Response to low-dose ketoconazole and subsequent dose escalation to high-dose ketoconazole in patients with androgen-independent prostate cancer

Abstract

BACKGROUND.

High-dose ketoconazole (HDK) in combination with steroids has been recognized as an effective secondary hormonal therapy in androgen-independent prostate cancer (AIPC). However, HDK causes more severe adverse events than low-dose ketoconazole (LDK). To the authors' knowledge, relatively little is known regarding the efficacy of LDK in AIPC. The efficacy of LDK and of subsequent dose escalation from LDK to HDK was evaluated as secondary hormonal therapy in patients with AIPC.

METHODS.

In a single institution, patients with AIPC treated with LDK (at a dose of 200 mg orally 3 times daily) as secondary hormonal therapy with or without concomitant steroids were retrospectively identified. In addition, patients were identified who received dose escalation to HDK (400 mg orally 3 times daily) after experiencing a rising prostate-specific antigen (PSA) level.

RESULTS.

Thirty-nine of 138 eligible patients (28.3%, 95% confidence interval [95% CI], 20.9–36.6%) treated with LDK experienced PSA declines ≥50%. The median time to disease progression or dose escalation on LDK was 3.2 months (range, 0.1⁺–61 months). Dose escalation to HDK was subsequently performed in 55 patients (39.9%), 7 of whom (12.7%) demonstrated a subsequent PSA decline ≥50%. A longer duration of primary androgen deprivation therapy and total duration of all previous hormonal therapies was associated with a longer time to progression with LDK (P < .05). The most common reversible adverse effect of LDK was NCI Common Toxicity Criteria Grade 1 or 2 fatigue (12.3%).

CONCLUSIONS.

LDK is associated with a PSA response rate comparable to HDK as secondary hormonal therapy in patients with AIPC, but with less toxicity. Although uncommon, additional durable responses occurred in some patients after dose escalation. Cancer 2006. © 2006 American Cancer Society.

Prostate cancer is the second most common cause of cancer death in men in the U.S. Standard initial management of prostate disease consists of androgen deprivation therapy (ADT) to achieve a castrate level of testosterone. Although the majority of patients respond to ADT, most will progress to androgen-independent disease. Androgen-independent prostate cancer (AIPC) occurs when there is evidence of cancer progression in the setting of castrate levels of testosterone. Often, the initial sign of androgen-independent progression is an asymptomatic increase in prostate-specific antigen (PSA), although patients may also have radiographic evidence of metastases.1 When initial ADT fails, options include secondary hormonal therapy and chemotherapy. Secondary hormonal therapies include antiandrogens, adrenal androgen inhibitors, and estrogenic therapies.2 Recently, docetaxel-based chemotherapy was shown to confer a survival benefit in patients with metastatic AIPC, although the median survival benefit was only 2–3 months.3, 4 Chemotherapy is also associated with toxicity, which may diminish quality of life, particularly in asymptomatic patients. Thus, it is reasonable to consider oral regimens with lower toxicity in appropriate clinical scenarios before initiating chemotherapy.

Ketoconazole, a broad-spectrum azole antifungal drug, is a potent inhibitor of gonadal and adrenal steroid synthesis, through the inhibition of both adrenocortical 11β-hydroxylase and cholesterol side-chain cleavage.5 High-dose ketoconazole (HDK) induces prolonged decreases in plasma testosterone and cortisol levels without affecting plasma estradiol concentrations.6, 7 Ketoconazole therefore achieves additional suppression of testosterone synthesis in the adrenal gland beyond testicular androgen suppression.

HDK combined with corticosteroids has been shown to have moderate activity as a secondary hormonal therapy in men with AIPC.8–11 A Phase III trial (CALGB 9583) demonstrated that HDK and hydrocortisone with antiandrogen withdrawal (AAWD) was associated with declines in PSA ≥50% in 32% of 128 patients with AIPC.12 The median time to PSA progression was 8.6 months in the HDK plus AAWD arm, compared with 5.9 months in the AAWD alone arm. Whereas HDK was active, approximately 21% of patients had NCI Common Toxicity Criteria Grade 3 and 4 toxicities including neurologic toxicity, fatigue, and hepatotoxicity. Such adverse effects often require dose reduction and mandate glucocorticoid replacement.

Low-dose ketoconazole (LDK) may offer a more favorable toxicity profile than HDK. A single, small Phase II study assessed the efficacy and safety of LDK with hydrocortisone (30 mg daily) in 28 patients with AIPC.13 Thirteen of 28 patients (46%) had a PSA decline of ≥50%, with a median duration of PSA decline of 7.5 months. Toxicity was mild, with 10.7% Grade 3 and no Grade 4 toxicity. In that study, patients were treated with HDK at the time of disease progression. Sixteen patients (57.1%) received HDK; however, none responded to dose escalation.

In our institution, we have used a standard regimen in men with progressive AIPC starting with LDK (200 mg 3 times daily), with or without a corticosteroid. In the event that the patient had disease progression, usually evidenced by a rising PSA, a subsequent increased dose was administered to some patients of 400 mg given 3 times daily (HDK), usually with the addition of a corticosteroid if they were not already receiving it. We conducted a retrospective study to evaluate the efficacy and toxicity profile of LDK as secondary hormonal therapy in patients with AIPC. In addition, we analyzed the efficacy of dose escalation in the subset of patients who were treated with HDK and investigated the predictors of PSA response and time to progression.

MATERIALS AND METHODS

Eligibility and Treatment

The Institutional Review Board of the Dana-Farber Cancer Institute approved this retrospective study. All patients provided written informed consent to allow analysis of clinical data for research purposes. Eligible patients were identified by utilizing our Prostate Cancer Clinical Research Information System (Prostate CRIS), which is a relational database created at Dana-Farber Cancer Institute. Eligible patients included those treated with luteinizing hormone-releasing hormone (LHRH) analogs or bilateral orchiectomy and demonstrating progressive disease, according the PSA Working Group Consensus Criteria.14 Patients receiving LHRH analog therapy were required to have continued this treatment during ketoconazole treatment. All patients had a detectable PSA at the initiation of ketoconazole therapy, but were not required to have radiographic evidence of disease. Any number of prior secondary hormonal therapies was allowed. Patients were not allowed to have prior chemotherapy for AIPC. Patients were excluded from the analysis if they received concurrent antiandrogens, estrogens, other adrenal steroid inhibitors, or chemotherapy.

Treatment consisted of a dose of ketoconazole of 200–400 mg given 3 times daily with or without replacement doses of glucocorticoids (hydrocortisone at a dose of 40 mg/day orally or prednisone 10 mg/day orally).

Response Criteria

The primary endpoint of the study was to define the percentage of patients who achieved a clinically significant decline in PSA after initiation of ketoconazole, defined as a ≥50% decrease. Patients whose PSA values showed a decrease in PSA <50% were considered nonresponders for this analysis. Routine radiographic imaging was not performed. Response to LDK and HDK were evaluated separately from the PSA value at the initiation of the respective dose. Duration of LDK was defined as the time from initiation to cessation of LDK.

A modified version of the PSA Working Group Consensus Criteria was used to evaluate the time to progression separately for each dose. Date of PSA progression was defined as either the date the PSA increased 50% above the nadir, if a ≥50% decline in PSA had been achieved, or the date the PSA increased 25% above the nadir if a <50% decline in PSA was experienced.14 In some patients with <50% decline, the date of progression was defined as the date ketoconazole was ended because of rising PSA. While on LDK, some patients similarly had the date of progression defined as the date LDK was ended because of rising PSA with subsequent escalation to HDK. Time to PSA progression on LDK and HDK were defined from start of the respective dose of ketoconazole until the date of PSA progression. It was censored at the date the patient stopped ketoconazole if the patient stopped because of side effects or comorbidity before PSA progression or censored at the last follow-up visit if still responding or lost follow-up. Adverse events were assessed using National Cancer Institue Common Toxicity Criteria (version 3.0; available at URL: http://ctep.cancer.gov/reporting/ctcnew/html [published December 12, 2003]).

Statistical Analysis

Variables were summarized as the number (percent) of patients or median and range of values. Exact binomial 95% confidence intervals (95% CI) were reported for the response rate. Categorical variables were compared between LDK responders and LDK nonresponders using the Fisher exact test and extensions; continuous variables were compared using Wilcoxon rank sum tests. Time to PSA progression was compared between groups using the log-rank test. The overall survival time was defined from the start of ketoconazole until date of death, which was censored at the last follow-up visit if the patient was still alive or lost to follow-up. The statistical analysis was undertaken using SAS software (version 9; SAS Institute Inc., Cary, NC) and P < .05 (2-sided) was considered as statistically significant.

RESULTS

Patient Characteristics

One hundred fifty-eight patients treated with ketoconazole were identified via an ongoing prospective institutional prostate cancer database. Additional retrospective chart review was performed in these patients, and 138 patients were found to be eligible for this study. Eleven patients were ineligible because they had received an initial dose of ketoconazole other than that defined in this analysis: 200 mg twice daily (n = 3 patients), 400 mg twice daily (n = 3 patients), and 400 mg 3 times daily (n = 5 patients). Nine patients were ineligible because they received an unknown dose of ketoconazole. Table 1 summarizes the baseline patient characteristics. Fifty-five patients (39.9%) had dose escalation to HDK; there were no statistically significant differences in the baseline characteristics including PSA, Gleason score, clinical stage, and status of metastases at diagnosis or at the initiation of ketoconazole between patients receiving dose escalation and patients without dose escalation. Two patients were still receiving HDK at the time of analysis. The median follow-up after starting ketoconazole therapy was 27.6 months (range, 0.8⁺–77.4 months).

Table 1. Patient Characteristics

	Total (n = 138)
PSA indicates prostate-specific antigen; RP, radical prostatectomy; EBRT, extrabeam radiation therapy; LHRH, luteinizing hormone-releasing hormone; LN, lymph node.
Median age, y (range)	64 (41–84)
PSA (ng/mL)
At diagnosis	19.0 (1.8–696)
At initiation of ketoconazole	37.4 (0.75–2279)
Gleason score, n (%)
4–6	25 (18.1)
7	38 (27.5)
8–10	60 (43.5)
Unknown	15 (10.9)
Clinical stage, n (%)
T1	26 (18.8)
T2	57 (41.3)
T3–4	14 (10.1)
Tx	41 (29.7)
Definitive local therapy, n (%)
RP	42 (30.4)
EBRT	50 (36.2)
Brachytherapy	1 (0.7)
None	45 (32.6)
Castration, n (%)
LHRH analog	108 (78.3)
Orchiectomy	21 (15.2)
Orchiectomy + LHRH analog	9 (6.5)
Metastases, n (%)
Yes	109 (79.0)
No	28 (20.3)
Unknown	1 (0.7)
Sites of metastases, n
Bone	83
Bone and LN	18
LN	7
Lung	2
Other	4
Previous hormonal therapy, n
Bicalutamide	107
Flutamide	59
Nilutamide	3
Estrogen	8
Corticosteroids	4

PSA Response

Of 138 patients, 39 (28.3%, 95% CI, 20.9–36.6%) patients demonstrated PSA declines ≥50% after therapy with LDK. Twenty-two (15.9%) patients experienced a PSA decline ≥80% after initiating LDK. The median duration of LDK therapy was longer in responders than nonresponders (10 months vs. 2.5 months). A higher PSA at initiation of ketoconazole was associated with a higher response rate to ketoconazole (56.0 ng/mL vs. 29.4 ng/mL; P = .0072).

Whereas all patients received LDK initially, only 55 (39.9%) patients had dose escalation to HDK. The median duration of LDK was 3.6 months (range, 0.5–32.2 months) in the dose-escalation group and 3.0 months (range, 0.1–62.4 months) in the nondose-escalation group (P = .85). Seven of 55 patients (12.7%) had a subsequent PSA decline ≥50% after dose escalation to HDK, and 13 (23.6%) patients had a PSA decline of ≥25%. As noted in Table 2, no differences in PSA declines ≥50% were noted in patients based on their response to LDK, but using a more moderate PSA decline of ≥25%, 35.3% of LDK responders and 18.9% of LDK nonresponders had a subsequent response to HDK. Overall, 44 of 138 patients (31.9%) responded to ketoconazole at either dose.

Table 2. Effects of Dose Escalation to HDK Based on Response to LDK

	LDK Nonresponders (n = 37)	LDK Responders (n = 18)
LDK indicates low-dose ketoconazole; HDK, high-dose ketoconazole; PSA, prostate-specific antigen; LDK responders, patients who had PSA decline ≥50% to LDK; LDK nonresponders, patients who had PSA decline <50% to LDK.
Response to HDK (≥50% decline), n (%)	5/37 (13.5%)	2/17 (11.8%)
Response to HDK (≥25% decline), n (%)	7/37 (18.9%)	6/17 (35.3%)
Time to HDK progression, mo	1.3	2.9
Duration of LDK, mo	2.9 (0.5–12.9) (n = 35)	6.4 (2.6–32.2) (n = 17)
Duration of ketoconazole, mo	4.7 (0.6–20.7) (n = 34)	12.3 (5.7–33.8) (n = 17)
PSA at start of LDK, ng/mL	29.4 (0.8–316.5) (n = 37)	29.8 (4.3–1653.9) (n = 18)
PSA at start of HDK, ng/mL	35.5 (0.97–314.0) (n = 33)	15.9 (0.3–622.0) (n = 17)

Prior Hormonal Therapy and Concomitant Steroid Use

In all patients the median duration of primary ADT was 27.7 months (range, 0.8–130.6 months). The median duration of all previous hormonal therapy before ketoconazole was 35.4 months (range, 2.8–148.7 months). The duration of all previous hormonal therapy in LDK responders was 43.4 months (range, 5.7–138.1 months), whereas it was 31.5 months (range, 2.8–148.7 months) in LDK nonresponders (P = .14).

Of the total of 138 patients, ketoconazole was used as second-line hormonal therapy in 35 patients (25.4%), as third-line hormonal therapy in 69 patients (50.0%), as fourth-line therapy in 30 patients (21.7%), and as fifth-line hormonal therapy in 2 patients (1.4%). Two patients had missing data. Responses to ketoconazole were observed when used as second- to fifth-line hormonal therapy. Among the patients previously treated with antiandrogens, 100 patients had AAWD and 24 (24%) experienced an AAWD response. Overall, 25.9% of LDK responders and 23.3% of LDK nonresponders achieved an AAWD response, suggesting that an AAWD response did not correlate with a subsequent ketoconazole response (P = .80).

Sixty-two patients (44.9%) received concomitant steroids. Twenty-one LDK responders (53.9%) received concomitant steroids, whereas 41 LDK nonresponders (41.4%) received concomitant steroids (P = .25).

Time to Progression

Ninety (65.2%) patients experienced PSA and/or disease progression during treatment with LDK and 26 patients (18.8%) were dose-escalated to HDK because of rising PSA. Eleven patients (8.0%) stopped ketoconazole because of side effects before progression, 2 patients were lost to follow-up, 1 was admitted to a hospital due to comorbidity, 1 stopped because of participation in a clinical trial, and 7 patients had missing information. The median time to disease progression on LDK or dose escalation to HDK was 3.2 months (range, 0.1⁺–61 months); in LDK responders, it was 9.5 months (range, 2.6–61 months) and in nonresponders, it was 1.8 months (range, 0.1⁺–16.8 months). The univariate associations of patients' disease characteristics and treatment with time to progression were also investigated. Time to progression on LDK was significantly associated with a duration of primary ADT and a duration of all previous hormonal treatments. The median time to progression on LDK was 4.3 months among patients with a duration of primary ADT ≥27.7 months, and 2.8 months among patients with a duration of primary ADT below the median value of 27.7 months (P = .0152). The median time to progression on LDK was 4.1 months among patients with a duration of all previous hormonal treatments ≥35.4 months and 2.8 months among patients with a duration of all previous hormonal therapies below the median value of 35.4 months (P = .0274).

Among the 55 patients (39.9%) who had dose escalation, the median time to progression on HDK was 1.7 months (range, 0.13–30.4⁺ months) (2.9 months in LDK responders and 1.3 months in LDK nonresponders) (Table 2). The addition of steroid replacement therapy was not associated with the response to HDK (P > .05).

Toxicity

Table 3 summarizes the toxicity noted with LDK and subsequent dose escalation to HDK. Of 138 evaluable patients, 133 patients were evaluable for toxicity data, whereas 5 patients did not have toxicity data. The most common side effect while receiving LDK was fatigue, which was observed in 18 patients (13.5%). Gastrointestinal symptoms included anorexia (6.0%), nausea (5.3%), and diarrhea (4.5%). Three patients (2.3%) had Grade 3 toxicities including diarrhea, fatigue, and skin rash covering the whole body. Four patients did not tolerate LDK treatment due to cumulative fatigue and/or mild nausea. Twenty-three patients (41.8%) in the dose-escalation group experienced either worsening of prior toxicities or development of new toxicities. Six patients (10.9%) who received subsequent dose escalation stopped treatment due to toxicity. All side effects appeared to be reversible. No Grade 4 toxicity was observed during ketoconazole therapy. No deaths were reported during therapy.

Table 3. Toxicities

	LDK (n = 133)		HDK (n = 55)
	Grade 3	Grade 1/2	Grade 3	Grade 3
LDK indicates low-dose ketoconazole; HDK, high-dose ketoconazole; LFTs, liver function tests.
Fatigue	17	1	8
Anorexia	8		2
Nausea	7		9
Diarrhea	5	1	3
Weight loss	4		2
Abdominal pain	4		2	1
Abnormal LFTs	2		0
Rash	2	1	2
Constipation	2		1
Night sweats	1		1
Peripheral edema	1		0
Hallucination	1		0
Peripheral neuropathy	1		0
Hot flashes	1		0
Dry mouth	0		1

Survival

There were 59 known deaths (42.7%) at the time of analysis. Twenty patients (14.5%) were alive at the time of last follow-up, and the survival status was unknown in 59 patients (42.7%). There were 12 patients with an unknown survival status in the responders group (30.8%) and 47 with unknown data in the nonresponders group (47.5%) (P = .09). The median overall survival time was 35.4 months. The median survival was 37.2 months in LDK responders and 35.4 months in LDK nonresponders. Furthermore, there was no statistically significant difference in the baseline characteristics including age, PSA, Gleason scores, and metastatic diseases status between patients with known and unknown survival status (P > .05).

DISCUSSION

Although HDK in combination with hydrocortisone has been shown to have a moderate activity as a secondary hormonal therapy in patients with AIPC, patients are more likely to suffer significant toxicity from treatment. A dose escalation regimen of ketoconazole has traditionally been used in our clinic because most patients tolerate LDK better than HDK. Our study demonstrated that 28.3% of patients had a PSA decline ≥50% with LDK. Interestingly, approximately 30% of patients had an overall PSA decline of >50% regardless of the dose. Although the PSA response rate observed is lower than some previously reported,12, 13, 15–17 it is comparable to reports from CALGB 9583. Also, unlike in a clinical trial, there was no mandate given that ketoconazole be taken on an empty stomach, and no restriction on the use of antacids, anticholinergics, or H2-blockers, although gastric acidity is known to improve absorption.18 In addition, steroids have been shown to induce PSA responses alone in up to 20% of patients with AIPC, and less than half of the patients in this cohort received steroids.19

Of note, Grade 3 toxicities were observed in only 2.3% of patients who received LDK. Considering that 21% of patients had Grade 3 or 4 toxicities in the CALGB 9583 study, it appears that toxicity is milder in patients treated with LDK compared with HDK. Interestingly, although speculative, a higher PSA level at the time of the initiation of ketoconazole was found to be a predictor of response to LDK, suggesting that responders had more differentiated cancer cells than nonresponders. In addition, the duration of primary ADT and total duration of all previous hormonal therapies significantly predicted a longer time to PSA progression on LDK, suggesting that patients with hormonally responsive cancers were more likely to have a delay in PSA progression with ketoconazole as well. In this study, no statistically significant association was noted between an AAWD response and response to ketoconazole, which is not consistent with other studies. However, approximately 28% of patients did not attempt AAWD.

Because only 39.9% of patients received dose escalation after LDK, conclusions are difficult to make. However, a response to HDK was noted in 12.7% of patients, distributed equally among those who did or did not respond to LDK. That said, PSA declines ≥25% were seen more commonly with HDK in LDK responders compared with LDK nonresponders. Also, the median time to progression on HDK was longer in LDK responders than LDK nonresponders (2.9 months vs. 1.3 months). Although speculative, this suggests that patients responsive to LDK who subsequently progress are more likely to have a period of disease stabilization to a minor response to dose escalation and should be considered for HDK.

The mechanism of action of ketoconazole in AIPC is not fully understood. It inhibits the synthesis of cholesterol in mammalian cells,20 and it also interferes with cytochrome P-450 in several organs, including the testis,7 adrenal gland,6 ovary,21 kidney,22 and liver.23, 24 Two hours after a single dose of 200 mg of ketoconazole, total and free plasma testosterone and androstenedione levels substantially decrease in normal men.7, 25 However, changes in plasma testosterone levels induced by ketoconazole vary among individuals.26 This is most likely due to the variable absorption of ketoconazole among individuals. Response to ketoconazole could be affected by this variability; however, a lack of standardization of measuring serum ketoconazole levels makes it difficult to assess this effect on response. In CALGB 9583, a rise in dehydroepiandrosterone (DHEA), DHEA sulfate (DHEAS), and androstenedione levels was noted at the time of disease progression, suggesting that the negative feedback mechanism induced by ketoconazole was partially reversed at the time of disease progression.12

In this retrospective study, we were unable to identify why 83 patients did not receive dose escalation in this study. Some patients likely did not agree to dose escalation when offered, although physicians may also have believed that their patients were not likely to benefit. Also, toxicity data may be incomplete because these were not collected prospectively. Although our clinical experience has been that LDK is better tolerated than HDK, only a randomized trial would answer this question adequately. Finally, the issue of concomitant steroids complicates matters, because fewer than half of the patients in this study received steroids. In fact, steroids can be a complicating factor and, in many patients receiving long-term ketoconazole therapy, steroid treatment is often the most bothersome with regard to side effects.

Conclusions

This is a largest retrospective study of which we are aware evaluating the efficacy of LDK and dose escalation in AIPC. LDK is a well-tolerated secondary hormonal treatment and represents an important management option for AIPC. A longer duration of primary ADT and total duration of all previous hormonal therapies were significantly associated with a longer time to progression with LDK. A prospective study is needed to compare the efficacy and tolerability of LDK followed by dose escalation to standard HDK with corticosteroids in patients with AIPC.

Acknowledgements

We thank Carolyn P. Evan for excellent administrative support and Mona Ghazi-Moghadam for excellent support in data collection.