Patient Selection

This prospective phase II clinical trial was conducted after obtaining protocol approval from the institutional review boards of the participating institutions. Eligible patients must have been 18 years or older and have had histologically confirmed metastatic breast cancer with documented progressive disease. Patients must have had assessable or bidimensionally measurable disease in at least one site. Patients with assessable bone-only disease were required to have a lytic lesion that had not been radiated previously. Ascites and pleural effusions were not considered as measurable or assessable disease. Patients may have had no more than three prior chemotherapy regimens. One adjuvant chemotherapy was permitted in addition to two regimens for metastatic disease. If no adjuvant chemotherapy was given, then as many as three chemotherapy regimens for metastatic disease were allowed. There were no limitations to previous hormonal or biologic therapies. Patients had to be ambulatory with Eastern Cooperative Oncology Group performance status of 0, 1, or 2. Patients had to have clinically adequate organ function as follows: WBC count ≥ 3,000 cells/μL; hemoglobin level ≥ 8 g/dL; platelet count ≥ 75,000 cells/μL; prothrombin time and partial thromboplastin time less than 1.25 times the upper limit of normal; bilirubin, AST, ALT, and alkaline phosphatase levels no greater than 1.5 times the upper limit of normal; magnesium level ≥ 1.8 mg/dL; and serum creatinine level no greater than 1.5 times the upper limit of normal. Patients must have recovered from the reversible side effects of any prior therapy. Negative serum pregnancy testing was required within 48 hours of the first dose and monthly thereafter for all women with childbearing potential.

Exclusion criteria included the following: recent major surgery within 21 days from starting treatment on protocol; frequent vomiting and severe anorexia; chemotherapy, radiotherapy, or hormonal therapy within 4 weeks of study entry; presence of brain metastasis, carcinomatous meningitis, or cardiomyopathy; and grade 2 or greater peripheral neuropathy. Chemotherapy, radiation therapy, hormonal therapy, immune therapy, and investigational drug therapy were prohibited during treatment on protocol.

Study Design

This investigation was a multicenter (Georgetown University, University of Chicago, Dana-Farber Cancer Institute, and Duke University), open-label, randomized phase II study of thalidomide in patients with metastatic breast cancer. The primary objective was to assess whether thalidomide has activity in this setting and to detect any suggestion of any differences in activity between the high- and low-dose arms of thalidomide by evaluating percentage of patients who remained progression free at 2 months of therapy and by evaluating time to progression for patients who continued on treatment beyond the first 2 months. Safety profiles of the high- and low-dose arms of thalidomide were also obtained. The secondary objectives were to determine the objective response rate (complete and partial), to analyze effects of thalidomide on serum expression of bFGF, TNF-α, VEGF, matrix metalloproteinase activity (MMP-2 and MMP-9), and to study thalidomide pharmacokinetics.

Patients were stratified according to the number of previous chemotherapy treatments, zero or one versus two or three, and then randomized to one of two thalidomide doses: low dose (200 mg/d) or high dose (800 mg/d). Escalation by 200 mg every 2 weeks at the 800-mg/d arm was permitted in patients with no toxicity to a maximum of 1,200 mg/d. Thalidomide was given at 9 pm. Patients were asked to start taking laxatives along with thalidomide and to taper off any sedative or hypnotics they were taking. Patients received thalidomide as long as there was no evidence of tumor progression and as long as there was no dose-limiting toxicity. In the absence of new symptoms, the first tumor assessment was performed at week 8 of therapy. Patients with progressive disease were taken off study. All others continued therapy if toxicity was acceptable. Tumor assessment thereafter was performed every 2 months. Complete response was considered as the disappearance of all clinical and laboratory signs and symptoms of the disease for at least 4 weeks. Partial response was defined as a minimum reduction of at least 50% in the sum of the products of the longest perpendicular diameters of all indicator lesions. Progressive disease was defined as the appearance of new lesions or an increase of at least 25% in the sum of the products of the longest perpendicular diameters of measurable lesions. Stable disease was considered failure of the patient to qualify for complete or partial response or progressive disease at 2 months or more. Patients were observed and evaluated for toxicity by history and physical examination every 2 weeks for the first 2 months and monthly thereafter.

Thalidomide was withheld for grade 2 neurotoxicities, except drowsiness and somnolence, until resolution to less than or equal to grade 1 and then restarted at a 25% dose reduction of the original dose. For recurrent grade 2 neurotoxicity, either thalidomide was restarted at 50% of the original dose after toxicity resolution to less than or equal to grade 1 or the patient was removed from the study. If the patient developed intolerable drowsiness or somnolence at the starting dose of 200 mg/d, the dose was reduced to 100 mg/d. If the patient developed intolerable drowsiness and somnolence at the starting dose of 800 mg/d, the dose was reduced to 600 mg/d, and if she continued to be drowsy, to 400 mg/d, then to 200 mg/d, and finally to 100 mg/d. If the patient could not tolerate the 100 mg/d, then she was removed from the study. Patients were taken off study for grade 4 toxicities except hematologic toxicities. For hematologic toxicities, only grade 4 toxicity was considered dose limiting. For patients with grade 4 hematologic toxicities, the drug was withheld until resolution to grade 1 and then restarted at 75% of the original dose. Patients with recurrent grade 4 hematologic toxicities were removed from the study. For other dose-limiting toxicities, the drug was withheld until toxicity resolved to grade 1 and then restarted at 75% of the original dose. Re-evaluation occurred at weekly intervals. If it was not possible to resume therapy after 3 weeks of delay because of persistent treatment-related toxicities, the patient was removed from the study. For recurrent toxicities, the drug was withheld until toxicity resolved to grade 1 and then restarted at 50% the original dose. For recurrent toxicities while at a 50% dose reduction, patients were removed from the study.

Pharmacokinetics and Correlative Studies

The first dose of thalidomide was given at 9:00 am on day 1. Subsequent doses were given at 9 pm. Plasma samples were obtained for pharmacokinetics immediately before the first dose on day 1 and then 1/2, 1, 1 1/2, 2, 3, 4, 5, 6, and 7 hours after ingestion and at 9 am and 1 pm on day 2. Pharmacokinetic specimens were collected every 2 weeks for the first 2 months and monthly thereafter. Thalidomide was assayed in plasma by modification of the method of Eriksson et al.³⁴ Plasma samples for assay of thalidomide were collected and diluted immediately into Sorenson’s buffer (25 mmol/L disodium citrate pH 1.5) and then stored at −70°C before assay. Samples were thawed and kept on ice before further processing. Two milliliters of each plasma sample were placed in polytetrafluoroethylene-lined, screw-cap centrifuge tubes, and 50 μL of a 100-μg/mL solution of phenacetin was added. Samples were then subjected to vigorous vortex, and 5 mL of diethyl ether was added to each sample and the tubes capped before being shaken in an Eberbach mechanical shaker (Eberbach, Ann Arbor, MI) for 5 minutes. They were then centrifuged at 4,000 rpm in a Beckman J-6M centrifuge (Beckman, Fullerton, CA) with a JS 4.0 rotor for 5 minutes, and the ether layer was aspirated and placed in 12 × 75 mL culture tubes. The ether was evaporated by Speed-Vac (Savant Speed-Vac, Farmingdale, NY), and samples were reconstituted with 100 μL of the mobile phase below. Thalidomide was measured by high-performance liquid chromatography using an ALLTECH Spherisorb ODS-2 (250 × 4.6 mm) column (ALLTECH, Deerfield, IL), with a Waters Nova-Pak C₁₈ guard column (Waters, Milford, MA) both equilibrated in a mobile phase of water and acetonitrile (65/35 vol/vol). Using a pump speed of 1 mL/min, thalidomide was detected at 300 nm at a retention time of approximately 5.4 minutes, whereas the internal standard phenacetin was detected at approximately 6.75 minutes. By use of this method, the limit of quantification for thalidomide was found to be 0.1 μg/mL and the inter- and intraday coefficients of variation at this concentration were less than 14.5% and less than 9.7%, respectively.

Serum specimens for angiogenic factors were obtained on day 1 immediately before the first dose of thalidomide, every 2 weeks during the first 2 months, and then monthly. Circulating levels of bFGF, VEGF, and TNF-α were determined using quantitative sandwich enzyme immunoassay kits according to the manufacturer’s instructions (Quantikine, R & D Systems, Minneapolis, MN). Cutoffs to distinguish elevated levels were determined as the mean + 1 SD (84th percentile) of a normal population. These cutoff values were 9.0 pg/mL for bFGF, 429 pg/mL for VEGF, and 3.9 pg/mL for TNF-α.

Levels of plasma MMP-2 and MMP-9 were measured using zymograms, as described previously.³⁵ Briefly, patient plasma samples were diluted in 1 × zymogram loading buffer and run without boiling on 10% SDS Page gels (Sigma, St Louis, MO) that contained 0.1% gelatin. Conditioned media, treated with P-amino-phenyl-mercuric-acetate, from Hs578t and MDA-MB-231 cells were run on each gel to provide standards for the latent and active forms of MMP-2 and MMP-9. The gels were subsequently washed twice for 30 minutes in Tris-buffered saline that contained 2% Triton X-100 and then incubated overnight at 37°C in 50 mmol/L Tris pH 7, 5 mmol/L calcium chloride, and 1% Triton X-100 (Sigma) to allow gelatin degradation. Areas of digestion were then visualized by staining with Coomassie blue R250 (Sigma) in 10% acetic acid/20% methanol, followed by de-staining in 10% acetic acid/20% methanol. The gels were then dried, and the areas of clearing caused by the action of MMP-2 and MMP-9 were measured by image analysis of the appropriate portion of the gels.

Thalidomide Supply

Thalidomide was supplied by the National Cancer Institute–Cancer Therapy Evaluation Program. Thalidomide was provided as 100-mg tablets.

Statistical Methods

A two-stage sample design was used to evaluate whether thalidomide had any activity in patients with progressive metastatic breast cancer who had previously received zero to three chemotherapy regimens. Thalidomide was to be considered active in this patient population if at least 20% of the patients remained progression-free after 2 months on either arm of the study. In the first stage, 14 patients were entered onto each of the two arms by random assignment. The randomization was stratified by the number of prior chemotherapy regimens received (zero or one v two or three). If all of the first 14 patients on each arm progressed within 2 months on study, then the arm was to be terminated and it would be concluded that there was 95% confidence that thalidomide at the administered dose level was not active.³⁶ Alternatively, if at least one of the first 14 patients had stable disease or an objective response after 2 months on study, then the trial was to proceed to the second stage.

Serial circulating growth factor levels were determined for all but two patients. To evaluate whether relative changes in serial factor levels were significant, patients whose angiogenic factor levels never exceeded the cutoffs distinguished by 1 SD above the mean of a normal population were considered noninformative. Noninformative patients were excluded from these analyses because normal biologic changes may cause small absolute changes in marker levels within normal level range, and these small absolute changes may falsely represent large relative changes in serial levels. Furthermore, the standard curves for these assays are increasingly flatter toward the normal ranges. Therefore, small changes in the assay readout (for example, absorbance) are reflected as large relative changes in marker levels in this part of the curve. For example, if a normal cutoff is 2 U/mL, then 84% of the normal population are expected to have levels less than 2 U/mL. For both technical reasons (flat standard curve in this range) and biologic reasons (normal variation of 1 or 2 U/mL), a 50% change (from 1 to 2 U/mL) would not be considered likely to be a result of a therapeutic intervention. In contrast, the standard curves become quite linear and steep in ranges above the normal cutoff. Furthermore, normal biologic changes continue to account for serial changes of only 1 or 2 U/mL. Therefore, a 50% change (from 10 to 20 U/mL) is most likely to result from therapeutic intervention and not from biologic or technical causes. For these reasons, a relative change from baseline to a subsequent sample level was only considered informative if one, the other, or both sample levels exceeded the normal cutoff. To normalize the data, the natural logarithm of each marker level measurement at baseline and at time of removal from study (because of progression or toxicity) was taken. The log of the baseline level was subtracted from the log of the value at time of removal from study for each patient. Combining data from both dose levels, a paired t test was performed for each growth factor to test whether the mean difference from baseline to time of removal from study was statistically significant.

Patients

Twenty-eight patients were accrued at the four centers between 1996 and 1998 (Table 1). Fourteen patients were accrued to each of the two dose levels. Patients’ ages varied from 30 to 85 years, with the largest number of patients between 41 and 60 years old (18 patients). Two patients on each arm had had zero or one chemotherapy regimens before enrollment, and twelve had had two or three prior chemotherapy regimens. Three patients on the low-dose arm and two on the high-dose arm had received high-dose chemotherapy with peripheral stem-cell support before enrollment onto the protocol. Patients on the two treatment arms were comparable with respect to age and the number of chemotherapy regimens.

Dose Modifications

One patient at the 200-mg dose required dose reduction because of grade 3 peripheral neuropathy. At the 800-mg dose, five patients had to reduce doses to 600 mg and two patients to 400 mg, all because of neurotoxicity (somnolence). Two patients continued at the 800-mg dose with no changes. One patient’s dose was increased to 1,000 mg, and four had their dose escalated to 1,200 mg.

Efficacy

No patient achieved a partial or complete response. Two patients at the 200-mg dose had stable disease at 8 weeks. The first patient had a 43% reduction in size of hilar and mediastinal lymphadenopathy (site of measurable disease) at 8 weeks. However, at the staging at 16 weeks, she had progressive disease and was removed from the study. This patient had previously received adjuvant chemotherapy with cyclophosphamide and doxorubicin and, later, paclitaxel and then vinorelbine for metastatic disease. The second patient had relatively indolent chest-wall disease that was slowly progressing on no treatment over the 20 months preceding thalidomide. At the staging at 8 weeks, she had stable disease, but she was removed from the study at week 11 because of grade 3 peripheral neuropathy.

As of August 26, 1998, all patients had been removed from the study, 26 because of progressive disease and two because of patient choice and/or unacceptable toxicity (Table 2). Of the two patients who did not progress, one was on the 200-mg arm and was removed from study at week 11 because of grade 3 peripheral neuropathy. The second patient was on the 800-mg arm and refused to continue treatment beyond week 4 because of drug-related somnolence. This patient refused dose reduction.

At the 200-mg level, one patient was taken off study at 2 weeks and a second patient at 4 weeks after starting treatment because of rapidly progressive disease. Ten patients were taken off at 8 weeks because of progressive disease detected by routine restaging. Two patients went beyond the first 8 weeks of staging. The first patient was removed from study at 11 weeks and the second at 16 weeks (Table 3).

At the 800-mg level, two patients were removed from the study at 4 weeks, one because of progressive disease and the second, who refused to continue treatment, because of somnolence. Four patients were taken off study at 6 weeks and eight patients at 8 weeks, all because of progressive disease. None of the patients at the 800-mg level continued beyond the first 8 weeks of treatment (Table 3).

Adverse Events

Only one patient was removed from the study because of grade 3 neurotoxicity (peripheral neuropathy). This patient was on the 200-mg dose and was removed at week 11. In the high-dose arm, the main dose-limiting toxicity was somnolence that required dose reduction in seven patients. The dose was reduced from 800 to 600 mg for five patients and from 800 to 400 mg for two. The other adverse events that did not require dose reduction or removal from the study included the following: constipation, somnolence, fatigue, peripheral neuropathy, dizziness and instability, dry mouth, skin rash, nausea, anorexia, arrhythmia, neutropenia, headaches, and hypotension (Table 4).

There were five serious adverse event reports during the study. Two patients on the 200-mg dose required hospital admissions for progressive shortness of breath. Both of these patients had progressive increase in pleural effusions that were not related to thalidomide. One patient on the 800-mg dose developed postural hypotension during the first treatment day. She became diaphoretic and light headed but improved rapidly after intravenous fluids were started, and she continued treatment on protocol without similar episodes. Another patient on the 200-mg dose developed dizziness and palpitation, and ECG showed sinus bradycardia with bigeminy that resolved spontaneously. Her echocardiogram was normal, and she continued treatment on the protocol with no similar episodes. The last patient, who was on the 800-mg dose, developed vomiting and headaches that required intravenous hydration at the end of her first day of treatment. She continued on the study with no similar episodes. Although the last three episodes were considered to be probably related to thalidomide, none recurred with continued treatment.

Correlative Studies

Circulating angiogenic factor levels.

Circulating baseline angiogenic growth factor levels were determined in all but one patient. Of 27 patients, five (18.5%), six (22.2%), and 13 (48.1%) had elevated levels (≥ mean + 1 SD in normal population) of bFGF, VEGF, and TNF-α, respectively. The proportion of patients with elevated TNF-α levels was significantly greater than that for bFGF or VEGF levels (P = .034).

Changes in serum bFGF, VEGF, and TNF-α levels from baseline to time of removal from study (either because of progression or toxicity) in 26 patients were determined. One patient did not have baseline specimens collected, and one had baseline but no follow-up specimens collected. The bFGF, VEGF, and TNF-α data are illustrated in Fig 1. Only informative patients were included in this analysis. For bFGF, VEGF, and TNF-α, there were 19, 13, and three noninformative patients, respectively. The data from both dose levels were combined for the paired t test performed on each growth factor. The results of these analyses indicate mean percentage changes from baseline of −37% for bFGF, +60% for VEGF, and +79% for TNF-α. Of these, only the increase in TNF-α levels was statistically significant (P = .017).

View larger version:

• In this window
• In a new window

• PowerPoint Slide for Teaching

Fig 1. Changes (%) in circulating bFGF, VEGF, and TNF-α levels from baseline to time of removal from treatment with thalidomide. Symbols: (– – – –), ≥ 25% change from baseline level; (•), 800 mg/d; (○), 200 mg/d.

Of interest, serial changes in circulating levels of bFGF and VEGF seemed random in the single patient who experienced a near-partial response. However, in contrast to all but two other patients, serial TNF-α levels from this patient decreased from baseline to each time point (2, 4, and 8 weeks).

Plasma levels of MMP-2 and MMP-9 were determined by zymography. This method allows the simultaneous measurement of the latent and active forms of the enzyme. Plasma MMP-2 levels remained relatively unchanged over the period of the study. Comparison with the standards revealed that all of the enzyme was apparently in the inactive proform (data not shown). There was considerable variation in the plasma levels of MMP-9 over time, with no discernible, consistent pattern to these alterations (data not shown). Statistical analysis of MMP-9 levels before treatment and at any point during treatment by paired t test did not show any significant trend. Like MMP-2, all of the MMP-9 in patient plasma seemed to be in the inactive proform.