Fax: (713) 794-4385
Phase 2 study of gemcitabine and irinotecan in metastatic breast cancer with correlatives to determine topoisomerase I localization as a predictor of response
Version of Record online: 29 SEP 2008
Copyright © 2008 American Cancer Society
Volume 113, Issue 10, pages 2646–2654, 15 November 2008
How to Cite
Moulder, S., Valkov, N., Neuger, A., Choi, J., Lee, J. H., Minton, S., Munster, P., Gump, J., Lacevic, M., Lush, R. and Sullivan, D. (2008), Phase 2 study of gemcitabine and irinotecan in metastatic breast cancer with correlatives to determine topoisomerase I localization as a predictor of response. Cancer, 113: 2646–2654. doi: 10.1002/cncr.23916
- Issue online: 3 NOV 2008
- Version of Record online: 29 SEP 2008
- Manuscript Accepted: 11 JUL 2008
- Manuscript Revised: 8 JUL 2008
- Manuscript Received: 14 APR 2008
- National Comprehensive Cancer Network (NCCN)
- Pfizer, Inc.
- breast cancer;
Gemcitabine incorporation into DNA enhances cleavage complexes in vitro when combined with topoisomerase I inhibitors and demonstrates synergy in cancer cells when given with irinotecan. Topoisomerase I inhibitors require that topoisomerase I interacts with DNA to exert activity.
Patients who had received previous anthracycline therapy or were not candidates for anthracycline therapy received gemcitabine at a dose of 1000 mg/m2 intravenously over 30 minutes followed by irinotecan at a dose of 100 mg/m2 over 90 minutes on Days 1 and 8 of a 21-day cycle. The primary endpoint was improvement in response from that historically observed with gemcitabine (from 25% to 45%) as measured by Response Evaluation Criteria in Solid Tumors. Correlative studies included characterization of cellular levels and nuclear distribution of topoisomerase I and pharmacokinetic analysis of gemcitabine and irinotecan.
Forty-nine patients were assessed for response. The response rate was approximately 25% (all partial responses [PRs], 12 patients; 95% confidence interval [95% CI], 13-39). Six patients had stable disease (SD) for ≥6 months for a clinical benefit rate (PR + SD) of 39%. The median time to disease progression was 3.7 months (95% CI, 2.5 months-4.6 months), and median survival was 11.6 months (95% CI, 8.9 months-15 months). Toxicities included neutropenia, nausea, and vomiting. Seven of 9 tissue biopsies were assessable for topoisomerase I. Tumors with the 2 lowest nuclear to cytoplasmic ratios demonstrated no response to irinotecan.
Gemcitabine and irinotecan are active in metastatic breast cancer, but response did not meet predetermined response parameters, and the null hypothesis was accepted. Topoisomerase I localization can be measured in metastatic breast cancer. Further validation is needed to determine whether this assay can predict response. Cancer 2008. © 2008 American Cancer Society.
Gemcitabine, a nucleoside analogue, and irinotecan, a topoisomerase I inhibitor, have both demonstrated efficacy as single agents in patients with metastatic breast cancer. Both gemcitabine and irinotecan exert their cytotoxic effects by interacting with genomic DNA. Gemcitabine is a nucleoside analogue whose incorporation into DNA inhibits DNA polymerase during replication or repair. Irinotecan inhibits topoisomerase I (topo I), an essential enzyme for eukaryotic DNA replication.1–3 Stabilization of topo I-DNA cleavage complexes with topo I inhibitors, such as camptothecin or irinotecan, interferes with advancing replication or transcription forks, resulting in DNA-strand breaks and cell death.4 Preclinical data indicate that the incorporation of gemcitabine into DNA enhances camptothecin-induced topo I cleavage complexes in vitro.5 In addition, the incorporation of gemcitabine slowed the kinetics of reversal for camptothecin-induced cleavage complexes, which may increase the frequency of collisions with replication and transcription complexes, leading to increased strand breaks and improved cytotoxic activity. Synergistic or additive effects of combination therapy with gemcitabine and irinotecan have been demonstrated in human breast and lung cancer cell lines.6, 7
Inhibitors of topo I require enzyme interaction with DNA for drug activity. Human topo I contains an N-terminal domain that carries essential sequences for nuclear localization.8 Although this N-terminal domain does not contribute to the enzyme's catalytic activity in vitro, it is very important for in vivo activity, as the enzyme must enter the nucleus to exert its function.9–11 Data have indicated that mutations within the N-terminus of top1 result in loss of nuclear localization and subsequent resistance to topoisomerase I inhibitors.12 Localization of topo I can be determined using immunofluorescence microscopy in patients with hematologic and solid tumors.13, 14 This technique can also be used quantitatively to measure total cellular levels of topo I as well as levels within the nucleus and cytoplasm to determine a nuclear to cytoplasmic ratio, which may be predictive of response to inhibitors of topo I.
At least 3 phase 1 studies have evaluated combination therapy with gemcitabine and irinotecan in patients with solid tumors.15–17 Lima et al examined a dosing schedule of Days 1 and 8, every 3 weeks, in patients with solid tumors.16 Grade 4 diarrhea was the dose-limiting toxicity, and the recommended dosing for phase 2 testing was gemcitabine at a dose of 1000 mg/m2 and irinotecan at a dose of 100 mg/m2. In the current study, 3 patients achieved a partial response, 2 with pancreatic cancer and 1 with cancer of an unknown primary site.
On the basis of these data, we performed a single-arm, phase 2 trial administering gemcitabine at a dose of 1000 mg/m2 followed by irinotecan at a dose of 100 mg/m2 on Days 1 and 8 of a 21-day treatment cycle to determine the efficacy in patients with metastatic breast cancer who had disease progression or were not candidates for anthracycline-based chemotherapy. We hypothesized that the combination therapy would improve the response rate historically noted with single-agent gemcitabine in this patient population from 25% to 45%. We also measured topo I localization using immunofluorescence microscopy in patient samples obtained by fine-needle aspiration (FNA) before the initiation of therapy to determine whether this ratio could be used to predict response to therapy.
MATERIALS AND METHODS
Eligible male or female patients were age ≥18 years with metastatic cancer; histologically confirmed adenocarcinoma of the breast (stage IV) who were not candidates for or had documented disease progression during anthracycline-based chemotherapy (including the adjuvant setting); had an Eastern Cooperative Oncology Group performance status of ≤2; were free of prior malignancies for >5 years with the exception of curatively treated basal or squamous cell carcinomas of the skin or carcinoma in situ of the cervix; had adequate hepatic, renal, and hematologic function; and had measurable disease by Response Evaluation Criteria in Solid Tumors (RECIST) criteria. Patients were excluded if they had received >3 prior chemotherapy regimens for metastatic breast cancer, had untreated or uncontrolled brain metastasis, or had received prior gemcitabine or irinotecan.
Study Design and Treatment
This study was conducted in accordance with the US Food and Drug Administration, the institutional review board and clinical research committee, and the Declaration of Helsinki. All patients provided informed consent before study participation.
Patients received gemcitabine at a dose of 1000 mg/m2 in 250 mL of normal saline intravenously over 30 minutes followed by irinotecan at a dose of 100 mg/m2 in 500 mL dextrose 5% in water over 90 minutes on Days 1 and 8 of a 21-day treatment cycle. Dosing continued every 3 weeks with no maximum number of cycles. Patients received therapy unless there was evidence of progression of disease, unacceptable toxicity, or withdrawal of consent for further therapy. Dose reductions were allowed based upon treatment tolerability. No additional anticancer treatments, such as endocrine therapies or anti-HER-2/neu therapies, were administered while the patients were receiving protocol treatment.
All supportive measures consistent with optimal patient care were given throughout the study, including antiemetics, bisphosphonate therapy (for patients with bone metastasis), and growth factor support for prophylaxis after a documented episode of febrile neutropenia. The administration of atropine and loperamide was allowed for the treatment of irinotecan-induced diarrhea. Loperamide was supplied to patients without charge, and patients were instructed to begin therapy at the first sign of diarrhea.
The primary objective of this study was to assess the clinical activity of the combination therapy as measured by tumor response rate. Tumor response evaluation (by physical exam and/or imaging studies) was performed after every 2 cycles of therapy using RECIST criteria. Best overall response was evaluated by the treating physician and not subject to review by an independent review committee. Patients were considered eligible for response evaluation if they completed 1 cycle of therapy.
Toxicity was evaluated continuously. Hematologic and serum chemistry evaluations occurred before the administration of each dose of therapy and as clinically indicated. All patients who received at least 1 dose of therapy were evaluated for toxicity using the National Cancer Institute Common Toxicity Criteria (version 2.0).
This was a single-arm, phase 2 trial comparing combination therapy with the historical response rate of ∼25%18 noted with single-agent gemcitabine. The study involved a 2-stage design, with the initial stage accruing 26 response-assessable patients. If ≥6 patients met response criteria, accrual continued until a total of 49 response-assessable patients were enrolled. If >17 patients responded, the null hypothesis would be rejected, and the combination therapy would be worthy of further study in a randomized phase 3 trial. The probability of erroneously concluding that the treatment is active (P ≥ .45) when it is actually ineffective (P ≤ .25) is less than .04 (alpha = .04). The probability of erroneously concluding that the treatment is ineffective (P ≤ .25) when the treatment is actually effective (P ≥ .45) is less than .1 (power = 90%).
Baseline demographic and clinical characteristics were summarized using descriptive statistics. Response rate and its 95% exact confidence interval (CI) were estimated using the binomial distribution. Toxicity events were summarized by grades with frequencies and percentages.
Kaplan-Meier curves were used to estimate overall survival, time to tumor progression, and time to treatment failure with their 95% CIs. Survival time was defined as time between registration and death. Time to tumor progression was defined as the time between registration and disease progression. Time to treatment failure was defined as time between registration and discontinuation of therapy for any reason. Response duration was defined as the time from randomization to the time of progressive disease in patients who received at least a partial response to therapy.
In patients who gave consent for optional tumor biopsies, FNA was performed before initiation of protocol therapy. Tumor cells were suspended in 10 mL of RPMI-1640 and 10% fetal bovine serum. After low-speed centrifugation, the cells were resuspended in phosphate-buffered saline and counted. Cytospin slides were created using 2 × 105 cells/slide. Slides were examined by a collaborating pathologist to verify the presence of breast cancer cells, then topo I localization and cellular enzyme levels were determined using immunofluorescence as previously described.14 Quantification of topo I was performed on 50 randomly selected tumor cells/sample using the Smart Capture program with measurements confirmed by Adobe Photoshop 5.0 (Adobe Systems, San Jose, Calif). Each compartment was quantified separately in pixels, and nuclear/cytoplasmic ratios were calculated individually for each cell. These analyses were performed in a single laboratory by collaborating investigators blinded to patient outcome.
Serial blood samples were collected in heparinized tubes predose, at 15 minutes and 30 minutes (during infusion) and at postinfusion times of 45, 60, 90, and 120 minutes on Day 1 and at 24 hours on Day 2 of the first cycle only. Blood samples (10 mL) were drawn from a catheter contralateral to that used for drug infusion and placed in a heparinized tube containing 5 μM of the cytidine deaminase inhibitor, tetrahydrouridine. Samples were placed in an ice bath immediately after being obtained. Plasma was separated from all blood samples by centrifugation (400 g for 5 minutes at 4°C) and stored at −20°C until assayed using high-performance liquid chromatography (HPLC) as previously described.19 Gemcitabine along with its deaminated metabolite dFdU were determined by the assay. Calculation of the pharmacokinetic variables Cmax (highest drug concentration observed in plasma), t1/2 (terminal plasma half-life), clearance, and area under the concentration time curve (AUC) was performed by noncompartmental analysis with the pharmacokinetic software WinNonlin (version 1.5; Scientific Consulting Inc/Pharsight Corp, Mountain View, Calif).
Blood samples were collected at predose and then 3, 3.5, 11, and 11.5 hours postdose. Blood samples (10 mL) were drawn from a catheter contralateral to that used for drug infusion and also placed in heparinized tubes. Plasma was separated from whole blood using standard techniques. Irinotecan and its major metabolite SN38 were measured by HPLC using previously described methods.20, 21 A limited sampling strategy was applied for the prediction of the total AUC for both irinotecan and SN38.22
Fifty-three patients provided informed consent and were enrolled between September 1, 2003 and January 30, 2006 at the H. Lee Moffitt Cancer Center. One patient did not meet eligibility requirements and did not initiate therapy. Three patients only received Cycle 1 Day 1 doses of therapy and were not considered evaluable for response as stipulated by the protocol. All of these patients withdrew consent for further therapy after the first dose. As per protocol, these 3 patients were included in the toxicity but not the efficacy analysis.
The baseline demographics and clinical characteristics of the patients are described in Table 1. Most patients had received prior anthracyclines (83%) and/or taxanes (77%) as either adjuvant therapy or therapy for metastatic disease. A majority of patients had also received prior chemotherapy for metastatic disease. All HER-2/neu-positive patients had received prior therapy with trastuzumab for treatment of metastatic breast cancer. All hormone-receptor-positive patients had received prior endocrine therapy in either the adjuvant (62.5%) or metastatic (91.7%) settings. Twelve (50%) patients had received prior endocrine therapy in both the adjuvant and metastatic settings. Sixteen hormone-receptor-positive patients (66.7%) had received prior tamoxifen, 23 (95.8%) had received prior therapy with an aromatase inhibitor, and 14 (58.3%) had received prior therapy with both.
|Age, y (median [range], 50 [26-73])|
|Hormone receptor status|
|Involved disease sites|
|Disease sites (all lesions)|
Patients received a median number of 4 cycles of therapy (range, 1 cycle -42 cycles). Forty-two patients discontinued treatment because of disease progression or recurrence. One patient discontinued treatment because of toxicity after maintaining stable disease for 8 cycles of therapy, and 9 patients discontinued drug treatment at their own request, most after maintaining stable disease for multiple cycles of therapy.
The response rates for combination therapy with gemcitabine and irinotecan are listed in Table 2. The objective response rate was 25% (95% CI, 13-39), with all responders achieving a partial response (PR). The median duration of response was 5.7 months (95% CI, 3.1 months-9.7 months).
|Response||ITT (N=52)||Assessed for Response (n=49)|
|No. of Patients||%||No. of Patients||%|
|Stable disease <6 mo||11||21.2||11||22.5|
|Stable disease ≥6 mo||7||13.5||7||14.3|
In 18 patients (38%), stable disease was reported as the best response. All of these patients received at least 4 cycles of therapy (median, 6 cycles; range, 5 cycles-12 cycles). Seven patients had stable disease for ≥6 months for a clinical benefit rate (PR + stable disease ≥6 months) of 39%.
Three patients chose to discontinue protocol therapy after Cycle 1, Day 1 of therapy. By including these patients in an intent to treat analysis, the response rate of the combination therapy would decrease to 23%, and the clinical benefit rate would decrease to 36%.
The median time to progression and time to failure for all patients (n = 52) receiving combination therapy was 3.7 months (95% CI, 2.5 months-4.6 months) (Fig. 1A) and 3.0 months (95% CI, 1.8 months-4.4 months) (Fig. 1B), respectively. The overall median survival was 11.6 months (95% CI, 8.9 months -15 months) (Fig. 1C).
Fifty-two patients were evaluable for toxicity. Treatment-related adverse events (TRAEs) were usually low grade (Table 3) and manageable with supportive care. The most frequent nonhematologic TRAEs included fatigue and nausea/vomiting. Only 1 patient experienced grade 3 diarrhea. Although 1 patient developed grade 3 hypokalemia, the most common metabolic laboratory abnormalities detected were low-grade elevations in transaminases, aspartate aminotransferase (21%), alanine aminotransferase (31%), and alkaline phosphatase (12%). No treatment-related fatal or life-threatening events occurred during protocol participation. One patient with a history of malignant pleural effusions died during Cycle 1 of therapy because of worsening of her bilateral pleural effusions. Her death was attributed to progression of disease by the treating physician.
|Adverse Event*||Grade 1||Grade 2||Grade 3||Grade 4||Total|
|No.of Patients||%||No. of Patients||%||No. of Patients||%||No. of Patients||%||No. of Patients||%|
|Platelet count decreased||13||25||3||6||2||4||0||0||18||35|
|Blood alkaline phosphatase||4||8||2||4||0||0||0||0||6||12|
|Blood albumin decreased||3||6||1||2||0||0||0||0||4||8|
Grade 3 and 4 hematologic toxicity was commonly observed and included neutropenia (grade 3 in 33% of patients and grade 4 in 8% of patients), anemia (grade 3 in 6% of patients and grade 4 in 0% of patients), and thrombocytopenia (grade 3 in 4% of patients and grade 4 in 0% of patients). No patients developed febrile neutropenia while receiving protocol therapy.
Nine (17%) patients consented to FNA of tumors before the initiation of therapy. Of these samples, 7 were evaluable for topo I localization. Total cellular levels of topo I varied widely among patient samples (Table 4) and did not appear to correlate with response to therapy. The 2 lowest levels of nuclear to cytoplasmic (N/C) ratios were associated with disease progression, and the highest N/C ratio with clinical response; however, the sample size was too small to adequately correlate topo I localization with treatment response.
|Patient No.||Nuclear Topo I PixD||Cytoplasm Topo I PixD||N/C Ratio*||SD||Response|
Gemcitabine and irinotecan pharmacokinetic analyses were performed in 10 patients. For gemcitabine, the mean Cmax value of 37 ± 23 μg/mL was slightly higher than literature suggests, but occurred between 15 and 30 minutes after the start of infusion as expected.23, 24 The averages for AUC (0-t) of 18.7 ± 11.1 μg/hour/mL and AUC (inf) of 19.0 ± 11.3 μg/hour/mL were also slightly higher than the reported values; this was likely because of the higher Cmax values. However, gemcitabine decreased below quantifiable levels after 2 hours (mean t1/2 of 0.3 ± 0.1 hours), and there was no evidence of gemcitabine after 24 hours in any of the patients sampled. The mean clearance value was affected by these slightly higher AUC values, as it was 123.6 ± 66.1 L/hour (2060 mL/min). This number is a similar, but slightly lower clearance value than the 2550 mL/min reported in the literature.23
The dFdU metabolite pharmacokinetic parameter estimates were also compared with literature results. The mean characteristics were as follows: Cmax = 44 ± 15 μg/mL, AUC(0-t) = 369 ± 79.9 μg/hour/mL, and AUC(inf) = 415 ± 85.6 μg/hour/mL. The t1/2 was determined to be 8.4 ± 1.3 hours, compared with literature containing individual values ranging from 5 to 16 hours.23, 24
The mean total AUC for irinotecan was calculated as 5640 ± 996 ng/hour/mL, and mean clearance was 33.5 ± 7.2 L/hour. The mean total AUC for the major metabolite SN38 was determined to be 295 ± 68 ng/hour/mL.
Among the 49 patients evaluable for response to therapy, 12 patients achieved PRs, for an objective response rate of 25%. Although this study did not meet statistical criteria for improvement of response from 25% to 45% as originally stipulated, it did include a majority of patients previously treated with both anthracyclines and taxanes. In such patient populations, recently published clinical trials have reported much lower response rates to single-agent gemcitabine (range, 0%-29%; median, 17%).25–29 In addition, 2 patients achieved PRs that remained unconfirmed. One patient developed a PR after 6 cycles of therapy; however, at the time of response confirmation, she developed symptoms of confusion and headache. Magnetic resonance imaging of the brain was suggestive of leptomeningeal disease, but lumbar puncture was negative for malignancy. The patient's mental status continued to decline, and she was placed into hospice care. Because of the possibility of disease progression within the central nervous system, the patient was not included as a responder in the efficacy analysis. A second patient also achieved a PR after 6 cycles of therapy, but subsequent scans failed to confirm this response. She received an additional 6 cycles of therapy (12 in total) before developing disease progression.
The combination of gemcitabine with irinotecan was well tolerated, with 41% of patients experiencing grade 3 or 4 neutropenia. This rate of neutropenia was slightly lower that noted with gemcitabine given in combination with taxanes (48%-85%) for the first-line treatment of metastatic breast cancer.30, 31 Grade 3 or 4 thrombocytopenia (4%) and nausea/vomiting (15%) occurred at rates similar to those observed with single-agent gemcitabine therapy.32, 33 Most patients with nausea and vomiting responded to supportive care and the use of antiemetic prophylaxis during subsequent cycles of therapy. It is interesting to note that very few patients developed diarrhea, with only 1 patient (2%) developing grade 3 diarrhea. The low incidence of this toxicity is likely related to the dosing schedule of irinotecan and the use of loperamide as early treatment for diarrhea. Thirteen (25%) patients tolerated prolonged (≥6 months) administration of therapy, with 1 patient receiving therapy for >31 months before developing disease progression.
As would be expected given the toxicity data described, there appeared to be minimal pharmacokinetic interaction between the 2 drugs, although the study's limited pharmacokinetic sampling design cannot completely eliminate the possibility of a significant interaction. The pattern of the concentration-time profile for gemcitabine was similar to the results of other studies conducted with a dose of 1000 mg/m2 with a 30-minute infusion. The mean Cmax observed is more than enough to allow the production of the maximum amount of intracellular triphosphate levels according to the literature. The lack of detectable gemcitabine after 24 hours concurs with the previously reported mean elimination half life (t1/2) of 18 minutes. This elimination was rapid and clearly because of deamination to the major metabolite dFdU.23 It is possible that these mean estimates of Cmax, AUC, and clearance were affected by 2 potential outliers in the data. Two of 10 patients had reported Cmax values >50 μg/mL, yet appeared to have very similar elimination rates when compared with the rest of the group. The average Cmax for the metabolite dFdU was similar to known literature values. The dFdU AUC values for the group of 10 patients also fell within the range of reported values by Abbruzzese et al.23
The limited sampling strategy model that was previously validated for evaluating the irinotecan pharmacokinetics was an acceptable predictor for this study as well. The mean total AUC for both irinotecan and the active metabolite SN38 fell within the range of published reports. These AUCs also correspond to the literature that suggests that the total AUC of SN38 is approximately 100-fold less than that of irinotecan, and therefore it does not appear that the gemcitabine administration had any pharmacokinetic influence on irinotecan.34 The pattern of clearance of the parent compound also suggests the same lack of influence on the pharmacokinetics of irinotecan.
Although only a minority of patients underwent optional tumor biopsies, we were able to characterize cellular concentrations and localization of topo I in the majority (78%) of collected samples. To our knowledge, this is the first description of the expression and localization of topo I in human breast cancer. Our sample size was too small to definitively correlate response with topo I localization, but interesting observations included a response in the patient with the highest N/C ratio, and progression of disease in the 2 patients with the lowest N/C ratios. Such results should be further validated in larger sample sets because assays of this type could be of clinical importance with the increasing use of topo I inhibitors for the treatment of metastatic breast cancer and other types of malignancy.
In summary, although combination therapy with gemcitabine and irinotecan did not reach the parameters required for statistical significance, this study involved a number of pretreated patients, with the majority having received prior therapy with anthracyclines and/or taxanes in either the adjuvant or metastatic setting. Whereas this study was designed to determine the benefit of adding irinotecan to gemcitabine, work published by Perez et al suggests that single-agent irinotecan administered weekly on a 4 out of 6-week schedule has considerable activity for the treatment of metastatic breast cancer (response rate of 23%).35 Thus, it is feasible that the benefits noted in this study were related to the weekly administration of irinotecan rather than the combination therapy. Notably, a higher incidence of grade 3 of 4 diarrhea was reported with single-agent irinotecan administered on a 4 out of 6-week schedule compared with that reported herein in combination with gemcitabine (17% vs 2%). It is also uncertain if a 2 of 3-week administration schedule of single-agent irinotecan would have similar efficacy without the addition of gemcitabine.
With the increasing use of taxanes in the adjuvant setting, further investigation into nontaxane-based regimens should be contemplated for the first-line treatment of metastatic breast cancer. In addition, combination chemotherapy and chemotherapy/biotherapy regimens have, for the most part, demonstrated improved response compared with single-agent therapy and should be considered in patients with symptomatic metastasis or significant visceral organ involvement. Taken together, the preclinical rationale, favorable toxicity profile, reasonable response, and clinical benefit rate noted in this clinical trial suggest that the combination of gemcitabine with irinotecan is a feasible option for the treatment of patients with metastatic breast cancer.
We thank Dr. Caio Rocha-Lima for his contribution to the clinical trial design and analysis of results.
- 16Irinotecan and gemcitabine in patients with solid tumors: phase I trial. Oncology (Williston Park). 2002; 16( 5 suppl 5): 19–24., , , et al.
- 30Gemcitabine plus docetaxel (GD) versus capecitabine plus docetaxel (CD) for anthracycline-pretreated metastatic breast cancer (MBC) patients (pts): results of a European Phase III study. J Clin Oncol. 2005; 23( 16 suppl): A851., , , et al.