Pride or prejudice: Does Phoenix flatter radiation therapy?

Brachytherapy 13 (2014) 299e303

Pride or prejudice: Does Phoenix flatter radiation therapy? W. James Morris1,2,*, Tom Pickles1,2, Mira Keyes1,2, Michael McKenzie1,2, Ingrid Spadinger3 1

Division of Surgery, Department of Radiation Oncology, Faculty of Medicine, University of British Columbia, British Columbia, Canada 2 Department of Radiation Oncology, Vancouver Cancer Centre, 600 West 10 Avenue, Vancouver, British Columbia, Canada V5Z 4E6 3 Department of Medical Physics, Vancouver Cancer Centre, Vancouver, British Columbia, Canada

ABSTRACT

PURPOSE: To compare disease-free survival (DFS) rates using a O0.4 ng/mL biochemical failure definition with the Phoenix (nadirþ2 ng/mL) failure definition by analyzing a consecutive cohort of 1006 patients treated with low-dose-rate prostate brachytherapy (LDR-PB) monotherapy. METHODS AND MATERIALS: Data for first 1006 consecutive LDR-PB implants (1998e2003) were extracted from a prospective database. Patients had low- (58%) or intermediate (42%)-risk disease. Three months neoadjuvant and 3 months concomitant androgen deprivation therapy were used in 65% of cases. The Phoenix definition was modified to ‘‘unfail’’ patients who had a benign prostate-specific antigen (PSA) bounce. RESULTS: The median followup is 7.5 years. The median PSA at latest followup for disease-free patients was 0.04 ng/mL. The Phoenix definition yielded 5- and 10-year KaplaneMeier DFS estimates of 96.5  1.2% and 93.7  2.0%, respectively. Applying the O0.4 ng/mL threshold reduced these estimates to 94.4  1.6% and 88.8  3.0% (log rank, p 5 0.015). CONCLUSIONS: Compared with Phoenix, applying a O0.4 ng/mL failure definition increased biochemical failure by ~2% at 5 years and ~5% at 10 years. These data show that Phoenix did not greatly exaggerate DFS estimates compared with a surgical-type threshold. However, this observation is a consequence of the exceptionally low residual PSA values characteristic of LDR-PB and cannot be generalized to other forms of radiation therapy. Ó 2014 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved.

Keywords:

Prostate brachytherapy; Biochemical failure; Phoenix threshold; Biologic equivalent dose; Nadir PSA values

Introduction Biochemical relapse is a surrogate for disease recurrence, and the Phoenix definition (nadirþ2 ng/mL) provides a multiple validated end point that is correlated with diminished overall survival in both intermediate- and high-risk patients (1). However, critics of the Phoenix definition argue that it introduces a systematic bias favoring radiation therapy over radical prostatectomy (RP) by combining a relatively high (nadirþ2 ng/mL) threshold with a lack of backdating to the start of an inevitable, but often slow, rise in prostate-specific antigen (PSA) that

Received 15 March 2013; received in revised form 10 June 2013; accepted 28 June 2013. Conflict of interest: None of the authors have a financial interest or any other conflicts of interest to report in relation to the products mentioned in this manuscript. * Corresponding author. Department of Radiation Oncology, Vancouver Cancer Centre, 600 West 10 Ave, Vancouver, British Columbia, Canada V5Z 4E6. Tel.: 604-877-6000; fax: 604-877-0505. E-mail address: [email protected] (W.J. Morris).

eventually crosses that threshold. This issue is greatly exacerbated in series with short followup and infrequent followup PSA measurements (2, 3). In contrast to external beam radiation therapy (EBRT), practitioners of low-dose-rate prostate brachytherapy (LDR-PB) sometimes dispense with the Phoenix definition, preferring to report biochemical no evidence of disease results based on a fixed threshold of O0.2 or O0.4 ng/mL attained beyond a suitable posttreatment interval (typically 48e60 months) to define recurrence (4, 5). Because it is known that the risk of triggering the Phoenix threshold in long-term followup is correlated with the PSA level after LDR-PB or EBRT (6, 7), the residual posttreatment PSA value may provide an important independent measure of the relevant biologic effect of therapeutic radiation that is independent of the method of delivery. In the current analysis, we have followed other authors in applying a PSA threshold (O0.4 ng/mL) to define biochemical recurrence and compared the results with those obtained using the nadirþ2 ng/mL (Phoenix) definition.

1538-4721/$ - see front matter Ó 2014 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.brachy.2013.06.006

300

W.J. Morris et al. / Brachytherapy 13 (2014) 299e303

Methods Patient population and treatment protocol This analysis consists of all LDR-PB patients treated on or before October 23, 2003 (N 5 1006). The treatment and followup protocol for the identical cohort were described in a recent publication (8). No patients received supplemental EBRT, but 65% received neoadjuvant and concomitant androgen deprivation therapy (ADT) as previously described (8).

received any secondary treatment for prostate cancer at any time after the implant. Statistical analysis The rates of DFS and their respective 95% confidence intervals were estimated using the KaplaneMeier (K-M) method. All statistics were done using SPSS (the Statistical Package for Social Sciences version 14.0.4, SPSS, Chicago, IL).

Results Applying the Phoenix definition The nadirþ2 ng/mL threshold defined biochemical relapse. Patients with early rises in PSA that triggered the Phoenix definition were considered benign rises and later ‘‘unfailed’’ if subsequent PSA values declined to !0.5 ng/mL without intervention. This modification has been undertaken by several authors previously who feel, as we do, that it is inappropriate to score as a failure, a man whose PSA is near undetectable without additional intervention (9). Applying the O0.4 ng/mL definition In addition to the Phoenix definition, we applied a much lower threshold that defined biochemical failure as a PSA O0.4 ng/mL at any time $48 months after implant. The $48-month caveat exists because, in the first 4 years after implant, 20%e40% of implant recipients experience temporary benign increases in serum PSA; these are false positives and are not predictive of future biochemical relapse (10, 11). At times O48 months, benign fluctuations like these are rare, and the median PSA values converge near the detection limit. Using the above criteria, biochemical failure using the O0.4 ng/mL threshold was determined as follows: All biochemical failures identified according to the Phoenix definition are, of course, automatically relapses according to the 0.4 ng/mL definition. Records for men that were disease free by the Phoenix definition and with PSA followup $48 months (N 5 745) were examined for failure to maintain a PSA at #0.4 ng/mL and assigned accordingly. For men with PSA followup !48 months (N 5 212), the Phoenix definition was used to signal biochemical recurrence. For all patients identified as having relapsed using the O0.4 ng/mL definition, the time to treatment failure was defined as the date of the first PSA value O0.4 ng/mL that was not followed by a decline (without intervention) to !0.4 ng/mL. Defining disease-free survival Disease-free survival (DFS) was defined as the absence of biochemical, clinical, histologic, or imaging evidence of recurrent or persistent prostate cancer and not having

Patient population The patient’s prognostic parameters and dose metrics are summarized in Table 1. As shown, 42% of patients had intermediate-risk disease using the criteria established by the National Comprehensive Cancer Network. Two-thirds of men in the cohort and 92% of the intermediate-risk patients received ADT by protocol (data not shown). DFS The K-M DFS outcomes using the nadirþ2 ng/mL and the O0.4 ng/mL failure definitions are compared in Table 2 and Fig. 1. Using the 0.4 ng/mL definition increases Table 1 Clinical characteristics, pretreatment risk factors, and dosimetry Variable/subgroup

Entire cohort, N 5 1006

Age (y) Median 66 Range 45e82 Pretreatment PSA (ng/mL) Median 6.4 Range 0.3e19 Gleason score #6 766 (76) 7 239 (24) Clinical stage T1 450 (45) T2 556 (55) Risk group Low 586 (58) Intermediate 419 (42) Percent positive cores !50 640 (64) $50 269 (27) Missing 97 (10) D90 (Gy) Median 151 Mean  SD 151.0  19.0 V100 (%) Median 92 Mean  SD 91.1  6.1

ADTþ, N 5 658

Non-ADT, N 5 348

67 47e82

65 45e79

7.1 0.44e19

5.1 0.3e12

419 (63) 239 (37)

347 (99) 1 (!1)

292 (44) 366 (56)

158 (45) 190 (55)

272 (41) 386 (59)

314 (90) 34 (10)

389 (59) 195 (30) 74 (11)

251 (72) 78 (22) 1 (6)

148 148.7  19.2

156 155.4  18.0

92 90.3  6.6

94 92.6  5.0

ADT 5 androgen deprivation therapy; PSA 5 prostate-specific antigen; SD 5 standard deviation; D90 5 the minimum dose received by 90% of the postimplant CTebased prostate volume; V100 5 the percent of the postimplant CTebased prostate volume that receives at least 100% of the prescription dose. Percentage values are given in parentheses.

W.J. Morris et al. / Brachytherapy 13 (2014) 299e303

301

Table 2 K-M disease-free survival estimates at 5, 7, and 10 years for both thresholds Threshold O0.4 ng/mL

Nadirþ2 ng/mL Time (y)

K-M estimate (%)

95% CI

K-M estimate (%)

95% CI

5 7 10

96.5 94.7 93.5

1.2 1.6 2.0

94.4 92.5 88.7

1.6 2.0 3.0

K-M 5 KaplaneMeier; 95% CI 5 95% confidence interval.

the number of biochemical recurrence events from 49 to 73 and results in a ~2% absolute reduction in the K-M DFS estimates from 96.5% (1.2%) to 94.4% (1.6%) at 5 years after implant and a ~5% reduction by 10 years from 93.7% (2.0%) to 88.7% (3.0%). Although relatively small in absolute terms, these differences easily reach statistical significance in log-rank testing ( p 5 0.015). The timing of biochemical relapse The effect of applying the O0.4 ng/mL failure definition is clearly seen in Fig. 2, which compares the proportion of recurrence events for both definitions as a function of time in 3-year bins. Note that, with the Phoenix definition, only 16% of events (8 of 49) occurred in the first 36 months, whereas applying the 0.4 ng/mL definition results in nearly half of all recurrence events (33 of 73) being recorded at times !36 months. Compared with the Phoenix definition, the O0.4 ng/mL failure definition resulted in a 12-month reduction in the median time to relapse from 55.3 to 42.7 months. On the other hand, late recurrences were uncommon with either definition. For example, although 39% of the

Fig. 1. KaplaneMeier disease-free survival plots comparing the nadirþ2 ng/mL (black line per markers) and the O0.4 ng/mL (gray line per markers) definitions to define biochemical recurrence (log rank, p 5 0.015). The number of patients at risk is as follows: Year N

0 1006

2 928

3 876

4 804

5 702

6 605

7 525

8 392

9 260

10 145

Fig. 2. The percent of all recurrence events (y-axis) as a function time since implant. All the recurrence events (N 5 49 for the nadirþ2 ng/mL threshold and N 5 73 for the O0.4 ng/mL threshold) are placed into one of four consecutive 36-month intervals (x-axis).

cohort have been followed for more than 8 years, only 10% of recurrence events using the Phoenix definition (5 of 49) and 15% of events using the O0.4 ng/mL failure definition (11 of 73) are recorded at times later than 8 years.

Discussion In contrast to the present study, Merrick et al. (12) found no difference in the actuarial rate of biochemical relapse when they compared the Phoenix definition to a O0.4 ng/ mL threshold in a cohort of 284 high-risk patients treated with combined EBRT and LDR-PB (ADT was also used in 64% of their patients). However, the authors provide few details on the specific application of either PSA threshold, and, in particular, it is not clear whether PSA bounces were unfailed. The practice of unfailing men who have a benign PSA bounce is controversial. It was specifically discussed at the Phoenix consensus meeting, but as false calls were anticipated to be less than 5%, it was not thought to be of particular importance (3). However, in the era of highly effective brachytherapy, the number of false calls that trigger the Phoenix definition may equal or even exceed the true failure rate (11). It is our contention that it would be an error to include bounces as failures for two main reasons. First, it artificially overstates the true failure rate; second, it will weaken any analysis of prognostic factors predicting outcomes. The validity of the Phoenix definition and the utility of the 0.4 ng/mL definition Critics of the Phoenix definition of biochemical failure argue that it cannot be compared with the much lower PSA thresholds used by surgeons after RP. Consistent with

302

W.J. Morris et al. / Brachytherapy 13 (2014) 299e303

this criticism, we show that using a O0.4 ng/mL failure definition increases the number of PSA recurrence events by nearly 50% and reduces the median time to relapse by more than 12 months. On the other hand, if the Phoenix definition is considered the standard of comparison, then the O0.4 ng/mL failure definition carries a substantial risk of false positives. For example, in a separate study (6), our group showed that only 50% of LDR-PB patients crossing the 0.4 ng/mL threshold at 48e60 months go on to trigger the Phoenix definition by 96 months of followup. Thus, relative to the O0.4 ng/mL definition, the Phoenix definition may provide less sensitivity but achieves much better specificity. For this reason, in the opinion of the authors, it should remain the biochemical end point of record for all forms of radiation therapy. However, the authors respectfully encourage the parallel use of the 0.4 ng/mL threshold for two reasons: First, because approximately 50% of men with a PSA O0.4 ng/ mL at 48e60 months will go on to manifest biochemical failure using the Phoenix definition by 96 months (6), early investigation of these men may identify a subset with local persistent/recurrent disease who would benefit from radical salvage procedures, which (depending on initial treatment and other patient-related factors) may include surgery, brachytherapy, cryotherapy, or experimental focal treatments such as high frequency focused ultrasound. Secondly it appears to provide a more straightforward comparison with RP. For example, using as input the prognostic parameters for a large consecutive cohort of LDR-PB patients and biochemical relapse as output, the widely used Kattan nomogram predicted an actuarial relapse rate of 13% at 5 years after RP by experienced surgeons operating in high-volume centers. This rate compared unfavorably with the observed 5-year actuarial rate of 7% (log rank, p ! 0.003) for LDR-PB patients using the O0.4 ng/mL threshold to define biochemical recurrence for both modalities (13). The predictive potential of PSA values In the study cited above (6), the negative predictive value (using the Phoenix definition as the gold standard) exceeded 99% when the PSA was !0.4 ng/mL at 48e60 months after implant, but this overwhelming odds ratio does not tell the whole story. Indeed, our study confirmed other studies showing that the higher the 48e60 months PSA value, the more probable Phoenix relapse became with further followup (7). It is, therefore, noteworthy that the median PSA at the latest followup for the nonrelapsed patients in this cohort is extremely low (by postradiotherapy standards) at 0.04 ng/mL. Therefore, the authors submit that the low relapse rate observed in this cohort was not because the 48e60 months PSA values were !0.4 ng/mL but because they were, on average, about 10 times lower than that.

Given that the median residual PSA after EBRT for nonfailing patients is typically in the 0.3e1.0 ng/mL range by 5 years (3), it is obvious that applying a O0.4 ng/mL failure definition to an EBRT series would likely result in a large decrement in the actuarial DFS estimates compared with the Phoenix threshold. Thus, critics of the Phoenix definition make a reasonable case; the high (nadirþ2 ng/mL) PSA threshold, especially if combined with short followup and infrequent followup PSA measurements, can (and probably does) overestimate the effectiveness of EBRT, especially in the case of local failure in which PSA velocity after recurrence may be slow (2). However, we would argue that this criticism of the Phoenix definition does not apply to LDR-PB with its ultralow followup PSA values. Posttreatment PSA values as radiobiological markers Using multiple data sets, it should be possible to construct an isoeffect curve plotting the residual PSA values of nonfailing patients at ~5 years after treatment as a function of dose and fraction size after EBRT. The quantitative relationship derived from this curve could act as an independent yardstick by which to estimate the biologic equivalent dose delivered by LDR-PB and its more recent competitors, high-dose-rate prostate brachytherapy, hypofractionated intensityemodulated, image-guided radiation therapy, and stereotactic body radiation therapy. Unlike existing analytical models, this estimate of biologic equivalent dose would not depend on the input of multiple radiobiological parameters, such as the half time of repair, the potential doubling time, or the a:b ratio (14), whose values are imprecisely known, especially in vivo. Conclusions Surgery and radiotherapy outcomes for prostate cancer have traditionally been reported using different definitions of biochemical relapse, a reality that undermines comparison of these two treatment modalities. Our data show that the Phoenix definition, contrary to its critics, did not greatly exaggerate the actuarial DFS estimates compared with a surgical type of definition. However, this observation appears to be a direct consequence of the exceptionally low followup PSA values recorded after LDR-PB and may not generalize to all forms of radiation therapy. References [1] Herbert C, Liu M, Tyldesley S, et al. Biochemical control with radiotherapy improves overall survival in intermediate and high risk prostate cancer who have an estimated 10-year overall survival of O90%. Int J Radiat Oncol Biol Phys 2012;83:22e27. [2] Ciezki JP, Reddy CA, Stephenson AJ, et al. The importance of serum prostate-specific antigen testing frequency in assessing biochemical and clinical failure after prostate cancer treatment. Urology 2010; 75:467e471. [3] Roach M, Hanks G, Thames H, et al. Defining biochemical failure following radiotherapy with or without hormonal therapy in men with

W.J. Morris et al. / Brachytherapy 13 (2014) 299e303

[4]

[5]

[6]

[7]

[8]

[9]

clinically localized prostate cancer: Recommendations of the RTOGASTRO Phoenix consensus conference. Int J Radiat Oncol Biol Phys 2006;65:965e974. Critz FA. Time to achieve a prostate specific antigen nadir of 0.2 ng/mL after simultaneous irradiation for prostate cancer. J Urol 2002;168:2434e2438. Taira AV, Merrick GS, Galbreath RW, et al. Natural history of clinically staged low-and intermediate-risk prostate cancer treated with monotherapeutic permanent interstitial brachytherapy. Int J Radiat Oncol Biol Phys 2010;76:349e354. Hayden A, Morris WJ, Lapointe V, et al. Predicting biochemical failure free survival: Prognostic utility of the PSA value 4-5 years after brachytherapy. Radiother Oncol 2010;96. sup 2(abstract 5, page S2). Yock TI, Zietman AL, Shipley WU, et al. Long-term durability of PSA failure-free survival after radiotherapy for localized prostate cancer. Int J Radiat Oncol Biol Phys 2002;54:420e426. Morris WJ, Keyes M, Spadinger I, et al. Population-based 10-year oncologic outcomes after low-dose-rate brachytherapy for low-risk and intermediate-risk prostate cancer. Cancer 2013;119:1537e1546. Sylvester J, Grimm P, Wong J, et al. Fifteen-year biochemical relapse-free survival, cause-specific, and overall survival following

125

[10]

[11]

[12]

[13]

[14]

303

I prostate brachytherapy in clinically localized prostate cancer: Seattle experience. Int J Radiat Oncol Biol Phys 2011;81:376e381. Pickles T. Prostate-specific antigen (PSA) bounce and other fluctuations: Which biochemical relapse definition is least prone to PSA false calls? An analysis of 2030 men treated for prostate cancer with external beam or brachytherapy with or without adjuvant androgen deprivation therapy. Int J Radiat Oncol Biol Phys 2006; 64:1355e1359. Thompson A, Keyes M, Pickles T, et al. Evaluating the Phoenix definition of biochemical failure after I-125 prostate brachytherapy: Can PSA kinetics distinguish PSA failures from PSA bounces? Int J Radiat Oncol Biol Phys 2010;78:415e421. Merrick G, Butler M, Galbreath R, et al. Prostate cancer death is unlikely in high-risk patients following quality permanent interstitial brachytherapy. BJU Int 2010;107:226e233. Pickles T, Morris WJ, Kattan W, et al. Comparative 5-year outcomes of brachytherapy and surgery for prostate cancer. Brachytherapy 2011;10:9e14. Morris WJ, Halperin R, Spadinger I. Point: The relationship between postimplant dose metrics and biochemical no evidence of disease following low dose rate prostate brachytherapy: Is there an elephant in the room? Brachytherapy 2010;9:289e292.