The impact of a delay in initiating radiation therapy on prostate-specific antigen outcome for patients with clinically localized prostate carcinoma
Article first published online: 6 APR 2005
Copyright © 2005 American Cancer Society
Volume 103, Issue 10, pages 2053–2059, 15 May 2005
How to Cite
Nguyen, P. L., Whittington, R., Koo, S., Schultz, D., Cote, K. B., Loffredo, M., McMahon, E., Renshaw, A. A., Tomaszewski, J. E. and D'Amico, A. V. (2005), The impact of a delay in initiating radiation therapy on prostate-specific antigen outcome for patients with clinically localized prostate carcinoma. Cancer, 103: 2053–2059. doi: 10.1002/cncr.21050
- Issue published online: 28 APR 2005
- Article first published online: 6 APR 2005
- Manuscript Accepted: 12 JAN 2005
- Manuscript Revised: 3 JAN 2005
- Manuscript Received: 9 SEP 2004
- Clinical Research Fellowship Program at Harvard Medical School offered by the Doris Duke Charitable Foundation
- Harvard PASTEUR Program
- Harvard Office of Enrichment Programs
- prostate carcinoma;
- radiation therapy;
- prostate-specific antigen outcome
To determine whether a delay in initiating external beam radiation therapy (RT) following diagnosis could impact prostate-specific antigen (PSA) outcome for patients with localized prostate cancer, 460 patients, who received 3D conformal RT to a median dose of 70.4 Gy for clinically localized prostate cancer between 1992 and 2001, were studied.
The primary endpoint was PSA failure (American Society for Therapeutic Radiology and Oncology definition). Estimates of PSA control were made using the Kaplan–Meier method. Delay was defined as the time between diagnosis and the start of RT. Risk groups were defined based on known predictors of PSA outcome, namely, baseline PSA level, clinical T-category, Gleason score, and percentage of biopsy cores positive for tumor. Cox multivariate regression analysis was used to determine the ability of treatment delay to predict time to PSA failure after adjusting for the other known predictors.
Treatment delay independently predicted time to PSA failure following diagnosis for high-risk (Adjusted Hazard Ratio = 1.08 per month; P = 0.029) but not low-risk patients (P = 0.31). Patients with high-risk disease (n = 240) had 5-year estimates of PSA failure-free survival of 55% versus 39% (Plog-rank = 0.014) for those with delay < 2.5 months versus ≥ 2.5 months respectively. The median delay was 2.5 months.
Treatment delay adversely affected PSA outcome for high-risk patients but not for low-risk patients following RT. Cancer 2005. © 2005 American Cancer Society.
A delay of 3 months between diagnosis of clinically localized prostate carcinoma and initiation of treatment is not uncommon for patients who ultimately choose radiation therapy (RT). This interval is significantly longer than delays commonly observed for other cancers such as head and neck, cervical, lung, or breast carcinoma.1 The length of delay before treatment begins may be due to a combination of factors, including logistic issues (e.g., scheduling tests, obtaining a referral to a radiation oncologist, treatment planning time), patient concerns (e.g., needing more time or a second opinion to decide among treatment options), and a general sense that, because prostate carcinoma is a slowly progressing disease, there is little urgency to initiate treatment. However, once a diagnosis of prostate carcinoma has been made, any delay in initiation of definitive therapy theoretically allows an opportunity for local tumor progression and possibly a decreased chance of cure. This effect has been observed in pooled analyses of retrospective studies of delays in initiating radiation therapy for primary head and neck tumors, which have demonstrated poorer local control rates after a delay.2–6
To our knowledge, there are no published studies on the potential impact of a delay in initiating radiation therapy on PSA outcome in prostate carcinoma, although two studies of surgically managed patients have suggested that a delay of approximately 3 months before a prostatectomy could decrease PSA control, particularly for patients with poorly differentiated cancer or PSA levels in excess of 20 ng/mL at diagnosis.7–8 The purpose of the current study is to determine whether delays in initiating radiation therapy for patients with clinically localized prostate carcinoma could have a similar adverse impact on PSA outcome (after statistically controlling known predictors of PSA outcome).
MATERIALS AND METHODS
The study cohort was comprised of 460 patients who received 3D conformal external beam radiation for clinical category (1992 AJCC staging criteria) T1c or T2 adenocarcinoma of the prostate between 1992 and 2001 at three locations: Saint Anne's Hospital (Fall River, MA) (n = 288), the Hospital of the University of Pennsylvania in Philadelphia (n = 122), or the Brigham and Women's Hospital in Boston (n = 50).9 Patients were selected from a list of all patients who had been treated by one of two clinicians (AVD at SAH and BWH, or RW at HUP) from 1992 to 2001. No patient had received neoadjuvant or adjuvant hormonal therapy. Inclusion criteria also required availability of complete staging information, including pretreatment PSA (at the time of diagnosis), 1992 AJCC Clinical T-Category, biopsy Gleason score, percentage of biopsy cores positive for tumor, date of biopsy, and date of treatment initiation. This study was conducted with approvals from internal review boards at all three institutions.
Diagnosis of prostate carcinoma was made based on a transrectal ultrasound-guided biopsy. A minimum of 6 biopsy cores was taken in 77% of patients (range, 2–14 cores). A genitourinary pathologist reviewed these biopsy slides at each institution. A history was obtained from all patients, each of whom had also had a physical examination, including digital rectal examination, serum PSA determination, and further imaging or bone scan if clinically warranted as previously described.10 Patients with a positive bone scan or CT or MRI evidence of enlarged pelvic lymph nodes (>1cm) were not included in the current study. Patient pretreatment baseline characteristics are listed in Table 1.
|All patients (N = 460)||Low-risk (n = 220)||High-risk (n = 240)|
|Median age (yrs)||72||72||72|
|Median PSA (ng/mL)||8.1||6.7||11.0|
|Median delay (mos)||2.5||2.6||2.4|
|a25/75 percentile of delay||1.9/3.6||2.0/4.0||1.9/3.1|
|PSA ≤ 4 ng/mL||40||8.7||27||12.3||13||5.4|
|PSA 4.1–10 ng/mL||248||53.9||147||66.8||101||42.1|
|PSA 10.1–20 ng/mL||128||27.8||46||20.9||82||34.2|
|PSA > 20 ng/mL||44||9.6||0||0.0||44||18.3|
|Clinical category T1c||187||40.7||114||51.8||73||30.4|
|Clinical category T2a||157||34.1||88||40.0||69||28.8|
|Clinical category T2b||69||15.0||18||8.2||51||21.2|
|Clinical category T2c||47||10.2||0||0.0||47||19.6|
|Biopsy Gleason ≤ 6||251||54.6||175||79.5||76||31.7|
|Biopsy Gleason 7 (3 + 4)||121||26.3||34||15.5||87||36.2|
|Biopsy Gleason 7 (4 + 3)||44||9.6||11||5.0||33||13.8|
|Biopsy Gleason 8–10||44||9.6||0||0.0||44||18.3|
|< 34% cores positive||223||48.5||185||84.1||38||15.8|
|34–50% cores positive||114||24.8||35||15.9||79||32.9|
|> 50% cores positive||123||26.7||0||0.0||123||51.3%|
Treatment and Follow up
All patients received 3D conformal radiation delivered by a 4-field technique with a median final dose of 70.4 Gy (range, 66.3–77.9 Gy) to the planning target volume (PTV) after 95% normalization. For patients with low-risk disease, the PTV generally encompassed the prostate alone. For patients with high-risk disease, both the prostate and seminal vesicles were generally included in the PTV for 47.4 Gy (after 95% normalization), and then the final 23 Gy was delivered to the prostate alone. No patient received neoadjuvant or adjuvant hormonal therapy.
Patients were typically followed up 1 month after the end of treatment, then at 3-month intervals for 2 years, and then at 6-month intervals thereafter, alternating between their radiation oncologist and urologist. Each follow-up visit included a physical examination and PSA measurement. The median estimated follow-up time for all patients was 4.5 years (range, 1.5–9.8 yrs).11
Delay time was calculated as the time between the date of biopsy confirmation of prostate carcinoma and the first day of radiation. The primary endpoint of this study was PSA failure, which was defined as three consecutive rises above a nadir, in accord with the 1997 American Society for Therapeutic Radiology and Oncology (ASTRO) consensus statement.12 An increase of at least 0.2 ng/mL was required for each rise. Time zero was defined as the date of diagnosis (biopsy date). Date of PSA failure was defined as the midpoint between the nadir and the first rise. If a patient had one substantial rise in PSA (n = 10) or two significant rises in PSA (n = 23) that caused physician concern and that prompted initiation of salvage hormonal therapy, then the score was recorded as a biochemical failure and dated at the midpoint between the nadir and the first rise.
Using a previously defined risk stratification, the patients were divided into low-risk and high-risk groups. The low-risk group included patients with Clinical T1c/T2a disease, PSA ≤ 10, and Gleason ≤ 6 with ≤ 50% of biopsy cores positive for tumor. Low-risk also included patients with either T2b disease, or PSA range of 10.1–20, or a Gleason score of 7, if they had fewer than 34% of cores positive for tumor.13 All remaining patients were placed in the high-risk group.
The Cox proportional hazards model was used to determine the ability of delay time to statistically predict time to PSA failure, with controls for pretreatment PSA (at time of diagnosis), clinical T-category, biopsy Gleason score, and percentage of biopsy cores positive for tumor. Pretreatment PSA, percentage of cores positive for tumor, and delay time (in months) were treated as continuous variables. Clinical T-category (T1c, T2a, T2b, T2c) was treated as a categoric variable, with clinical T1c as the baseline group. Gleason score was also treated as a categoric variable and grouped as Gleason ≤ 6, Gleason 7 (3 + 4), Gleason 7 (4 + 3), and Gleason 8–10, with Gleason ≤ 6 as the baseline group. These analyses were performed for all patients, and then repeated for low and high-risk groups. Analysis was performed using the STATA statistical package (Version 8.0, Stata Corp., College Station, Texas). The assumptions of the Cox proportional hazards model were evaluated and satisfied.
The median delay for all patients was 2.53 months (range, 0.27–24.9 mos). The method of Kaplan and Meier was used to estimate the PSA failure-free survival of patients who had delays greater than or equal to the median delay of 2.53 months versus less than the median delay, stratified by risk group. Pair-wise comparisons of PSA survival estimates were made using the log-rank test. To assess for potential imbalances in predictive factors among patients with a delay value ≥ 2.53 months versus < 2.53 months, the chi-square test was used to compare the distribution of pretreatment PSA, biopsy Gleason scores, clinical T-category, and percentage of biopsy cores positive for tumor. In addition, the median age at diagnosis was compared by Wilcoxon rank-sum test, and medians of estimated distributions of follow-up time, estimated differences in median follow-up time, and 95% confidence intervals (CI) were computed.11
Cox Multivariable Regression
Results of the Cox multivariable regression analyses are displayed in Table 2. For all study patients (n = 460), treatment delay was not a significant predictor of time to PSA failure (P = 0.39), although pretreatment PSA (Adjusted Hazard Ratio (AHR) =1.03 [95% CI: 1.02–1.04], P < 0.001), percentage of biopsy cores positive for tumor (AHR =1.02 [95% CI: 1.01–1.02], P < 0.001), biopsy Gleason 4 + 3 (AHR = 2.52 [95% CI: 1.60–3.97], P < 0.001), biopsy Gleason 8–10 (AHR = 3.02 [95%CI: 1.92–4.75], P < 0.001), and clinical category T2b (AHR =1.81 [95% CI: 1.21–2.69], P = 0.004) were significant predictors.
|All patients (N = 460, 143 events) final model|
|Clinical predictor||AHR||95% CI||P value|
|Prediagnosis PSA (ng/mL)||1.03||1.02–1.04||< 0.001|
|% positive cores||1.02||1.01–1.02||< 0.001|
|Biopsy Gleason 7 (4 + 3)||2.52||1.60–3.97||< 0.001|
|Biopsy Gleason 8–10||3.02||1.92–4.75||< 0.001|
|Clinical category T2b||1.81||1.21–2.69||0.004|
|Variables not in all patients final model|
|Biopsy Gleason 7 (3 + 4)||—||—||0.890|
|Clinical category T2a||—||—||0.590|
|Clinical category T2c||—||—||0.121|
|Low-risk (n = 220, 37 events) final model|
|Clinical predictor||AHR||95% CI||Pvalue|
|Prediagnosis PSA (ng/mL)||1.14||1.07–1.22||< 0.001|
|Biopsy Gleason 7 (4 + 3)||3.52||1.36–9.07||0.009|
|Variable not in low-risk final model|
|% positive cores||—||—||0.193|
|Biopsy Gleason 7 (3 + 4)||—||—||0.514|
|Clinical category T2a||—||—||0.576|
|Clinical category T2b||—||—||0.416|
|High-risk (n = 240, 106 events) final model|
|Clinical predictor||AHR||95% CI||Pvalue|
|No. mos delaya||1.08||1.01–1.16||0.029|
|Prediagnosis PSA (ng/mL)||1.02||1.01–1.04||< 0.001|
|% positive cores||1.02||1.01–1.03||< 0.001|
|Biopsy Gleason 7 (4 + 3)||2.41||1.43–4.04||< 0.001|
|Biopsy Gleason 8–10||3.03||1.84–4.98||< 0.001|
|Clinical category T2b||1.89||1.21–2.95||0.005|
|Variables not in high-risk final model|
|Biopsy Gleason 7 (3 + 4)||—||—||0.865|
|Clinical category T2a||—||—||0.695|
|Clinical category T2c||—||—||0.078|
For low-risk patients (n = 220), treatment delay was not a significant predictor of time to PSA failure (P = 0.31), whereas pretreatment PSA (AHR = 1.14 [95% CI: 1.07–1.22], P < 0.001) and biopsy Gleason 4 + 3 (AHR = 3.52 [95% CI: 1.36–9.07], P = 0.009) were significant.
In the cohort of high-risk patients (n = 240), treatment delay was a significant predictor of time to PSA failure (AHR = 1.08 [95% CI: 1.01–1.16] per month, P = 0.029), in addition to pretreatment PSA (AHR = 1.02 [95% CI: 1.01–1.04], P < 0.001), percentage of biopsy cores positive for tumor (AHR = 1.02 [95% CI: 1.01–1.03], P < 0.001), biopsy Gleason 4 + 3 (AHR = 2.41 [95% CI: 1.43–4.04], P < 0.001), biopsy Gleason 8–10 (AHR = 3.03 [95% CI: 1.84–4.98], P < 0.001), and clinical category T2b (AHR = 1.89 [95% CI: 1.21–2.95], P = 0.005).
To illustrate, estimates of time to PSA failure following diagnosis are shown in Figures 1, 2, and 3, for the entire study cohort, low, and high-risk patients, respectively. When stratified by the median value of delay (2.53 months), results showed a significant difference in estimates of PSA outcome in the high-risk cohort (P = 0.014) but not in the low-risk cohort (P = 0.62) or in the total study cohort (P = 0.15). Specifically, in the high-risk cohort, the PSA failure-free survival estimates at 5 years following diagnosis for patients with a delay < 2.53 months was 55%, whereas for patients with a delay ≥ 2.53 months the estimate was 39%.
Confounding Factor Assessment
Table 3 illustrates the comparison of baseline PSA, clinical T-category, Gleason score, percentage of positive biopsy cores, age at diagnosis, and median follow up for patients in the high-risk cohort whose delay was less than versus greater than or equal to the median delay. Within the high-risk cohort, there were no statistically significant imbalances in the distribution of baseline PSA at diagnosis (P = 0.15), clinical T-category (P = 0.22), Gleason score (P = 0.39), percentage of biopsy cores positive for tumor (P = 0.99), median age at diagnosis (73 yrs vs. 72 yrs; P = 0.14), or median estimated follow up (4.8 yrs vs. 4.7 yrs; P = 0.92), for patients whose delays were less than vs. greater than or equal to the median delay, respectively.
|Delay < 2.5 mos (n = 126)||Delay ≥ 2.5 mos (n = 114)||P value||Type of test|
|Median age (yrs)||73||72||0.14||Mann-Whitney|
|Median follow up (yrs)||4.8||4.7||NS||a|
|PSA ≤ 4 ng/mL||5||4.0||8||7.0|
|PSA 4.1–10 ng/mL||61||48.4||40||35.1|
|PSA 10.1–20 ng/mL||41||32.5||41||36.0|
|PSA > 20 ng/mL||19||15.1||25||21.9||0.15||Chi-square|
|Clinical category T1c||34||27.0||39||34.2|
|Clinical category T2a||40||31.7||29||25.4|
|Clinical category T2b||31||24.6||20||17.6|
|Clinical category T2c||21||16.7||26||22.8||0.22||Chi-square|
|Biopsy Gleason ≤ 6||34||27.0||42||36.8|
|Biopsy Gleason 7 (3 + 4)||50||39.7||37||32.5|
|Biopsy Gleason 7 (4 + 3)||19||15.1||14||12.3|
|Biopsy Gleason 8–10||23||18.2||21||18.4||0.39||Chi-square|
|< 34% cores positive||20||15.9||18||15.8|
|34–50% cores positive||41||32.5||38||33.3|
|> 50% cores positive||65||51.6||58||50.9||0.99||Chi-square|
This study examined the possibility that delays in initiating radiation therapy for clinically localized prostate carcinoma could adversely affect biochemical control. The rationale for this hypothesis was that radiation can cure only a localized tumor, and any delay in starting treatment theoretically gives a tumor more opportunity to progress locally and/or permits the growth of occult micrometastases that may be beyond the treatment field (Fig. 4). Conversely, because prostate carcinoma generally progresses slowly, and can remain undetected for years, it is unclear whether a treatment delay of a few months is enough time to adversely affect outcomes, although 2 recent surgical studies have suggested that prostatectomy delays of approximately 3 months may decrease PSA control, particularly for high-risk patients.7, 8
Similar to studies of surgical results, the current study supports the hypothesis that delays in initiating radiation therapy adversely impact PSA outcome in patients with high-risk disease. Specifically, Cox multiple regression analyses provide evidence that, after statistically adjusting the value of pretreatment PSA, clinical T-category, biopsy Gleason score, and percentage of biopsy cores positive for tumor, treatment delay remained significantly associated (AHR = 1.08 per month, P = 0.029) with time to PSA failure following diagnosis for patients with high-risk and clinically localized prostate carcinoma. Whereas patients with low-risk disease had no significant decrement in PSA outcome due to a delay, patients with high-risk disease had a 5-year estimate of PSA failure-free survival of 55% vs. 39% (Plog-rank = 0.014) for those who had a delay less than versus greater than or equal to the median delay of 2.53 months, respectively. The fact that there were no statistically significant differences (all P > 0.05) between the distributions of baseline PSA, Gleason score, T-category, percentage of positive biopsy cores, age, and median follow up between high-risk patients with a delay less versus greater than or equal to 2.53 months provides further evidence that the observed difference in PSA outcome may be attributed to the delay itself.
The clinical implication for patients with high-risk disease is that a delay of only a few months may lead to disease progression that reduces biochemical control after radiation. Therefore, patients with high-risk disease who choose radiation therapy may need to begin treatment as soon as possible to maximize PSA outcome. It remains to be seen whether the adverse effect of delay on PSA outcome for high-risk prostate carcinoma will remain significant for patients who receive higher-dose irradiation14–17 or irradiation plus hormonal therapy.18–19
For patients with low-risk disease, the current study did not discern a significant impact on PSA outcome that could be attributed to delay. However, this lack of difference may be limited by the number of events in the low-risk cohort. Specifically, given a sample size of n = 220 patients with a 10-year accrual period, this study had 80% power at a 0.05 significance level to detect a difference in hazard rates that would produce a difference in the survival percentage at 6 years of about 18% in the low-risk group when stratified by the median delay. Therefore, the maximum period of “safe” delay in low-risk patients treated with RT remains unknown.
There were several limitations to the current study. First, given that this was not a randomized trial, confounding factors yet to be discovered in addition to those studied could contribute to the difference seen in high-risk patients. Second, not all patients with PSA failure experienced progression to distant disease and death from prostate carcinoma. Therefore, further follow up is required to determine whether these clinically relevant endpoints can be impacted by treatment delay in patients with high-risk disease.20
In conclusion, similar to the results of patients who experienced approximately a 3-month delay before radical prostatectomy, a treatment delay of approximately 3 months before receiving RT appeared to adversely affect PSA outcome in patients with high-risk, but not low-risk disease. Analyzed continuously, treatment delay appeared to increase the hazard rate by a rate of 8% per month of delay. Whether this effect will remain after the use of higher-dose RT or RT plus concomitant hormonal therapy in patients with high-risk disease requires further study.
- 3The influence of delay in the initiation of definitive radiotherapy in carcinoma of the tonsillar region. Int J Radiat Oncol Biol Phys. 1998; 42: S97., , , et al.
- 7Delay in surgical therapy for clinically localized prostate cancer and biochemical recurrence after radical prostatectomy. Can J Urol. 2003; 10: 1891–1898., , , et al.
- 8How long can radical prostatectomy (RP) safely be delayed? CPDR's experiences from 3324 cases. J Urol. 171(Suppl 4): 312., , , et al.
- 9American Joint Committee on Cancer. AJCC Cancer Staging Manual. 4th ed. Philadelphia: Lippincott–Raven, 1992.
- 12American Society for Therapeutic Radiology and Oncology Consensus Panel. Consensus statement: guidelines for PSA following radiation therapy. Int J Rad Onc Biol Phys. 1997; 37: 1035–1041.