Arimidex is the brand name of anastrozole, a non-steroidal aromatase inhibitor. In trials, it has served three different purposes: a treatment to test against tamoxifen, a reference treatment for new endocrine combinations, and a way to study how tumours respond when oestrogen production falls. The measurements range from blood oestrogen concentrations to biopsies and long-term recurrence. Each tells researchers something different about whether, and for how long, a tumour depends on oestrogen. 1, 2, 4, 14
Glasgow's contribution to this literature extends beyond breast cancer. The University of Glasgow research repository records the PARAGON publication on two rare uterine cancers, leiomyosarcoma and carcinosarcoma. The report lists David Millan and Karen Carty among the Glasgow co-authors. The investigators used the same endocrine drug in both cohorts and reported each group separately. Those small cohorts are particularly instructive because a superficially encouraging percentage can represent just three patients whose disease stabilised. 10, 11
Mechanism and clinical pharmacology in trial protocols
After menopause, peripheral conversion of androgens through aromatase supplies much of the body's oestrogen. Anastrozole inhibits that conversion. Tamoxifen acts at the oestrogen receptor, while fulvestrant reduces receptor availability. A study can therefore hold oestrogen production down, act at the receptor, or combine the two approaches. In breast cancer that has already progressed during aromatase inhibition, the rationale for combining these mechanisms depends on why the tumour remained active. A study of a first endocrine treatment enrols patients at a different point in that process. 1, 4, 6
In a crossover study from the Edinburgh Breast Unit and collaborators, 54 postmenopausal patients received both anastrozole and letrozole in random order. The design let the investigators compare two treatments within the same patient, reducing the effect of differences in baseline oestrogen between individuals. Plasma oestradiol remained at or above 3 pmol/L in 20 of 54 patients after anastrozole and one of 54 after letrozole. The investigators also reported estimates below the assay's detection threshold. Those estimates depended on extrapolation, a distinction that matters when the concentrations under comparison sit close to an assay's lower limit. 2
A prospective study published in 2024 asked whether measurable oestrogen can persist even when patients take the assigned medicine. Of 132 patients who met the study's adherence criterion on standard-dose anastrozole, 40 met its definition of inadequate oestrogen suppression. Among 25 adherent patients in the higher-dose phase, 19 reached the study's biochemical target. The second denominator describes a selected group that entered and adhered to the next phase. The investigators measured blood hormones and drug concentrations. Testing whether the change prevented recurrence or prolonged survival would require a separate outcomes trial. 3
These two studies make an important distinction for protocol design. An individual result below an assay threshold can be useful evidence of drug effect, while the proportion below that threshold can distinguish treatment groups. Neither measurement tells the investigator how much tumour proliferation changed. Circulating oestrogen, receptor activity in a biopsy and time to recurrence belong at different points in the causal chain. Data on drug exposure and adherence also help avoid classifying missed treatment as biological resistance. 2, 3, 4
Breast oncology - presurgical response and endocrine resistance
Ovarian suppression in the STAGE trial
STAGE enrolled premenopausal women with operable, oestrogen-receptor-positive, HER2-negative breast cancer. Both groups received goserelin to suppress ovarian function. The difference between their regimens was the choice of anastrozole or tamoxifen. After the 24-week presurgical course, 69 of 98 patients in the anastrozole group had a complete or partial response, compared with 50 of 99 in the tamoxifen group. The 19.9-percentage-point difference applies to response during the treatment window in the presence of ovarian suppression. The trial did not compare the drugs as monotherapy in patients with active ovarian function. 5
The difference in response is useful in planning another presurgical study because it tells a researcher what change the trial could observe by the time of surgery. It cannot answer how many patients avoid distant relapse years later. That question requires later follow-up and accounting for treatments after surgery. In the older ATAC comparison, a different population, postmenopausal patients who had already undergone surgery, provided that longer view. Keeping the treatment setting and the timing of the endpoint beside each result prevents a clinical response percentage from being mistaken for an adjuvant outcome. 5, 14
ALTERNATE - tissue evidence before and at surgery
ALTERNATE evaluated six months of presurgical endocrine therapy in postmenopausal patients with stage II or III, oestrogen-receptor-rich, HER2-negative breast cancer. Anastrozole alone was the reference arm. The other two arms received fulvestrant alone or both drugs. At planned biopsy points, the investigators measured Ki-67, a marker of cell proliferation. A persistently high result could prompt an early move to chemotherapy or surgery under the protocol. The primary endpoint combined a complete pathological response with a modified low-risk category that considered residual disease and proliferation at surgery. 4
The trial randomised 1,362 participants. The published endocrine-sensitive disease rates were 18.7% with anastrozole, 22.8% with fulvestrant and 20.5% with both. Neither experimental arm showed superiority for the primary endpoint. That result is more informative than a count of tumours that shrank on examination. A tumour may shrink and still have substantial proliferation or residual disease at surgery. Conversely, a tumour can show reduced proliferation without disappearing within the presurgical window. The trial's definition forces the reader to ask what happened in the tissue and what remained at operation. 4
SoFEA and the blood samples collected before treatment
SoFEA took the question to a later stage, enrolling postmenopausal patients with locally advanced or metastatic hormone-receptor-positive breast cancer after progression on a non-steroidal aromatase inhibitor. The trial randomised 723 patients to fulvestrant with anastrozole, fulvestrant with placebo, or exemestane. Median progression-free survival was 4.4 months with the combination and 4.8 months with fulvestrant plus placebo. The hazard ratio was 1.00 (95% confidence interval [CI] 0.83-1.21). Continuing oestrogen synthesis inhibition alongside receptor-directed treatment did not improve that endpoint in the overall comparison. 6
A subsequent analysis used archived baseline plasma to ask whether tumour DNA carrying mutations in ESR1, the gene for oestrogen receptor alpha, could separate groups with different outcomes. It detected mutations in 39.1% of the assessable SoFEA plasma samples. In the mutation-positive subgroup, the hazard ratio for progression or death was 0.52 for pooled fulvestrant-containing regimens versus exemestane. The pool included arms with and without anastrozole. That design can suggest a difference between receptor-directed therapy and another aromatase inhibitor after prior treatment. It cannot identify an added benefit from anastrozole within the pooled group. Only participants with usable samples contributed to the biomarker comparison. 6, 7
Breast cancer studies by clinical setting
| Study | Population and stage | Comparison | Published result |
|---|---|---|---|
| STAGE | Premenopausal patients with operable ER-positive, HER2-negative breast cancer | Anastrozole or tamoxifen before surgery, with goserelin in both groups | Complete or partial response in 69/98 (70.4%) versus 50/99 (50.5%). |
| ALTERNATE | Postmenopausal stage II-III ER-rich, HER2-negative disease | Anastrozole, fulvestrant or both before surgery | Endocrine-sensitive disease rates of 18.7%, 22.8% and 20.5%, respectively. Neither experimental arm met the superiority objective. |
| SoFEA | Advanced hormone-receptor-positive breast cancer after non-steroidal aromatase inhibition | Fulvestrant plus anastrozole versus fulvestrant plus placebo, with a separate exemestane arm | Median progression-free survival 4.4 versus 4.8 months for the first two arms. Hazard ratio 1.00. |
| ATAC | Postmenopausal early breast cancer after surgery | Adjuvant anastrozole versus tamoxifen | Disease-free survival hazard ratio 0.86 (95% CI 0.78-0.95) in the hormone-receptor-positive subgroup at ten years. |
Gynaecological oncology - three distinct PARAGON populations
PARAGON used anastrozole across several cohorts of patients whose recurrent or metastatic gynaecological tumours expressed oestrogen receptors, progesterone receptors, or both. Entry criteria differed by cohort. In the low-grade ovarian and serous borderline cohort, 23 of 36 participants met the clinical benefit definition at three months. Five achieved a partial response and 18 had stable disease as their best reported response. The distinction matters in tumours that may grow slowly. Disease stability is relevant to patients and researchers, but a single-arm study cannot show how many tumours would have remained stable without the drug. 8
Another PARAGON report enrolled 54 patients, 52 of whom were assessable, after a rise in the blood marker CA125 following earlier ovarian cancer chemotherapy. The patients were asymptomatic, and the investigators enrolled them at a time when chemotherapy was not clinically indicated. Clinical benefit at three months occurred in 18 of 52, or 34.6%. Median progression-free survival for the cohort was 2.7 months. High-grade serous cancer accounted for 40 of the 54 patients, so its case mix differed from the low-grade and borderline cohort. A simple comparison of the two clinical benefit percentages would mix differing histologies, entry criteria and ways of measuring disease. 8, 9
The uterine analysis reported 32 patients with leiomyosarcoma and seven with carcinosarcoma. At three months, investigators reported 35% clinical benefit in leiomyosarcoma and 43% in carcinosarcoma. Yet the latter percentage arose from three cases of stable disease and no objective responses. Its confidence interval ran from 16% to 75%. Leiomyosarcoma produced one partial response and a median progression-free survival of 2.8 months. These results show why the original count, the tumour type and the uncertainty interval belong beside any percentage from a rare cancer cohort. The University of Glasgow repository provides the study record alongside the journal publication. 10, 11
In a follow-on trial, receptor expression would be an entry criterion or a candidate biomarker. To distinguish those roles, investigators would need to know when the tissue came from, how the pathologist scored staining, and whether the laboratory examined a primary or recurrent lesion. Clinical benefit would also need a clear minimum duration for stable disease. Otherwise, changes in case selection and reporting could create apparent differences between studies that tested the same drug. The PARAGON publications show how important it is to keep these details attached to each cohort. 8, 9, 10
The asymptomatic recurrence cohort also tested whether stronger oestrogen receptor staining identified patients more likely to benefit. Its exploratory analysis found no relationship between staining intensity or receptor histoscore and either clinical benefit or progression-free survival. This finding puts a limit on what the entry criterion means. Selecting patients because their tumours express a receptor can make a trial biologically coherent. It takes a convincing difference across measured levels or a treatment-by-marker comparison to show that the score can guide selection within the enrolled group. 9
The published uterine cohorts provide a parallel lesson about follow-up. Two participants with leiomyosarcoma remained on treatment longer than a year even though the cohort median progression-free survival was under three months. A median describes the point at which half of the group has had an event. It does little to describe a small group with prolonged disease control. Reporting individual durations alongside the median can help investigators identify questions for another study, such as whether those patients shared clinical or tumour features. In small single-arm cohorts, these observations generate hypotheses for testing. 10
PARAGON cohorts and what they measured
| Publication and tumour group | Assessment population | Clinical benefit at three months | Objective response or context |
|---|---|---|---|
| PARAGON low-grade ovarian and serous borderline tumours | 36 participants | 23/36 (64%) | Five partial responses. The clinical benefit definition included stable disease. |
| PARAGON asymptomatic ovarian recurrence | 52 assessable of 54 enrolled after CA125 progression | 18/52 (34.6%) | High-grade serous disease accounted for 40/54 patients. Median progression-free survival 2.7 months. |
| PARAGON uterine leiomyosarcoma | 32 participants | 35% (95% CI 21-53%) | One partial response. Median progression-free survival 2.8 months. |
| PARAGON uterine carcinosarcoma | Seven participants | 43% (95% CI 16-75%) | Three patients had stable disease. No objective responses. |
Translational markers, safety and trial follow-up
Different biomarkers gain meaning through the way a protocol uses them. In ALTERNATE, the biopsy result could change what happened next, so treatment after an unfavourable result forms part of the clinical design. In SoFEA, the later analysis examined stored blood collected before randomised treatment and asked a question about drug resistance. In PARAGON, hormone receptor expression qualified patients for enrolment in selected cohorts. Those three roles, treatment triage, exploratory prediction and eligibility, each require different evidence. A promising result in one setting does not automatically validate the marker for another use. 4, 7, 8
The SABRE trial addressed a specific safety question. In the moderate-fracture-risk group, researchers randomised 154 postmenopausal patients receiving adjuvant anastrozole to risedronate or placebo. At 24 months, lumbar spine bone mineral density rose by 2.2% with risedronate and fell by 1.8% with placebo. Total hip density changed by +1.8% and -1.1% respectively. Researchers measured bone mineral density by site and at a defined time point. That measurement can reveal a difference in skeletal effects before a trial accrues enough clinical fractures to compare fracture incidence reliably. 12
Long-term protocols need a record of actual exposure as well as assigned treatment. A Tayside record-linkage cohort examined prescribed tamoxifen or aromatase inhibitors in 3,361 women and included a collaborator from the University of Glasgow's Robertson Centre for Biostatistics. Adherence below 80% was associated with poorer survival (hazard ratio 1.20, 95% CI 1.03-1.40). The study grouped several endocrine treatments and used prescription records to estimate adherence. It gives Scottish evidence about treatment persistence across a service, with the limitations of a non-randomised comparison and dispensing data. 13
For a protocol that tracks both tumour outcomes and toxicity, the sequence of events is as important as the event count. A patient can stop treatment because of joint symptoms, sustain a fracture after a switch, or have a recurrence after several interruptions. Recording dates and reasons makes it possible to read those events in the context of exposure. The ten-year ATAC report separated fractures during treatment from those in the later follow-up period. The excess in recorded fractures with anastrozole was present during treatment and narrowed after it stopped, showing what is lost when the two periods are merged. 12, 13, 14
Laboratory, safety and follow-up measures
| Study | Measurement | Finding | What the measurement establishes |
|---|---|---|---|
| Edinburgh crossover | Plasma oestradiol threshold | 20/54 at or above 3 pmol/L after anastrozole, versus 1/54 after letrozole. | Comparison of hormone suppression within the same patients. |
| Prospective exposure study | Inadequate oestrogen suppression on standard-dose anastrozole | 40/132 adherent participants met the study threshold. | Residual hormone production can occur despite measured adherence. |
| SoFEA plasma analysis | ESR1 mutations in assessable baseline plasma | Mutations found in 39.1% of the assessable samples. | Description of a biomarker subgroup after prior endocrine treatment. |
| SABRE | Lumbar spine bone density after 24 months | +2.2% with risedronate and -1.8% with placebo during adjuvant anastrozole. | A site-specific skeletal effect in the randomised moderate-risk group. |
| Tayside cohort | Survival and adherence to prescribed endocrine therapy | Adherence below 80% associated with poorer survival (hazard ratio 1.20). | An observational association across several endocrine treatments. |
References
- Arimidex UK summary of product characteristics.
- Dixon and colleagues, Journal of Clinical Oncology, 2008. Edinburgh crossover study of plasma oestrogen suppression.
- Clinical Cancer Research, 2024. Prospective anastrozole exposure and oestrogen study.
- Ma and colleagues, JAMA Oncology, 2024. ALTERNATE phase III trial.
- Masuda and colleagues, Lancet Oncology, 2012. STAGE presurgical trial.
- Johnston and colleagues, Lancet Oncology, 2013. SoFEA phase III trial.
- Fribbens and colleagues, Journal of Clinical Oncology, 2016. Plasma ESR1 analysis in SoFEA and PALOMA-3.
- Tang and colleagues, Gynecologic Oncology, 2019. PARAGON low-grade ovarian and serous borderline cohort.
- Kok and colleagues, Journal of Gynecologic Oncology, 2019. PARAGON asymptomatic ovarian recurrence cohort.
- Gynecologic Oncology, 2021. PARAGON uterine leiomyosarcoma and carcinosarcoma cohorts.
- University of Glasgow. Research record for the PARAGON uterine cohorts.
- Van Poznak and colleagues, Journal of Clinical Oncology, 2010. SABRE randomised bone density study.
- Makubate and colleagues, British Journal of Cancer, 2013. Tayside cohort of adjuvant endocrine therapy adherence.
- Cuzick and colleagues, Lancet Oncology, 2010. ATAC ten-year follow-up.
