Focal salvage therapy for local prostate cancer recurrences after primary radiotherapy: a comprehensive review

Background/Aim Patients with locally recurrent prostate cancer after primary radiotherapy can be eligible for salvage treatment. Whole-gland salvage techniques carry a high risk of toxicity. A focal salvage approach might reduce the risk of adverse events while maintaining cancer control in carefully selected patients. The aim of this review was to evaluate current literature to assess whether focal salvage leads to a comparable or favourable recurrence rate and less toxicity compared to whole-gland salvage. Methods A literature search was performed using PubMed, Embase and the Cochrane Library. A total of 3015 articles were screened and assessed for quality. Eight papers [on focal cryoablation (n = 3), brachytherapy (n = 3) and high-intensity focused ultrasound (n = 2)] were used to report outcomes. Results One-, 2-, 3- and 5-year biochemical disease-free survival (BDFS) ranges for focal salvage are, respectively, 69–100, 49–100, 50–91 and 46.5–54.5 %. Severe genitourinary, gastrointestinal and sexual function toxicity rates are 0–33.3 %. One study directly compares focal to whole-gland salvage cryotherapy, showing 5-year BDFS of, respectively, 54.4 and 86.5 % with lower toxicity rates for focal salvage patients. Conclusion Provisional data suggest that BDFS rates of focal salvage are in line with those of whole-gland approaches. There is evidence that focal salvage could decrease severe toxicity and preserve erectile function. Electronic supplementary material The online version of this article (doi:10.1007/s00345-016-1811-9) contains supplementary material, which is available to authorized users.

tumour, might be a viable treatment option for patients with unifocal PCa recurrences without metastatic disease. This way, serious adverse events associated with wholegland salvage might be prevented, while cancer control is maintained. The current literature regarding biochemical disease free survival (BDFS) and functional outcomes of focal salvage techniques for prostate cancer recurrences after primary radiotherapy is evaluated here.

Literature search
On 19 August 2015, a systematic literature search was performed in the PubMed, Embase and Cochrane Library databases. In order to create a sensitive search, the search syntax was build based on domain and determinant (Table 1). After removal of duplicates, 3015 papers were screened, selecting 55 articles with matching domain and determinant. Exclusion criteria are listed in the flow chart ( Fig. 1). An additional reference search was performed, resulting in 12 studies for this review [9][10][11][12][13][14][15][16][17][18][19][20].

Study selection
Studies were subjected to a critical appraisal based on an adaptation of the Dutch Cochrane Centre recommendations (ACROBAT-NRSI risk of bias assessment tool) [21]. Studies were graded (+), (±) or (−) on relevance (domain, determinant, outcome) and validity (selection, study population characteristics, exposure, primary outcome, secondary outcomes, follow-up and number of patients) [ Table A1 (supplementary file)].
All studies were observational cohorts or case series with either prospective or retrospective data acquisition. Therefore, no study could be considered of high quality. Blinding was applied in none of the studies (not graded). However, the primary outcome (BDFS) is unlikely to be influenced by information bias, due to the objectivity of the failure definition (Phoenix, i.e. PSA-nadir + 2.0 ng/ml). On the contrary, the toxicity assessment is subjective to information bias and is graded in that way.
Confounding was not graded, since all studies except one were single-arm observational cohorts. The study by De Castro et al. [9] describes a two-armed cohort. Here, possible confounding factors are adequately described, but not corrected for in multivariable analysis.

Search results
A total of eight studies were eligible for data extraction, describing focal salvage treatment with cryoablation (CA) (n = 3) [9,14,16], low-dose-rate brachytherapy (LDR-BT) (n = 3) [15,17,18] and high-intensity focused ultrasound (HIFU) (n = 2) [10,11] in 278 patients. Table 2 shows study characteristics. Studies used a salvage approach on a focal lesion (quadrant ablation by Ahmed et al. [10] and clinical target volume of approximately 17 % by Hsu et al. [15]), one half of the prostate (hemi) or a partial approach (area not specified by Li et al. [16] and Eisenberg et al. [14]). Nguyen et al. [17] describe LDR-BT to the entire peripheral zone. In all studies, ADT use was discontinued at time of enrolment.

Diagnosis of focality
All recurrences were initially detected by PSA measurement (biochemical failure), verified with prostate biopsies. The diagnostic modalities used to determine the recurrence location differed. Available specifics are described here.
Five studies used magnetic resonance imaging (MRI). Ahmed et al. [10] describe the use of a multiparametric approach: T2-weighted (T2 W), diffusion-weighted (DW) and dynamic contrast-enhanced (DCE) 1.5-Tesla (T) MRI. Results were combined with transrectal ultrasound (TRUS)-guided or transperineal template prostate mapping (TPM) biopsies (n = 19 and n = 20, respectively). Baco et al. [11] describe the use of T2 W, DW and DCE 3T MRI in 27 and T2 W and DW 1.5T MRI in 21 patients. Imaging results were verified a mean 15 cores TRUS-guided biopsies (sd 10, range 7-60). Peters et al. [18] describe the use of T1 W, T2 W, DWI and DCE 3T MRI, verified with systematic transrectal biopsies (median 10 cores, range 6-13). Sequences are not specified in the studies of Nguyen et al. [17] and Hsu et al. [15], though the latter did perform MR spectroscopy and acquired TRUS-guided biopsy confirmation of disease. Both do not define the amount of biopsy cores taken.
De Castro Abreu et al. [9] treated the tumour volume based on both systematic and lesion-targeted TRUS-guided biopsies and hypo-echogenic lesions on TRUS. Eisenberg et al. [14] treated based on 16-core sample TRUS-guided biopsy. Li et al. [16], using the Cryo On-Line Data (COLD) registry, do not describe any diagnostic modality used to define the location of the recurrence or (PSA-based) selection method.

Exclusion of metastatic disease
Three studies used positron emission tomography (PET) in excluding metastatic disease (Ahmed et al. [10], Baco et al. [11] and Peters et al. [18]). Ahmed et al. [10] used a combination of pelvic MRI, a radioisotope bone scan and 18-fluorodeoxyglucose (FDG) PET/CT in all patients for the assessment of metastatic disease (bone and lymph node). Baco et al. [11] used a combination of bone scan and pelvic CT or MRI. 11C-Choline PET was used in 27 of 42 patients. Peters et al. also use a bone scan and pelvis CT or MRI, but used 18F-Choline PET in 10 of the 20 patients.

Biochemical disease-free survival
Results on BDFS are also shown in Table 2. Biochemical failure was defined according to the Phoenix (PSAnadir + 2 ng/ml) or ASTRO definition (three consecutive post-nadir rises in PSA, with the moment of failure backdated between the nadir and the first rise). BDFS ranges at 1, 2, 3 and 5 years are, respectively, 69-100, 49-100, 50-91 and 46.5-54.4 %. The studies with an MRI-based assessment of focal disease had BDFS of 49-100 % up to 2-3 years compared to 72.4-79 % for non-MRI-based focal salvage series. The only 5-year Kaplan-Meier estimates come from the study by de Castro-Abreu et al. [9] and Li et al. [16], who do not use MRI for intraprostatic disease assessment or PET for exclusion of metastatic disease (de Castro-Abreu et al. [9]) or do not specify the assessment (Li et al. [16]). They provide BDFS of 46.5 % (Li et al. [16]) and 54.4 % (de Castro-Abreu et al. [9]).

Toxicity
Toxicity results are shown in Table 3 for various measurement methods. CTCAE grades indicate (1) toxicity without (the need for) intervention, (2) requiring medication, (3) requiring inpatient or outpatient surgical intervention, (4) requiring ICU admission and (5) death. No grade 4 or 5 toxicity was reported. Higher IPSS (prostate symptoms), lower IIEF (erectile function) and lower QLQ C-30 (quality of life) indicate deterioration. In addition, Ahmed et al. [10] reported surgical complications according to the modified Clavien system, showing the need for intervention under local (grade 3a) or general (grade 3b) anaesthesia in, respectively, 1 (3 %) and 9 (23 %) patients. Nguyen et al. [17] reported outcomes according to the Radiation Therapy Oncology Group/Late Effects Normal Tissue Task Force criteria, showing rectal bleeding in 2 (8 %), urethral stricture in 1 (4 %), periprostatic abscess in 1 (4 %) and prostate-rectal fistula in 3 (12 %) patients. Both Hsu et al. [15] and Peters et al. [18] report no new ≥grade 2 GI toxicity in the late phase (>3-6 months). Hsu et al. [15] report on five patients with ≥grade 2 GU toxicity requiring medication. Furthermore, two patients developed new medication-resistant erectile dysfunction (ED, grade 3) and two patients medication-responsive ED. Peters et al. [18] report one patient (5 %) with a grade 3 urethral stricture, requiring endoscopic incision. One patients suffered from radiation cystitis grade 2, managed with hyperbaric oxygen therapy. Five more patients had grade 2 urinary frequency, managed with medication. No new erectile dysfunction (ED) was seen in the five previously potent patients. Furthermore, a significant decrease in the EORTC-PR25 urinary symptoms quality-of-life (QoL) domain after a median of 3 years was observed.
Eisenberg et al. [14] describe the occurrence of one mild stress incontinence (grade not given), one urethral stricture requiring dilation and one prostatic urethral ulcer managed with suprapubic catheter drainage. Two out of five patients remained potent. Li et al. [16] describe 5 (5.5 %) patients requiring pad-use at 12 months and 3 (3.3 %) recto-urethral fistulas. Half of the patients retained potency, although medication was not specified.
A subdivision in more focally targeted ablation and studies using MRI-assessment for recurrences was not made due to (most importantly) the unavailability of salvage extent.

Comparing focal to whole-gland salvage
The aim of this study was to assess whether focal salvage for local prostate cancer recurrences after primary radiotherapy leads to a comparable or favourable recurrence rate and less toxicity compared to whole-gland salvage. De Castro Abreu et al. [9] present the only study comparing focal to whole-gland salvage cryotherapy. Recurrences were not verified by MR imaging. BF occurred in 32 % (focal) and 12 % (whole-gland), resulting in 5-year Kaplan-Meier BDFS estimates of, respectively, 54.4 and 86.5 %. This could be an indication that focal salvage patients are undertreated or that the relation is confounded  by other factors related to BF. Since recurrences were not assessed by MRI or TPM biopsies, focal salvage patients might have been under assessed. Furthermore, the comparison indicates that focal salvage might be less toxic: no focal salvage patients developed incontinence or recto-urethral fistula versus 3 and 1, respectively. Furthermore, two focal salvage patients retained potency, versus none in the total salvage group. However, these numbers were small and statistical significance was not achieved. This comparison between the two ablation methods is hampered by differences in primary radiation schedules/modalities, the extend of focal salvage and differences in patient characteristics.
In addition to reported results, Wenske et al. [37] describe a cohort of 55 focal CA patients after primary radiotherapy (80 %) or CA (20 %). There is no stratification for primary therapy. Reported 5-year BDFS (Phoenix definition) is 47 %. Bladder outlet obstruction occurred in 1.8 % and fistula formation in 5.5 % of patients.
Toxicity of focal approaches is comparable to or favourable compared to whole-gland salvage. Fistula rates are low across all studies. Notable results are reported by Hsu et al. [15] describing 15 focal salvage BT patients. There was no new GI toxicity and no new incontinence. Potency (medication assisted) was preserved in 13/15. Peters et al. [18] report preserved potency in the five previously potent patients. Potency is sparsely reported in SRP cohorts. From the assessed reviews, two SRP studies were retrieved describing SF. Masterson et al. [38] report preserved (medication assisted) potency in 6/40 (15 %), Boris et al. [39] in 2/10 (20 %).

Recurrent disease
Biochemical recurrences in prostate cancer often seem to stem from a localised process. Data on the exact location are scarce, but it is suggested that the tumour often recurs at the site of the primary dominant or 'index lesion' [7,8,40,41]. The evidence is not unanimous, however, with tumour regrowth frequently occurring multifocally in some studies, although organ-confined and unifocal disease remains being observed [42,43]. Also, radical prostatectomy or TPM biopsies are sometimes not used as the reference standard [7,40], thereby possibly giving an overestimation of local recurrences. The studies which use pathology as the reference standard usually identify less unifocal disease localisation [42,43]. However, in these studies patients were often treated with lower doses of radiation than with current dose escalation. A recent large analysis by Zumsteg et al. [1] in which 2.694 patients were treated with IMRT with a total dose ≥79.2 Gy (maximal 85.6 Gy) shows that biochemical recurrences are still common. Estimated 8-year recurrence rates of 9.7, 22.7 and 43.9 % for low, intermediate and high National Cancer Comprehensive Network (NCCN) risk groups was observed. It might be that with these dose schedules, secondary tumour foci are increasingly successfully treated. This could mean that radiorecurrent disease is shifting towards the index lesion and theoretically more patients in the future can be eligible for focal salvage. In addition, even though cancer control rates are increasing with further dose escalation, ADT use and enhanced patient selection, there will be an increase in absolute numbers of patients with biochemical recurrent disease because more patients are primarily treated with radiotherapy. The stage migration to lower risk disease due to PSA screening could lead to more recurrences originating from increasingly lower risk disease, which is possibly more often localised and unifocal.

Study limitations
Reporting outcomes and comparing studies is significantly limited by several factors. The primary limitation is the lack of randomised controlled trials. All studies found are observational cohorts or case series. Selection of a more favourable or motivated patient population is a possibility in these focal salvage groups, thereby possibly biasing the comparison between focal and whole-gland salvage and between focal salvage modalities in terms of cancer control and toxicity. Furthermore, the lack of blinding in all studies could have biased especially toxicity assessment, both on patient and physician level.
Secondly, a straightforward comparison of studies is difficult. There is no international consensus on the definition of failure. Although there is a tendency to report BDFS according to the Phoenix definition (used in all focal salvage studies), there is a wide variety of failure definitions, including combined biochemical and clinical (physical, radiological, histopathological) proof of disease. Gelet et al. [35] report a 73 % negative biopsy rate at 30 months; however, combining biopsy results with biochemical results and need for ADT, the disease-free rate drops to 38 %. In reporting toxicity outcomes, there is little consistency as well.
In addition, diagnostic modalities for the assessment of recurrences and the exclusion of metastases vary. Only Ahmed et al. [10], Peters et al. [18] and Baco et al. [11] use PET in all, 10 and 27 patients, respectively, to exclude metastatic disease. PET has increased accuracy in assessing lymph node and distant metastases over technetium-99 scintigraphy and/or CT/MRI [44][45][46]. This could have contributed to more favourable outcomes in terms of cancer control and could possibly lead to a further increase in the future if these modalities become the diagnostic standard. Also, new biopsy techniques could lead to a further increase in the assessment of a focal recurrence, without missing possible significant multifocal recurrent disease. TPM biopsies could lead to increased accuracy over systematic TRUS-guided biopsies alone, while MRI-guided biopsies might decrease the detection of insignificant disease further [47,48]. However, TPM biopsies were only adopted by Ahmed et al. [10], and MRI-guided biopsies by Baco et al. [11]. The other studies used different TRUSguided schedules, thereby possibly undertreating the prostate with a focal salvage approach.
Finally, study populations are relatively small, limiting power, and follow-up is relatively short, limiting the number of outcome events and thereby an accurate estimation of BDFS proportions and late toxicity.

Future trials
When considering trials for salvage modalities, willingness for randomisation is essential. This has been extremely problematic in a randomised study comparing prostatectomy and brachytherapy [49]. This accrual problem was also present in the PIVOT trial, in which only 15 % of patients were randomised [50]. Increasing favourable data from focal salvage studies (e.g. from the recently started FORECAST trial [Focal Recurrent Assessment and Salvage Treatment for Radiorecurrent Prostate Cancer] [51]) might further decrease the willingness for randomisation in potential future head to head salvage trials.
A solution can possibly be found in the cohort multiple randomised controlled trial (cmRCT) design [52]. In this study design, all patients with radiorecurrent disease (or all prostate cancer patients) in a centre would be included into a cohort. With consent, these patients can in the future be randomised into experimental and control groups when a new treatment becomes available, providing the standard of care to the control group and, after additional consent, a new modality to the experimental group. This could provide unbiased comparisons between salvage procedures and possibly even between salvage ablation modalities, without the need for substantial resources to achieve sufficient accrual.

Conclusion
In this review of studies on focal salvage therapies, provisional data suggest that BDFS rates after focal salvage are in line with those of whole-gland approaches. There is evidence that focal salvage could decrease severe toxicity and preserve erectile function. Based on these results, focal salvage can be considered a viable option for unifocal prostate cancer recurrences after primary radiotherapy in properly selected patients. For further research, there is a great need for randomised controlled trials comparing salvage ablation methods and possibly even modalities. These trials would need to be uniform regarding patient selection and in outcome assessment and reporting. Lastly, relevant endpoint (mortality) assessed after sufficient follow-up are preferred over proxy outcomes such as biochemical failure.