The clinical significance of biochemical failure in the first 2 years in patients treated with neo-adjuvant androgen deprivation and radiotherapy for prostate cancer

Introduction: Biochemical failure in the first 2 years after radical prostatectomy confers an adverse prognosis compared to later biochemical relapse. This supplementary analysis of the Irish Clinical Oncology Research Group Trial 97-01 assess if this is also the case for patients managed with neo-adjuvant hormonal treatment and radiotherapy. Methods: The Irish Clinical Oncology Group initiated a prospective randomised trial (ICORG 97-01) that compared 4 versus 8 months of neo-adjuvant androgen deprivation therapy (NADT) for patients treated with 3D conformal radiotherapy (RT) for prostate cancer. This supplementary study describes a Cox regression analysis of data from 253 participants in that trial. Results: In patients with EBF (early biochemical failure) i.e., within 2 years, median survival time after completion of radiotherapy is 8.9 years, compared to 12.6 years in patients with LBF (late biochemical failure). Prostate cancer specific mortality risk was higher in the EBF cohort (Hazard ration (HR)= 6.75, 95% Confidence Interval (C.I.)= 3.1–14.7, P-value < 0.0001) compared to those with LBF. The risk of EBF was higher for patients with stage T2c-T4 disease compared those with stage T1-T2b disease (HR= 2.93, 95% CI= 1.30–6.58, P-value= 0.009). On multivariate analysis the estimated hazard of risk of EBF increased 1.02 fold for each increase of 1ng/mL in initial PSA. In the multi-variate model Gleason score also predicted EBF. Conclusion: Patients with prostate cancer treated with neo-adjuvant androgen deprivation therapy and radiation therapy that develop EBF have a shorter median duration of prostate cancer specific mortality (PCSS) than those with LBF.


Introduction
Androgen deprivation therapy (ADT) combined with external beam radiotherapy (RT) improves outcome for patients with prostate cancer. Randomised trials of ADT and RT for locally advanced prostate cancer include RTOG 85-31, EORTC 22863, RTOG 86-10, RTOG 92-02, RTOG 94-13, TROG 96.01, and a study from Brigham and Women's Hospital, Boston [1]. We report an unplanned secondary analysis of ICORG 97-01, a randomised trial comparing 4 versus 8 months of neo-adjuvant ADT (NADT) prior to RT in localised prostate cancer. In a population of correlative interest, patients that fail biochemically within 2 years of a radical prostatectomy have an unfavourable prognosis [2]. We examined the hypothesis that this subgroup would also have an unfavourable prognosis in patients treated with neo-adjuvant hormonal therapy and radiation. The relationship between time of biochemical failure (BF) following RT and prostate-cancer-specific survival (PCSS) is assessed. Age, initial PSA (last PSA pre NADT), Gleason score, T-stage and duration of NADT were also assessed for correlation with EBF.

Methods
The ICORG 97-01 trial was a randomised phase III clinical trial, comparing 4 months versus 8 months (arms 1 and 2 respectively) of NADT followed by RT (70Gy). 276 patients with prostatic adenocarcinoma were accrued. Patients had localised, node negative disease.

Eligibility criteria
Inclusion criteria: PSA >20ng/mL, Gleason score greater or equal to 7, or stage T3 or more, Karnofsky performance status (KPS) of greater or equal to 60. Exclusion criteria: Previous treatment for prostate cancer (other than transurethral resection of the prostate), bilateral orchidectomy, prior ADT, previous malignancy (other than non-melanoma skin cancer), and uncontrolled severe medical illnesses. Evaluation included history and physical examination, KPS assessment, complete blood count, liver function tests, chest radiograph, bone scan, and computed tomography (CT) pelvis to exclude metastasis.

Follow-up
Year 1, monthly. Subsequent follow-up, 3 monthly. Included history, rectal examination, PSA, and toxicity assessment. Treatment failure was defined as clinical evidence of local or distant recurrence or BF. neo-adjuvant hormonal therapy followed by radiotherapy. All patients were node negative and had no metastases. The neo-adjuvant hormonal therapy was the LHRH agonist (Triptorelin 3.75mg im once monthly) combined with antiandrogen therapy (flutamide 250mg po 3 times daily) (Figure 1).

Statistical analysis
The definition of biochemical (PSA) failure in the trial protocol was in accordance with the American Society for Therapeutic Radiation statement but data were analysed using the new definition of BF, defined as PSA >nadir after RT +2ug/L measured at least 18 months after completion of RT [3,4,5]. PSA doubling time was calculated as; PSADT (months) = ln 2 × [(month 2 -month 1) / (ln PSA2 -ln PSA1)] [6]. Cause of death was taken from the death certificate and by a review of patient records if the death certificate was inadequate. To record death as caused by prostate cancer, there had to be documented metastatic prostate cancer or hormone-refractory prostate cancer (HRPC) and evidence that the PSA level was increasing at the time of the last follow-up visit despite the use of second-line hormones before death.
Time to PCSS was calculated from the date enrolled or date of consent (whichever was earliest) to the date of death if the death was disease related, or date of last contact. The Kaplan-Meier method was used to estimate survival times [7].
Cox regression analysis was used to determine univariate predictors of PCSS and time to EBF. Variables with a P ≤ 0.10 were included in the multivariate models. The proportional hazards assumption of the Cox model was tested through the graphical examination of the log-log plots of the variables used in the model. These plots formed approximate parallel straight lines as required. All statistical tests were two-sided and assessed for significance at the 0.05 level, except in analyses in which a Bonferroni adjustment for multiple comparisons was applied, in which case the significance level was 0.05, divided by the number of comparisons. All  Data are for number of patients unless otherwise specified. Abbreviations: PSA= prostate-specific antigen; SHT= salvage hormone therapy; PSADT= PSA doubling time (from BF to initiation of salvage hormones); HRPC= hormone-refractory prostate cancer (progression on SHT as evidenced by three consecutive PSA increases taken at least one week apart or findings on physical examination or imaging studies).

Results
The median follow-up, from consent to last contact, for all living patients was 111 months (range 54-158). The median follow-up, from the end of RT to last PSA test, for patients living free from BF was 96 months (range 57-141). Median initial PSA was higher in patients with EBF, 22 ng/ ml compared with 14ng/ml in those with LBF. None of those with T1 disease had EBF. Twenty-nine percent of those with T2 disease had EBF.
In addition, 58% of those with PCSM had EBF compared to 15% of those with LBF. Salvage hormone treatment was   (Figure 2). PCSS curves by EBF/ LBF are curtailed at 120 months when 7 and 28 cases were remaining for EBF and LFB respectively (Figure 2). Patients with EBF had a median survival time of 107 months compared to 151 months for those with LBF. Using a Cox proportional hazards models, risk factors for EBF were evaluated as potential prognostic indicators by univariate analyses. We found initial PSA group, T-stage group and Gleason score group to be predictors of EBF, whereas age and arm were not predictive of EBF (Table 4).
In the multivariate model, after adjustment for the other variables, initial PSA, T-Stage group and Gleason score group remained as independent predictors for EBF ( Table 4). The hazard (risk of EBF) is higher and thus the prognosis worse, for subjects with higher values of initial PSA (HR= 1.019, 95% CI= 1.01-1.02, P-value < 0.0001). The estimated hazard or risk of EBF increases by 1.02 times for each increase of 1ng/mL in PSA, after adjustment for the effects of T-stage and Gleason score (e.g. there would be an estimated 10 fold increase in hazard for a patient with a PSA of 21ng/mL compared to a patient with a PSA of 11ng/mL). The hazard (risk of EBF) is higher and thus the prognosis worse, for subjects with T2c-T4 disease (HR= 4.38, 95% CI= 1.90-10.06, p= 0.001). Compared to those with T1-T2b, patients with T2c-T4 had over 4 times the risk of EBF, after adjustment for initial PSA and Gleason score. The hazard (risk of EBF) is higher and thus the prognosis worse, for subjects with Gleason score ≥ 7. Compared to those with Gleason score <7, patients with Gleason score >7 had 4.55 times the risk of EBF, after adjustment for initial PSA group and T-Stage group.

Discussion
This analysis of the ICORG 97-01 trial assesses the clinical significance of biochemical failure in the first 2 years post   Furthermore, de Crevoisier et al., reported that a PSA decline 6 weeks after the start of EBRT when used as monotherapy and 3 months after the start of ADT in patients treated with combined ADT and EBRT is predictive of progression and survival [18]. Patients with a T stage T2c-4 compared to T1-2B were 4 times more likely to have EBF. We report an association between EBF and death from prostate cancer. While this association is not unexpected, it's quantification is nonetheless important. This observation is consistent with The Transman study TROG 96.01 [19,20]. Our study gives clinicians further insight into the natural history of biochemically recurrent prostate cancer after RT and NADT.
It had previously been established that the natural history of prostate cancer relapsing post radical prostatectomy is that the median interval from biochemical relapse to the identification of metastasis is 8 years and the median time from metastatic disease to death is 5 years [21]. The ICORG study by comparison has shorter intervals. The 'Pound study' , in which 44.7% of patients had early biochemical failure, found that the time to biochemical relapse was predictive of the probability of and time to development of metastatic disease (p< .001). It would appear from the ICORG 97-01 study that EBF identifies an important subgroup with a more aggressive natural history and short survival. In patients treated with neo-adjuvant androgen deprivation and radiation EBF had a median PCSS time of 8.9 years. Those experiencing LBF had a PCSS time of 12.6 years. Therefore EBF defines a high-risk cohort for PCSM.
Predictors of biochemical recurrence are well established and we sought to establish the association of some of the  parameters (initial PSA, Gleason score and T-stage) with biochemical recurrence in the first two years post RT and NADT. Walz et al., (2009) devised a predictive nomogram of recurrent disease using Cox regression models of 2911 patients post RRP. They found that pathological Gleason sum 7 or greater, seminal vesicle invasion and lymph node involvement were the most powerful predictors of biochemical recurrence [22]. A further study of biochemical recurrence after radical prostatectomy found that Gleason grade and pathologic stage (p<0.002) were predictive of biochemical recurrence [23]. In our study, established predictive factors: initial PSA, Gleason score and T-stage were assessed to see if they were associated with EBF. We found that initial PSA, T-Stage group and Gleason score group predicted EBF. Compared to those with T1-T2b stage disease patients with T2c-T4 prostate cancer had a greater risk of EBF, after adjustment for initial PSA and Gleason score group. A study of similar design to the ICORG 97-01 was that of the Trans-Tasman Radiation Oncology Group (TROG) 96.01 Trial [20]. That study of 802 men with locally advanced prostate cancer compared treating prostate cancer with RT alone to 3 or 6 months of maximum short-term ADT before and during RT. In a subsequent analysis they examined time to BF (TTBF) and PSADT as surrogate endpoints for PCSM. TTBF was found to be a useful vicarious endpoint, and more particularly TTBF in a cut point band from less than 1.5 to less than 2.5 years predicted the endpoints of the trial. All four Prentice surrogacy criteria [24] were satisfied using the less than 1.5, less than 2, and less than 2.5 year cut-offs. The distinction between prognostic factors and those that predict response to therapy is also clinically important.
Men with a PSA doubling time after radical prostatectomy of less than 6 months have a worse prognosis overall than those with a PSA doubling time of 6 months or longer, but they derive a greater benefit from salvage therapy relative to untreated patients [2]. This may indicate that many men with a PSA doubling time of 6 months or longer have biologically less aggressive disease that is less likely to result in metastatic disease or death during their lifetime. An analysis of the RTOG 92-02 study of short versus long term ADT in T2c-4 prostate cancer found that time to treatment failure did not satisfy the 4th Prentice criterion (i.e., that the surrogate mediates the effect of initial treatment on the true endpoint) [25,26]. That study was designed so that all patients received goserelin and flutamide for 4 months before and during RT and were then randomised to no further ADT or 24 months of goserelin. The postulated reason that the 4th Prentice criterion was not satisfied was that a proportion of patients on long term ADT that had recurrent disease were hormone insensitive at time of PSA relapse and therefore were insensitive to salvage ADT [27]. The TROG 96.01 study randomised patients to 0, 3 or 6 months ADT before and during RT and the ICORG 97-01 study randomised to 4 versus 8 months NADT.
Because of the long natural history of prostate cancer, randomized trials designed to detect statistically significant differences in mortality often are prohibitive in terms of size, duration, and cost. The use of surrogate endpoints substantially decreases the size and duration of clinical trials, allowing more rapid prospective testing of hypotheses and potentially accelerating development of improved cancer treatment. Therefore an analysis of the natural history of EBF in this doi: 10.7243/2052-6199-1-2 context is even more important. Ultimately this study further supports the TROG 96.01 study that BF at or before 2 years post RT is associated with an increased risk of PCSM. Perhaps before these finding of time to biochemical failure and risk of PCSM are incorporated into routine practice a prospective clinical trial of time to biochemical failure and PCSM should be initiated. That study should obviate the inherent flaws in the clinically convenient but somewhat arbitrary selection of the 2-year cut-off in the ICORG 97-01 supplementary analysis or the categorical cohort based approach in the TROG 96.01 study. A continuous model that discovers the optimal abscission time point and then extended actuarial documentation of the longitudinal PCSM outcomes are required.
Finally a previous Cochrane review did compare neoadjuvant hormone therapy with radiotherapy versus radiotherapy alone for localized and locally advanced prostate cancer [28]. Outcome measures included biochemical free survival, and clinical disease free survival. Ultimately only individualized patient data from other previous studies of neo-adjuvant hormonal therapy and radiotherapy will provide meaningful comparator data to out study and that of Denham in the Trans-Tasman study. LBF: Late biochemical failure. Defined for the purposes of this study as biochemical relapse greater than 2 years post completion of 3Dconformal radiotherapy. As per the definitions A-C above. A 2-year cut off was decided upon prior to data analysis as practical duration to assist the physician in the clinic.