Phase 1/2 study of preoperative docetaxel and mitoxantrone for high-risk prostate cancer

Authors


Abstract

BACKGROUND:

A study was conducted to determine the 5-year recurrence-free survival in patients with high-risk prostate cancer after neoadjuvant combination chemotherapy followed by surgery. Secondary endpoints included safety, pathologic effects of chemotherapy, and predictors of disease recurrence.

METHODS:

Fifty-seven patients were enrolled in a phase 1/2 study of weekly docetaxel 35 mg/m2 and escalating mitoxantrone to 4 mg/m2 before prostatectomy. Patients were treated with 16 weeks of chemotherapy administered weekly on a 3 of every 4 week schedule. A tissue microarray, constructed from the prostatectomy specimens, served to facilitate the exploratory evaluation of biomarkers. The primary endpoint was recurrence-free survival. Disease recurrence was defined as a confirmed serum prostate-specific antigen (PSA) >0.4 ng/mL.

RESULTS:

Of the 57 patients, 54 received 4 cycles of docetaxel and mitoxantrone before radical prostatectomy. Grade 4 toxicities were limited to leukopenia, neutropenia, and hyperglycemia. Serum testosterone levels remained stable after chemotherapy. Negative surgical margins were attained in 67% of cases. Lymph node involvement was detected in 18.5% of cases. With a median follow-up of 63 months, 27 of 57 (47.4%) patients recurred. The Kaplan-Meier recurrence-free survival at 2 years was 65.5% (95% confidence interval [CI], 53.0%-78.0%) and was 49.8% at 5 years (95% CI, 35.5%-64.1%). Pretreatment serum PSA, lymph node involvement, and postchemotherapy tissue vascular endothelial growth factor expression were independent predictors of early recurrence.

CONCLUSIONS:

Preoperative chemotherapy with docetaxel and mitoxantrone is feasible. Approximately half of the high-risk patients remain free of disease recurrence at 5 years, and clinical and molecular predictors of early recurrence were identified. Cancer 2010. © 2010 American Cancer Society.

Recent advances have lead to improvements in diagnostic and therapeutic strategies for prostate cancer; however, treatment failure after primary therapy occurs in 35% to 40% of cases.1, 2 The implementation of risk stratification models has allowed for the identification of prostate cancer patients who are at greatest risk for recurrence after primary therapy.3, 4 After radical prostatectomy, disease recurrence may occur locally, at distant sites, or both.5, 6 Effective systemic therapy regimens capable of eradicating microscopic metastases and improving local control are needed to improve the outlook for patients with high-risk, localized prostate cancer.

Randomized controlled trials of neoadjuvant androgen-suppressive therapy have shown that 3 months of preoperative androgen suppression significantly alters tumor phenotype and surgical margin status, but does not reduce the risk of prostate cancer recurrence.7-9 Complete pathologic responses, a validated marker in other tumor types, have only rarely been observed with androgen suppression alone. Longer androgen suppression before prostatectomy has not shown a clinical benefit to date.10 These results strongly suggest that castration-resistant prostate cancer clones are present at the time of initial diagnosis. Therapeutic agents that are active against disseminated prostate cancer, and in particular, castration-resistant prostate cancer, are likely to be necessary to improve outcomes in high-risk localized prostate cancer.

The use of systemic chemotherapeutic agents has only recently become standard treatment for patients with metastatic castration-resistant prostate cancer. Mitoxantrone and docetaxel are both active in advanced prostate cancer.11, 12 The availability of active cytotoxic agents stimulated their experimental application in combination with surgery in high-risk prostate cancer patients.13 Preoperative treatment also has the unique advantage of allowing assessment of tumor response and collection of pre- and post-treatment tumor tissue for molecular interrogation.

To date, several groups have tested the safety and preliminary efficacy of preoperative taxane-based chemotherapy in patients with high-risk disease.14-18 These studies have consistently shown that taxanes can be safely administered in the preoperative setting. In studies of single-agent docetaxel, prostate-specific antigen (PSA) reductions were observed in the majority of patients, although pathologic complete responses were not observed. The addition of androgen suppression therapy to systemic taxane treatment has also been evaluated.16, 18 In these studies, both PSA and pathologic changes have been typical of the expected responses with hormonal therapies. It is somewhat encouraging that, in a recent study by Chi et al, complete pathologic responses were observed in 3% of patients treated with a combination of androgen suppression and docetaxel.19 The contribution of chemotherapy to this outcome is unclear, because it was studied in combination with hormonal therapy and not in isolation.

As docetaxel and mitoxantrone exert antitumor effects through distinct cellular mechanisms, this combination has potential for synergistic activity against prostate cancer. On the basis of the experience with similar agents in breast cancer, incomplete cross-resistance between these 2 agents would be expected. Indeed, evidence suggests that docetaxel is active in patients previously exposed to mitoxantrone, and that mitoxantrone retains some activity in patients whose cancer has progressed after docetaxel.20 On the basis of these lines of evidence, we developed a regimen that combines both of these agents. Previously we reported the results of mitoxantrone dose escalation to 4 mg/m2 with a fixed dose of docetaxel (35 mg/m2) in our initial phase 1 cohort. Here we report the pathologic, biomarker, and clinical outcomes for this multimodality approach to the treatment of high-risk prostate cancer.

MATERIALS AND METHODS

Patients

Eligibility criteria for this study included histologically confirmed adenocarcinoma of the prostate, prostatectomy planned as primary therapy, 10-year minimum life expectancy, and at least 1 of the following high-risk features: clinical stage T2c or surgically resectable T3a, or serum PSA ≥15 ng/mL, or Gleason grade ≥4 + 3 (ie, 4 + 3, 4 + 4, or any 5 elements). Additional requirements included a negative radionuclide bone scan and a pelvic computerized tomographic scan to rule out pelvic lymph node involvement (computed tomography required only in patients with a PSA ≥40 ng/mL), Eastern Cooperative Oncology Group (ECOG) performance status ≤1, and left ventricular ejection fraction ≥50% by multigated acquisition technetium scan. Exclusion criteria were any prior therapy for prostate cancer, any significant active medical illness, a second malignancy other than nonmelanoma skin cancer within 5 years, grade ≥2 peripheral neuropathy, hypersensitivity to drugs formulated with polysorbate-80, significant contraindications to corticosteroids, white blood cell count <3000/mm3, neutrophil count <1500/mm3, platelet count <100,000/mm3, conjugated bilirubin >upper limit of normal (ULN), alkaline phosphatase >4.0 × ULN, and alanine transaminase (ALT) >2.0 × ULN or ALT >1.5 × ULN concomitant with alkaline phosphatase >2.5 × ULN.

Written informed consent was obtained from all patients. The protocol was approved by the institutional review boards of all participating institutions.

Treatment

Docetaxel was administered at a fixed dose of 35 mg/m2, whereas mitoxantrone was dose escalated in the first 10 patients, who received mitoxantrone at doses of 2 to 5 mg/m2,14 and was given at a dose of 4 mg/m2 in all remaining patients. Patients received 4 28-day cycles of chemotherapy with docetaxel and mitoxantrone administered as 3 weekly doses followed by a 1-week break. Dexamethasone 4 mg was given orally 12 hours and 1 hour before and 12 hours after treatment. In patients who weighed >130% of their ideal body weight (IBW, defined as 50 kg + 2.3 kg/inch for each inch of height over 5 feet),21 body surface was calculated based on adjusted weight, defined as IBW + 0.5(actual weight − IBW).

Dose Modifications

Hematologic events

Scheduled doses with both chemotherapeutic agents were withheld for platelet count <75,000/mm3 or neutrophil count <1000/mm3 until count levels recovered to above these parameters. The mitoxantrone dose was reduced by 25% if recovery took >1 week, if counts were below these parameters for 2 consecutive doses, or for any grade 4 hematologic toxicity. The dose of docetaxel was reduced by 25% if a patient again met these criteria despite the mitoxantrone dose reduction. Patients who had a platelet count <75,000/mm3, a neutrophil count <1000/mm3 that persisted >2 weeks, or had any grade 4 hematologic toxicity despite the 25% dose reduction of both agents received no additional chemotherapy. Hematologic growth factors were allowed at the treating physician's discretion; however, growth factor support could not serve as a replacement for the preplanned dose reductions.

Nonhematologic events

The dose of docetaxel and mitoxantrone was to be reduced by 25% for aspartate transferase of 1.6 to 5 × ULN. Therapy was withheld for bilirubin > ULN, alkaline phosphatase >5 × ULN, or ALT >5 × ULN. Therapy was resumed at a 25% dose reduction in patients whose liver function tests recovered within 3 weeks. Mitoxantrone therapy was to be discontinued for any clinical evidence of cardiotoxicity that was associated with a >10% decline in ejection fraction. For all other persistent and clinically significant grade ≥3 nonhematologic toxicities, therapy was to be withheld until resolution of toxicity and treatment resumed with a 25% reduction in the dose of both agents.

Patient Monitoring

At Week 1 of each 4-week cycle, a chemistry panel, PSA, and complete blood count with automated differential were collected. A complete blood count was also obtained at week 2 and 3, during each drug treatment visit. Serum testosterone was measured before chemotherapy and after 4 cycles of treatment. A physical examination and assessment of toxicity was performed every 4 weeks. Perioperative surgical morbidities, operative time, blood loss, and pathologic findings were recorded in all patients. After completion of chemotherapy and prostatectomy, patients are to be observed with serum PSA tests every 3 months and regular clinic visits for 5 years or until evidence of recurrence.

Translational Studies

Pre- and postchemotherapy plasma vascular endothelial growth factor (VEGF) were measured. Tumor expression of VEGF, nuclear Ki67, nuclear and cytoplasmic p16, and CD10 were characterized in a tissue microarray constructed from formalin-fixed prostatectomy specimens. Detailed methods for these analyses are provided below.

Methods

Patient samples

Before the initiation of chemotherapy, each patient underwent a transrectal ultrasound-guided prostate biopsy (10 cores) for translational studies. Fresh biopsy samples were embedded in Tissue-Tek optimal cutting temperature (OCT) compound (Sakura, Torrence, Calif) and immediately snap frozen in liquid nitrogen, then transferred to a −80°C freezer. Plasma samples were obtained at enrollment and before each cycle of chemotherapy and stored at −80°C. At prostatectomy, surgical specimens were immediately placed in a sealed specimen bag and placed on ice. Unfixed portions of both cancer and noncancer tissue from the prostatectomy specimen were identified by microscopic examination of frozen sections and rapidly frozen after embedding in OCT.

Tissue microarray

To facilitate the analysis of biomarker studies, a tissue microarray was constructed from each prostatectomy specimen. Tissue cores were removed from the paraffin-embedded blocks and placed in a master paraffin block using a precision tissue arrayer (MTA-1, Beecher Instruments, Sun Prairie, Wis). Three cores each of prostate cancer tissue, normal prostate, and where applicable, lymph nodes with metastatic cancer were placed in the master paraffin block. To determine the appropriate location to sample for these 3 tissue types, H & E slides of each donor block were examined microscopically. Dispersed among the study cores were control tissues from nonstudy patients (ie, liver, prostate, lymph node, salivary gland, kidney, and testis tissue), and untreated prostate cancer cell lines (DU-145, PC-3, LNCaP). After sample placement, the master block was heated to 37°C for 30 minutes and then cut into 5-μm sections.

Immunohistochemistry

Slides were incubated overnight at 4°C in a humid chamber with primary antibodies diluted in 0.3% bovine serum albumin in tris-buffered saline (TBS) with 0.1% sodium azide (NaN3). Primary antibodies included anti-Ki-67 (Ventana Medical Systems, Tucson, Ariz), p16 (Ventana Medical Systems), CD10 (Ventana Medical Systems), and VEGF (1:200, Zymed, San Francisco, Calif). Endogenous peroxidase was quenched by immersing the tissue microarray slides in 3% H2O2 in 80% methanol with 0.1% NaN3 for 15 minutes. The slides were then washed in 0.05% Tween in TBS (2 × 5 minutes), rinsed in TBS alone, and incubated with avidin-biotin complex (Vector Laboratories, Burlingame, Calif) for 30 minutes at room temperature. Next, the slides were washed in 0.05% Tween in TBS (2 × 5 minutes), in TBS alone, and then incubated in 0.1% 3,3′-diaminobenzidine solution (DAKO, Carpentaria, Calif). The slides were counterstained with Gill hematoxylin for 1 minute, then rinsed in tap water until clear. The slides were then dehydrated in graded alcohols and xylene, and coverslipped. Positive and negative control cores were examined to ensure the adequacy of staining. For Ki-67 and p16 assessment, the percentage of positively staining nuclei was reported regardless of staining intensity. CD10 expression was scored using a semiquantitative intensity scale. The 4-point scale (0-3) represented an overall level of staining characterized as none, weak, moderate, or strong by a pathologist; this was assessed separately for benign and neoplastic tissues. For tissue VEGF staining, cancer and normal epithelial cells separately were each assigned a score based on the number of cells stained according to the following definition: 0 = <5%, 1+ = 6% to 25%, 2+ = 26% to 75%, or 3+ = 76% to 100%. As each patient was represented by 3 core samples on the tissue microarray, an average score was then calculated.

Plasma VEGF assays

Pre- and postchemotherapy plasma VEGF levels were quantified using a human VEGF Quantikine Immunoassay (R&D Systems, Minneapolis, Minn) according to the manufacturer's recommendations. Briefly, samples were added to a microplate coated with a mouse monoclonal anti-VEGF antibody. After an incubation period, the samples were aspirated, the microwells were washed, and a polyclonal conjugate against VEGF was applied. Samples were again aspirated, and microwells were washed. The optical density of each well was determined at 450 nm after addition of a substrate.

Findings

Tissue Ki-67 expression

The Ki-67 protein is a well-known marker of proliferation in cancer cells that has been shown to correlate with the biological behavior of clinical prostate cancer.22, 23 The mean number of positive cancer cells was 5.6% (standard deviation [SD], 6.7%; range, 0%-44.4%).

CD10 tissue expression

CD10 is a transmembrane zinc metallopeptidase that mediates signal transduction and local concentrations of peptide factors.24 The mean CD10 intensity score for all patients was 1.32 (SD, 1.02; range, 0-3) in neoplastic tissues and 2.70 (SD, 0.50; range, 1-3) in benign epithelium.

p16 tissue expression

p16 is a member of the INK4a cell cycle inhibitor protein family, which has significant prognostic capability in several tumor types. Expression of nuclear p16 was seen in 47.7% of cells in cancer tissue (SD, 33.2%; range, 0%-100%) and 17.4% of cells in normal tissue (SD, 20.8%; range, 0%-80%).

Tissue VEGF expression

VEGF, a promoter of angiogenesis, is frequently overexpressed in prostate cancer, and its expression correlates with more advanced stage as well as higher Gleason scores. The VEGF score averaged 2.46 in cancer tissue (SD, 0.61%; range, 0.5-3) and 2.17 in normal tissue (SD, 0.74; range, 0-3).

Plasma VEGF levels

Considerable heterogeneity in plasma VEGF was observed. Mean pretreatment VEGF was 78.6 pg/mL (SD, 80.7; range, 10.6-314.7). Post-treatment VEGF was 97.6 pg/mL (SD, 110.1; range, 10.6-687.6). No statistically significant change in plasma VEGF with chemotherapy was seen.

Statistical Analysis

The primary endpoint for this study was to determine 5-year recurrence-free survival after combination chemotherapy and prostatectomy. Disease recurrence was defined as the detection and confirmation of a serum PSA >0.4 ng/mL or any other clinical evidence of disease or initiation of any prostate cancer-directed therapy. Sample size and power calculations were previously described.14

For exploratory outcomes analyses, Gleason scores were categorized into either 3 categories (6 vs 7 vs 8 and higher) or 2 categories (6 and 7 vs 8 and higher). Pathologic and clinical T classifications were each split into T1-2 versus T3-4. Continuous variables were dichotomized into high and low groups based on median values. Nuclear and cytoplasmic p16 were split into groups with greater or less than 20% staining. In addition, each of these variables was analyzed after stratification for surgical lymph node status. Pre- and postchemotherapy testosterone levels and serum VEGF levels were compared using a t test for paired data. PSA fluctuations during chemotherapy were calculated by determining the percentage change from baseline. Values that remained between +20% and −20% from baseline were considered stable.

Time-dependent analyses were carried out using the Kaplan-Meier method. The effect of covariates on time-dependent outcome variables was examined using the log-rank test in univariate analyses and the Cox proportional hazards ratio in multivariate analyses. Predictor variables tested included patient age, lymph node status, surgical margin status, biopsy Gleason score, pathological Gleason score, clinical and pathologic T stage, cancer tissue VEGF percent intensity, prechemotherapy plasma VEGF, baseline PSA, PSA density (serum PSA/prostate volume), PSA pattern of change on therapy, prostate volume, nuclear Ki67, nuclear and cytoplasmic p16, and CD10 staining.

RESULTS

Patients

Between January 2001 and November 2004, 57 patients with high-risk localized prostate cancer were recruited from the clinics at all study sites. All screened patients were eligible. The median age was 63 years (range, 49-74 years). Median ECOG performance status was 0. The median serum PSA was 12.2 ng/mL (range, 1.4-58.6 ng/mL), and the median biopsy Gleason score was 8 (range, 6-10). The pretreatment characteristics are summarized in Table 1. The median follow-up for recurrence-free patients was 63 months (range, 7-88 months). The median follow-up for all patients (until last visit or until recurrence) was 55 months (range, 7-88 months).

Table 1. Patient Characteristics Before Therapy
  1. ECOG indicates Eastern Cooperative Oncology Group; PS, performance score; PSA, prostate-specific antigen; AJCC, American Joint Committee on Cancer.

No.57
Age, y 
 Median63
 Range49-74
ECOG PS 
 054
 13
PSA, ng/mL 
 Median12.2
 Range1.4-58.6
Clinical stage (AJCC 2002 criteria) 
 T1c7
 T2a7
 T2b10
 T2c19
 T3a11
 T3b2
 T41
Biopsy Gleason score 
 65
 722
 817
 912
 101

Treatment

Of the 57 patients, 54 completed 4 cycles of therapy and had surgery. Two patients completed <4 cycles of therapy before going on to surgery because of the following factors: 1 patient had persistent grade 2 nausea after 8 of 12 planned doses, and 1 patient developed grade 2 neuropathy after 9 of 12 planned doses of treatment. One patient was incarcerated out of the reach of our healthcare system and was lost to follow-up.

Among the 54 patients who received 4 cycles of therapy and had surgery, 218 of 228 (96%) doses of chemotherapy were delivered as scheduled. All 10 of the missed doses were because of neutropenia. Four patients received erythropoietin (40,000 U weekly) for anemia, and no patient received granulocyte colony-stimulating factors.

Toxicity

Treatment-related toxicity for the entire 4-cycle course of therapy is detailed in Table 2. Neutropenia was the most common treatment-related hematologic toxicity. Hyperglycemia, expected with dexamethasone premedications, and onycholysis, commonly reported with weekly docetaxel, were the most common nonhematologic toxicities. One patient suffered a calf venous thrombosis below the popliteal vein while receiving chemotherapy.

Table 2. Treatment-Related Toxicities Grade >1 (n=57)
ToxicityGrade 2Grade 3Grade 4% Occurrence
  1. URI indicates upper respiratory tract infection.

Hematological    
 Neutropenia911750%
 Leukopenia1516567%
 Febrile neutropenia0204%
 Thrombocytopenia1002%
 Anemia3006%
Nonhematological    
 Hyperglycemia115233%
 Onycholysis170031%
 Alopecia100019%
 Fatigue80015%
 Hypophosphatemia52013%
 Diarrhea33011%
 Hyperlacrimation4007%
 Taste disturbance4007%
 Vision disturbance3006%
 Dermatology-skin toxicity3006%
 Neuropathy2004%
 Hyperphosphatemia2004%
 Abdominal pain1104%
 Anorexia2004%
 Diaphoresis1002%
 Epistaxis1002%
 Extrapyramidal reaction1002%
 Fingertip pain1002%
 Gastric reflux1002%
 Hearing loss1002%
 Indigestion1002%
 Nail pain1002%
 Night sweats1002%
 Pain due to biopsy1002%
 Paronychia1002%
 Pneumonia1002%
 Stomatitis1002%
 URI1002%
 Vomiting1002%
 Weight loss1002%
 Cellulitis0102%
 Calf thrombosis0102%
 Injection site reaction0102%
 Nausea0102%

Surgery After Chemotherapy

Acknowledging the limitations of sample size as well as variations in individual pelvic anatomy and in operative technique among different surgeons, observed operative morbidity was comparable to most published data for prostatectomy without prior chemotherapy.25-27 In the absence of direct measures of technical difficulty, we evaluated commonly used surrogates. The median operative time was 4.3 hours (mean, 4.4 hours; range, 2.4-6.5 hours). The median estimated blood loss was 900 mL (mean, 1023 mL; range, 80-3100). There was 1 intraoperative complication, a rectal injury that was thought to be possibly related to chemotherapy. Postoperative complications were limited to a wound infection in 1 patient. There were no postoperative deep venous thromboses.

Effect of Chemotherapy on Serum Testosterone and PSA Values

To ensure that any changes in PSA or tumor characteristics were not because of an effect on androgenic steroid production, pre- and postchemotherapy testosterone measurements were compared. There was no change in testosterone with therapy; the mean serum testosterone concentration was 343 ng/dL before therapy (95% confidence interval [CI], 303.7-382.1 ng/dL) and 352 ng/dL after therapy (95% CI, 304.7-399.3 ng/dL) (n = 53, P = .71).

The use of PSA as an indirect measure of antitumor activity was facilitated by the finding that therapy did not result in reduced testosterone levels. Considerable fluctuation in individual PSA measurements were seen, and conventional measures developed in metastatic hormone-refractory prostate cancer are not readily applicable in this setting. Hence, we examined the overall effect on PSA in each patient. Three distinct patterns emerged. The first group (n = 28, 52%) had a decreasing PSA pattern defined as sustained decline >20%. The second group (n = 12, 22%) had PSA levels that neither increased nor decreased by >20% during chemotherapy. The third group (n = 14, 26%) had both increases and decreases of >20% from 1 time point to the next. No patient developed sustained PSA progression defined as a confirmed increase of >20%.

Initial Surgical Outcome

Pathologic outcome at surgery is summarized in Table 3. Negative margins were attained in 36 of 54 patients (67%; 95% CI, 53%-79%). There were no pathologic complete responses. Postoperative PSA was <0.2 ng/mL in 43 of 54 patients (80%; 95% CI, 66%-89%).

Table 3. Pathologic Outcomes After Neoadjuvant Chemotherapy (n=54)
  • PSA indicates prostate-specific antigen; AJCC, American Joint Committee on Cancer.

  • a

    Surgical Gleason score was not assigned in 1 patient.

Negative surgical margin36 (67%)
Postoperative serum PSA <0.2 ng/mL43 (80%)
Lymph node metastases present10 (18.5%)
Pathologic stage (AJCC 2002) 
 T2a4
 T2b5
 T2c16
 T3a10
 T3b9
 T3c7
 T43
Surgical Gleason scorea 
 68
 725
 88
 911
 101

Biomarker Studies

Biomarker results are presented in supplementary data. Briefly, the expression of Ki-67, CD10, p16, and plasma VEGF concentrations were not predictive of clinical outcomes, whereas postchemotherapy tissue VEGF expression was an independent predictor of early recurrence.

Cancer Recurrence

Twenty-seven of 57 (47.4%) patients have experienced disease recurrence. The Kaplan-Meier progression-free survival for the entire study group was 65.5% (95% CI, 53.0%-78.0%) at 2 years and 49.8% (95% CI, 35.5%-64.1%) at 5 years. For lymph node-negative patients, 2- and 5-year recurrence-free survival probabilities were 77.9% and 58.7%, respectively.

Independent predictors of disease recurrence are shown in Table 4. Biopsy Gleason score, PSA density, increased VEGF expression in surgical specimens, and lymph node involvement with tumor were independently associated with risk of recurrence. The profound effect of persistent disease in lymph nodes is illustrated in Figure 1. The effect of tissue VEGF expression is shown in Figure 2. Because the presence of disease in lymph nodes after induction chemotherapy was a dominant predictor of outcome, we also examined predictors of recurrence in the subset of patients who did not have lymph node involvement identified at surgery. Only tumor VEGF expression after chemotherapy was associated with early recurrence in this subset. PSA decline during chemotherapy was not predictive of VEGF expression or time to recurrence.

Figure 1.

Progression-free survival in patients with pathologically involved versus those with uninvolved lymph nodes is shown (log-rank test, P < .001).

Figure 2.

Progression-free survival in patients with tumor vascular endothelial growth factor (VEGF) expression above and below the median is shown (log-rank test, P = .02).

Table 4. Predictors of Disease Recurrence on Multivariate Analyses
VariableHazard Ratio (95% CI)P
  1. CI indicates confidence interval; PSA, prostate-specific antigen; VEGF, vascular endothelial growth factor.

All patients  
 Lymph node metastases9.7 (2.8-33.4)<.001
 Biopsy Gleason score1.8 (1.1-3.1).03
 PSA density6.4 (1.8-22.7).004
 Increased cancer VEGF expression3.5 (1.2-10.4).023
Lymph node-negative patients  
 Increased cancer VEGF expression3.8 (1.2-12.0).023

DISCUSSION

Studies of androgen suppression therapy before surgery have shown a near uniform resistance to androgen suppression illustrated by the presence of residual tumor in pathologic specimens.7-9 Several groups have therefore explored the use of chemotherapy before or after surgical removal of the prostate.13, 22

The regimen developed in this study combines the 2 most active chemotherapy drugs in prostate cancer and was well tolerated and feasible in a multicenter study of neoadjuvant chemotherapy for high-risk prostate cancer.

Patients who participated in our study were similar to those who took part in other contemporary taxane-based treatment trials developed for high-risk, localized prostate cancer.16, 19 Negative surgical margins, a favorable pathologic finding in men undergoing radical prostatectomy,23 were attained in 67% of our cases. These results compared favorably to high-risk patients undergoing surgery alone, where the rates of negative margins range only from 35% to 54%.7-9 Our negative margin rates in patients treated with chemotherapy alone were comparable to the results obtained with chemohormonal treatment, where negative margin rates were 73% to 78%.16, 19 Further studies are needed to determine whether preoperative cytotoxic chemotherapy has a direct role in improving surgical margin status. Positive lymph nodes were detected in 18.5% of cases in our trial, compared with 6% in 2 other chemohormonal neoadjuvant trials.16, 19 There were no pathologic complete responses in the current study, a finding that has been observed in the majority of other neoadjuvant chemotherapy trials.15-18

With a median follow-up of 63 months for nonrelapsing patients, the Kaplan-Meier progression-free survival rates at 2 years and 5 years for all patients were 65.5% and 49.8%, respectively, comparable to those reported from studies of neoadjuvant chemohormonal therapy. Chi et al19 reported 70% recurrence-free survival with a follow-up of 43 months, whereas Konety et al16 saw a 55% recurrence-free survival with a median follow-up of 29 months. As was the case with neoadjuvant hormonal therapy, larger studies and long-term follow-up would be required to determine whether preoperative chemotherapy is beneficial and whether the addition of hormonal therapy to chemotherapy in this setting adds any benefit.

We studied predictors of recurrence because early recurrences after surgery are associated with increased mortality.24 In this study, the strongest predictor of treatment failure was the presence of lymphatic disease. These findings support the pursuit of improved imaging methods with the capacity to identify these high-risk patients so alternate strategies can be investigated. Not unexpectedly, pretreatment PSA was also predictive of disease recurrence.

Prostate cancer biomarkers are needed to improve the prediction of disease recurrence for high-risk patients. We identified increased tissue VEGF expression as a new independent predictor of recurrence in patients undergoing neoadjuvant chemotherapy followed by prostatectomy. This novel finding merits confirmation. This finding supports the hypothesis that VEGF signaling may be important in prostate cancer treatment resistance. Although identified in a different setting and not yet confirmed, the finding is encouraging for the ongoing Cancer and Leukemia Group B trial of docetaxel with or without the VEGF antibody, bevacizumab.

An important limitation of this study is our inability to isolate the effects of tumor biology, surgical skill, and chemotherapy. Only a randomized clinical trial where the systemic therapy is the sole variable can measure its impact on recurrence-free survival. The clinical and biomarker risk factor analyses should be interpreted as exploratory.

In conclusion, we demonstrated that the 2 cytotoxic drugs that are most commonly used in the treatment of advanced prostate cancer can be safely combined. The novel regimen developed here may warrant evaluation in patients with advanced metastatic disease. We also demonstrated the feasibility of delivering multiagent chemotherapy before prostatectomy. The 5-year recurrence-free survival results are encouraging despite the absence of complete pathologic responses. Consideration should be given to the evaluation of this multimodality strategy in a randomized controlled trial.

Acknowledgements

We thank Drs. Bruce Lowe, Raul Parra, Jeffrey Johnson, Douglas Ackerman, William J. Ellis, Thomas Takayama, Roger Wicklund, and Robert Skinner who, in addition to Drs. Garzotto, Lange, and Lieberman, performed surgery on patients enrolled in the study; Vasko Kaimaktchiev, and Nina Katovic for technical assistance; Dr. Larry True for support in constructing the tissue microarray; and Susanne McGlothlin for editorial assistance.

CONFLICT OF INTEREST DISCLOSURES

This study was supported in part by grant GIA US 16,080 from Aventis Pharmaceuticals, grant 031.G0008 from Serono, Inc, and grants 3M01RR00334-33S2 and 1 R01 CA119125-01 from the National Institutes of Health. Dr. Beer has received speaker's honoraria from Aventis, a company that may have a commercial interest in the results of this research. This potential conflict of interest has been reviewed and managed by the Veterans Administration Conflict of Interest in Research Committee.

Ancillary