Pretreatment PSA Velocity and Risk of Death From Prostate Cancer Following External Beam Radiation Therapy FREE

The impact of prostate-specific antigen (PSA) screening on mortality remains under study,^1- 2 but since the introduction of PSA testing the presentation of prostate cancer has shifted toward younger men with less aggressive and smaller-volume disease.^3- 4 Concomitant with stage migration the median PSA level at presentation has also decreased such that PSA levels greater than 10.0 ng/mL at diagnosis have become infrequent.⁵ Moreover, benign prostatic hypertrophy increases serum PSA levels and is commonly seen in men of prostate cancer–bearing age. As a result, the prognostic significance of any single value of PSA below 10.0 ng/mL is becoming more limited.

Despite declining PSA values at presentation, the information obtained from serial PSA values in the form of a PSA velocity has been shown to be significantly associated with tumor stage,⁶ grade,⁶ time to PSA failure,⁷ and time to prostate cancer–specific mortality following radical prostatectomy.⁸ Specifically, based on a secondary analysis of a prospective screening study, the observation was made that a PSA velocity—ie, the change in PSA level during the year prior to diagnosis—of greater than 2.0 ng/mL per year is significantly associated with prostate cancer–specific mortality despite radical prostatectomy. Therefore, the basis on which to distinguish which patient with a PSA level of 4.0 ng/mL at presentation has good- vs poor-prognosis disease may be made based on whether the PSA level the year prior was 3.5 ng/mL or 1.5 ng/mL, respectively.

While the PSA velocity appears to be useful, particularly in the growing population of men diagnosed with prostate cancer at a PSA level less than 10.0 ng/mL, some questions remain. First, the cutpoint of 2.0 ng/mL per year that represented the upper quartile of PSA values in the surgical study⁸ showing the association with prostate cancer–specific mortality requires validation using an independent data set before it can be used in clinical practice to assess prognosis. Second, the significance of the PSA velocity in an otherwise low-risk patient (clinical tumor category T1c or T2a and PSA level <10.0 ng/mL and Gleason score ≤6) is unknown. Therefore, in this study we evaluated whether the spectrum of continuous values that the PSA velocity had during the year prior to diagnosis was significantly associated with prostate cancer–specific mortality following radiation therapy (RT) after controlling for prognostic factors available at diagnosis and evaluated the specific value of 2.0 ng/mL per year and the special case of men with low-risk disease.

Pretreatment and follow-up information was compiled on 480 men who were treated with external beam RT at St Anne’s Hospital in Fall River, Mass, a Harvard Medical School affiliate, from January 1, 1989, to December 1, 2002, for clinical tumor categories T1c (nonpalpable) and T2 (palpable) prostate cancer. A total of 44 men who had only a single measurement of PSA prior to RT were excluded from the study, as were 78 men whose pre-RT PSA measurements were spaced less than 6 months apart. The remaining 358 men comprised the study cohort. No patient received neoadjuvant, concurrent, or adjuvant hormonal therapy. Each man provided written informed consent before study entry; the St Anne’s Hospital institutional review board approved the study. The median age of the men at the time of initial therapy was 71.2 (range, 43.2-83.5) years. Forty-four percent of patients were diagnosed on the basis of an increased PSA level, and 68% had a PSA level of 10.0 ng/mL or less. The median PSA level was 8.0 (range, 0.5-124.5) ng/mL. All biopsy slides were reviewed by a single genitourinary pathologist (A.A.R.). Table 1 shows the clinical characteristics of the men before treatment stratified by the pretreatment risk group. The preradiation staging has been previously described.⁹

Treatment consisted of 3-dimensional and conformal RT from 1994-2002 with computed tomography–based treatment planning. Prior to 1994, shaped blocks were used and treatment planning involved transferring computed tomography–defined volumes onto plain films of the pelvis. A 4-field box technique was used to treat the prostate and seminal vesicles to 4500 rad (45.0 Gy) in twenty-five 180-rad (1.8-Gy) fractions followed by a boost to the prostate of 2200 rad (22.0 Gy) in eleven 200-rad (2.0-Gy) fractions using a 1.5-cm margin throughout. All doses were prescribed to the planning target volume, and a 95% normalization was used so the minimum dose received by the prostate was 7035 rad (70.35 Gy). The seminal vesicles were not treated in patients with low-risk disease (ie, PSA level <10.0 ng/mL, biopsy Gleason score of ≤6, and clinical tumor category of T1c or T2a), but the total dose delivered to the prostate gland was the same.

The median follow-up was 4.0 (range, 0.2-13.5) years, and follow-up started on the last day of RT and concluded on March 1, 2005, or the date of death, whichever was sooner. No patient was lost to follow-up. Before PSA-defined recurrence, as specified by the American Society for Therapeutic Radiology and Oncology consensus definition,¹⁰ patients had a serum PSA measurement at a median of every 6 (interquartile range, 5-7) months and an annual digital rectal examination. After PSA recurrence, PSA levels were measured at a median of 4 (range, 1-12) months. Patients sustaining PSA recurrence started salvage hormonal therapy at a PSA level of approximately 10.0 ng/mL. The median PSA level was 9.6 (interquartile range, 8.8-9.8) ng/mL when the salvage hormonal therapy was initiated. Overall, there were 160 PSA recurrences and 79 deaths, of which 30 (38%) were from prostate cancer. The attending oncologist determined the cause of death. For a death to be recorded as being due to prostate cancer, it had to occur in a patient with a documented history of hormone-refractory metastatic prostate cancer and evidence that the PSA level was increasing at the time of the last follow-up visit before death.

Calculation of PSA Velocity. Using the PSA value closest in time to diagnosis (median, 1 [range, 0.5-3] months) and all prior PSA values that were within 1 year of diagnosis and separated by at least 6 months from the PSA value at diagnosis, the PSA velocity during the year prior to diagnosis was calculated using a linear regression analysis.¹¹ A minimum of 2 and a maximum of 3 PSA values were used to calculate the PSA velocity.

Assessment of Recurrence and Mortality. A Cox regression analysis¹¹ was used to test whether the PSA velocity prior to diagnosis and the PSA level, biopsy Gleason score, and clinical tumor stage at diagnosis were significantly associated with the time to post-RT PSA recurrence, prostate cancer–specific mortality, and all-cause mortality. For the purpose of the Cox regression multivariable analyses, the PSA velocity was considered first as a continuous and then as a categorical variable (>2.0 ng/mL per year vs ≤2.0 ng/mL per year). The PSA level at diagnosis was evaluated as a continuous variable, whereas the biopsy Gleason score and the clinical tumor category were analyzed as categorical variables defined as Gleason score 8 to 10 vs 7 vs 6 or less and T2 vs T1c, respectively. The covariate of age at the end of RT was evaluated as a continuous variable in the analyses of all-cause mortality. Time zero was defined as the last day of RT.

A Cox regression analysis¹¹ adjusting for age at diagnosis was used to test whether the pretreatment risk group was associated with time to post-RT PSA recurrence, prostate cancer–specific mortality, and all-cause mortality and to assess whether there was an interaction between the pretreatment risk group and the PSA velocity for these end points.

For all categorical variables, the cutpoints were determined prior to examining the database on established¹² or recently defined⁸ strata. For all Cox regression analyses, the assumptions of the proportional hazards model were tested and met, and all statistical tests were 2-sided. The hazard ratios (HRs) for PSA recurrence, prostate cancer–specific mortality, and all-cause mortality with the associated 95% confidence intervals (CIs) were calculated for all covariates using the proportional hazards model. Illustrations of the estimates of time to PSA recurrence, prostate cancer–specific mortality, and all-cause mortality following RT were made using Kaplan-Meier¹³ cumulative incidence¹⁴ and Kaplan-Meier¹³ plots, respectively, for men with low-risk and higher-risk prostate cancer. There were no deaths that occurred prior to PSA recurrence. Analyses were performed using SAS version 9.1.3 (SAS Institute Inc, Cary, NC); P <.05 was used to determine statistical significance.

The median PSA velocity for all patients was 1.5 (interquartile range, 0.74-3.8) ng/mL per year, with 2.0 ng/mL per year representing the 58th percentile, whereas for the 125 low-risk and 233 higher-risk patients, 2.0 ng/mL per year represented the 77th and 48th percentiles, respectively. As shown in Table 2, for all study patients the categorical covariate of a PSA velocity greater than 2.0 ng/mL per year as well as the continuous spectrum of PSA velocities were significantly associated with the time to PSA recurrence and prostate cancer–specific mortality after adjusting for the PSA level, biopsy Gleason score, and clinical tumor category at diagnosis. Of 30 prostate cancer deaths observed, 28 occurred in men whose PSA velocity was greater than 2.0 ng/mL per year. After adjusting for age at the end of RT in addition to the PSA level, Gleason score, and clinical tumor category, the categorical covariate of a PSA velocity greater than 2.0 ng/mL per year and the continuous spectrum of PSA velocities also remained significantly associated with all-cause mortality. The adjusted HRs of PSA recurrence, prostate cancer–specific mortality, and all-cause mortality for all patients with a PSA velocity greater than 2.0 ng/mL per year compared with 2.0 ng/mL per year or less were 1.8 (95% CI, 1.3-2.6; P = .001), 12.0 (95% CI, 3.0-54.0; P = .001), and 2.1 (95% CI, 1.3-3.6; P = .005), respectively.

The sample, event size, and comparison of the estimates of PSA recurrence, prostate cancer–specific mortality, and all-cause mortality when stratified by the pretreatment risk group and PSA velocity are shown in Table 3. The pretreatment risk group was significantly associated with PSA recurrence (P = .008), prostate cancer–specific mortality (P = .007), and all-cause mortality (P = .02), adjusting for age as shown in Table 4. Table 5 illustrates a significant interaction between PSA velocity and pretreatment risk group for the end points of PSA recurrence, prostate cancer–specific mortality, and all-cause mortality (P<.001 for all). Specifically, the adjusted HRs of PSA recurrence were 1.4 (95% CI, 1.2-1.6) and 1.03 (95% CI, 1.02-1.05) for patients with low- and higher-risk disease, respectively. These values were 2.4 (95% CI, 1.6-3.5) and 1.08 (1.05-1.10), respectively, for prostate cancer–specific mortality and 1.5 (95% CI, 1.2-1.8) and 1.04 (95% CI, 1.02-1.06), respectively, for all-cause mortality. It is important to note that the adjusted HRs for all 3 outcomes were larger in men with low- vs higher-risk disease. This means that for a given increase in PSA velocity the risk of PSA recurrence, prostate cancer–specific mortality, and all-cause mortality will increase at a faster rate in a man with low-compared with higher-risk disease.

The Figure illustrates a significant difference between the estimates of PSA recurrence (P = .005), prostate cancer–specific mortality (P<.001), and all-cause mortality (P = .003) rates, respectively, when stratified by the pre-RT PSA velocity value of greater than 2.0 ng/mL per year vs 2.0 ng/mL per year or less for men with low-risk disease, as well as for those with higher-risk disease.

Among men presenting with low-risk disease (n = 125), the 7-year estimates of PSA recurrence were 78% (95% CI, 57%-99%) vs 54% (95% CI, 40%-69%) for those with PSA velocity greater than 2.0 ng/mL per year vs 2.0 ng/mL per year or less, respectively; for prostate cancer–specific mortality, the corresponding estimates were 19% (95% CI, 2%-39%) vs 0%; and for all-cause mortality, were 53% (95% CI, 23%-81%) vs 14% (95% CI, 5%-24%). Among men presenting with higher-risk disease (n = 233), the 7-year estimates of PSA recurrence were 87% (95% CI, 74%-100%) vs 60% (95% CI, 46%-74%) for those with PSA velocity greater than 2.0 ng/mL per year vs 2.0 ng/mL per year or less, respectively; for prostate cancer–specific mortality, the corresponding estimates were 24% (95% CI, 12%-37%) vs 4% (95% CI, 0%-11%); and for all-cause mortality, were 44% (95% CI, 29%-59%) vs 31% (95% CI, 16%-46%).

Among the 125 men with low-risk disease, the adjusted HR of prostate cancer–specific mortality for those with PSA velocity greater than 2.0 ng/mL per year vs 2.0 ng/mL per year or less could not be estimated because there were no prostate cancer–specific deaths in men with low-risk disease and a PSA velocity of 2.0 ng/mL per year or less. For the 233 men with higher-risk disease, this adjusted HR of prostate cancer–specific mortality was 6.8 (95% CI, 1.5-32.0; P = .02).

The statistically and clinically significant associations between the PSA doubling time and prostate cancer–specific mortality in the setting of PSA failure following radical prostatectomy or RT in the treatment of localized prostate cancer have been described by several investigators.^15- 16 However, only a single study⁸ has suggested that pretreatment changes in PSA levels over time are significantly associated with prostate cancer–specific mortality following definitive local therapy for patients with localized prostate cancer. Specifically, a greater than 2.0-ng/mL increase in PSA level during the year prior to diagnosis was found to be significantly associated with a nearly 10-fold higher rate of cancer death following radical prostatectomy when compared with men whose PSA level increased by 2.0 ng/mL per year or less.⁸

In the current study, evidence is provided to validate the finding that an increase in PSA level of more than 2.0 ng/mL during the year prior to diagnosis is associated with a significantly higher rate of death from prostate cancer. Specifically, similar to the prior report of surgically managed patients,⁸ the majority (28/30) of the observed prostate cancer deaths occurred in men whose PSA velocity exceeded 2.0 ng/mL per year, translating into a 12-fold increase in the rate of prostate cancer–specific mortality after adjusting for the known prognostic factors at diagnosis for men whose PSA velocity was greater than 2.0 ng/mL per year as compared with all others. Moreover, a significant interaction was observed between the pretreatment risk group and the PSA velocity. Specifically, for a given increase in PSA velocity the corresponding increment in the risk of PSA recurrence, prostate cancer–specific mortality, and all-cause mortality was significantly higher in men with low- compared with higher-risk disease.

The clinical significance of these findings when considered in conjunction with the results of the prior report⁸ is that an increase in PSA level of more than 2.0 ng/mL during the year prior to diagnosis places a patient into a high-risk category for cancer death, despite radical prostatectomy or RT and despite “apparent” low-risk disease, presumably on the basis of occult micrometastatic disease present at the time of diagnosis. Evidence to support this statement is provided by the comparison of the prostate cancer–specific mortality plots for low- and high-risk disease in the Figure (center plots). Specifically, the short time course (7 years) and substantial magnitude (approximately 20%) of the cumulative incidence estimates of prostate cancer–specific mortality shown in the upper plot for men with low-risk disease and a PSA velocity greater than 2.0 ng/mL per year are strikingly similar to the data in the lower plot for men with more advanced disease and a high risk of harboring occult micrometastatic disease. Therefore, a PSA velocity of greater than 2.0 ng/mL per year alone is sufficient to confer a high risk of prostate cancer–specific mortality following RT. Today, the standard of practice for men with higher-risk prostate cancer is RT and androgen suppression therapy based on the survival benefit reported from 2 prospective randomized trials.^17- 18 Therefore, for men with low-risk disease and a pre-RT PSA velocity greater than 2.0 ng/mL per year who are planning to undergo RT and are in good health, offering RT and androgen suppression therapy could be viewed as a reasonable option.

Several points from this study require clarification. First, not all men who had a PSA velocity greater than 2.0 ng/mL per year experienced recurrence and prostate cancer–specific mortality. Therefore, while men who experienced an increase in PSA level of more than 2.0 ng/mL during the year prior to diagnosis had a 12-fold higher risk of experiencing prostate cancer–specific mortality on average, the risk to any individual man can vary from 3-fold to 50-fold, as noted by the 95% CIs in Table 2. Second, as noted in the footnotes following Table 2, PSA velocity, as a continuous variable, was also significantly associated with time to PSA recurrence, prostate cancer–specific mortality, and all-cause mortality. As a result, larger values of PSA velocity within the greater than 2.0 ng/mL per year category would be expected to be associated with a shorter median time to PSA recurrence, prostate cancer–specific mortality, and all-cause mortality; however, the exact functional relationship (eg, linear, exponential) between PSA velocity and these end points requires further study. Third, while observation of patients with a PSA velocity greater than 2.0 ng/mL per year would not appear to be warranted, the question of whether observation would lead to a shorter survival time compared with that found among patients offered radical prostatectomy, RT, or RT plus androgen suppression therapy remains unanswered and requires a randomized study such as the Prostate Cancer Intervention vs Observation Trial.¹⁹ Fourth, higher doses of radiation^20- 21 and treatment of the pelvic lymph nodes²² have been shown to confer a PSA control benefit in predominately low- and higher-risk patients, respectively. Whether a higher RT dose, treatment of the pelvic lymph nodes, or both could decrease prostate cancer–specific mortality in men with a PSA velocity greater than 2.0 ng/mL per year remains to be studied. Fifth, we conducted 2 Cox regression analyses to evaluate whether the treatment received by the men in this study or in the prior surgical study⁸ for low- or higher-risk prostate cancer was significantly associated with time to prostate cancer–specific mortality, adjusting for known prognostic factors including the PSA velocity. The result was that initial treatment was not significantly associated with time to prostate cancer–specific mortality for men in either risk group (Table 6). These results imply that treatment choice did not significantly impact the time to prostate cancer–specific mortality. However, this is a hypothesis-generating statement limited by the retrospective nature of this comparison, in that unknown prognostic factors may exist that can confound the comparison. Finally, PSA velocity in men diagnosed with prostate cancer has been shown to increase over time.⁶ As a result, by using PSA values dating back several years the calculated value of the PSA velocity can be less than the actual value at diagnosis, which more accurately reflects the clinical scenario at the time when management decisions and counseling are occurring. Therefore, when estimating the PSA velocity it is important to use information from the year-prior diagnosis and PSA values spaced approximately 6 months apart, or the baseline variation in the PSA assay may produce an erroneous result. For these reasons in this study we used a minimum of 2 and a maximum of 3 PSA values that were within 1 year of diagnosis but separated by 6 months.

In conclusion, an increase in PSA level greater than 2.0 ng/mL during the year prior to diagnosis is associated with a significantly higher risk of death due to prostate cancer following RT and despite low-risk disease. Such men who are planning to undergo RT and are in good health could be considered for RT combined with androgen suppression therapy because this approach improves survival in men with higher-risk disease.