Biochemical results from a prospective Phase II trial of intermittent androgen suppression for recurrent prostate cancer after radiotherapy were analyzed for correlations to the onset of hormone-refractory disease.
Biochemical results from a prospective Phase II trial of intermittent androgen suppression for recurrent prostate cancer after radiotherapy were analyzed for correlations to the onset of hormone-refractory disease.
Patients with histologically confirmed adenocarcinoma of the prostate and a rising serum prostate-specific antigen (PSA) level after external beam irradiation of the prostate were treated intermittently with a 36-week course of cyproterone acetate and leuprolide acetate. Then, patients were stratified according to their serum PSA range at the start of each cycle and were followed with further biochemical testing until disease progression was evident.
The mean PSA reduction was 95.2% irrespective of stratification group. A baseline serum PSA level <10 μg/L and a serum PSA nadir ≤0.2 μg/L were associated with the longest time off treatment. The overall mean nadir PSA value in the progression group at 1.40 ± 0.19 μg/L was 2.6-fold greater than the value of 0.55 ± 0.88 μg/L in the no-progression group (P = .0002). Recovery of serum testosterone to a level of ≥7.5 nmol/L was observed in 75%, 50%, 40%, and 30% of men in Cycles 1 to 4, respectively, and was sufficient to normalize the level of hemoglobin in each cycle, which dropped by an average of 10.8 g/L during treatment (P < .0001).
The length of the off-treatment interval during cyclic androgen withdrawal therapy was related inversely to baseline and nadir levels of serum PSA. Nadir PSA was a powerful predictor of early progression to androgen independence. Cancer 2007 © 2007 American Cancer Society.
The concept that intermittent androgen suppression will improve quality of life and delay the evolution of androgen resistance in patients with prostate cancer has been the focus of several clinical studies.1, 2 Although it is unlikely that any type of hormone therapy will cure the malignancy,3 there is growing appreciation of the potential benefits of preserving the hormone-dependent state of a tumor for an extended time; this includes the possibility of changing a metastatic and fatal disease into a chronic state that is amenable to long-term control.4, 5 The prospect of maintaining apoptotic potential through cyclic exposure to androgens has been evaluated in a Canadian prospective Phase II trial that examined intermittent androgen suppression as a form of therapy for men in biochemical failure after they received external-beam irradiation for locally advanced disease.6 Conducted between 1995 and 2001, the study afforded an extensive database for detailed examination not only of the clinical parameters related to intermittent androgen suppression6 but also of the biochemical effects, as described in this report.
Patients with histologically confirmed adenocarcinoma who had rising serum prostate-specific antigen (PSA) levels after external-beam irradiation of the prostate were accepted into the study. Patient characteristics were as follows: The mean age (± standard error) at baseline was 73.3 ± 0.7 years; 76% of patients had clinical stage T2b and T3 tumors; 47.6% of patients had Gleason scores ≥7, and 52.4% had Gleason scores ≤6; 83.7%, of patients had positive biopsies, and 16.3% had negative biopsies; the mean time to PSA failure (± standard error) after irradiation was 4.6 ± 0.3 years; the mean pretreatment serum PSA (± standard error) was 21.2 ± 1.6 μg/L; and the mean testosterone level (± standard error) was 13.2 ± 0.4 mol/L. This clinical trial antedated the introduction of conformal radiotherapy and dose escalation in Canada; conventional practice would have entailed from 65 grays (Gy) to 66 Gy in 1.8-Gy to 2.0-Gy fractions using a standard 4-field box. PSA testing was not available generally prior to 1992; thus, data on serum PSA levels in the trial population before the start of radiation (range, 10–20 μg/L) are incomplete. The mean and median follow-up were 3.7 years and 4.2 years, respectively.6
This was a prospective, noncomparative, open-label, multicenter study. Of 109 patients who were accrued to the study, 103 patients were evaluable.6 Patients were followed on study until the development of androgen-independent disease (time to progression), which we defined as 3 sequential increases in serum PSA >4.0 μg/L despite castrate levels of serum testosterone. On this basis, patients were divided into a no progression group and a progression group. Treatment in each cycle consisted of cyproterone acetate given as lead-in therapy for 4 weeks, followed by a combination of leuprolide acetate and cyproterone acetate, and ending after a total of 36 weeks. If the serum PSA level was <4.0 μg/L at both 24 weeks and 32 weeks with a constant or decreasing slope, then therapy was stopped at 36 weeks. If the serum PSA level was not <4.0 μg/L at both 24 weeks and 32 weeks, then the patient was not eligible for interruption of therapy and was taken off study. Serum PSA and testosterone levels were monitored every 4 weeks, and therapy was resumed when the serum PSA level reached ≥10 μg/L. At the conclusion of the clinical trial, 103 patients had been initiated into Cycle 1, 86 patients had been initiated into Cycle 2, 56 patients had been initiated into Cycle 3, 26 patients had been initiated into Cycle 4, and 7 patients had been initiated into Cycle 5. Details of the final disposition of patients as well as the complete panel of clinical and biochemical evaluations have been described previously.6
Whole venous blood (5 mL) routinely was collected in the morning, allowed to clot, and centrifuged, and the separated serum was used for the assay. Serum PSA was measured initially by using the Abbott AxSYM Total PSA poly/mono assay, which soon was supplanted by the Abbott AxSYM Total PSA mono/mono assay. The reference interval was ≤4.0 μg/L and encompassed 94.7% of specimens from normal males. The lower reportable limit of detection was <0.02 μg/L. Pairwise statistical comparisons of PSA stratification groups were based on the Mann-Whitney test.
Serum testosterone was measured by the DPC Coat-A-Count total testosterone radioimmunoassay method. The reference ranges were from 8.6 to 64 nmol/L for men ages 20 to 49 years and from 6.3 to 27 nmol/L for men aged ≥50 years. Note that blunting of the circadian rhythm in aging men may reduce daily variations within the reference range.7 For the purposes of the current analyses, we assumed that a value of 7.5 nmol/L approximated the lower limit of the normal range for all ages. The lower reportable limit of detection was <0.3 nmol/L.
This study was approved for implementation by the Health Protection Branch of the Ministry of Health of Canada and by the Clinical Research Ethics Review Board of each participating center. All patients provided written informed consent in accordance with institutional guidelines.
The data presented in Figure 1 indicate that the pattern of response to androgen withdrawal for both serum PSA and testosterone was biphasic in each of 4 consecutive cycles. Typically, there was an initial steep slope followed by a shallow slope. For the serum PSA level, it is believed that this represents an initial inhibition of PSA synthesis over the first 8 to 10 weeks, after which apoptotic cell loss accounts for further reduction in serum PSA.4 The mean serum PSA values at the beginning of each cycle were 18.5 μg/L, 10.5 μg/L, 11.5 μg/L, and 11.9 μg/L, for Cycles 1 thorough 4, respectively. The overall reduction of serum PSA during the on-treatment period amounted to 98.2% in Cycle 1, 91.8% in Cycle 2, 96.3% in Cycle 3, and 94.5% in Cycle 4 for an average reduction of 95.2%.
The mean serum testosterone values at the beginning of each cycle were 12.8 nmol/L, 9.2 nmol/L, 8.2 nmol/L, and 7.2 nmol/L for Cycles 1 through 4, respectively. The corresponding PSA nadir values at 36 weeks of treatment were 0.45 nmol/L, 0.40 nmol/ L, 0.38 nmol/L, and 0.50 nmol/L, respectively. Thus, the overall reduction of serum testosterone during the on-treatment period amounted to 96.5% in Cycle 1, 95.6% in Cycle 2, 95.3% in Cycle 3, and 93.6% in Cycle 4 for an average reduction of 95.3%. These observations indicate that there is a reproducible inhibitory effect of treatment on testicular function even as the number of cycles increases. A trend toward reduced baseline levels of serum testosterone at the start of each cycle is in keeping with a declining ability of the testis to recover and synthesize a normal amount of testosterone.8, 9
The percentage reduction in serum PSA versus the duration of treatment for each of 4 consecutive cycles is plotted in Figure 2. In all 4 cycles, a plateau was achieved at approximately 95% after 40 weeks of treatment. The plateau level in Cycle 1 was attained after 20 to 24 weeks of treatment, whereas a longer period of treatment, 36 to 40 weeks, was required to reach the same level in Cycles 2, 3, and 4.
To determine whether the reduction in serum PSA in response to androgen withdrawal was influenced by PSA range and to establish whether there may be some predictive value inferred from such a relation, patients were stratified according to PSA ranges of 4 < × ≤10 μg/L, 10 < × ≤20 μg/L, and >20 μg/L, as indicated in Table 1. The magnitude of PSA response was calculated relative to the baseline PSA at the beginning of each cycle. Patients in Cycle 1 with a PSA range of 4 < × ≤10 μg/L showed a mean reduction of 7.4 ± 0.3 μg/L. Patients with a PSA range of 10 < × ≤20 μg/L had a mean 2-fold greater reduction of 14.0 ± 0.5 μg/L (P < .0001). Another approximate doubling to 34.6 ± 3.5 μg/L (P < .0001) was observed in patients with a serum PSA range >20 μg/L. Despite the greater absolute reductions in serum PSA with increasing baseline serum PSA at the beginning of each cycle, the percentage reduction was relatively constant between 95.8% and 97.2%. Similar trends in absolute and percentage reductions in serum PSA were observed in Cycles 2, 3, and 4 with an overall mean percentage reduction that incorporated all ranges of 94.6%.
|Cycle||PSA range, μg/L*|
|4 < × ≤10||10 < × ≤20||>20|
|Reduction, μg/L†||%||No.||Reduction, μg/L†||%||No.||Reduction, μg/L†||%||No.|
|Cycle 1||7.4 ± 0.3||95.8||18||14 ± 0.5||96||45||34.6 ± 3.5||97.2||23|
|Cycle 2||8.1 ± 0.3||93.5||19||12.5 ± 0.5||93.8||39||26.8 ± 5.5||94.1||5|
|Cycle 3||6.9 ± 0.4||93.1||2||13.2 ± 0.6||93||23||26.7 ± 5||95.7||3|
|Cycle 4||—‡||0||0||12.5 ± 0.8||93.5||11||20.8 ± 0.5||94.6||2|
The results presented in Table 2 show the relation between the baseline serum PSA level and the mean time off treatment when patients were stratified into 3 different groups according to their serum PSA range: 4 < × ≤10 μg/L, 10 < × ≤20 μg/L, and >20 μg/L. In Cycle 1, the mean time off treatment in the first group was 90.7 ± 11.5 weeks, which decreased by a statistically significant 29% to 65.2 ± 5.4 weeks (P = .0393) in the second group and by 57% to 39.4 ± 4.6 weeks (P < .0001) in the third group. In Cycle 2, the mean time off treatment in the first group was 66.5 ± 9.5 weeks, which decreased by a nonsignificant 39.9% to 40 ± 3.4 weeks (P = .0677) in the second group, and no further decrease was observed in the third group (P = .2144). In Cycle 3, the mean time off treatment in the first group was considerably lower at 26.2 ± 5 weeks and also was lower than the values in the other 2 groups (35.5 ± 4.2 weeks and 40.7 ± 4.5 weeks for the second and third groups, respectively). The mean time off treatment in the third group (men with a baseline PSA level >20 μg/L) was the same in Cycles 1 through 3, (ie, approximately 40 weeks) before it decreased to 24.2 ± 2.8 weeks in Cycle 4.
|Cycle||Baseline PSA, μg/L|
|4 < × ≤10||10 < × ≥ 20||>20|
|Weeks off treatment*||No.||Weeks off treatment*||No.||Weeks off treatment*||No.|
|Cycle 1||90.7 ± 11.5||21||65.2 ± 5.4||46||39.4 ± 4.6||26|
|Cycle 2||66.5 ± 9.5||16||39.9 ± 3.4||35||41.1 ± 4.8||20|
|Cycle 3||26.2 ± 5||7||35.5 ± 4.2||16||40.7 ± 4.5||14|
|Cycle 4||22.7 ± 0||1||29.4 ± 4.9||9||24.2 ± 2.8||5|
To determine whether the nadir level of serum PSA after 36 weeks of androgen withdrawal therapy was related to the baseline serum PSA level at entry to trial, patients were stratified according to PSA range, and mean nadir levels of serum PSA were calculated for each category. The resultant data (not shown) indicated that, in Cycle 1, in the first group (men with baseline serum PSA 4 < × ≤10 μg/L), the mean nadir level of serum PSA was 0.3 ± 0.1 μg/L. This increased to 0.6 ± 0.1 μg/L in the second group (men with baseline serum PSA 10 < × ≤20 μg/L) and to 1.0 ± 0.3 μg/L in the third group (men with baseline serum PSA >20 μg/L). Thus, there was a 3-fold difference in the mean nadir level of serum PSA (P = .05) between men with the lowest baseline serum PSA range and men with the highest baseline serum PSA range. Similar trends were observed in Cycles 2, 3, and 4 but were not statistically significant.
The inverse correlation between the nadir level of PSA and time off treatment is illustrated by the data provided in Table 3. For men in Cycle 1 who had nadir levels of serum PSA ≤0.2 μg/L, the mean time off treatment was 75.7 ± 7.1 weeks. This time off treatment was reduced by a relatively constant 30% to 53.5 ± 5.0 weeks for men who had nadir levels of serum PSA 0.2 < × ≤1.0 μg/L, to 37.2 ± 6.7 weeks for men who had nadir levels of serum PSA 1.0 <× ≤2.0 μg/L, and to 26.5 ± 8.4 weeks for men who had nadir levels of serum PSA >2.0 μg/L. The mean time off treatment for men who achieved nadir levels of serum PSA ≤0.2 μg/L was 3 times greater (P = .0046) than the time off treatment for men who had nadir levels of serum PSA >2.0 μg/L. A similar but weaker trend was observed in Cycle 2; and, by Cycle 3, the differences in time off treatment among the different stratification groups, for the most part, had disappeared. Even in the most favorable stratification groups (ie, nadir levels of serum PSA ≤0.2 μg/L, 0.2 < × ≤1.0 μg/L, and 1.0 < × ≤2.0 μg/L), the mean time off treatment decreased as the number of cycles increased.
|Cycle||PSA nadir, μg/L|
|≤0.2||0.2 < × ≤1||1 < × ≤2||>2|
|Weeks off Treatment*||No.||Weeks off treatment*||No.||Weeks off treatment*||No.||Weeks off treatment*||No.|
|Cycle 1||75.7 ± 7.1||44||53.5 ± 5||31||37.2 ± 6.7||5||26.5 ± 8.4||6|
|Cycle 2||59.7 ± 6.9||22||41.6 ± 4.1||26||42.3 ± 11||9||27.3 ± 6.8||7|
|Cycle 3||39.4 ± 4.8||8||38.8 ± 5.5||12||28.2 ± 1.9||5||24.5 ± 4.7||3|
|Cycle 4||†||32.6 ± 4.4||8||12.4 ± 1.3||2||25.2 ± 3.9||2|
To investigate a possible relation between nadir levels of serum PSA and treatment outcomes, the nadir levels achieved in different stratification ranges were compared between progression-free patients and patients who developed disease progression. The results of that comparison are shown in Table 4. In all cycles (except Cycle 1 in the 4 < × ≤10 stratification range), the absence of disease progression was associated with a mean nadir level of serum PSA less than the corresponding level in the same stratification range associated with disease progression.
|Cycle||PSA range, μg/L*|
|4 < × ≤10||10 < × ≤20||>20|
The mean nadir level of serum PSA obtained by combining the nadir levels of all cycles within a given stratification range was compared in the no-progression and progression groups (Table 5). If the PSA range was 4 < × ≤10 μg/L, then the mean nadir was 0.41 ± 0.08 μg/L in patients without disease progression, and it was 2.6-fold greater (1.06 ± 0.43 μg/L) in patients with disease progression (P = .0325). If the PSA range was 10 < × ≤20 μg/L, then the mean nadir values corresponding to the same 2 groups were 0.56 ± 0.06 μg/L and 1.35 ± 0.24 μg/L, respectively, for a 2.4-fold significant difference (P = .0004). A PSA range >20 was characterized by mean nadir values of 0.75 ± 0.20 μg/L and 1.67 ± 0.40 μg/L, respectively, for a 2.3-fold significant difference (P = .0111) between the groups. Combining nadir values for all stratification ranges of serum PSA, the mean nadir associated with no progression was 0.55 ± 0.05 μg/L compared with 1.40 ± 0.19 μg/L with progression, a statistically significant, 2.6-fold difference (P = .0002).
|PSA Range, μg/L||No progression||Progression||P|
|Nadir, μg/L*||No. of nadirs||Nadir, μg/L*||No. of nadirs|
|4 < × ≤10||0.41 ± 0.08||36||1.06 ± 0.43||4||.0325|
|10 < × ≤20||0.56 ± 0.06||92||1.35 ± 0.24||27||.0004|
|>20||0.75 ± 0.20||22||1.67 ± 0.40||11||.0111|
|All ranges||0.55 ± 0.05||150||1.40 ± 0.19||42||.0002|
The data presented in Figure 3 show the rates of increase in serum levels of testosterone and PSA after the interruption of androgen withdrawal therapy for each of 4 cycles. In Cycle 1, serum testosterone levels recovered to a peak of 10 nmol/L approximately 50 weeks after the beginning of the off-treatment part of the cycle. The mean levels then decreased gradually over the next 150 weeks and plateaued at approximately 2.5 nmol/L, because the patients who remained in the off-treatment period of Cycle 1 beyond 1 year comprised a group with delayed, incomplete, or no testosterone recovery. Serum PSA levels also increased, but the velocity of increase lagged behind that of serum testosterone, plateaued at 7.5 μg/L approximately 80 weeks after the start of the off-treatment period, and remained at that level beyond 250 weeks. This plateau effect occurred because the only patients whose PSA was still being measured in Cycle 1 at these time points were men with low serum testosterone levels whose PSA levels had not gone high enough to move into Cycle 2. In Cycles 2 through 4, the gap between the rate of increase in serum testosterone levels and the rate of increase in serum PSA levels became smaller. The serum PSA level achieved a more durable plateau in men whose serum testosterone levels remained in the vicinity of 5 nmol/L, resulting in a longer time off treatment.
The ability of testis to restore serum testosterone levels to the normal range was examined by plotting the percentage of patients who had serum testosterone levels ≥7.5 nmol/L as a function of time after the interruption of androgen withdrawal therapy (Fig. 4). In Cycle 1, recovery of testicular function was complete after approximately 5 months, and from 70% to 75% of men showed serum testosterone levels in this range. The recovery of serum testosterone in Cycle 2 was characterized by a longer lag period before the level of testosterone began to increase; and, by 5 months after the start of the off-treatment period, only approximately 50% of men had achieved serum testosterone levels ≥7.5 nmol/L. In Cycles 3 and 4, there was a further reduction in the percentage of men whose serum testosterone levels recovered into the normal range.
We expected that intermittent androgen suppression would influence blood hemoglobin levels; and, indeed, it was possible to document fluctuations in each of 4 cycles of therapy, as shown in Figure 5. From a baseline level of 143.3 ± 1.8 g/L, hemoglobin levels decreased by 7.6 g/L to 135.7 ± 3.4 g/L at the end of the period of androgen suppression in Cycle 1. After interruption of therapy, the hemoglobin increased to the baseline level of 143.8 ± 1.9 g/L; then, it dropped by 14.7 g/L to 129.1 ± 5.3 g/L at the end of treatment in Cycle 2 before recovering to 139.9 ± 2.4 g/L after therapy was interrupted. Further reductions and recoveries of hemoglobin of similar magnitude were observed in Cycles 3 and 4. The overall mean baseline hemoglobin level at the start of each cycle was 143.1 ± 1.7 g/L (n = 140 patients); this was reduced to 132.3 ± 2.3 g/L (n = 36 patients) at the time that therapy was interrupted, a difference of 10.8 g/L (P < .0001). Thus, despite declining levels of serum testosterone, as documented in Figure 4, these were sufficient to restore hemoglobin levels to normal in each successive cycle of intermittent therapy.
Biochemical recurrence after irradiation for localized prostate cancer proved amenable to cyclic androgen withdrawal therapy and showed a reproducible pattern of response for levels of serum PSA and testosterone in 4 consecutive cycles (Fig. 1). The overall mean reduction of serum PSA levels during the on-treatment period was 95.2%. Reductions of this magnitude were associated with an approximately 40% decrease in the volume of the prostate in Cycles 1 and 2, but there were no detectable changes in volume in Cycles 3 and 4.6
The observations presented in Table 1 indicate that the total reduction in serum PSA is a function of the serum PSA value at the start of each cycle of treatment. Although a patient may be stratified into a different PSA range with increasing cycle number, and although the magnitude of the PSA reduction may increase or decrease accordingly, the actual percentage change in serum PSA remains the same (overall mean reduction, 94.6%). Thus, a shift from 1 stratification level to another will affect the net reduction in serum PSA, but this would not be accompanied by any alteration in the percentage decrease, which remains almost constant at 95%. It follows that the stratification level of serum PSA in itself has no predictive value insofar as all patients will experience a drop in serum PSA in the vicinity of 95%.
The rate of serum PSA reduction decreased as the number of treatment cycles increased (Fig. 2). A maximum reduction in serum PSA levels of 95% was observed in all 4 cycles after 40 weeks of treatment. However, the plateau level in Cycle 1 was observed after from 20 weeks to 24 weeks of treatment, whereas a longer period (from 36 weeks to 40 weeks) was required to reach the same plateau level in Cycles 2, 3, and 4. This lengthening may reflect the selection of increasingly testosterone-sensitive tumors in successive cycles (Fig. 3) and the need for a greater, time-dependent reduction of serum testosterone (Fig. 1). Because the maximum achievable reduction in serum PSA was 95%, it is plausible that the on-treatment period could be shortened and limited to the length of time required to bring about a 95% reduction in the level of serum PSA.10, 11
The overall mean reduction in the serum testosterone level during the on-treatment periods for all cycles of therapy was 95.3% (Fig. 1). There was progressive reduction in the baseline levels of serum testosterone at the start of each cycle, suggesting that the ability of the testis to recover and synthesize a normal level of testosterone declined with repeated cycling. Whether this represented primary gonadal failure or failure secondary to pituitary insufficiency was not assessed in our study.
In relating mean time off treatment to the baseline serum PSA level at trial entry (Table 2), it was demonstrated that a serum PSA level <10 μg/L was associated with the longest time off treatment; this treatment-free interval was 27% shorter in Cycle 2, 71% shorter in Cycle 3, and 74% shorter in Cycle 4. The same trend toward a shorter time off treatment was observed in the group of patients who had baseline serum PSA levels in the 10 < × ≤20 μg/L range. For the patients who had baseline serum PSA levels >20 μg/L, the time off treatment remained constant in the first 3 cycles before it became shorter in the Cycle 4. The finding that, in Cycle 1, the time off treatment in the group of men with serum PSA in the 4 < × ≤10 μg/L range, relative to the men with serum PSA levels >20 μg/L, was 2.3 times longer (P < .0001) suggests that intermittent androgen suppression therapy should be started before the serum PSA reaches a level of 10 μg/L. We speculate that the mean time off treatment is significantly longer in the lower serum PSA range, because patients have a smaller residual tumor burden after androgen withdrawal. Thus, the subsequent regrowth of tumor based on PSA, a variable that depends on the number of cells and the level of PSA up-regulation, should take place at a slower rate.
Because the data shown in Table 1 indicate that androgen withdrawal routinely will produce a constant 95% reduction in the serum PSA level independent of the range of serum PSA at baseline, it is clear that there will be a direct relation between the serum PSA level at the start of androgen withdrawal and the achievable nadir PSA level. This relation was most apparent in Cycle 1, in which there was a 3-fold difference in nadir values between men who had baseline serum PSA levels in the lowest range compared with men who had baseline serum PSA levels that fell within the highest range. Furthermore, there was an inverse relation between PSA nadir and time off treatment (Table 3), with the longest time off treatment observed in Cycle 1 among men who had nadir values of serum PSA ≤0.2 μg/L. Because the serum PSA level will decrease by 95% over a treatment period of 36 weeks, we infer that the time off treatment will be superior if androgen withdrawal is introduced when the serum PSA level is approximately 4.0 μg/L.
The mean nadir level of serum PSA tended to increase with more cycles of therapy in each of the 3 stratification ranges, an effect that was slightly more pronounced in the group of men who progressed to androgen independence during the course of this study (Table 4). The overall mean nadir value was 0.55 ± 0.05 μg/L in the no-progression group compared with a 2.6-fold greater value of 1.40 ± 0.19 μg/L (P = .0002) in the progression group (Table 5). This is in keeping with the data reported by Scholtz and colleagues,12 who observed that a PSA nadir of >0.1 ng/mL was an important factor for predicting the early onset of androgen independence. It is possible that the successive increases in PSA nadir levels in all stratification ranges are the results of recruiting alternative pathways of androgen receptor activation under conditions of testosterone deprivation, an early step in progression to androgen independence.5
Univariate analysis of progression-free survival data revealed that the Cycle 1 PSA nadir was a significant prognostic factor (P < .001), whereas no significance could be attached to age (P = .1), clinical stage at initial diagnosis (P = .08), Gleason score (P = .8), baseline prostatic volume (P = .7), baseline PSA (P = .6), baseline testosterone (P = .09), or time from irradiation to the start of treatment (P = .1). Multivariate analysis of selected factors confirmed the significance of Cycle 1 PSA nadir (hazard ratio, 2.2; P = .001) but not of age, clinical stage, baseline PSA, baseline testosterone, or time from irradiation to the start of treatment.
There was a close relation between the recoveries of serum testosterone and PSA (Fig. 3). Men who had serum testosterone levels that recovered quickly experienced a more rapid increase in serum PSA levels, affording a shorter off-treatment period in the range from 50 weeks to 100 weeks. In approximately 10% of patients, the mean levels of serum testosterone in Cycle 1 decreased after 100 weeks to plateau at 2.5 nmol/L. The abnormally low serum testosterone levels in such men resulted in a prolonged time off treatment that occasionally lasted up to 5 years.
Closing of the gap between the rate of increase in serum testosterone and the rate of increase in serum PSA likely reflects the selection of more androgen-sensitive tumors.5 Men with such tumors will experience a rapid increase in serum PSA as testosterone recovers, triggering the resumption of androgen withdrawal therapy. The finding that the mean durations of the off-treatment period in Cycles 2, 3, and 4 (46.2 ± 3.3 weeks, 35.7 ± 2.8 weeks, and 27.2 ± 3.2 weeks, respectively) were considerably shorter than in Cycle 1 (63.7 ± 4.4 weeks)6 supports this hypothesis.
After the interruption of androgen withdrawal therapy in Cycle 1, 75% of men demonstrated recovery of serum testosterone into the low normal range (7.5 nmol/L) within 5 months after the interruption of therapy (Fig. 4), in agreement with the observations of Nejat et al.8 This percentage decreased in succeeding Cycles 2, 3, and 4 to 50%, 40%, and 30%, respectively. Despite such evidence of declining testicular function, the levels of serum testosterone were able to restore hemoglobin to normal in successive cycles in which the hemoglobin had dropped by an average of 10.8 g/L during treatment. The reversal of anemia by intermittent androgen suppression has been noted by others.13
The current data highlight several correlations between serum PSA and the outcome of intermittent androgen suppression in men with locally advanced prostate cancer who have progressed after radiotherapy. We observed that the nadir level of serum PSA was a strong determinant of time off treatment and a powerful predictor of hormone-refractory tumor progression. The nadir level of serum PSA itself is determined by the baseline serum PSA level and a 95% limit on the reduction of serum PSA in response to androgen withdrawal. From these observations, we conclude that the benefit of intermittent androgen suppression is most apparent if the serum PSA level is ≤4 μg/L when therapy is started. An advantage of early hormone therapy in the treatment of recurrent prostate cancer also can be implied from our data. The effects of intermittent androgen suppression on quality of life and on progression-free and overall survival will be discussed in a subsequent report.
We thank the staff of the Tumor Marker Laboratory at the British Columbia Cancer Agency for expert biochemical testing, Jonas Abersbach for information technology support and Alla Sekunova for typing and editing the article.