematic review of the available literature and created a dose-response TCP model for the endpoint of freedom from biochemical relapse (FFBR). Results were stratified by prostate cancer risk group.

RESULTS

Twenty-five published cohorts were identified for inclusion with a total of 4,821 patients (2,235 low-risk, 1,894 intermediate-risk, and 446 with high-risk disease, when reported) treated with a variety of dose/fractionation schemes, permitting dose-response modeling. Five studies had a median follow-up over 5 years. Dosing regimens ranged from 32-50 Gy in 4-5 fractions, with total BED (α/β=1.5 Gy) between 183.1-383.3 Gy. At 5 years, we found in patients with low-intermediate risk disease, an EQD2 of 71 Gy (31.7 Gy in 5 fractions) achieved a TCP of 90% and an EQD2 of 90 Gy (36.1 Gy in 5 fractions) achieved a TCP of 95%. In patients with high-risk disease, an EQD2 of 97 Gy (37.6 Gy in 5 fractions) can achieve a TCP of 90% and an EQD2 of 102 Gy (38.7 Gy in 5 fractions) can achieve a TCP of 95%.

CONCLUSIONS

We found significant variation in the published literature on target delineation, margins used, dose/fractionation, and treatment schedule. Despite this variation, TCP was excellent. Most prescription doses range from 35-40 Gy, delivered in 4-5 fractions. The literature did not provide detailed dose-volume data, and our dosimetric analysis was constrained to prescription doses. There are many areas in need of continued research as SBRT continues to evolve as a treatment modality for prostate cancer, including the durability of local control with longer follow up across risk groups, the efficacy and safety of SBRT as a boost to IMRT, and the impact of incorporating novel imaging techniques into treatment planning.

SUMMARY

We performed a review of the available literature and modeled a dose-response tumor control probability (TCP) for the endpoint of freedom from biochemical relapse. We found significant variation in the published literature on target delineation, margins used, dose/fractionation, and treatment schedule. Despite this variation, TCP was excellent. Limitations are a relative lack of data in high-risk patients and the need for longer follow up.

1. CLINICAL SIGNIFICANCE

Multiple randomized trials have consistently demonstrated that higher dose radiation treatment improves disease control for patients with localized prostate cancer,^1-5 and have established the modern standard of at least 75.6 Gy conventionally-fractionated external beam radiotherapy (EBRT) for the treatment of this disease.⁶ Additionally, a recent randomized trial using EBRT followed by a brachytherapy boost suggests additional dose escalation further improves disease control.⁷ Dose escalation with conventional EBRT has traditionally been constrained by concerns related to toxicities from nearby organs such as the rectum and bladder.

The long duration of EBRT has also led to the exploration of shortening the treatment course, and there are now nine published randomized trials suggesting that moderately hypofractionated EBRT delivered in 4-5 weeks can be as safe and effective as 8-9 weeks of conventionally fractionated EBRT.^8-16 These shortened treatments improve patient convenience¹⁷ and reduce health care costs. There is accumulating evidence from single-arm prospective trials exploring the safety and efficacy of ultra-hypofractionated stereotactic body radiation therapy (SBRT), usually delivered over only 1-2 weeks.^18,19 Additionally, results from the HYPO-RT-PC randomized trial comparing a 7-fraction SBRT-like regimen to conventional fractionation found similar rates of cancer control and toxicity at 5 years.²⁰ Further, prostate cancer appears to have a low α/β ratio²¹, a radiobiologic parameter value that suggests increasing fraction sizes could result in a higher biologically effective dose and may therefore be even more effective than dose-escalation with conventional fractionation, although this has not been definitively demonstrated.^22,23

Thus, the combination of improved patient convenience, reduced costs, and possibly improved cancer-control efficacy have fueled intense interest in the clinical studies and adoption of prostate SBRT. As prostate SBRT is still an evolving treatment paradigm, there is currently no standard dose fractionation regimen. This variation in the doses used across studies allows tumor control probability (TCP) modeling to assess if there is a dose-effect relationship. Therefore, the purpose of this study was to perform a systematic review of the available literature and characterize the probability of prostate cancer tumor control in patients treated with SBRT.

2. ENDPOINTS

The primary efficacy endpoint reported by most series is biochemical control using the PSA nadir + 2 ng/mL Phoenix definition of biochemical failure and reported as freedom from biochemical relapse (FFBR); this is what we used for this analysis.²⁴ Some studies have additionally reported clinical recurrences (e.g. local, regional, or distant) and PSA nadir.^25,26

Because localized prostate cancer is often a slow-growing disease, long-term endpoints are important. Five-year FFBR outcomes were used in our analysis whenever possible. Further, we analyzed FFBR separately for low, intermediate, and high-risk patient groups, which are defined using National Comprehensive Cancer Network (NCCN) ⁶ or the similar D'Amico criteria.²⁷

3. CHALLENGES DEFINING AND SEGMENTING ANATOMIC VOLUMES

Prostate SBRT can be planned using CT images alone. Consensus guidelines can help with consistent target volume delineation.²⁸ Adding MRI may help define the prostatic apex compared to CT²⁹, but adds challenges as well. A high degree of accuracy is required for the CT-MRI image fusion, as imprecision during this process can negate any benefit of added imaging data from MRI. Bladder and rectum volume differences between the time of CT and MRI acquisition can make precise imaging fusions especially challenging.³⁰ Further, studies have shown that a delineated prostate volume is usually smaller on MRI compared to CT and that MRI (and transrectal ultrasound) volumes are consistent with prostatectomy specimens^31,32 While theoretically treating a smaller volume can reduce treatment-related toxicity, it could also risk higher rates of cancer recurrence, especially for patients with higher-risk disease. It is unknown how MRI- vs. CT-based target structure delineation during prostate radiation treatment planning affects treatment-related toxicity and cancer control.

To aid in delineation of the prostatic urethra, CT simulation and MR images can be obtained with a Foley catheter in place. The urethra can also be identified with MRI alone.^33,34 Accurate urethral delineation is important to avoid radiation hot spots in the urethra during planning. The presence of the Foley catheter can also serve as a useful reference structure for CT and MRI image fusion. However, a potential drawback to the use of a Foley catheter during image acquisition is that removal of the Foley can cause changes in the prostate position.³⁵ This could result in planning images not accurately representing the patient’s anatomy during treatment, and have dosimetric consequences.³⁵

Another challenge in prostate SBRT planning is that there is no uniform standard for target definition (specifically, different studies have not consistently included the seminal vesicles) and volume expansions. Because SBRT plans are highly conformal, it is possible that even small variations in PTV margin or SV coverage may affect the probability of disease control, particularly in the setting of extracapsular extension.³⁶ Some studies have used a risk-adapted approach, with higher-risk patients receiving more seminal vesicle treatment and larger PTV margins; but there is significant variation across institutions (Supplemental Table). These variations are expected, because prostate SBRT continues to be a relatively new and evolving treatment modality; clinical experience and research data continue to accumulate to help refine optimal use of this treatment.

4. REVIEW OF OUTCOMES DATA

A PubMed search was performed using combinations of the following search terms: (prostate OR prostate cancer) AND (stereotactic OR SBRT OR hypofractionated OR hypofractionation). Series not in English were excluded. Series published through March 2018 were selected for initial review. Articles that did not report clinical data on definitive treatment using extreme hypofractionation monotherapy (≥ 6 Gy/fraction) for localized prostate cancer were excluded. We further excluded studies with fewer than 40 patients and studies using protons. Proton therapy was beyond the scope of this HyTEC (High Dose per Fraction, Hypofractionated Treatment Effects in the Clinic) initiative commissioned by the American Association of Physicists in Medicine Working Group on Biological Effects of Hypofractionated Radiotherapy/SBRT (WGSBRT). Studies with more than 5 fractions were excluded. In cases where cohorts were followed with two or more reports over a period of years, outcomes cited in the most recent publication were selected, though earlier publications were sometimes used if they provided more detail about treatment or analytic methods. Twenty-five published cohorts (2 described in a single publication³⁷), describing treatment and FFBR in a total of 4,821patients were ultimately selected for inclusion (Table 1).

Table 1.

Summary of included SBRT series.

	Cohort	N	Risk Category (%)			Platform	Dose/fractionation				ADT (%)	Med f/u (yrs)	FFBR (%)
			Low	Int	High		Gy	Fx	EQD2Gy	Schedule			2-yr	5-yr
1	RSSR39	437	43	49	8	CK, LINAC	35-38	4-5	85.0-119.4	NA	11	1.7	96.1	-
2	RPCR40	1743*	41	42	10	CK	35-40	4-5	102.5-108.6	NA	-	2.0	92.0	-
3	McBride et al.41	45	100	0	0	CK	36.25-37.5	5	90.6-96.4	4-20 days	0	3.7	97.7	-
4 z	Hannan et al.42	91	36	64	0	Tomo, LINAC	45-50	5	135.0-164.3	NA	17	4.5	100	98.6
5	Jackson et al.43	66	49	51	0	LINAC	37	5	94.1	q3d	0	3.0	100	-
6	Virginia Mason51	40	100	0	0	LINAC	33.5	5	78.5	qd	-	3.4	90.0	-
7	Royal Marsden44***	51	20	69	12	CK	36.25	5	90.6	NA	0	1.2	-	-
8	Western Australia34	45	24	62	13	CK	36.25	5	90.6	qod	16	1.5	100	-
9 z	pHART337	84	100	0	0	LINAC	35	5	85.0	qw	1	9.6	100	97.5
10 z	pHART637	30	60	40	0	LINAC	40	5	108.6	qw	0	6.9	100	96.7
11 z	21st Century Oncology46	102	100	0	0	LINAC	40	5	108.6	qod	-	5.0a	100	100
12 z	Genesis Healthcare Partners47	79	51	49	0	CK	38	4	119.4	NA	-	5.0a	100	95
13 z	Stanford60	67	100	0	0	CK	36.25	5	90.6	qd/qod	0	2.7	100	94
14	Erasmus MC52	50	60	40	0	CK	38	4	119.4	qd	0	1.9	100	-
15	Milan55	90	59	41	0	LINAC	35	5	85.0	qod	13	2.3	97.8	-
16	Gliwice50	400	53	47	0	CK	36.25	5	90.6	qod	58	1.3	99.5**	-
17	Olsztyn49	68	10	90	0	LINAC	33.5	5	78.5	biw	77	2.0	100	-
18	Naples45	112	NA	NA	NA	CK	35-36	5	85.0-92.6	qd	19	2.0	97.4	-
19 z	Seoul61	44	11	23	66	CK	32-36	4	86.9-108.0	qd	89	3.3	100 (low/int risk) 96.0 (high risk)	100 (low/int risk) 90.9 (high risk)
20	Finland56	218	22	27	51	CK	35-36	5	85.0-92.6	qod	65	1.9	95.4	-
21	Georgetown54	100	37	55	8	CK	35-36.25	5	85.0-90.6	qod	11	2.3	99.0	-
22 z	Flushing48	515	63	30	7	CK	35-36.25	5	85.0-90.6	qd	14	7.0	98.0 (low/int risk)b 72.0 (high risk)b	94.7 (low/int risk)b 68.6 (high risk)b
23	Vicenza53	100	41	42	17	CK	35	5	85.0	qd	29	3.0	96.0c	94.4c
24 z	Philadelphia62	142	43	44	13	CK	35-37.5	5	85.0-96.4	qod	28	3.3	97.9 (low/int risk) 86.7 (high risk)	94.4 (low/int risk) 86.7 (high risk)
25	Virginia Hospital Center57	102	36	55	8	CK	36.25	5	90.6	qd	9	4.3	100	-

^a5 years minimum of follow up

^bEstimated from published Kaplan Meir plot.

^cEstimated from published Kaplan Meir plot. Manuscript reports 94.4% at 3 years.

^*8% of patients not classified by risk group

^**Reported in earlier publication of this series⁷⁷

^***No tumor control statistics reported

^zPatient cohorts having at least 5 years of follow-up post-radiation and used in the final Tumor Control Probability model

Abbreviations: ADT, Androgen deprivation therapy; FFBR, freedom from biochemical relapse (FFBR); int, intermediate; EQD2Gy, equivalent dose in 2 Gy per fraction using an α/β=1.5; LINAC linear accelerator; CK, Cyberknife; RSSR, Radiosurgery Society Registry; RPCR, Registry for Prostate Cancer Radiosurgery; Tomo, tomotherapy; pHART, prostate Hypofractionated Accelerated RadioTherapy; min, minimum; qd, daily; qod, every other day; q3d every third day; qw, weekly; biw, twice a week ; NA, not available

One series of historical interest but ultimately not used in the final modeling is notable.³⁸ This study reported results for 232 patients treated from 1964-84 in the pre-PSA era with 2-D RT planning. The majority received 36 Gy prescribed to isocenter in 6 fractions over 18 days. 41% (treated prior to 1977) were treated with a 3-field technique or rotation using Co-60; thereafter, a 4-field box technique on a linear accelerator (linac) was used. Field sizes were seldom more than 10 x 10 cm. The median overall survival was 5.4 years and median cancer-specific survival was 8.1 years. For 115 patients treated after 1975, the “local response” based on an urologist’s examination of the prostate, was 67.6%.

The 25 cohorts included in the modeling were published after 2006. Two cohorts^39,40 were registry studies of 437 (Radiosurgery Society Registry, RSSR) and 1743 (Registry for Prostate Cancer Radiosurgery, RPCR) patients, 3 cohorts were multi-institutional,^41-43 and the rest were essentially single institutional cohorts. All treatments were image-guided, based on fiducials implanted in the prostate. Sixteen used the Cyberknife® system, 7 used conventional linacs, one⁶ used either linac or tomotherapy, and one²¹ used either Cyberknife or linac. FFBR defined by the Phoenix “nadir+2” criterion was used in all but 3 cohorts;^34,44,45 two used PSA decline trends,^34,44 and the other study described failure only as a rising PSA followed by salvage therapy.⁴⁵ Most studies had relatively short follow-up; 5 studies had median-follow-up of 5 years or more.^37,46-48

The vast majority of patients treated in published literature were low-risk (N=2,235) or intermediate-risk (N=1,894); only 446 patients had high-risk prostate cancer (Table 2). Given so few high-risk patients, these are likely highly-selected individuals and may not be representative of the overall patient population with high-risk prostate cancer.

Table 2.

Summary of series with patients with high-risk prostate cancer.

Cohort	High-risk patients (N)	Years post treatment	FFBR rate (%)
Royal Marsden44	6	1.2	NR
Western Australia34	6	2.0	100.0
Georgetown54	8	2.0	87.5*
Virginia Hospital Center57	8	2.0	100.0
Vincenza53	17	3.0	94.4
Philadelphia62	18	5.0	86.7
Seoul61	29	5.0	90.9
RSSR39	33	2.0	89.8
Flushing48	38	5.0	68.6
Finland56	111	1.9	92.8
RPCR40	172	2.0	87.0

^*Crude estimate

Abbreviations: FFBR, freedom from biochemical relapse; NR, not reported; RSSR, Radiosurgery Society Registry; RPCR, Registry for Prostate Cancer Radiosurgery

There are a variety of dose/fractionation and treatment schedules for prostate SBRT. Prescriptions ranged from 32-38 Gy in 4 fractions and 33.5-50 Gy in 5 fractions; total BED (α/β=1.5 Gy) for the reviewed studies ranged between 183.1-383.3 Gy. A commonly used regimen is 36.25 Gy in 5 fractions; however, some studies used a daily (QD) treatment schedule while others used every other day (QOD) in an attempt to reduce toxicity. Other studies used every-3-days,⁴³ twice a week,⁴⁹ or once a week treatments.³⁷ An underappreciated source of variation is that some series use “heterogeneous” planning, intentionally including hot spots (often 50% or higher, thus doses as high as 60 Gy) within the prostate to mimic brachytherapy.⁴⁷ The prescription dose typically included at least 95% of the PTV: exceptions were where coverage metrics were not specified^39,40,50 and where prescription was to isocenter with the prostate covered by the 90% isodose line.⁵¹ Androgen deprivation therapy is not commonly used with prostate SBRT.

Despite significant variations described above, published studies have shown favorable FFBR outcomes. Most series combined FFBR results for low- and intermediate-risk patients, with reported 2-year rates ranging from 90-100%, and reported 5-year results of 92-100%. Results from high-risk patients are lacking (Table 2), but one of the larger high risk studies (38 patients) having 5-year results showed a FFBR of 69%.⁴⁸

Other PSA endpoints reported include PSA bounce and nadir. A benign PSA bounce was reported in 13 cohorts and observed in approximately 10-45% of patients.^{37,39,55-57,40-42,44,46,52-54} These bounce rates appear similar-to-slightly higher than the 12-20% seen in conventional EBRT.^26,58 In cohorts with low ADT use (<20%) that reported PSA nadirs, values ranged from 0.1-0.5 ng/mL.^{37,42,43,52,54,55}

A review of the normal tissue (e.g. rectum, bladder) complication probabilities following prostate SBRT will be addressed in a separate manuscript.

5. FACTORS AFFECTING OUTCOMES

Broadly speaking, multiple factors (described below) have been demonstrated to affect tumor control outcomes following conventional RT. These same factors are likely important in the setting of SBRT.

1. Prostate cancer aggressiveness: This is defined using the patient’s PSA level, biopsy Gleason score, and clinical stage using the D’Amico criteria²⁷ or its later adaptation NCCN criteria.⁶ More aggressive cancer is associated with worse cancer control outcomes regardless of treatment modality, but whether cancer aggressiveness has an impact on sensitivity to dose per fraction, and consequently the α/β ratio for radiation therapy, is not fully understood. This is directly relevant for hypofractionation and SBRT.

2. Treatment planning factors: Relevant factors include whether and how much of the seminal vesicles are included, CT vs. MRI target delineation, margins used, dose/fractionation, prescription isodose line and hot spots within the prostate, and treatment schedule.

3. Systemic therapies: For higher-risk prostate cancers, multiple randomized trials have demonstrated that adding androgen deprivation therapy (ADT) to conventionally-fractionated EBRT improves cancer control and overall survival.⁵⁹ If indeed SBRT delivers a higher biological dose than conventionally fractionated EBRT, whether ADT is still necessary and improves cancer-control outcomes is unknown. Existing published studies have included mostly low- and intermediate-risk patients, and as a result, few patients have received ADT with SBRT.

6. MATHEMATICAL/BIOLOGICAL MODELS

The prostate TCP model was built with data from 24 of the 25 studies summarized in Table 1; one study did not provide tumor control statistics.⁴⁴ Prescription doses were used as model inputs because individual patient-level dose-volume histograms (DVH) were unavailable; individual patient-level DVHs would strengthen future modeling efforts. Prescription isodose lines covering the PTV were reported in 13 studies^{34,41,60-62,46,48,49,51,53,54,56,57} (mean prescription isodose line: 85%), 10 of which had prescription isodose lines of 70-90%. Available details regarding prescription and target delineation for each study are summarized in the Supplemental Table. The model results are stratified by risk group into either 1) low/intermediate or 2) high risk. This is because most studies report the outcomes of the low and intermediate risk patients together; and when they are provided separately, outcomes of low and intermediate-risk patients are similar. In studies that provided different FFBR rates for low and intermediate risk patients, a weighted average was used. Treatment time was not included in the model since the duration of therapy for the various schedules are generally short and similar to each other.

The TCP Model

Doses were converted to equivalent doses of 2 Gy per fraction (EQD_2Gy) based on the linear quadratic model.⁶³ An α/β value of 1.5 Gy was used to account for the effect of fractionation.⁶⁴ The EQD_2Gy was calculated from the prescription dose with the following:

$${\text{EQD}}_{2\text{Gy}}=D\cdot \frac{\alpha ∕\beta +d}{\alpha ∕\beta +2}$$ (1)

where D is the total dose, and d is the dose per fraction (both expressed in Gy).

The Poisson model was used to calculate the TCP, mathematically expressed as follows^65,66:

$$\text{TCP}=2^{-\text{exp}\left[e\gamma \left(1-EQ{D}_{2\text{Gy}}∕D_{50}\right)\right]}$$ (2)

The basic parameters of each model are: D₅₀, which is the equivalent dose for a FFBR rate of 50%; and γ, which is the maximum normalized slope (gradient) of the dose response curve. The values of the parameters of the TCP model were determined using the maximum likelihood method. The fitting process weighted each study according to its size. The profile likelihood method was used to determine the 95% confidence intervals of model parameters.

Model Results

We examined the data at 2 and 5 years following treatment, which included 3659 and 1154 patients, respectively (Table 1). Given the long natural history of prostate cancer, we present the more relevant 5-year outcomes and the data were observed to fit better for the 5-year time point (as indicated by a significance of fit p value; 0.037-0.039 for 5 years, whereas for the 2 year time point both the low-intermediate risk and the high-risk model had p values > 0.05); the 5-year model parameters are shown in Table 3. The 5-year dose-response curves from the model, with their 95% confidence intervals are shown in Figure 1. The reviewed studies did not include equivalent prescription doses below approximately 80 Gy, represented by the faded regions of the curves – therefore these regions are only suggestive in the model and clinical FFBR conclusions should not be made in these regions that extend outside the available data. The contributing sample size of each cohort included in the model is reflected by the proportional corresponding circle diameter on the Figure. Further, for studies that reported a single outcome for treatments with several different fractionations, the x-coordinate is the median prescription EQD_2Gy.

Table 3.

Summary of the Poisson model parameters. These were derived by fitting the data for the two risk groups with at least 5 years follow-up post-radiation. The D₅₀, is the equivalent dose in 2 Gy fractions for a tumor control rate of 50%; and γ, is the slope (gradient) of the dose response curve. The values in the parentheses represent the 95% confidence intervals of the corresponding parameters. The p value of the significance of fit was calculated based on the likelihood ratio test.

Risk group	D50 (Gy)	γ	p
Low-Intermediate	20.6 (18.9-22.3)	0.15 (0.13-0.17)	0.039
High	84.2 (81.4-86.8)	4.50 (2.82-6.53)	0.037

Figure 1.

The dose response curves of the Poisson prostate tumor control probability model for the Low-Intermediate risk (left) and High risk (right) patient cohorts having at least 5 years for follow-up post-radiation. The dashed lines indicate the 95% confidence intervals of the curves. The dose-response data of each patient cohort is plotted in the form of open circles, whose size is proportional to the size of the patient cohort from each study relative to the size of the overall risk cohort; 95% CI error bars are shown. The plot for the Low-Intermediate risk patients was based on data from 9 studies* (1069 patients)^{37,42,46-48,60-62}; the plot for the high-risk patients was based on 3 studies (85 patients).^48,61,62 (*One study is not shown in the plot because it lies outside the x-axis dose range⁴²). The error bars were calculated assuming a binomial distribution of the data. FFBR, freedom from biochemical relapse (FFBR). The reviewed studies did not include equivalent prescription doses below approximately 80 Gy, represented by the faded regions of the curves – therefore these regions are only suggestive in the model and clinical FFBR conclusions should not be made in these regions that extend beyond the available data.

7. SPECIAL SITUATIONS

These findings are most applicable to patients with intact and untreated localized prostate cancer. The efficacy of SBRT as a boost or in the salvage setting after prior radiation or other definitive local therapies such as radical prostatectomy, high-intensity focused ultrasound, or cryotherapy is unknown and we recommend caution generalizing these findings to those clinical situations. There are emerging data suggesting SBRT is an option as a salvage treatment after prior radiotherapy but it remains preliminary.^67,68 Additionally, with randomized data showing that conventional EBRT with a brachytherapy boost has superior disease control compared to EBRT alone in patients with higher-risk prostate cancer⁷, the efficacy of an SBRT boost instead of brachytherapy is under investigation.^69,70

8. IMPLICATIONS FOR BIOLOGICAL DOSE EVALUATIONS

The modeling results in Section 6 (Table 3) and Figure 1 at five years reflect a significant dose-response fit. Based on the low-intermediate risk and the high risk models, in patients with low-intermediate risk, an EQD2 of 71 Gy (31.7 Gy in 5 fractions) can achieve a TCP of 90% and an EQD2 of 90 Gy (36.1 Gy in 5 fractions) can achieve a TCP of 95% at 5 years. In patients with high risk, an EQD2 of 97 Gy (37.6 Gy in 5 fractions) can achieve a TCP of 90% and an EQD2 of 102 Gy (38.7 Gy in 5 fractions) can achieve a TCP of 95% at 5 years. However, the authors urge caution in interpreting the data for high-risk patients because few such patients are included in the published literature and these are likely highly-selected patients.^19,71 The majority of the patients in the reviewed studies had the prescription isodose line covering ≥95% of the planned target volume.

There is much variation in the published literature regarding target delineation, margins used, dose/fractionation, and treatment schedule (summarized in Table 1 and Appendix). Despite this variation, reported tumor control outcomes are excellent. Most of the prescription doses range from 35-40 Gy, delivered in 4-5 fractions. A published dose-escalation trial demonstrated high rates of rectal fistula in its highest dose level of 50 Gy in 5 fractions.⁷² With higher doses, tradeoffs may occur between PTV coverage and normal tissue constraints, and we recommend reporting of prescription isodose lines and structure margins, as detailed in Section 10. With excellent tumor control from 35-40 Gy in 4-5 fractions, and based on the modeling results summarized above, we do not recommend higher SBRT doses outside of protocols due to toxicity concerns and likely lack of further tumor control benefit.

9. FUTURE STUDIES

With favorable tumor control outcomes in the published literature, prostate SBRT is recognized in existing guidelines as a standard option for patients with low and intermediate risk disease.^6,73 However, prostate SBRT is still an evolving treatment modality and follow up for many cohorts continues to mature – the tumor control and toxicity outcomes with long-term follow-up are highly anticipated. Many other opportunities for future studies remain:

1. How does the tumor control efficacy of SBRT compare with more conventionally-fractionated IMRT? BED calculations suggest that ultra-hypofractionated treatment with SBRT delivers a higher biological dose than conventional fractionation. Existing and ongoing randomized trials including NRG GU005 (36.25 Gy in 5 treatments vs. 70 Gy in 28 treatments; NCT 03367702) and Prostate Advances in Comparative Evidence (PACE)(36.25 Gy in 5 treatments vs. 62 Gy in 20 treatments or 78 Gy in 39 treatments; NCT 01584258) examine whether this higher BED translates to improved cancer control.^19,20

2. Efficacy of SBRT in the treatment of unfavorable intermediate-risk and high-risk prostate cancer. The literature is relatively sparse on the outcomes of these patients with more aggressive prostate cancer, and included patients in published series are likely highly selected. Component questions include whether α/β ratio is similar for high vs. low-risk prostate cancer, what is the optimal treatment volume (e.g. how much of the seminal vesicles should be included) and PTV margin, and whether adding ADT to SBRT improves cancer control and survival outcomes.

3. Efficacy and safety of SBRT as a boost to IMRT. A randomized trial for patients with intermediate- and high-risk prostate cancer demonstrated that combination IMRT plus brachytherapy significantly improved long-term FFBR compared to dose-escalated IMRT.⁷ This trial has led to increased enthusiasm for research combining IMRT (which can include prophylactic pelvic treatment) with SBRT instead of a brachytherapy boost in the same patient population.^69,70

4. The optimal dose/fractionation for prostate SBRT is yet undefined, and head-to-head comparisons of different regimens on cancer control outcomes might be helpful.

5. With increasing use of multi-parametric MRI during the prostate cancer diagnosis and work-up process, more tumors within the prostate are being visualized. This presents an opportunity for incorporation of the tumor location information during SBRT treatment planning. Research on possible GTV delineation based on visualized tumor(s) on the MRI and dose-escalation to the GTV, while still treating the entire prostate to some “minimum prescription dose”, is necessary to determine if this approach further improves the therapeutic ratio of prostate SBRT. Additionally, the toxicity and cancer control outcomes of MRI- vs. CT-based prostate definition during the treatment planning process also requires study.

6. The efficacy and safety of SBRT as a salvage treatment – after prior prostate radiation, cryotherapy, or high-intensity focused ultrasound – is mostly unknown. Further, SBRT may represent another technology for focal therapy – which is currently deemed investigational.⁷³ Existing imaging technology, including multi-parametric MRI, may not be sufficiently accurate to allow a high degree of confidence that treating only part of the prostate will eradicate all of the cancer.

7. PSA control is used as a surrogate for primary tumor clearance, with the understanding that some patients have biochemical recurrence due to distant failure. The studies modeled herein were largely done in the pre-prostate cancer specific positron emission tomography era, and so perhaps future studies will add precision to the TCP estimates by incorporating modern imaging that may facilitate differentiation between local and regional or distant failure.⁷⁴

10. REPORTING STANDARDS FOR OUTCOMES

In order to facilitate future analyses of pooled data, a uniform approach and minimum standards for reporting outcomes and treatment is recommended.⁷⁵

• Outcomes. Although originally developed and validated for outcomes after conventionally-fractionated EBRT or brachytherapy, the Phoenix definition of biochemical failure has been used as the de facto standard for biochemical outcomes with SBRT as well.²⁴ As SBRT series report long-term outcomes, additional endpoints become more relevant, including metastasis-free survival, prostate-cancer specific survival, and overall survival.

• Clinical characteristics. Pretreatment prognostic variables should include at a minimum, age at diagnosis, clinical stage (TNM), Gleason score, PSA, percent positive biopsy cores, as well as NCCN risk category.

• Therapies. The use, duration, and timing of adjuvant or concurrent treatments, including ADT, should be reported.

• Radiation planning. Treatment planning details should be provided including how target volumes are defined, margins utilized, imaging modalities used for simulation, prescription isodose line and its relation to the target, dose/fractionation, and treatment schedule. Adopting standardized nomenclature as recommended by American Association of Physicists in Medicine’s Task Group 263 (AAPM TG-263) will be particularly helpful for future data pooling.⁷⁶ Most journals (e.g., the International Journal of Radiation Oncology, Biology, Physics) allow online appendices where raw outcome and dose volume data can be reported. Reporting raw data will allow future efforts for combined analysis and TCP modeling to further gain insights on this evolving treatment modality.