- © 2002 by American Society of Clinical Oncology
Comparison of the Systemic and Intratumoral Effects of Tamoxifen and the Aromatase Inhibitor Vorozole in Postmenopausal Patients With Primary Breast Cancer
- Catherine L. Harper-Wynne,
- Nigel P.M. Sacks,
- Karyn Shenton,
- Fiona A. MacNeill,
- Paul Sauven,
- Ian J. Laidlaw,
- Zen Rayter,
- Stephanie Miall,
- Angela Howes,
- Janine Salter,
- Margaret J. Hills,
- Frances M. Lowe,
- Roger A’Hern,
- Nazar Nasiri,
- Debbie Doody,
- Jhangir Iqbal and
- Mitchell Dowsett
- From the Departments of Biochemistry, Surgery, Histopathology, and Statistics, Royal Marsden Hospital, London; Breast Unit, Essex County Hospital, Colchester; Department of Surgery, Frimley Park Hospital, Surrey; Department of Surgery, Broomfield Hospital, Essex; Department of Surgery, Bristol Royal Infirmary, Somerset; and Janssen-Cilag Ltd, High Wycombe, Buckinghamshire, United Kingdom.
- Address reprint requests to M. Dowsett, PhD, Academic Department of Biochemistry, Royal Marsden Hospital, Fulham Rd, London SW3 3JJ, United Kingdom; email: mitch{at}icr.ac.uk
Abstract
PURPOSE: To determine biologic differences, if any, between presurgical endocrine treatment with an aromatase inhibitor (vorozole) and tamoxifen in patients with postmenopausal primary breast cancer.
PATIENTS AND METHODS: Randomization was to 12 weeks of 2.5 mg of vorozole per day or 20 mg of tamoxifen per day, both orally. Clinical response was assessed monthly together with serum sex hormone binding globulin (SHBG), luteinizing hormone (LH), follicle-stimulating hormone (FSH), estrogens (E1, E2, and E1S), lipids, insulin-like growth factor-1 (IGF-1), and bone metabolites (CrossLaps CTx). Tissue samples for Ki67, apoptotic index (AI), estrogen receptor, and progesterone receptor were collected at 0, 2, and 12 weeks.
RESULTS: Ki67 fell by 58% and 43% (means) at 2 weeks in the vorozole and tamoxifen patients, respectively (P = .13). In the vorozole group, the correlations of proportional changes in Ki67 at 2 weeks with tumor volume changes and clinical response at 12 weeks were not significant (P = .09) and marginally significant (P = .04), respectively. Serum lipids did not differ between groups. Serum levels of EI, E2, and E1S were suppressed markedly by vorozole, whereas levels of SHBG increased and LH and FSH fell significantly with tamoxifen. IGF-1 levels fell significantly with tamoxifen (P = .001) compared with the nonsignificant rise with vorozole. Twelve-week CTx values fell by 19% with tamoxifen (P = .006) and rose by 11% with vorozole (P = .15).
CONCLUSION: The correlation with vorozole of Ki67 with volume and clinical response supports this as an intermediate marker. The nonsignificant effects on bone and lipid metabolism by the aromatase inhibitor may be important to consider for adjuvant and potential prevention strategies.
OVER RECENT YEARS, aromatase inhibitors have become increasingly prominent in the treatment of postmenopausal breast cancer in patients with advanced disease.1 Their excellent tolerability and clinical effectiveness as second-line treatment has led to their ongoing evaluation as first-line agents versus tamoxifen in metastatic disease and in a series of large ongoing adjuvant trials. The eventual application of aromatase inhibitors in these settings will depend on their effects on several estrogen-dependent metabolic systems (eg, bone and cardiovascular) as well as their clinical effectiveness. The scenario of presurgical therapy is attractive for making comparisons of both metabolic and clinical end points. This study describes a randomized comparison of tamoxifen and an aromatase inhibitor in this clinical setting.
Treatment of primary breast cancer conventionally involves initial complete excision of the tumor, with most patients receiving some form of systemic adjuvant therapy, because it is now well established that the systemic use of adjuvant chemotherapy and endocrine treatment significantly reduces the risk of relapse and death.2,3 Over recent years, however, there has been an interest in the potential advantages (eg, downstaging of tumor and ability to monitor effect of intended adjuvant treatment) of neoadjuvant therapy. In the National Surgical Adjuvant Breast and Bowel Project B-18 trial, neoadjuvant chemotherapy was associated with equivalent survival compared with adjuvant treatment and was noted to have the benefits of allowing tumor response and biomarker changes to be assessed in parallel.4 There have been no equivalent randomized trials with survival as the primary end point comparing a fixed period of neoadjuvant endocrine treatment to no presurgical treatment. Nonetheless, over recent years, several neoadjuvant hormonal studies have been conducted. In postmenopausal patients, the antiestrogen tamoxifen has been the predominant drug used in these settings, but recent phase II neoadjuvant studies have reported promising responses with aromatase inhibitors.5 In addition, a randomized trial of the aromatase inhibitor letrozole versus tamoxifen has recently reported the greater efficacy of the inhibitor in this scenario.6 The clinical status of aromatase inhibitors has recently been summarized by Goss and Strasser.7 In brief, aromatase inhibitors have their effect by inhibiting the conversion of androgens to estrogens by the cytochrome P-450 enzyme aromatase.1 The third generation of inhibitors of aromatase include anastrozole, letrozole, vorozole, and exemestane,1 which show aromatase inhibition of more than 96%.7-10 Each of these agents showed improvements in one or more clinical efficiency end points in phase III trials against megestrol acetate or aminoglutethimide, together with improved tolerability.1,7 Recent data indicate that the nonsteroidal inhibitors anastrozole and letrozole may have advantages over tamoxifen.11,12
Estrogen receptor (ER)–positive patients treated with adjuvant tamoxifen have improved survival benefit over ER-negative patients,13 but a substantial proportion of patients will relapse and die from recurrent disease. It would be desirable for both primary treatment of patients and adjuvant treatment if biomarkers could be established that act as intermediate markers for response and survival benefit. Advances in molecular endocrinology offer better understanding of the mechanism of action of endocrine therapies and of response and resistance to them. Clinical application of these findings may enable improved treatment strategies to be developed. The neoadjuvant and primary treatment scenario provides a unique opportunity to evaluate such mechanisms for the following reasons: (1) there are no potential carryover effects from previous treatments that often confound results in advanced breast cancer patients; (2) biomarker data and clinical response are determined from the same lesion; (3) multiple biopsies can be taken with minimal trauma to both patient and tumor, allowing the dynamics of biomarker change to be assessed; and (4) concurrent therapy is less frequent than in advanced disease.
Therefore, the trial described was designed to compare, in a prospective randomized fashion, the effect of two primary treatments, tamoxifen and vorozole, on a key set of biomarkers, selected in parallel with clinical response. Endocrine treatments have been considered to act primarily in a cytostatic manner, but our earlier data showed that some antiestrogens can also induce programmed cell death.4 This supports xenograft studies reporting that estrogen deprivation and antiestrogens can lead to a profound induction of apoptosis.15-17 The primary objective of this study, therefore, was to compare the effects of an aromatase inhibitor and tamoxifen on tumor cell proliferation and death using Ki67 as the marker of proliferation and the terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate-biotin nick end-labeling (TUNEL) method for highlighting apoptotic cells. Ki67 has previously been investigated in primary treatment studies of tamoxifen, where reductions in Ki67 after treatment were observed.18,19 A small nonrandomized study showed that the decrease in Ki67 in the first 14 days of tamoxifen treatment was related to clinical response.20 Secondary objectives included evaluation of the pathologic markers ER and progesterone receptor (PgR), which are both important predictors for response to endocrine therapy.21 Other markers to be assessed included insulin-like growth factor-1 (IGF-1), which has been reported to act as both a mitogenic and survival factor in breast cancer,22 as well as changes in lipid profiles (cholesterol, low-density lipoprotein [LDL], and high-density lipoprotein [HDL]) and biologic markers of bone metabolism. Consideration of changes in these metabolic markers is likely to be significant in assessing the relative acceptability of aromatase inhibitors and tamoxifen in the adjuvant and prevention settings.
PATIENTS AND METHODS
Patients
From November 1996 to August 1998, patients with primary operable breast cancer were recruited from five hospitals that were part of the Vorozole Study Group into a randomized trial of primary tamoxifen therapy versus vorozole. The trial aimed to recruit 50 patients. The study was not formally powered; rather, it was sized on the basis that previous randomized neoadjuvant studies had reported statistically significant changes in the two primary endpoints to be studied using a similar number of patients.14,23
Local ethics committee approval was obtained at each hospital. All patients received approved information sheets before entry into the study and provided written consent to participate in the study. Inclusion criteria were as follows: (1) postmenopausal status (≥ 12 months amenorrhea or, if < 56 years of age and < 1 year amenorrhea, ovariectomized or hysterectomized before menopause confirmation with follicle-stimulating hormone [FSH] and luteinizing hormone [LH] at local laboratory was required); (2) tumor size of 2 cm or larger by clinical examination; (3) malignancy and ER positivity (ie, ≥ 20 on H-score) confirmed by core-cut biopsy; (4) life expectancy of 6 months or longer; and (5) written informed consent. Exclusion criteria were as follows: (1) metastatic breast cancer; (2) any prior or concomitant hormonal therapy (eg, tamoxifen, HRT taken < 1 month before entry into trial, and systemic glucocorticosteroids); (3) any concurrent or previous malignant disease within the last 5 years (except treated basalcell carcinoma of the skin); (4) significant hematologic abnormalities (WBC count < 3 × 109/L or platelet count < 100 × 109/L); (5) significant renal (serum creatinine ≥ 1.5 times upper limit of normal) or hepatic dysfunction (total bilirubin ≥ two times upper limit of normal, all transaminases ≥ two times upper limit of normal, or only one ≥ 2.5 times upper limit of normal); (6) Eastern Cooperative Oncology Group performance score higher than 1; and (7) inability to comply with study protocol.
Patients randomized to vorozole received 2.5 mg, and those randomized to tamoxifen received 20 mg, each to be taken orally once a day for 12 weeks. Patients were assessed at 4, 8, and 12 weeks for tumor response (manually with calipers and by ultrasound) and tolerability of medication. If there was evidence of progressive disease at 4 or 8 weeks, the patient was withdrawn from the study, and additional appropriate management was considered. At 12 weeks, the type of surgery was decided locally. The results of definitive surgery were used locally to determine patients’ suitability for subsequent adjuvant radiotherapy or chemotherapy. Patients continued on treatment until surgery, which could be performed up to 1 week later. Where surgery was not indicated, a core-cut biopsy was taken at 12 weeks. Patients from either arm who achieved a good clinical response continued adjuvant treatment with tamoxifen for 5 years after surgery.
Staging and Response
Clinical staging of patients, according to local protocols, was carried out before commencement of the study treatment. Percent change in tumor volume using ultrasound was calculated by using ultrasound measurements at commencement of study and week 12. When three dimensions were available, the product of the three was divided by two for the volume, and for two dimensions only the following calculation was used: larger diameter × (smaller diameter)2/2. Radiologists reading ultrasound scans were blinded to the treatment assignment.
Tumor response was assessed at intervals of 4 weeks. Clinical response to treatment was ascribed using International Union Against Cancer definitions using the two largest diameters from ultrasound measurements, ie, complete response (CR) was defined as no residual disease on ultrasound at 12 weeks, partial response (PR) was defined as greater than 50% reduction in bidimensional measurements (bdm) between start and week 12, and stable disease (SD) was defined as between 50% reduction and 25% increase in bdm from start to 12 weeks. Progressive disease (PD), however, was ascribable at 4, 8, and 12 weeks and defined as greater than 25% increase in bdm from the start of treatment or the development of new lesions. Any patients showing PD were to be immediately withdrawn from trial therapy, but biologic data collected up to the point of withdrawal would be included in the appropriate analysis.
Sample Collection
Before study medication was started, a 20-mL clotted blood sample was collected for endocrine, lipid, and bone product analysis, and collections were repeated at 4, 8, and 12 weeks. It was collected at approximately the same time of day on each occasion and before the day’s medication. The sample was allowed to clot for 2 hours at room temperature and was then centrifuged for 10 minutes, and the serum was stored at −20°C. Together with the diagnostic core cut, an additional core cut was taken only at 2 weeks and at 12 weeks if indicated (see above).
Laboratory Methods
Analyses were conducted with the analyst blinded to the trial therapy.
Immunohistochemical analyses.
Tumor material obtained before treatment, at 2 weeks, and at 12 weeks was fixed in neutral buffered formalin and embedded in paraffin wax. Sections of 3-μm thickness were cut and mounted on aminopropyltriethoxysilane-coated slides. General reagents were purchased from Sigma (Dorset, United Kingdom). The method for Ki67 staining using MIB-1 antibody (DAKO, Ely, United Kingdom) and ER and PgR staining involved microwave antigen retrieval and three layers with avidin-biotin complex technique, as previously described by this laboratory.15 Primary antibodies for ER and PgR, however, were purchased from Novocastra and used at 1:40 dilution (monoclonal clone 6F 11 and 1A6 respectively; Novocastra Newcastle-On-Tyne, United Kingdom). Scoring of Ki67 utilized a previously described system,15 and ER and PgR staining was scored using the H-score method.23 Apoptotic index (AI) was measured using the TUNEL assay24 and was expressed as the percent of malignant cells staining positive where at least 3,000 cells were scored.14
Serum analyses.
Sex hormone binding globulin (SHBG) and IGF-1 were measured using the standard kits Orion Diagnostica (Pharmacia Limited, Milton Keynes, United Kingdom) and DSL 2800 (Diagnostic Systems Laboratories, London, United Kingdom), respectively. LH and FSH were measured by enzyme immunoassay on an Axsym analyzer (Abbott Diagnostics, Maidenhead, United Kingdom). Estrone (E1), estradiol (E2), and estrone sulphate (E1S) were measured using previously described methods from this laboratory.11,25 Lipid analyses, cholesterol, and HDL used an enzymatic method on a Synchron CX9ALX analyzer (Beckman, High Wycombe, United Kingdom). LDL was calculated using the Friedwald equation. Degradation products of type 1 collagen were measured using a new ELISA, Serum CrossLaps (CTx) (Osteometer BioTech A/S, Herlev, Denmark). This test uses two highly specific monoclonal antibodies against amino acid sequences in the products from the C-terminal telopeptides of type 1 collagen in the serum.
Statistical Methods
When data were not normal and skewed, it was log-transformed. The significance of changes in parameter values between two time points was assessed by the Wilcoxon signed rank test, which investigated whether the log of the proportional change was not zero (which is equivalent to examining whether the proportional change differs significantly from 1). Before this, if more than two time points were being statistically reported for a parameter, the Friedman test was used to ensure statistical significance over time. To compare the changes between groups, eg, vorozole and tamoxifen, the magnitudes of the logs of the proportional changes were compared using the Mann-Whitney test. Confidence intervals were calculated on the basis of the logs of the proportional changes, and these were then back transformed to express them as proportional changes. Two-sided tests were used in every instance. Data were analyzed using Minitab Release 7 (1989) (Minitab Inc, State College, PA).
RESULTS
Fifty-three patients were recruited into the study, 26 receiving vorozole and 27 receiving tamoxifen. The demographic characteristics are listed in Table 1. Three patients were not assessable for all end points for the following reasons: two patients withdrew before 4 weeks (one because of toxicity and one because of inability to comply with study medication) and one patient’s clinical record and samples were lost. No patients showed PD before the 12-week point, and none, therefore, required early withdrawal.
Demographic Characteristics of Study Patients
Response
Four patients were not assessable for radiologic response for the following reasons: three patients had no radiologic assessment before commencement of study medication and one patient could not have accurate radiologic assessment of tumor because of its size. Figure 1 shows individual tumor volumes after 12 weeks as a percentage of pretreatment volume in both groups. The median reduction in volume for tamoxifen-treated patients was 58% (range, −74% to +95%) and for vorozole-treated patients was 37% (range, −80% to +95%) (P = .11).
Fig 1. Volume changes using ultrasound (US) where volume at 12 weeks is plotted as a percentage of pretreatment volume. P = .11 for the difference between the two treatment groups.
Nine (39%) and five (22%) patients showed a PR to tamoxifen and vorozole, respectively. This gave a 17% difference in the response rates (95% confidence interval, −9% to 44%). There were no complete responses, and three patients had progressive disease (one tamoxifen and two vorozole).
Biologic Marker: Immunohistochemical Markers
Patients were considered assessable for Ki67, ER, PgR scoring, or AI if they had sufficient tissue from their pretreatment and one other sample. Forty-two patients were assessable for Ki67, ER, and PgR scoring (22 vorozole and 20 tamoxifen) and 31 for AI (15 vorozole and 16 tamoxifen). Table 2 summarizes the results for Ki67, ER, PgR, and AI at pretreatment, 2 weeks, and 12 weeks after treatment. The 2- and 12-week values were expressed as a percent of pretreatment value, and mean results are listed in Table 2.
Immunocytochemical Results
Ki67.
Individual changes in Ki67 for patients in each treatment group are shown in Figs 2A and 2B. There were mean falls of 58% and 43% at 2 weeks in the vorozole and tamoxifen patients, respectively (P = .002, P = .004). Both groups showed significant falls by 12 weeks (P = < .001 vorozole, P = .015 tamoxifen), and again there was no significant difference between the groups at this time (Fig 3). There was also a significant drop between the 2-week and 12-week results in the tamoxifen patients (P = .03) but not vorozole patients. There was no significant relationship between baseline Ki67 and proportional fall in individuals for either treatment group. Thus, the greater fall seen with vorozole at 2 weeks does not seem to have been attributable to the higher baseline value in that group.
Fig 2. Individual patient Ki67 (using MIB-1) changes over the first 2 weeks of treatment for vorozole (A) and tamoxifen (B).
Fig 3. Ki67 recorded for both treatments at 2 weeks and 12 weeks and plotted as percentage of pretreatment values. Bars indicate 95% confidence intervals of the means. Bars not crossing 100% mark, ie, baseline values, indicate that these within-group changes are significant.
A positive correlation of borderline significance was found between the proportional changes in Ki67 at 2 weeks and the tumor volume at 12 weeks for vorozole (r = .49, P = .09) but not for tamoxifen (r = .376, P = .17) (Figs 4A and 4B). In the vorozole group, clinical responders at 12 weeks showed a significantly greater fall in Ki67 at 2 weeks than nonresponders (P = .04).
Fig 4. Correlation of the 2-week Ki67 value as a percent of pretreatment with the 12-week volume as a percentage of pretreatment volume for vorozole (A) and tamoxifen (B). P = .09 for the vorozole correlation.
AI.
There was a significant fall in AI in the vorozole group (P = .03) but not the tamoxifen group at 2 weeks, leading to a borderline significant difference between the two groups at this time (P = .06). After 12 weeks, AI was not statistically different from pretreatment values for either drug (Table 2, Fig 5). The ratio of Ki67 to AI was calculated at each time point, and the on-treatment values were expressed as a proportion of pretreatment values, which we have termed the growth index. The 2-week and 12-week growth indices revealed reductions of 46% (mean, P = .005) and 63% (P = .006), respectively, for tamoxifen and 36% (P = not significant) and 69% (P = .007), respectively, for vorozole. There were no significant differences between the treatment groups in the growth index at either of the on-treatment time points (Table 2).
Fig 5. AI recorded for both treatments at 2 and 12 weeks plotted as percentage of pretreatment values. Bars indicate 95% confidence interval of the means. P = .06 for the difference between the two groups at 2 weeks.
ER and PgR.
There were significant falls in ER at 2 weeks for tamoxifen (P = .001) and for both treatment groups at 12 weeks (P = .001 for each group, Table 2). There were no significant differences between the two groups. PgR fell significantly over 2 weeks in the vorozole group (P = .004), which was significantly different (P = .002) from the nonsignificant rise in the tamoxifen group. This fall in the vorozole group was maintained at 12 weeks (P = .001) and was again significantly different from the nonsignificant fall at 12 weeks in the tamoxifen group (P = .004, Table 2).
Biologic Markers: Serum Markers
Patients were assessable for these analyses if they had both a pretreatment and a 12- or 8-week value. In the few instances (maximum of three patients) where the 12-week value was not available, the 8-week result was used. In all cases, data were compared between the treatment groups using proportional changes as above.
SHBG levels increased significantly in the tamoxifen group and showed a nonsignificant fall in the vorozole group (Table 3). LH and FSH both fell significantly in the tamoxifen group, and there was also a significant rise in FSH in the vorozole group (Table 3). E1, E2, and E1S levels all fell markedly and significantly in the vorozole group. There was a nonsignificant rise in E1S over 12 weeks in the tamoxifen arm. IGF-1 levels fell significantly in the tamoxifen group, whereas vorozole patients showed a nonsignificant increase; the difference between the groups was significant (P = .002, Table 3).
Serum Hormone Results
Total cholesterol and LDL both fell significantly in the tamoxifen group. Although vorozole patients showed no significant change, there was no significant difference between the groups (Table 4). HDL showed no significant change in either group (Table 4). Serum levels of the bone resorption biomarker, CTx, fell significantly by a mean 19% in the tamoxifen group. The mean level was increased by 11% in the vorozole group, but this was not statistically significant. The difference between the groups was statistically significant (P = .001, Table 4).
Serum Lipids and Bone CrossLaps (ctx) Results
DISCUSSION
Aromatase inhibitors are a widely used hormonal treatment for postmenopausal breast cancer and are presently under evaluation in several large comparative trials with tamoxifen in both advanced and early disease.7 Recently, some of the advanced disease studies have indicated the improved efficacy of aromatase inhibitors over tamoxifen.6,11,12 This study was designed to assess the comparative effects of the drugs on several tissue and blood biomarkers, which may influence the application of the drugs, eg, in adjuvant and prevention settings. In this limited-size study, it was particularly important that the end points should be unaffected by the confounding effects of previous or concurrent therapy; the neoadjuvant setting provides an ideal opportunity to achieve this and to make such comparisons. The study was the first randomized trial comparing neoadjuvant and primary treatment of an aromatase inhibitor, in this case vorozole, with tamoxifen in postmenopausal breast cancer. The primary goal was to compare the effects of the two treatments on cell proliferation and apoptosis, the major determinants of tissue growth, and to relate these effects to clinical response.
Data have recently been published on a large randomized neoadjuvant study of letrozole versus tamoxifen, which revealed the aromatase inhibitor to be more effective in this scenario.6 The clinical outcome data from the present study should not be compared with this, because it was not powered for valid clinical outcome comparisons. Although vorozole is no longer in clinical development, its clinical activity in a phase III trial against megestrol acetate was similar to that of other third-generation inhibitors.7,26
A significant fall occurred in the proliferation marker Ki67 after 2 and 12 weeks with both vorozole and tamoxifen. Our results confirm the inhibition of tumor cell proliferation seen by ourselves and others in both the short term18,20 and long term19 with tamoxifen. However, such changes have not been reported previously with an aromatase inhibitor. The results suggest that there may be a lesser or a slower effect on proliferation with tamoxifen than with aromatase inhibitor. This may be related to a different mechanism of response between the two compounds; for example, the partial agonist effect of tamoxifen may lead to a reduced impact on proliferation, particularly in the short term (see also discussion on apoptosis below). However, a lesser effect of tamoxifen may also be related to the pharmacokinetic differences that exist between the two drugs. Tamoxifen has a half life of 7 to 10 days,27 leading to its reaching steady state after about 5 weeks. In contrast, the half-life of vorozole is approximately 11 hours, resulting in steady state being after 3 days, well before the 2-week assessment.28 Although Ki67 is widely used as a marker of proliferation, it is not universally accepted. We have previously found Ki67 to be strongly and significantly related to S phase in breast cancer,29 which supports its use as a surrogate for proliferation.
Our earlier preliminary results suggested that apoptosis was increased in ER-positive patients treated with 2 weeks of tamoxifen,30 and in a subsequent series we showed a small but significant increase in AI in ER-positive patients after 21 days of neoadjuvant tamoxifen.14 These effects are consistent with studies showing an increase in apoptosis with tamoxifen in MCF-7 xenograft models.15 The results with tamoxifen from the present study do not strengthen, but are nonetheless consistent with, the earlier observation of a small increase in apoptosis after 2 weeks. The fall in apoptosis after 2 weeks of vorozole is in substantial contrast to the earlier data on other endocrine therapy and in particular with xenograft studies of estrogen deprivation, where a four-fold increase of apoptosis was seen.15-17 There are several potential explanations for this unexpected finding.
It has been noted by several authors that there is a direct relationship between proliferation and apoptosis in breast carcinomas, suggesting that there may be mechanistic regulation between the two processes. This is strongly supported by more definitive molecular studies in model systems showing c-myc to be a potent inducer of both proliferation and apoptosis.31 Thus in circumstances of profound suppression of proliferation (as with vorozole), apoptosis may be downregulated.
Another possible explanation for the difference in the apoptotic effects seen with the two drugs may lie in their different effects on IGF1 levels. IGF1, as well as being mitogenic, is an important cell survival factor for many cell types.22,32 Our findings are consistent with previous studies reporting that tamoxifen reduces IGF-1 levels.33 Previous clinical studies have shown an increase in IGF-1 levels during treatment with several aromatase inhibitors.33-36 Our data on vorozole are consistent with this, albeit that the increase seen in this study was not statistically significant. The complexity of the synergistic relationship between estrogen and the IGF family in breast cancer cells suggests that it is simplistic to consider that IGF1 levels alone explain the difference in apoptosis. However, additional detailed investigation of the IGF signaling pathway as well as of regulators of apoptosis, such as the Bcl2 family, in tissues derived in this study and in other similar ongoing studies may provide a mechanism for this difference.
The trend toward the reduction in Ki67 at 2 weeks being correlated with 12-week tumor volume reduction with vorozole and the significant association of change in Ki67 with clinical response to vorozole are consistent with our previous unrandomized data showing that such Ki67 reductions at 2 weeks were significantly related to clinical response to tamoxifen.20,37 It would seem that these relationships are not close; establishing the power of the change in Ki67 to predict for response will require substantially larger series. The greater number of responders to tamoxifen than vorozole in this series may be in part attributable to the opposing differential effects on apoptosis. A first approximation to assessing this interaction may be attempted by the calculation of changes in the Ki67 to AI ratio over time. In this study, the effect on this ratio was similar for each agent. Overall, the data indicate that changes in proliferation predominate in determining response to both tamoxifen and the aromatase inhibitor. Although change in proliferation may prove to be a useful intermediate end point, it seems likely that the drug- and time-related effects of hormonal therapy on apoptosis will preclude this being possible with AI. It is possible that assessment of changes in proliferation and apoptosis might allow the optimal duration of therapy to be identified in individual patients, but it would require a different study design to evaluate this.
Steroid receptors are widely used in clinical practice to predict patients most likely to respond to endocrine therapies. Significant reductions occurred in ER expression in both groups by 12 weeks. This is consistent with the data of Bajetta et al19 using tamoxifen treatment. The results of estrogen deprivation using vorozole differ, however, from the data seen in vitro and in xenograft models with MCF-7 cells, where increased ER expression is seen after estrogen withdrawal.17 It is unclear why the clinical and model system data differ in this regard, but it may be that treatment-induced changes in the cell cycle may themselves influence ER expression. Our earlier work with tamoxifen has reported rises in PgR at 14 days,20,37 which is presumed to reflect the initial agonist effect of tamoxifen.38 The present study also found an early, but again nonsignificant, increase in PgR. This was in marked contrast to the vorozole group, where there was a marked and persistent fall in PgR levels. This is consistent with expectations of the deprivation of estrogen signaling to this estrogen-dependent protein. This contrast emphasizes the biologic differences achieved with these different types of agent.
Serum hormone results obtained in this study are consistent with previous work in breast cancer. The significant reductions achieved in E1, E2, and E1S with vorozole have been reported frequently in many phase II trials of this and other aromatase inhibitors.1,7 All patients showed good estrogen suppression, indicating good compliance and that the biologic effects can therefore be interpreted with confidence. An increase in E1S in the absence of any effect on E1 and E2 has previously been reported in tamoxifen-treated patients.39
Suppression of LH and FSH and increased SHBG levels have previously been reported with tamoxifen therapy, mostly in advanced breast cancer trials,39 and are generally interpreted as reflecting its partial or selective estrogen agonist activity. We and other investigators have previously reported reduction of SHBG levels and a rise in LH and FSH with vorozole treatment, but in both studies the effects were attributed to a wash-out effect from previous tamoxifen treatment.40,41 In the present study similar results for SHBG and FSH were observed after vorozole treatment in a circumstance in which there was no prior tamoxifen. It would therefore seem that these changes are effects of vorozole itself.
The measurement of lipids in this study was made on nonfasting samples, which is likely to reduce the precision of any estimated changes. Nonetheless, previously reported reductions in both LDL and total cholesterol with tamoxifen42 were confirmed and indicate that the measurements made were valid. The few data which exist on the effect of aromatase inhibitors on lipids indicate that, consistent with our data, there are no significant changes in lipids that would be of concern in relation to cardiovascular risk.43,44 To detect any subtle changes would require larger studies than those conducted to date.
Maintenance of bone density in postmenopausal women with tamoxifen45 is regarded as a substantial advantage, particularly in the prevention setting and in women with low risk of relapse. Serum CTx, a degradation product of type 1 collagen, was measured using unfasted samples, which are associated with circadian changes with a nadir plateau extending from 10 am to 4 pm.46 Our patients had their samples taken within this nadir. The significant reduction in levels of CTx seen in this study with tamoxifen is in keeping with its bone-protective effects. In contrast, there is concern that the profound estrogen deprivation achieved with third-generation aromatase inhibitors will have detrimental effects on bone turnover.47,48 There are no data on the effects of bone density with aromatase inhibition. Although there was an upward trend in CTx levels with vorozole, this was not statistically significant, and this may indicate that any induced bone loss may be limited.
In conclusion, it is clear that these two widely used methods of endocrine therapy show distinct differences in their effects on blood and tumor biomarkers. In this series, the lack of significant changes in cholesterol or LDL and in the degradation product of type 1 collagen (CTx) are relatively reassuring in relation to the large ongoing trials of aromatase inhibitors in the adjuvant setting and for any future evaluation as potential preventive agents, although the relatively small number of patients studied did not allow subtle changes to be detected. Response-related changes in Ki67 after 2 weeks support this marker being a valid intermediate end point of clinical response for hormonal therapy; much larger studies are required to assess its predictive power. It will be important to determine the mechanism underlying the difference in the early effects of the agents on apoptosis, such that its potential influence in combination therapy may be evaluated.
Acknowledgments
Jannsen-Cilag Ltd sponsored this study before commercial decisions resulted in vorozole no longer being marketed.
ACKNOWLEDGMENT
We thank Drs Valerie Thomas, S. Al-Sam, Paul Conn, and Chris Smith and their teams. We also thank the research nurses and data managers, especially Kathy Rooke and Wendy Nicholas.
- Received July 27, 2000.
- Accepted October 5, 2001.
