Prognostic value of initial prostate-specific antigen levels after salvage cryoablation for prostate cancer


  • David A. Levy,

    1. Department of Regional Urology, Glickman Urological and Kidney Institute, Cleveland Clinic Foundation, Cleveland, OH, and Department of Urology, University of Texas M. D. Anderson Cancer Center, Houston, TX, USA
    Search for more papers by this author
  • Louis L. Pisters,

    1. Department of Regional Urology, Glickman Urological and Kidney Institute, Cleveland Clinic Foundation, Cleveland, OH, and Department of Urology, University of Texas M. D. Anderson Cancer Center, Houston, TX, USA
    Search for more papers by this author
  • J. Stephen Jones

    1. Department of Regional Urology, Glickman Urological and Kidney Institute, Cleveland Clinic Foundation, Cleveland, OH, and Department of Urology, University of Texas M. D. Anderson Cancer Center, Houston, TX, USA
    Search for more papers by this author

David A. Levy, Department of Urology, Glickman Urological and Kidney Institute, Cleveland Clinic Foundation, 18200 Lorain Avenue, Cleveland, OH 44111, USA.



To assess the prognostic value of initial prostate-specific antigen (PSA) levels after salvage cryoablation (SCA) for the long-term biochemical progression-free survival (bPFS) in patients with prostate cancer.


In all, 455 hormone-naïve patients from the Cryo On Line Data Registry, and treated with whole-gland SCA were assessed for PSA-based bPFS using the Phoenix criteria. The initial PSA level measured after SCA was considered the nadir. Kaplan-Meier plots of bPFS for initial PSA level of <0.6, ≥0.6–≤5.0 and >5.0 ng/mL were constructed and plotted to 60 months.


In all, 280 patients had an initial PSA level of <0.6 ng/mL after SCA. At 12, 24 and 36 months 80%, 73.6%, and 67% of patients, respectively, were progression-free. For 118 patients with an initial PSA level after SCA of ≥0.6–≤5 ng/mL, 28% and 50% of these patients at 6 and 12 months, respectively, had PSA progression. Of 57 patients with an initial PSA level of ≥5 ng/mL, 64% progressed at 6 months. The PSA level before SCA and Gleason score correlated with bPFS by Spearman correlation (P < 0.001 and 0.002), respectively.


Curative therapy in prostate cancer not responding to radiotherapy is extremely challenging. There is no definition of success for cryosurgical treatment. The available data indicate that an initial PSA level of <0.6 ng/mL after SCA portends a favourable (67% at 36 months) bPFS. Individuals with initial PSA levels of ≥0.6 ng/mL after SCA are at risk of short-term biochemical progression (50% at 12 months).


salvage cryoablation


(biochemical) disease-free survival


Cryo On Line Data (registry)


(biochemical) progression-free survival


American Society for Therapeutic Radiology and Oncology.


In the late 1990s cryosurgery of the prostate was used predominantly as a salvage procedure for patients with recurrent prostate cancer after radiotherapy [1–3]. While biochemical standards of treatment success have been established for patients undergoing radiotherapy [4] and surgical extirpation [5], there are no such criteria for patients treated with primary or salvage cryoablation (SCA).

There are numerous studies that detail outcomes after salvage therapy for recurrent prostate cancer, from both the radiotherapy and surgical fields [6–11]. For SCA, outcomes have been reported using various endpoints of assessment. In 2001, short-term biochemical disease-free survival (bDFS) rates of 86% and 74% at 12 and 24 months, respectively, were reported from 38 patients treated at Columbia University, based upon Kaplan-Meier estimates [2]. In 2001, Izawa et al.[12] reported the effect of initial stage of disease and PSA level before SCA on biopsy-confirmed SCA failure from 145 patients treated at MD Anderson Cancer Center. Variables associated with the outcomes of SCA treatment were subsequently reported by Izawa et al.[13] in 2002. Using an increasing PSA level of ‘greater than nadir plus 2 ng/mL’ after SCA, as devised by the authors, that study reported 5-year DFS rates of 57% and 23% for patients with pre-treatment PSA levels of ≤10 and ≥10 ng/mL, respectively, and 90% and 69% for men with pre-radiotherapy clinical T1/T2 and T3/T4 disease, respectively, who had SCA. In 2006, Spiess et al.[14] reported a statistically significant difference in PSA doubling times (12.3 vs 5.6 months, respectively; P= 0.02) for men with lower (<10 ng/mL) vs higher (≥10 ng/mL) PSA levels before SCA. A PSA level of >10 ng/mL before SCA (P= 0.002) and a shorter (≤16 months) PSA doubling time after SCA (P= 0.06) were indicators of eventual biochemical failure, as defined by the Phoenix definition.

Currently, there are no studies which identify specific PSA-based guidelines of treatment outcomes after prostate SCA. Here we attempt to identify an evidence-based correlation of initial PSA levels after SCA and PSA progression based on Phoenix criteria after such therapy.


We identified 455 patients from the Cryo On Line Data (COLD) Registry who were treated with whole-gland SCA for biopsy-confirmed locally recurrent prostate cancer in the absence of hormonal influence. All patients had initial PSA levels measured at 6–12 weeks after SCA, and thereafter as indicated by the treating physician’s protocol. The patients were assessed for PSA-based biochemical progression-free survival (bPFS) based on Phoenix criteria, as applied to the initial PSA level recorded after SCA and follow up PSA level for up to 60 months. Kaplan-Meier plots of bPFS for initial PSA levels of <0.6, ≥0.6–≤5.0 and >5.0 ng/mL were constructed and plotted with available patient data.


The descriptive statistics of the patients are detailed in Table 1; 280 had an initial PSA level of <0.6 ng/mL after SCA. Within this subgroup, the associated bPFS was 80% (149 of 186) at 12 months, 73.6% (81 of 110) at 24 months and 67% (53 of 79) at 36 months (Fig. 1). In all, 118 patients had an initial PSA level after SCA of ≥0.6–≤5 ng/mL, 72% (61 of 85) of whom had bPFS at 6 months and 50% (24 of 48) of whom had PSA progression based on Phoenix criteria at 12 months (Fig. 1). For the 57 patients with an initial PSA level after SCA of >5 ng/mL, the 6-month bPFS rate was 63%. A log-rank test showed significant differences in PSA progression among the three PSA ranges (P < 0.001; Fig. 1, Table 1).

Table 1.  The demographics of the three groups based on specific initial PSA ranges after treatment, and estimates for DFS
Variable (n)PSA level, ng/mLStatistic, P
No. of men (455)28011857 
Mean (sd, range) age, years 70.2 (6.4, 46–86)69.5 (7.1, 49–90)70.9 (6.0, 57–88)t-test, 0.36
PSA level at diagnosis, ng/mL   Spearman r = 0.334, <0.001
 <4 (143)10633 4 
 4–9.9 (194)1324616 
 10–19.9 (71) 322118 
 >20 (41)  81518 
 Missing (7)  2 2 3 
T stage   Pearson chi2, 0.253
 <T2b (354)2258742 
 ≥T2b (101) 553115 
Gleason score   Spearman r = 0.178, <0.001
 ≤6 (191)1324316 
 7 (124) 812815 
 ≥8 (121) 574321 
 Missing (19) 10 4 5 
Estimates for DFS percentile (months), with log-rank test    
 25% 42 9 6<0.001
 50%  021 6 
 75%  06012 
Figure 1.

A Kaplan-Meier 60-month bPFS curve based on the Phoenix criteria for 455 patients with initial PSA levels after SCA of <0.6 (blue, 280), ≥0.6–≤5.0 (green, 118) and >5 ng/mL (orange, 57).

Lower PSA levels before SCA correlated with lower initial PSA levels afterward (Spearman correlation coefficient, 0.334, P < 0.001; Table 1). Also, pre-treatment Gleason scores correlated with PSA outcomes after SCA (Spearman correlation coefficient, 0.178, P= 0.002; Table 1).


PSA levels are a major component of assessing treatment outcomes for prostate cancer. In 1997, The American Society for Therapeutic Radiology and Oncology (ASTRO) Consensus Panel published criteria indicative of the failure of radiotherapy [15]. In 2006, ASTRO published a revised definition of biochemical failure for patients undergoing radiotherapy as the nadir PSA level plus 2 ng/mL, now commonly referred to as the Phoenix definition. Notably, the authors of that report specifically recommended that the definition not be used for treatments other than radiotherapy [4]. For the definition of surgical outcomes, in 2007 the AUA Prostate Cancer Guidelines Panel recommended a standardization of the definition of biochemical recurrence after radical prostatectomy. The panel listed an initial serum PSA level of ≥0.2 ng/mL, with a second confirmatory level of ≥0.2 ng/mL as the criterion for biochemical failure [5]. In the November 2008 publication of the ‘Best Practice Statement on Cryosurgery for the Treatment of Localized Prostate Cancer’ by the AUA [16] the panel was unable to make a statement about endpoints by which cryosurgical treatment success could be measured, due to a lack of available data.

For the published results from cryosurgical reports, both the 1997 ASTRO definition and the revised Phoenix criteria have been applied to cryosurgical patients as a means of assessing treatment outcomes. Based on the similarity to radiotherapy, in that the prostate gland remains in situ, we chose to use the Phoenix definition in the present study. Identification of a specific PSA level after SCA that can be correlated with favourable bDFS after such therapy is currently lacking, and is a major objective as one component of developing a definition of treatment success.

We recently reported the prognostic value of the initial PSA level after cryoablation for long-term bDFS in a risk-stratified cohort of 2427 patients treated with primary whole-gland cryoablation studied from the COLD Registry [17]. That study identified a PSA threshold of <0.6 ng/mL as prognostic for a favourable 60-month bDFS, with 86% low-, 67% intermediate- and 51% high-risk patients, respectively (by the criteria of D’Amico et al.[18]). A subsequent study delineated the prognostic value of relative disease burden in the prostate gland, as defined on univariate analysis by the number of positive cores (P= 0.031) and maximum percentage of cores positive (P= 0.024), and on multivariate analysis as the number of cores positive (P= 0.010), maximum percentage of cores positive (P= 0.034) and ratio of the number of positive cores to prostate gland volume (P= 0.023) as prognostic for favourable PSA outcomes, based on the identical PSA criteria after treatment, <0.6 ng/mL [19].

In an effort to begin a similar process for patients treated with SCA we conducted the present study. The present findings support the concept that an initial PSA threshold of <0.6 ng/mL portends favourable biochemical progression after SCA, similar to that reported in patients treated with primary whole-gland cryoablation. Although there was a statistically lower rate of PFS for the salvage patients than the primary cryoablation patients, the differences between these two groups cannot be over-emphasized and direct comparisons between them are unreasonable.

Patients with prostate cancer in whom radiotherapy fails are by definition a high-risk population in whom disease eradication is challenging. Risk stratification (D’Amico et al.[18]) criteria are not applicable in these men and were not used here. The current data indicate that a PSA level of <0.6 ng/mL after SCA portends a 67% 36-month bPFS, contrasted with a 12-month 50% bPFS rate, exceeding the Phoenix criteria, if the initial PSA level was ≥0.6 ng/mL. While these data support the use of a PSA threshold of <0.6 ng/mL as a potential benchmark for favourable outcomes in patients treated with SCA, and are consistent with what was reported in our earlier national report on primary whole-gland cryotherapy as described above [17], they are not intended to represent a definition of treatment success but perhaps one component of a developing definition of such.

The current study differs somewhat from our previous COLD Registry study, in that the PSA ranges used in the SCA population had to be broadened to provide for meaningful statistical interpretation of the data. This is primarily related to a slightly more erratic pattern of available follow-up data and represents one of the limitations inherent in a registry-based study. However, the number of evaluable patients at specified intervals was sufficient for data analysis and conclusions as stated. Treatment success or failure cannot be assessed based on PSA data alone.

PSA level, Gleason score and clinical stage are the main clinical measures used for assessing a man’s prognosis before intervention, and there have been critical assessments of such [18,20,21]. After therapy, several studies have incorporated these factors, as well as specific PSA kinetics, in an effort to identify prognostic factors for treatment outcomes [22,23]. In 2005, Lin et al.[24] reported a statistically significant correlation of high-risk disease, Gleason score 8–10 and short-term (<6 month) PSA doubling time after prostatectomy, for 205 of 1785 patients who had PSA failure. Teeter et al.[25] found that stage, relative preoperative PSA variables, high-grade disease (Gleason ≥7) and extraprostatic disease/seminal vesicle invasion were independent predictors of a short (<9 months) PSA doubling time and aggressive recurrence. Similarly, Naselli et al.[26] reported that a preoperative PSA of >20 ng/mL was a statistically significant independent predictor of a detectable PSA, as well as a predictor of PSA progression (P < 0.001) after radical prostatectomy in a cohort of 318 patients.

The finding of a statistically significant correlation of the pre-treatment PSA level (P= 0.001) and Gleason score (P < 0.001) with the PSA outcome after SCA is consistent with what has been reported in primary prostate cancer studies, as detailed above. While our institutional study of 122 patients treated with primary whole-gland cryoablation showed that only the relative disease burden was prognostic of favourable PSA outcomes, a subsequent registry-based study is clearly warranted to examine a potential relationship between biopsy-related data and PSA outcomes, to better delineate the role of prognostic variables in counselling treatment candidates.

The COLD Registry provides the potential for comprehensive data analysis of cryosurgical outcomes. Limitations of the current investigation include the somewhat erratic follow-up of submission of salvage data by participating physicians, and that it is an analysis of retrospectively collected registry data dependent on entry by physicians or ancillary staff. In addition, the specific surgical technique used by each surgeon is not known, so some variability of technique is likely to occur, which could affect treatment outcomes. Biopsy data after cryotherapy, currently being added to the Registry, will be an important correlate to the present data to better delineate the effect of the reported PSA findings. However, based on the available follow-up PSA data, the implications remain consistent with the previously published national study.

Every effort has been made to confirm all data points with each site, through ongoing support from Watermark Research Partners Inc. (Indianapolis, IN, USA), a company contracted to separate Endocare Inc. (Irvine, CA, USA), the study sponsor, as much as possible from the data, to improve integrity.

Several questions arise from our data analysis: What are the patterns of failure for those men who exceed the Phoenix definition? Will these men have progressively increasing or stable PSA levels? What is the biopsy status for these patients? What will hormonal therapy response patterns show? We will make an effort to extract that data when mature and report them accordingly. Also, metastasis-free survival will be an important variable to include in these studies.

In conclusion, based on published data, after primary whole-gland prostate cryoablation, initial PSA levels of <0.6 ng/mL are prognostic for bDFS. The present study shows consistent findings in patients treated with SCA with regard to a 0.6-ng/mL PSA threshold, although a different interval to biochemical progression was evident. A PSA level of <0.6 ng/mL after SCA continues to show prognostic value for bPFS after SCA of the prostate. PSA levels below this threshold cannot be used as a criterion of DFS until these data are correlated with biopsy data after treatment, and disease-specific and metastasis-free survival.


David A. Levy is a Proctor for Endocare. Louis L. Pisters and J. Stephen Jones are Board Members for Endocare.