Prostate Cancer Resource Center
Surrogate Endpoints in Localized Prostate Cancer

The Cancer Journal

Holly E. Hartman, MS, and William C. Jackson, MD


Randomized clinical trials assessing novel therapies in men with localized prostate cancer frequently require large patient numbers and more than a decade of follow-up to demonstrate improvements in overall survival. As the landscape of treatment options for prostate cancer is rapidly changing, clinical trials requiring long follow-up threaten to impede treatment improvements and run the risk of results being obsolete by the time that they are reported in publication. To address these issues, there has been tremendous interest in identifying an intermediate clinical endpoint that can be assessed earlier in the disease course to serve as a robust surrogate for overall survival in men with localized prostate cancer. Herein we review the relevant data for surrogate endpoints in localized prostate cancer, highlighting the work performed by the Intermediate Clinical Endpoints in Cancer of the Prostate Working Group identifying metastasis-free survival as a valid surrogate for men treated for localized prostate cancer.

Prostate cancer is the most common cancer in men, and it is estimated in 2019 that roughly 1 in 9 men will ultimately be diagnosed with prostate cancer in their life.1–3 While the number of men who die of prostate cancer each year has steadily declined since the 1990s because of improvements in therapies for localized and metastatic disease, death from prostate cancer remains the second leading cause of cancer death in men behind lung cancer.1–3 Because the vast majority of men initially present with localized disease,2 most prostate cancer–related deaths occur in men who previously received definitive local therapy with either prostatectomy or radiation therapy. Thus, clinical trials are motivated to assess novel definitive and adjuvant therapies for men diagnosed with localized prostate cancer.

Randomized clinical trials with a primary endpoint of overall survival are the criterion standard method by which to assess the efficacy of novel therapeutic options in all of medicine. Overall survival is a nonambiguous endpoint, easy to assess and interpret, and clinically meaningful. For randomized clinical trials assessing treatments for localized prostate cancer, overall survival as a primary endpoint necessitates considerable patient follow-up. The median time from the enrollment of the first patient on trial to publication of the primary outcome in this setting is often close to 12 years.4 This duration of time does not include the years spent upfront designing and implementing the clinical trial. Thus, the time from study concept to publication of overall survival results for trials assessing treatments for men with localized prostate cancer may be at least 15 years with 3 years for conception and implementation of a clinical trial. Furthermore, in order to be sufficiently powered to demonstrate differences in overall survival, these trials typically require enrolling hundreds to thousands of men.5–7 The need for large patient numbers and long-term follow-up to assess overall survival in men with localized prostate cancer is multifactorial. First, the median age at diagnosis of localized prostate cancer is 66 years.3 As such, most men will die of causes other than prostate cancer.3 The competing risks of death dilute the ability to detect changes in survival attributable to the therapy being assessed in the trial. Additionally, there are an increasing number of effective salvage therapies available for men who develop recurrent or metastatic disease following local therapy lending to improved long-term survival. While clearly beneficial to patients, this further contributes to the long follow-up needed to detect differences in overall survival between treatment arms. The required long-term follow-up and large number of patients increase the cost and resources necessary to perform trials. This usually restricts trials to being performed in a cooperative group setting and has also limited the number of trials performed in localized prostate cancer over the past few decades. Consequently, treatment of localized prostate cancer has made limited progress in the last decade.

As the landscape of treatment options for prostate cancer continues to grow, the long natural history of localized prostate cancer and the abundance of deaths from competing risks preclude expeditious assessment of novel therapies when trials have a primary endpoint of overall survival. This has led to enthusiasm over identification of an intermediate clinical endpoint, which could serve as a robust surrogate for overall survival in men with localized prostate cancer. Herein we discuss the use of surrogate endpoints, present statistical methods by which surrogacy is assessed, review intermediate clinical endpoints that have been evaluated for surrogacy to date, and highlight the groundbreaking work performed by the Intermediate Clinical Endpoints in Cancer of the Prostate (ICECaP) Working Group.4,8 This innovative work has identified metastasis-free survival as valid surrogate endpoint in men treated for localized prostate cancer with prostatectomy or radiation therapy.

Defining a Surrogate Endpoint

The Biomarkers Definitions Working Group defines a surrogate endpoint as “a biomarker that is intended to substitute for a clinical endpoint,” which “is expected to predict clinical benefit (or harm or lack of benefit or harm) based on epidemiologic, therapeutic, pathophysiologic, or other scientific evidence.”9 Ideally, a surrogate endpoint is able to be measured earlier or more conveniently than the true endpoint, adequately demonstrates the benefit of the treatment being assessed, and has a justifiable biological mechanism. Earlier endpoint assessment has the potential to shorten the duration of follow-up needed to demonstrate potential efficacy. Additionally, if the surrogate has a higher event rate and occurs more often relative to the true endpoint within a set follow-up time, then the sample size needed to detect a difference is reduced. In therapeutic trials for localized prostate cancer, intermediate clinical endpoints are most commonly assessed as potential surrogates for overall survival. These intermediate clinical endpoints are endpoints that occur prior to death from any cause. These include prostate-specific antigen (PSA)-based metrics, any form of recurrence, progression-free survival, and disease-free survival, among others. It is important to highlight that surrogate endpoints are both disease and treatment dependent and, as such, cannot be extrapolated from different treatment settings.

Assessing Surrogacy

Statistical identification and validation of a surrogate endpoint can be an arduous process. Two main approaches have been utilized to assess the potential surrogacy of intermediate clinical endpoints: fulfillment of the Prentice criteria,10 and the 2-stage meta-analytic approach proposed by Buyse et al.11 The Prentice criteria are an approach to assessing surrogacy of an intermediate endpoint within the results from a single randomized trial. The 4 criteria that must be satisfied to prove surrogacy are (1) the treatment must be significantly associated with the true endpoint; (2) the treatment must be associated with the occurrence of the surrogate; (3) the surrogate must be significantly associated with the true endpoint; and (4) the surrogate should fully account for the impact of the treatment effect on the true endpoint.10 As 3 of the 4 criteria require knowledge of the impact of a treatment on the true endpoint or the surrogate, the Prentice criteria can only be met when assessing prospective randomized data from a trial with a positive outcome.

A surrogate satisfies the Prentice criteria if (A) the treatment only affects the outcome through the surrogate so that the surrogate fully captures the treatment effect, and (B) the disease has an effect on the surrogate, which, in turn, has an effect on the outcome. With an ideal surrogate, the disease and treatment do not have an effect on the outcome except through the surrogate (Fig. 1A).

This can go awry several different ways. First, if there is an effect of the disease on the outcome that is not captured by the surrogate (Fig. 1B). Because the treatment only affects the outcome through the surrogate, the treatment may not actually be very effective. However, this is still a valid surrogate under the Prentice criteria. Another possibility is that the treatment has an effect on the outcome that is not captured by the surrogate (Fig. 1C). This effect could be positive or negative (e.g., a treatment lengthens PSA doubling time but negatively affects overall survival). This effect violates statement (A) above and makes the surrogate invalid. One extreme example of this occurred in the Nocturnal Oxygen Therapy Trial.12 There were no treatment differences on any surrogate outcomes, but there was nearly a 2-times increase in mortality in the treatment group. This suggests that there was only a treatment effect on the outcome, not on the surrogates, and that the disease may have additional effects on the outcome not captured by the surrogates (Fig. 1D). Problems can also arise if the treatment only affects the surrogate and this effect is not seen in the outcome (Fig. 1E). This violates (A) and (B) and also results in an invalid surrogate.

An alternative approach is the 2-stage meta-analytic approach proposed by Buyse et al.11 This methodology requires that 2criteria be met, one at the individual patient level and the second at the trial level. Individual-level surrogacy is that the surrogate is prognostic for the true endpoint. This can be assessed by examining the correlation between the surrogate and the true endpoint. Trial level surrogacy is that there is a strong correlation between the treatment effect on the surrogate and the treatment effect on the true endpoint of interest.3 Trial-level surrogacy is analogous to statement (A) of the Prentice criteria above, and individual-level surrogacy is analogous to statement (B).

Potential Surrogates for Localized Prostate Cancer

To date, multiple intermediate clinical endpoints have been proposed as surrogate endpoints for overall survival in men with localized prostate cancer, but until recently, none have been implemented into clinical practice. Intermediate endpoints that have been evaluated as potential surrogates in data from randomized clinical trials in localized prostate cancer include PSA posttreatment nadir, the interval to biochemical failure, PSA doubling time, clinical treatment failure, distant metastases, metastasis-free survival, and disease-free survival.5,13–18

RTOG 9202 was a randomized clinical trial comparing radiation therapy plus short-term androgen deprivation therapy (4 months) versus radiation therapy plus long-term androgen deprivation therapy (28 months) in 1520 men with high-risk prostate cancer.5 This trial has mature follow-up with 15-year results recently published, with an overall survival advantage associated with the use of long-term androgen deprivation therapy in this cohort.5 Multiple intermediate clinical endpoints have been assessed for surrogacy in the data from this trial. Most recently, the interval to biochemical failure was assessed as a potential surrogate.16 The interval to biochemical failure was defined as the time from completion of radiation therapy, to the time of biochemical failure using the Phoenix definition (PSA nadir +2 ng/mL).19 Using the Prentice criteria, the authors established that an interval to biochemical failure of 2 to 4 years satisfied all 4 Prentice requirements for surrogacy in this cohort, noting that 50% of men with biochemical failure within 3 years of radiation therapy died of prostate, compared with only 19% of men who experience biochemical failure more than 3 years posttreatment.5 Other intermediate clinical endpoints assessed in data from this trial include the posttreatment PSA doubling time, general clinical treatment failure, and development of distant metastases.14,15 Prostate-specific antigen doubling time was found to be associated with cause-specific survival, but failed to meet Prentice's criteria. Thus, PSA doubling time was not validated as a surrogate endpoint in this trial.14 Development of distant metastasis and general clinical treatment failure (defined as a local, regional, or distant recurrence; initiation of androgen deprivation therapy after completion of protocol treatment; or a PSA level >25 ng/mL after completion of radiation therapy) were also assessed as potential surrogate endpoints. At 3 years posttreatment, both endpoints fulfilled all 4 Prentice criteria as surrogate endpoints for prostate cancer–specific survival at 10 years posttreatment.15 Given that these endpoints were assessed only in a single trial, the authors stress the need for validation of their findings prior to broad adoption of these potential surrogate endpoints.

The interval to biochemical failure was also assessed as a potential surrogate endpoint for the true outcome of prostate cancer–specific mortality in data from TROG 96.01, which randomized 802 men with locally advanced prostate cancer to radiation alone, or with 3 or 6 months of androgen deprivation therapy.13 The authors demonstrated that an interval to biochemical failure of less than 1.5 to 2.5 years met all 4 Prentice criteria.5 They also assessed PSA doubling time and found that a PSA doubling time of less than 12 or less than 15 months similarly satisfied all 4 Prentice criteria. However, the interval biochemical failure correlated better with prostate cancer–specific mortality, the primary endpoint, leading the authors to favor this intermediate clinical endpoint.13 Again, the authors highlight that external validation is needed before either of these potential surrogate endpoints is implemented into clinical trial design.

Lastly, 4 intermediate clinical endpoints were assessed as potential surrogates for all-cause mortality in a smaller randomized trial out of the Dana-Farber Cancer Institute. This trial randomized men with localized prostate cancer to either radiation therapy alone or radiation therapy with 6 months of androgen deprivation therapy.17 The intermediate clinical endpoints assessed were PSA failure, a posttreatment PSA nadir greater than 0.5 ng/mL, a PSA doubling time of less than 9 months, and an interval to biochemical failure of less than 30 months.17 The authors demonstrate that a posttreatment PSA nadir greater than 0.5 ng/mL, a PSA doubling time of less than 9 months, and an interval to biochemical failure of less than 30 months all satisfied the 4 Prentice criteria. General PSA failure did not meet all the criteria. Overall, the proportion of treatment effect explained by the surrogates was greatest for a posttreatment PSA nadir of greater than 0.5 ng/mL, leading the authors to advocate for this endpoint as the most robust surrogate for all-cause mortality.17 This intermediate clinical endpoint had also previously been identified as a potential surrogate meeting the 4 Prentice criteria in a combined analysis between the TROG 96.01 trial and the Dana-Farber Cancer Institute trial.18

Overall, analyses from these 4 randomized clinical trials demonstrate multiple intermediate clinical endpoints meeting the Prentice criteria for surrogacy in men with localized prostate cancer. However, given that these findings were all in data from a single trial, these endpoints have not yet been incorporated as a means of expediting assessment of treatment efficacy in clinical trials. Many of the authors also highlight the need for validation in multiple randomized trials, preferably using a meta-analytic technique for statistical assessment of surrogacy. Additionally, because surrogates are treatment-specific, caution is required when implementing a trial with a novel treatment and a surrogate outcome to ensure there are no unintended effects of the treatment as discussed above.

Intermediate Clinical Endpoints in Cancer of the Prostate Working Group

The largest and most comprehensive effort to date to identify a surrogate endpoint for localized prostate cancer was performed by the ICECaP Working Group.4,8 This monumental effort sought to identify an intermediate clinical endpoint that would serve as a robust surrogate for men treated for localized prostate cancer with either prostatectomy or radiation therapy and independent of salvage therapies delivered in the setting of subsequent disease recurrence following local therapy. The authors performed a systematic review of randomized controlled trials for treatment of localized prostate cancer.4,8 In total they identified 43 trials meeting inclusion criteria and with data suitable for analysis. Individual patient data were provided for 28 of the 43 trials (n = 22,825 patients).8 These trials enrolled patients from 1987 to 2011. The ICECaP Working Group assessed both disease-free survival and metastasis-free survival as potential surrogates for overall survival using the meta-analytic 2-stage validation model of Buyse et al.11 Disease-free survival was defined as neither evidence of any recurrence (local, regional, or distant) nor death from any cause. Metastasis-free survival was similarly defined but included only distant recurrences. More individual patient data were available for the assessment of disease-free survival as a surrogate (n = 21,140) than for metastasis-free survival (n = 12,712), as disease-free survival was more commonly reported in the available trials. Notably, the majority of patients (90%) in the metastasis-free survival analysis were treated with radiation as their local therapy.8

The authors demonstrated a strong correlation between disease-free survival and overall survival, as well as between metastasis-free survival and overall survival. They subsequently demonstrated a strong correlation between treatment effect for both metastasis-free survival and disease-free survival on overall survival.8 From this analysis, metastasis-free survival and disease-free survival meet the 2 criteria for surrogacy when using the meta-analytic technique. As such, the authors claimed disease-free survival and metastasis-free survival as valid surrogates for overall survival in this cohort of men with localized prostate cancer and a 10-year 85% disease-specific survival.8 The correlation between metastasis-free survival and overall survival was stronger than that between disease-free survival and overall survival, leading the authors to favor metastasis-free survival as the preferable surrogate.8 Given the small number of surgically treated patients with metastasis-free survival data available, metastasis-free survival was not able to be independently assessed as a surrogate in the surgical subgroup, so these findings may be more applicable to men treated with definitive radiation therapy for their localized disease.

These comprehensive findings from the ICECaP Working Group have the potential to expedite future clinical trial completion. As an example, we present the results of RTOG 9601, a randomized trial of salvage after prostatectomy radiation with or without 2 years of bicalutamide. This trial reported improved overall survival in patients receiving 2 years of bicalutamide after 12 years of follow-up. However, they were able to report improved metastasis-free survival after only 7 years of follow-up.6 While in general we would not expect metastasis-free survival to shorten the duration of a clinical trial by 5 years, this example nonetheless highlights the potential as to how using metastasis-free survival as the primary endpoint may significantly shorten the required follow-up needed to demonstrate efficacy of new therapies in clinical trials for men with localized prostate cancer.

The ongoing ATLAS trial, a randomized trial of definitive radiation therapy and long-term androgen deprivation therapy in men with high-risk localized or locally advanced prostate cancer with or without the addition of apalutamide, is the first large phase III trial to use metastasis-free survival as the primary efficacy outcome.20 Based on the findings from ICECaP, it is expected that more trials will take this same approach moving forward.

Limitations of Surrogate Endpoints for Localized Prostate Cancer

A need clearly exists for surrogate endpoints for localized prostate clinical trial design. Advances in our understanding of cancer biology and continuing innovation of new effective therapies for prostate cancer have the potential to dramatically alter the way in which localized prostate cancer is treated in the coming years. However, clinical trials requiring more than a decade of follow-up threaten to slow needed progress and waste valuable resources because the questions being asked in clinical trials initiated now may be nonpertinent a decade from now.

While the overall concept of surrogate endpoints for survival in localized prostate cancer is appealing, surrogate endpoints are not without limitations. Prostate-specific antigen–based endpoints (such as biochemical failure, PSA doubling time, and PSA nadir) are attractive because they almost universally precede other intermediate clinical endpoints in prostate cancer. However, these endpoints have several shortcomings. First, PSA values can be significantly impacted by the use of androgen deprivation therapy, and this may not be consistent among different forms of androgen deprivation therapy. Even when androgen deprivation therapy has been discontinued, men can experience vastly different rates of testosterone level recovery, which can influence PSA values.21 Also, expected posttreatment PSA values are dependent on whether primary local therapy is delivered with prostatectomy or radiation therapy. Values such as PSA doubling time can also be influenced by how often PSA values are measured, and PSA doubling time can only be calculated retrospectively. Additionally, many men will experience rising PSA, but very few of these men will ultimately die of prostate cancer, given the large number of competing causes of death in this population. Because of the large number of confounding factors associated with interpretation of PSA-based endpoints, these potential surrogates have not been widely adopted.

Another limitation of surrogate endpoints is that they are treatment dependent. The meta-analytic approach utilized by the ICECaP Working Group limits this concern to some extent because patients were treated with radical prostatectomy or radiation therapy as local therapy. However, it is possible that surrogacy will not be maintained in the setting of new upfront and salvage therapies. Because patients included in the ICECaP analysis were enrolled on trials up to 2011, there were likely few patients who received abiraterone, enzalutamide, or apalutamide as part of their treatment course. It is unclear how these improvements in salvage therapies for men who fail localized therapy will impact the validity of metastasis-free survival as a surrogate endpoint. Additionally, the ICECaP analysis was performed on patients treated in the pre–molecular-based imaging era. As the use of prostate-specific ligand-based positron emission tomography imaging increases, so too does our ability to detect previously radiographically occult metastatic disease. It is possible that the surrogacy of metastasis-free survival may differ as our ability to detect low-volume metastatic disease earlier continues to improve. Thus, while there is an apparent need for surrogate endpoints for survival in men with localized prostate cancer, care must be taken to ensure that the surrogate being used is valid in the planned treatment setting.


Localized prostate cancer has a long natural history with a variable disease course. The majority of men treated for localized prostate cancer will die of causes other than prostate cancer. The long disease course and high competing risk of death from non–prostate cancer–related causes pose challenges to performing expeditious clinical trials with the criterion standard primary endpoint of overall survival. Such trials require large numbers of patients and frequently require more than a decade of follow-up to demonstrate improvements in survival. Intermediate clinical endpoints that have been validated as surrogate endpoints for overall survival have the potential to shorten the required duration of randomized trials and thereby allow earlier assessment of efficacy of novel therapies. To date, the most comprehensive assessment of surrogate endpoints in localized prostate cancer was performed by the ICECaP Working Group, with identification of metastasis-free survival as a valid surrogate for overall survival in men treated with surgery or radiation therapy. Metastasis-free survival has started to be incorporated as the primary efficacy endpoint in randomized trials for men with localized prostate cancer. The use of this surrogate as a primary endpoint will likely continue to increase in the coming years. Because of continued improvement in systemic and local therapies for prostate cancer and the increasing use of prostate-specific ligand-based positron emission tomography imaging allowing for earlier detection of low-volume metastatic disease, ongoing research is needed to ensure appropriate surrogate endpoints are being used in the proposed treatment setting.


1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin. 2019;69:7–34.

2. Negoita S, Feuer EJ, Mariotto A, et al. Annual report to the nation on the status of cancer, part II: recent changes in prostate cancer trends and disease characteristics. Cancer. 2018;124:2801–2814.

3. Miller KD, Siegel RL, Lin CC, et al. Cancer treatment and survivorship statistics, 2016. CA Cancer J Clin. 2016;66:271–289.

4. Sweeney C, Nakabayashi M, Regan M, et al. The development of Intermediate Clinical Endpoints in Cancer of the Prostate (ICECaP). J Natl Cancer Inst. 2015;107:djv261.

5. Lawton CAF, Lin X, Hanks GE, et al. Duration of androgen deprivation in locally advanced prostate cancer: long-term update of NRG oncology RTOG 9202. Int J Radiat Oncol Biol Phys. 2017;98:296–303.

6. Shipley WU, Seiferheld W, Lukka HR, et al. Radiation with or without antiandrogen therapy in recurrent prostate cancer. N Engl J Med. 2017; 376:417–428.

7. Bolla M, de Reijke TM, Van Tienhoven G, et al. Duration of androgen suppression in the treatment of prostate cancer. N Engl J Med. 2009;360:2516–2527.

8. Xie W, Regan MM, Buyse M, et al. Metastasis-free survival is a strong surrogate of overall survival in localized prostate cancer. J Clin Oncol. 2017; 35:3097–3104.

9. Biomarkers Definitions Working Group. Biomarkers and surrogate endpoints: preferred definitions and conceptual framework. Clin Pharmacol Ther. 2001;69:89–95.

10. Prentice RL. Surrogate endpoints in clinical trials: definition and operational criteria. Stat Med. 1989;8:431–440.

11. Buyse M, Molenberghs G, Burzykowski T, et al. The validation of surrogate endpoints in meta-analyses of randomized experiments. Biostatistics. 2000;1:49–67.

12. Nocturnal Oxygen Therapy Trial Group. Continuous or nocturnal oxygen therapy in hypoxemic chronic obstructive lung disease: a clinical trial. Ann Intern Med. 1980;93:391–398.

13. Denham JW, Steigler A, Wilcox C, et al. Time to biochemical failure and prostate-specific antigen doubling time as surrogates for prostate cancer– specific mortality: evidence from the TROG 96.01 randomised controlled trial. The Lancet Oncology. 2008;9:1058–1068.

14. Valicenti RK, DeSilvio M, Hanks GE, et al. Posttreatment prostatic-specific antigen doubling time as a surrogate endpoint for prostate cancer–specificsurvival: an analysis of Radiation Therapy Oncology Group Protocol 92–02. Int J Radiat Oncol Biol Phys. 2006;66:1064–1071.

15. Ray ME, Bae K, Hussain MH, et al. Potential surrogate endpoints for prostate cancer survival: analysis of a phase III randomized trial. J Natl Cancer Inst. 2009;101:228–236.

16. Dignam JJ, Hamstra DA, Lepor H, et al. Time interval to biochemical failure as a surrogate end point in locally advanced prostate cancer: analysis of randomized trial NRG/RTOG 9202. J Clin Oncol. 2019; 37:213–221.

17. Royce TJ, Chen M-H, Wu J, et al. Surrogate end points for all-cause mortality in men with localized unfavorable-risk prostate cancer treated with radiation therapy vs radiation therapy plus androgen deprivation therapy: a secondary analysis of a randomized clinical trial. JAMA Oncol. 2017;3: 652–658.

18. D'Amico AV, Chen M-H, de Castro M, et al. Surrogate endpoints for prostate cancer–specific mortality after radiotherapy and androgen suppression therapy in men with localised or locally advanced prostate cancer: an analysis of two randomised trials. Lancet Oncol. 2012;13:189–195.

19. Roach M 3rd, Hanks G, Thames H Jr., et al. Defining biochemical failure following radiotherapy with or without hormonal therapy in men with clinically localized prostate cancer: recommendations of the RTOG-ASTRO Phoenix consensus conference. Int J Radiat Oncol Biol Phys. 2006;65: 965–974.

20. Sandler HM, McKenzie MR, Tombal BF, et al. ATLAS: a randomized, double-blind, placebo-controlled, phase 3 trial of apalutamide (ARN-509) in patients with high-risk localized or locally advanced prostate cancer receiving primary radiation therapy. J Clin Oncol. 2016;34(15 suppl): TPS5087–TPS5087.

21. Nascimento B, Miranda EP, Jenkins LC, et al. Testosterone recovery profiles after cessation of androgen deprivation therapy for prostate Cancer. J Sex Med. 2019;16:872–879.