Outcome Predictors for the Increasing PSA State After Definitive External-Beam Radiotherapy for Prostate Cancer
http://www.100md.com
《临床肿瘤学》
the Department of Radiation Oncology, Department of Epidemiology and Biostatistics, Solid Tumor Service Department of Medical Oncology, and Department of Urology, Memorial Sloan-Kettering Cancer Center, New York, NY
ABSTRACT
MATERIALS AND METHODS: A total of 1,650 patients with clinical stage T1 to T3 prostate cancer were treated with high-dose three-dimensional conformal radiotherapy. Of these, 381 patients subsequently developed three consecutive increasing PSA values and were characterized as having a biochemical relapse. The median follow-up time was 92 months from the completion of radiotherapy.
RESULTS: The 5-year incidence of DM after an established PSA relapse was 29%. In a multivariate analysis, PSA doubling time (PSA-DT; P < .001), the clinical T stage (P < .001), and Gleason score (P = .007) were independent variables predicting for DM after established biochemical failure. The PSA-DT for favorable-, intermediate-, and unfavorable-risk patients who developed a biochemical failure was 20.0, 13.2, and 8.2 months, respectively (P < .001). The 3-year incidence of DM for patients with PSA-DT of 0 to 3, 3 to 6, 6 to 12, and more than 12 months was 49%, 41%, 20%, and 7%, respectively (P < .001). Patients with PSA-DT of 0 to 3 and 3 to 6 months demonstrated a 7.0 and 6.6 increased hazard of developing DM or death, respectively, compared with patients with a DT more than 12 months.
CONCLUSION: In addition to clinical stage and Gleason score, PSA-DT was a powerful predictor of DM among patients who develop an isolated PSA relapse after external-beam radiotherapy for prostate cancer. Patients who develop biochemical relapse with PSA-DT ≤ 6 months should be considered for systemic therapy or experimental protocols because of the high propensity for rapid DM development.
INTRODUCTION
The prognosis of patients with an increasing PSA profile after failed primary radiotherapy is not well defined. A better understanding of the natural history of biochemical relapse and which specific increasing PSA pattern would most reliably predict for a tangible impact on the clinical outcome (such as distant metastatic spread) have not been well explored. The dominant challenge is to determine more effectively whether the source of the increasing PSA level after radiotherapy represents persistent local disease, systemic failure, or both. This information would have obvious implications for deciding the most appropriate therapy. Some have advocated earlier treatment interventions for patients who experience biochemical relapse only if the clinical characteristics before initial therapy were consistent with a more aggressive tumor phenotype. Others have used the PSA kinetics and response after radiotherapy, such as the PSA doubling time (PSA-DT), whereas other investigators have relied on arbitrary PSA levels or thresholds to trigger the initiation of therapy.1–3
We retrospectively reviewed the outcomes of 381 patients who developed a biochemical relapse after three-dimensional conformal radiotherapy (3D-CRT) and who were observed subsequently. In these patients, the PSA kinetics was carefully evaluated to determine which parameters influenced the development of DM. Our findings suggest that among other variables identified, the PSA-DT represents the most important variable predicting for DM and survival after a biochemical relapse, and we suggest that this parameter should be used as an integral part of the decision-making process for initiating therapy in those who develop a PSA relapse after primary therapy.
MATERIALS AND METHODS
The pretreatment characteristics of these 381 patients are listed in Table 1. In general, patients were treated with a five- or six-field conformal treatment approach as previously described.4 The median radiotherapy dose was 75.6 Gy (range, 64.8 to 81 Gy). Doses were prescribed to the maximum isodose line, which completely encompassed the planning target volume. The planning target volume was defined as the prostate and seminal vesicles with a circumferential 1-cm margin except at the prostate-rectal interface, where a 6-mm margin was used. The techniques for treatment planning and delivery of 3D-CRT and intensity-modulated CRT have been described in detail elsewhere.4 Treatment was delivered with 15 MV x-rays in daily fractions of 1.8 Gy. One hundred fifty-three patients (40%) with enlarged prostate glands were given a 3-month course of neoadjuvant total androgen deprivation therapy (ADT), and this therapy was discontinued at the completion of radiotherapy. Patients were classified into prognostic risk groups according to the National Comprehensive Cancer Network practice guidelines recurrence risk groups (www.nccn.org). Patients were classified as having favorable risk with clinical stages T1 to T2a, a Gleason score ≤ 6, and a pretreatment PSA less than 10 ng/mL. Patients with the presence of clinical stage T2b or T2c, a Gleason score of 7, or a pretreatment PSA of 10 to 20 ng/mL were classified as having intermediate risk. If patients had clinical stage T3a or higher, a Gleason score between 8 and 10, or a pretreatment PSA more than 20 ng/mL, they were classified as having unfavorable risk.
Follow-up evaluations after treatment were performed at intervals of 3 to 6 months for 5 years, and yearly thereafter. The median follow-up time was 92 months (range, 12 to 159 months) from the completion of radiotherapy. The median time from the completion of radiotherapy to the date of established biochemical relapse was 33 months (range, 3 to 177 months), and the median follow-up time from the date of established biochemical relapse among patients who did not experience DM was 43 months (range, 0 to 145 months). At the time of a documented relapse, patients routinely underwent a restaging evaluation that included imaging of the pelvis (computed tomography or magnetic resonance imaging) and bone scan. These tests were often repeated at 6-month intervals as part of the patient’s continued evaluation.
The primary end point of this analysis was the incidence of DM in patients with a documented biochemical relapse after 3D-CRT. Survival outcomes were measured from the date when PSA relapse was established (the date of the third increase in PSA). For this analysis, we applied stringent criteria to the definition of biochemical relapse using the American Society for Therapeutic Radiology and Oncology definition of three consecutive increases in PSA; however, we did not backdate the time of recurrence to the midpoint between the date of the nadir PSA value and the date of the first increasing PSA value. We defined time zero to be the date of the third increase in PSA, which is the clinically recognized time of the established biochemical relapse. This avoids a lead time bias, which artificially inflates the survival time that can be associated with backdating the PSA relapse date. The median time interval between the date of the first to the third consecutive increase in PSA was 16 months (range, 2 to 79 months). The incidence of DM was calculated using a competing-risk analysis accounting for death as a competing risk.4 Time to DM or death was calculated from the date of biochemical relapse to the date of DM, death, or last follow-up. The hazard ratios were computed using competing-risk regression and compared using a {chi}2 test.5
The relationship between PSA and time was modeled using a log linear form; the logarithm of PSA was modeled as a linear function of time. The PSA-DT was calculated as the reciprocal of the slope of this regression equation, for which the logarithm was taken with a base of 2. The PSA values used in the regression model to calculate the DT were only those PSA values from the start to the end of the three consecutive increases used to determine the biochemical relapse. Some of the PSA values before the first increasing value were zero or close to zero, which would lead to short DTs even when, in fact, the increase was incrementally small. To correct for this, a sensitivity analysis was performed and 0.2 was added to every PSA value before the logarithm was calculated. The PSA values used in the regression model to calculate the doubling time were only those PSA values from the start to the end of the three consecutive increases used to determine the biochemical relapse. The association between the DT and the pretreatment variables was assessed with the Kruskal-Wallis test.6 A multivariate generalized linear regression model was built with the log of PSA-DT as the dependent variable, and clinical T stage, Gleason score, use of neoadjuvant ADT, and log of the starting PSA value as the independent variables.
RESULTS
Given that PSA-DT was determined to be an important clinical variable reflecting the prognosis of patients with PSA relapse after 3D-CRT, variables were identified that influenced the PSA-DT, as shown in Table 4. These variables included the pretreatment PSA, biopsy Gleason score, and the clinical T stage. The PSA-DT for favorable-, intermediate-, and unfavorable-risk patients who developed a biochemical failure was 18.0, 12.3, and 8.8 months, respectively (P < .001). A multivariate analysis of variables that influenced the PSA doubling time is shown in Table 5.
The strong relationship of the PSA-DT with incidence of DM is shown in Figure 2. The 3-year incidence of DM rates for patients with PSA-DT of 0 to 3, 3 to 6, 6 to 12, and more than 12 months were 49%, 41%, 20%, and 7%, respectively (P < .001). Patients with PSA-DT of 0 to 3, 3 to 6, and 6 to 12 months demonstrated a 7.0-, 6.6-, and 2.8-fold increased risk of developing DM or death, respectively, compared with patients with DT greater than 12 months.
During the follow-up evaluation, 110 patients were subsequently treated with salvage ADT for their biochemical relapsing disease. The incidence of DM among patients who were treated with salvage hormonal therapy was reduced during the early follow-up period compared with patients who were observed without salvage ADT, but no obvious differences were apparent between these groups (data not shown) with additional follow-up.
DISCUSSION
Scher and Heller7 have adopted a clinical state model to describe the natural history of prostate cancer. Unlike the standard TNM staging system in which the patient is characterized with a clinical stage at the point of his diagnosis, the clinical state model relates to the entire spectrum of the disease during the lifetime of the patient and describes untreated and treated prostate cancer in sequential phases from prediagnosis to death. At any time point, a patient can reside in only one clinical state. At each clinical encounter the patient is assessed using clinical, biochemical, or other biologic determinants to predict the probability of progressing to the next clinical state, or more importantly, of developing symptoms or of succumbing to disease. The increasing PSA state can be viewed as a separate clinical state in which patients have experienced treatment failure after a primary therapy, yet have not manifested evidence of a clinical relapse. Within the increasing PSA state, the natural history can exhibit extraordinary variation from one patient to another. Our study suggests that the PSA-DT would be a useful prognostic tool for patients in this state to identify those patients who are most likely to migrate rapidly to the next clinical state of systemic disease compared with patients who are likely to stay within this state for prolonged periods and remain relatively asymptomatic.
In contrast to other DT analyses,9 for this study we used the American Society for Therapeutic Radiology and Oncology definition of PSA relapse without backdating the date of relapse to the midpoint between the nadir PSA and the first increasing PSA value. We defined time zero to be the date of the third increase in PSA (the time of the established biochemical relapse) to avoid a lead time bias that artificially inflates the survival time when backdating the PSA relapse date. Another challenging aspect of the calculation of the PSA-DT is whether to consider the PSA-DT as a time-dependent variable or a fixed quantity. If the PSA-DT were analyzed as a time-dependent variable, then by definition the value of the PSA-DT varies over time as more PSA measurements are collected. At each time point at which a new PSA measurement is taken, the PSA-DT must be recalculated on the basis of all of the measurements to that date, and would require knowledge of the entire PSA history of a patient until the study end point (ie, distant metastasis or death) is reached. This type of analysis has many confounding features and is difficult to interpret.
It remains unclear how long a patient’s PSA should be observed to determine his PSA-DT. Furthermore, using varying numbers of PSA measurements would give rise to varying estimates of DT for an individual patient. For these reasons, our approach considered the PSA-DT as a fixed covariate when we examined its effect on the probability of a future event such as the incidence of distant metastasis. Applying a landmark analysis to this data set by calculating PSA-DT using only those measurements from the start to the end of the three consecutive increases in PSA provides an easily computed estimate of PSA-DT for an individual patient who had a biochemical relapse (three consecutive increases in PSA), and is a clearer way of interpreting its effect on the probability of the incidence of DM.
As part of the calculation of the PSA-DT, some have recommended subtracting the PSA nadir from subsequent PSA values.10 This modification was performed to make the estimate comparable to the surgical literature, in which PSA values typically begin at undetectable levels. However, this can represent a confounding factor in the calculation of the DT because it can drastically change the interpretation of the DT for the patients receiving radiation therapy. PSA-DT is calculated by modeling the relationship between PSA and time using a log linear form; the logarithm of PSA is modeled as a linear function of time. This suggests that the PSA grows exponentially or that the log of the PSA values grows linearly. The nadir value varies from patient to patient. Hence, if the nadir value, a random amount, is subtracted from the subsequent PSA values, it may violate the log linear relationship that is modeled. For this reason in the current analysis we did not subtract the PSA nadir from subsequent PSA values in the calculation of the PSA-DT.
Our study is consistent with the findings of D’Amico et al,8 who reported the results of a multi-institutional study that comprised 840 patients with a documented PSA recurrence after radiotherapy, and demonstrated the importance of PSA-DT and its impact on prostate cancer-specific survival. In their report, PSA-DT of less than 3 months was associated with a significant increased risk of prostate cancer-specific deaths compared with slower PSA-DT. Investigators from the Fox Chase Cancer Center (Philadelphia, PA) also reported on 136 patients who developed PSA relapse after external-beam radiotherapy and demonstrated that ADT, longer PSA-DT, lower post-treatment PSA nadir levels, and a longer interval to achieve a nadir PSA from the start of treatment were independent predictors of improved DM-free survival.9 Similar findings have been reported by others.10 In our study, which was limited to patients who developed PSA relapse only with negative bone scans, there seemed to be an initial benefit of salvage ADT; however, with additional follow-up, these differences were no longer apparent. Given the biases associated with our retrospective study, we do not have conclusive information to determine the value of salvage ADT in this clinical setting.
Patients in the state of a increasing PSA after definitive radiotherapy for clinically localized prostate cancer represent a therapeutic challenge. How does one discriminate effectively patients for whom the increase of the PSA is of no clinical significance, and those destined to develop clinical metastases and ultimately succumb to their disease Our results show that the higher T stage and Gleason score at diagnosis, and more rapid PSA-DT after recurrence, predict for a shorter transition time to the state of clinical metastases. Although the results confirm previous reports of the prognostic significance of PSA-DT and disease progression, they also show that this is not a stand-alone predictor, given that characteristics of the tumor at the time of diagnosis (as well as other yet undetermined factors) also affect outcomes. Additional refinement of our ability to predict outcomes for the heterogeneous group of patients in this clinical state is crucial for patient counseling and clinical trial design.
Authors’ Disclosures of Potential Conflicts of Interest
NOTES
Authors’ disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Catton C, Milosevic M, Warde P, et al: Recurrent prostate cancer following external beam radiotherapy: Follow-up strategies and management. Urol Clin North Am 30:751-763, 2003
2. Syed S, Petrylak DP, Thompson IM: Management of high-risk localized prostate cancer: The integration of local and systemic therapy approaches. Urol Oncol 21:235-243, 2003
3. Duchesne GM, Millar JL, Moraga V, et al: What to do for prostate cancer patients with a rising PSA A survey of Australian practice. Int J Radiat Oncol Biol Phys 55:986-991, 2003
4. Gray RJ: A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 16:1141-1154, 1988
5. Fine JP, Gray RJ: A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 94:496-509, 1999
6. Hollander M, Wolfe DA: Nonparametric Statistical Methods. New York, NY, John Wiley, 1973, pp 115-119
7. Scher HI, Heller G: Clinical states in prostate cancer: Toward a dynamic model of disease progression. Urology 55:323-327, 2000
8. D’Amico AV, Moul JW, Carroll PR, et al: Surrogate end point for prostate cancer-specific mortality after radical prostatectomy or radiation therapy. J Natl Cancer Inst 95:1376-1383, 2003
9. Lee WR, Hanks GE, Hanlon A: Increasing prostate-specific antigen profile following definitive radiation therapy for localized prostate cancer: Clinical observations. J Clin Oncol 15:230-238, 1997
10. Sandler HM, Dunn RL, McLaughlin PW, et al: Overall survival after prostate-specific-antigen-detected recurrence following conformal radiation therapy. Int J Radiat Oncol Biol Phys 48:629-633, 2000(Michael J. Zelefsky, Leah)
ABSTRACT
MATERIALS AND METHODS: A total of 1,650 patients with clinical stage T1 to T3 prostate cancer were treated with high-dose three-dimensional conformal radiotherapy. Of these, 381 patients subsequently developed three consecutive increasing PSA values and were characterized as having a biochemical relapse. The median follow-up time was 92 months from the completion of radiotherapy.
RESULTS: The 5-year incidence of DM after an established PSA relapse was 29%. In a multivariate analysis, PSA doubling time (PSA-DT; P < .001), the clinical T stage (P < .001), and Gleason score (P = .007) were independent variables predicting for DM after established biochemical failure. The PSA-DT for favorable-, intermediate-, and unfavorable-risk patients who developed a biochemical failure was 20.0, 13.2, and 8.2 months, respectively (P < .001). The 3-year incidence of DM for patients with PSA-DT of 0 to 3, 3 to 6, 6 to 12, and more than 12 months was 49%, 41%, 20%, and 7%, respectively (P < .001). Patients with PSA-DT of 0 to 3 and 3 to 6 months demonstrated a 7.0 and 6.6 increased hazard of developing DM or death, respectively, compared with patients with a DT more than 12 months.
CONCLUSION: In addition to clinical stage and Gleason score, PSA-DT was a powerful predictor of DM among patients who develop an isolated PSA relapse after external-beam radiotherapy for prostate cancer. Patients who develop biochemical relapse with PSA-DT ≤ 6 months should be considered for systemic therapy or experimental protocols because of the high propensity for rapid DM development.
INTRODUCTION
The prognosis of patients with an increasing PSA profile after failed primary radiotherapy is not well defined. A better understanding of the natural history of biochemical relapse and which specific increasing PSA pattern would most reliably predict for a tangible impact on the clinical outcome (such as distant metastatic spread) have not been well explored. The dominant challenge is to determine more effectively whether the source of the increasing PSA level after radiotherapy represents persistent local disease, systemic failure, or both. This information would have obvious implications for deciding the most appropriate therapy. Some have advocated earlier treatment interventions for patients who experience biochemical relapse only if the clinical characteristics before initial therapy were consistent with a more aggressive tumor phenotype. Others have used the PSA kinetics and response after radiotherapy, such as the PSA doubling time (PSA-DT), whereas other investigators have relied on arbitrary PSA levels or thresholds to trigger the initiation of therapy.1–3
We retrospectively reviewed the outcomes of 381 patients who developed a biochemical relapse after three-dimensional conformal radiotherapy (3D-CRT) and who were observed subsequently. In these patients, the PSA kinetics was carefully evaluated to determine which parameters influenced the development of DM. Our findings suggest that among other variables identified, the PSA-DT represents the most important variable predicting for DM and survival after a biochemical relapse, and we suggest that this parameter should be used as an integral part of the decision-making process for initiating therapy in those who develop a PSA relapse after primary therapy.
MATERIALS AND METHODS
The pretreatment characteristics of these 381 patients are listed in Table 1. In general, patients were treated with a five- or six-field conformal treatment approach as previously described.4 The median radiotherapy dose was 75.6 Gy (range, 64.8 to 81 Gy). Doses were prescribed to the maximum isodose line, which completely encompassed the planning target volume. The planning target volume was defined as the prostate and seminal vesicles with a circumferential 1-cm margin except at the prostate-rectal interface, where a 6-mm margin was used. The techniques for treatment planning and delivery of 3D-CRT and intensity-modulated CRT have been described in detail elsewhere.4 Treatment was delivered with 15 MV x-rays in daily fractions of 1.8 Gy. One hundred fifty-three patients (40%) with enlarged prostate glands were given a 3-month course of neoadjuvant total androgen deprivation therapy (ADT), and this therapy was discontinued at the completion of radiotherapy. Patients were classified into prognostic risk groups according to the National Comprehensive Cancer Network practice guidelines recurrence risk groups (www.nccn.org). Patients were classified as having favorable risk with clinical stages T1 to T2a, a Gleason score ≤ 6, and a pretreatment PSA less than 10 ng/mL. Patients with the presence of clinical stage T2b or T2c, a Gleason score of 7, or a pretreatment PSA of 10 to 20 ng/mL were classified as having intermediate risk. If patients had clinical stage T3a or higher, a Gleason score between 8 and 10, or a pretreatment PSA more than 20 ng/mL, they were classified as having unfavorable risk.
Follow-up evaluations after treatment were performed at intervals of 3 to 6 months for 5 years, and yearly thereafter. The median follow-up time was 92 months (range, 12 to 159 months) from the completion of radiotherapy. The median time from the completion of radiotherapy to the date of established biochemical relapse was 33 months (range, 3 to 177 months), and the median follow-up time from the date of established biochemical relapse among patients who did not experience DM was 43 months (range, 0 to 145 months). At the time of a documented relapse, patients routinely underwent a restaging evaluation that included imaging of the pelvis (computed tomography or magnetic resonance imaging) and bone scan. These tests were often repeated at 6-month intervals as part of the patient’s continued evaluation.
The primary end point of this analysis was the incidence of DM in patients with a documented biochemical relapse after 3D-CRT. Survival outcomes were measured from the date when PSA relapse was established (the date of the third increase in PSA). For this analysis, we applied stringent criteria to the definition of biochemical relapse using the American Society for Therapeutic Radiology and Oncology definition of three consecutive increases in PSA; however, we did not backdate the time of recurrence to the midpoint between the date of the nadir PSA value and the date of the first increasing PSA value. We defined time zero to be the date of the third increase in PSA, which is the clinically recognized time of the established biochemical relapse. This avoids a lead time bias, which artificially inflates the survival time that can be associated with backdating the PSA relapse date. The median time interval between the date of the first to the third consecutive increase in PSA was 16 months (range, 2 to 79 months). The incidence of DM was calculated using a competing-risk analysis accounting for death as a competing risk.4 Time to DM or death was calculated from the date of biochemical relapse to the date of DM, death, or last follow-up. The hazard ratios were computed using competing-risk regression and compared using a {chi}2 test.5
The relationship between PSA and time was modeled using a log linear form; the logarithm of PSA was modeled as a linear function of time. The PSA-DT was calculated as the reciprocal of the slope of this regression equation, for which the logarithm was taken with a base of 2. The PSA values used in the regression model to calculate the DT were only those PSA values from the start to the end of the three consecutive increases used to determine the biochemical relapse. Some of the PSA values before the first increasing value were zero or close to zero, which would lead to short DTs even when, in fact, the increase was incrementally small. To correct for this, a sensitivity analysis was performed and 0.2 was added to every PSA value before the logarithm was calculated. The PSA values used in the regression model to calculate the doubling time were only those PSA values from the start to the end of the three consecutive increases used to determine the biochemical relapse. The association between the DT and the pretreatment variables was assessed with the Kruskal-Wallis test.6 A multivariate generalized linear regression model was built with the log of PSA-DT as the dependent variable, and clinical T stage, Gleason score, use of neoadjuvant ADT, and log of the starting PSA value as the independent variables.
RESULTS
Given that PSA-DT was determined to be an important clinical variable reflecting the prognosis of patients with PSA relapse after 3D-CRT, variables were identified that influenced the PSA-DT, as shown in Table 4. These variables included the pretreatment PSA, biopsy Gleason score, and the clinical T stage. The PSA-DT for favorable-, intermediate-, and unfavorable-risk patients who developed a biochemical failure was 18.0, 12.3, and 8.8 months, respectively (P < .001). A multivariate analysis of variables that influenced the PSA doubling time is shown in Table 5.
The strong relationship of the PSA-DT with incidence of DM is shown in Figure 2. The 3-year incidence of DM rates for patients with PSA-DT of 0 to 3, 3 to 6, 6 to 12, and more than 12 months were 49%, 41%, 20%, and 7%, respectively (P < .001). Patients with PSA-DT of 0 to 3, 3 to 6, and 6 to 12 months demonstrated a 7.0-, 6.6-, and 2.8-fold increased risk of developing DM or death, respectively, compared with patients with DT greater than 12 months.
During the follow-up evaluation, 110 patients were subsequently treated with salvage ADT for their biochemical relapsing disease. The incidence of DM among patients who were treated with salvage hormonal therapy was reduced during the early follow-up period compared with patients who were observed without salvage ADT, but no obvious differences were apparent between these groups (data not shown) with additional follow-up.
DISCUSSION
Scher and Heller7 have adopted a clinical state model to describe the natural history of prostate cancer. Unlike the standard TNM staging system in which the patient is characterized with a clinical stage at the point of his diagnosis, the clinical state model relates to the entire spectrum of the disease during the lifetime of the patient and describes untreated and treated prostate cancer in sequential phases from prediagnosis to death. At any time point, a patient can reside in only one clinical state. At each clinical encounter the patient is assessed using clinical, biochemical, or other biologic determinants to predict the probability of progressing to the next clinical state, or more importantly, of developing symptoms or of succumbing to disease. The increasing PSA state can be viewed as a separate clinical state in which patients have experienced treatment failure after a primary therapy, yet have not manifested evidence of a clinical relapse. Within the increasing PSA state, the natural history can exhibit extraordinary variation from one patient to another. Our study suggests that the PSA-DT would be a useful prognostic tool for patients in this state to identify those patients who are most likely to migrate rapidly to the next clinical state of systemic disease compared with patients who are likely to stay within this state for prolonged periods and remain relatively asymptomatic.
In contrast to other DT analyses,9 for this study we used the American Society for Therapeutic Radiology and Oncology definition of PSA relapse without backdating the date of relapse to the midpoint between the nadir PSA and the first increasing PSA value. We defined time zero to be the date of the third increase in PSA (the time of the established biochemical relapse) to avoid a lead time bias that artificially inflates the survival time when backdating the PSA relapse date. Another challenging aspect of the calculation of the PSA-DT is whether to consider the PSA-DT as a time-dependent variable or a fixed quantity. If the PSA-DT were analyzed as a time-dependent variable, then by definition the value of the PSA-DT varies over time as more PSA measurements are collected. At each time point at which a new PSA measurement is taken, the PSA-DT must be recalculated on the basis of all of the measurements to that date, and would require knowledge of the entire PSA history of a patient until the study end point (ie, distant metastasis or death) is reached. This type of analysis has many confounding features and is difficult to interpret.
It remains unclear how long a patient’s PSA should be observed to determine his PSA-DT. Furthermore, using varying numbers of PSA measurements would give rise to varying estimates of DT for an individual patient. For these reasons, our approach considered the PSA-DT as a fixed covariate when we examined its effect on the probability of a future event such as the incidence of distant metastasis. Applying a landmark analysis to this data set by calculating PSA-DT using only those measurements from the start to the end of the three consecutive increases in PSA provides an easily computed estimate of PSA-DT for an individual patient who had a biochemical relapse (three consecutive increases in PSA), and is a clearer way of interpreting its effect on the probability of the incidence of DM.
As part of the calculation of the PSA-DT, some have recommended subtracting the PSA nadir from subsequent PSA values.10 This modification was performed to make the estimate comparable to the surgical literature, in which PSA values typically begin at undetectable levels. However, this can represent a confounding factor in the calculation of the DT because it can drastically change the interpretation of the DT for the patients receiving radiation therapy. PSA-DT is calculated by modeling the relationship between PSA and time using a log linear form; the logarithm of PSA is modeled as a linear function of time. This suggests that the PSA grows exponentially or that the log of the PSA values grows linearly. The nadir value varies from patient to patient. Hence, if the nadir value, a random amount, is subtracted from the subsequent PSA values, it may violate the log linear relationship that is modeled. For this reason in the current analysis we did not subtract the PSA nadir from subsequent PSA values in the calculation of the PSA-DT.
Our study is consistent with the findings of D’Amico et al,8 who reported the results of a multi-institutional study that comprised 840 patients with a documented PSA recurrence after radiotherapy, and demonstrated the importance of PSA-DT and its impact on prostate cancer-specific survival. In their report, PSA-DT of less than 3 months was associated with a significant increased risk of prostate cancer-specific deaths compared with slower PSA-DT. Investigators from the Fox Chase Cancer Center (Philadelphia, PA) also reported on 136 patients who developed PSA relapse after external-beam radiotherapy and demonstrated that ADT, longer PSA-DT, lower post-treatment PSA nadir levels, and a longer interval to achieve a nadir PSA from the start of treatment were independent predictors of improved DM-free survival.9 Similar findings have been reported by others.10 In our study, which was limited to patients who developed PSA relapse only with negative bone scans, there seemed to be an initial benefit of salvage ADT; however, with additional follow-up, these differences were no longer apparent. Given the biases associated with our retrospective study, we do not have conclusive information to determine the value of salvage ADT in this clinical setting.
Patients in the state of a increasing PSA after definitive radiotherapy for clinically localized prostate cancer represent a therapeutic challenge. How does one discriminate effectively patients for whom the increase of the PSA is of no clinical significance, and those destined to develop clinical metastases and ultimately succumb to their disease Our results show that the higher T stage and Gleason score at diagnosis, and more rapid PSA-DT after recurrence, predict for a shorter transition time to the state of clinical metastases. Although the results confirm previous reports of the prognostic significance of PSA-DT and disease progression, they also show that this is not a stand-alone predictor, given that characteristics of the tumor at the time of diagnosis (as well as other yet undetermined factors) also affect outcomes. Additional refinement of our ability to predict outcomes for the heterogeneous group of patients in this clinical state is crucial for patient counseling and clinical trial design.
Authors’ Disclosures of Potential Conflicts of Interest
NOTES
Authors’ disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Catton C, Milosevic M, Warde P, et al: Recurrent prostate cancer following external beam radiotherapy: Follow-up strategies and management. Urol Clin North Am 30:751-763, 2003
2. Syed S, Petrylak DP, Thompson IM: Management of high-risk localized prostate cancer: The integration of local and systemic therapy approaches. Urol Oncol 21:235-243, 2003
3. Duchesne GM, Millar JL, Moraga V, et al: What to do for prostate cancer patients with a rising PSA A survey of Australian practice. Int J Radiat Oncol Biol Phys 55:986-991, 2003
4. Gray RJ: A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 16:1141-1154, 1988
5. Fine JP, Gray RJ: A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 94:496-509, 1999
6. Hollander M, Wolfe DA: Nonparametric Statistical Methods. New York, NY, John Wiley, 1973, pp 115-119
7. Scher HI, Heller G: Clinical states in prostate cancer: Toward a dynamic model of disease progression. Urology 55:323-327, 2000
8. D’Amico AV, Moul JW, Carroll PR, et al: Surrogate end point for prostate cancer-specific mortality after radical prostatectomy or radiation therapy. J Natl Cancer Inst 95:1376-1383, 2003
9. Lee WR, Hanks GE, Hanlon A: Increasing prostate-specific antigen profile following definitive radiation therapy for localized prostate cancer: Clinical observations. J Clin Oncol 15:230-238, 1997
10. Sandler HM, Dunn RL, McLaughlin PW, et al: Overall survival after prostate-specific-antigen-detected recurrence following conformal radiation therapy. Int J Radiat Oncol Biol Phys 48:629-633, 2000(Michael J. Zelefsky, Leah)