Part XI · Hormone receptor-positive, HER2-negative disease · Chapter 62
Neoadjuvant therapy
What preoperative treatment settles in this subtype once pathological complete response has stopped carrying the argument.
1 · Scope, and why the neoadjuvant setting is harder in this subtype
This chapter covers preoperative systemic therapy for operable and locally advanced hormone receptor-positive, HER2-negative breast cancer. Hormone receptor-positive HER2-positive disease is a different entity with its own algorithm. It has its own chapter in Early triple-positive disease.
The organising fact is uncomfortable. Everywhere else in early breast cancer the neoadjuvant setting earns its place because a pathological complete response is both achievable and informative. Here it is neither, because complete responses in this subtype are uncommon. The pooled analysis that established the prognostic value of the endpoint found the association weakest in exactly this group1. A separate pooled analysis of seven randomised trials found no association at all between complete response and disease-free survival in luminal A tumours2.
Preoperative treatment here is therefore an experiment at least as much as a treatment. Nearly every endpoint read before the operation is a proliferation measurement standing in for an outcome. Nearly every question in this chapter turns on whether that substitution holds.
Each section below opens with the question its trials were built to answer, taken from the [Chapter Stories](stories.html) map, and closes with a table filtered on that same question. One warning belongs at the front rather than at the back. A table filtered on this subtype also prints every trial that enrolled without restriction by receptor status, because such a trial legitimately contained this cell. The rate printed on those rows is a mixed-population rate. Which rows are receptor-restricted and which are not is argued in each section, and it changes what each row is worth.
2 · Neoadjuvant chemotherapy and the low pathological complete response rate
Early disease · HR+/HER2- · neoadjuvant
The preoperative window as a measurement and as a treatment
Can the tumour's response to preoperative treatment be used to withhold chemotherapy?
Rationale
Most of this cell gains little from cytotoxic treatment. The difficulty is telling who. A baseline genomic score infers endocrine dependence from the untreated tumour. A short preoperative course tests it instead. A tumour whose proliferation collapses under an aromatase inhibitor has demonstrated the dependence the score was trying to predict.
Experiment
Response-adapted designs: a short endocrine run-in, a central Ki67 read against a prespecified threshold, and a route to chemotherapy for the tumours that stay above it. One programme combined that window with a genomic score and tested whether the group given endocrine therapy alone was noninferior. Its successor randomises a CDK4/6 inhibitor against chemotherapy after the same kind of window, with invasive disease-free survival as the endpoint. Others dropped chemotherapy in favour of a CDK4/6 inhibitor and compared what was left at surgery, by molecular risk score or by cell-cycle arrest. One trial now recruiting carries the same idea through to distant metastasis-free survival.
Finding
Response does separate the population, and in the direction the biology predicts. Patients with an intermediate genomic score whose tumours responded in the window did as well as the control group whose score was low to begin with. Only a minority of tumours reach the endocrine-sensitive state these designs define, whichever endocrine drug is used to produce it. The trials that replaced chemotherapy with a CDK4/6 inhibitor reached much the same molecular and proliferative state at surgery as chemotherapy did, at a lower cost in serious adverse events. The randomised trial that would settle the substitution on invasive disease-free survival has not reported.
Limitation
In the response-guided programme the groups were allocated by score and by response rather than by randomisation. The comparison is therefore between two differently selected populations and not between two ways of treating the same patient. Its window is preoperative. The therapy decision it drives is adjuvant. It does not test preoperative treatment at all. In the chemotherapy-replacement trials the arms produced similar estimates. Similar estimates are not noninferiority. None of those trials carried a noninferiority margin or the size to test one. The trial powered to show that endocrine and CDK4/6 treatment arrests the cycle in far more tumours than chemotherapy did not show it. It also randomised only the tumours whose proliferation stayed up after an endocrine run-in, a deliberately endocrine-resistant group and a small fraction of those registered. The whole family of designs turns on a Ki67 threshold that is read centrally in the trials and locally in clinic. The successor trial's own concordance analysis found that the two readings of one specimen do not always agree. The trial built to settle the question on distant control has published a design paper and nothing else. Nothing on its record is a result.
Next question
If the endocrine backbone sets the ceiling on how many tumours respond, would a better receptor-directed drug raise it?
Two separate facts with different causes and different consequences are usually run together here.
The first is that complete responses are uncommon. That is a property of the disease rather than of the regimen. Slowly proliferating, oestrogen-driven tumours are not eradicated by cytotoxic therapy at the rate that highly proliferative ones are. Intensifying the regimen moves the rate a little. It does not move it into the range where a binary endpoint discriminates usefully between patients.
The second is that, within this subtype, achieving a complete response separates outcomes less sharply than it does elsewhere1. The two facts compound each other. Few patients reach the category, and reaching it says less about them than it would in triple-negative disease.
Most of the rows below were not generated in this cell. NSABP B-18, NSABP B-27, GeparSepto and I-SPY 2 enrolled without restriction by receptor status. ETNA, GeparOLA, INFORM and the I-SPY 2 pembrolizumab arm enrolled HER2-negative disease of both receptor states. Two of those selected on germline BRCA status or on homologous recombination deficiency rather than on the receptor at all. Only CheckMate 7FL, KEYNOTE-756, NeoPAL and CORALLEEN required hormone receptor positivity to enter. That distinction, and not the regimen, explains most of the spread in the result column.
Where the randomised record agrees is on timing. Giving the same chemotherapy before the operation rather than after it changes the operation and leaves disease-free and overall survival where they were. NSABP B-18 is the trial that decision rests on, because it randomised the sequence and nothing else3,4. NSABP B-27 then added the finding that makes this chapter necessary. Preoperative docetaxel doubled the pathological complete response rate without moving disease-free or overall survival5. A response endpoint moved, and the outcome it is supposed to stand for did not.
The two nab-paclitaxel trials look like a disagreement about a drug and are not one. GeparSepto / GBG 69 found nab-paclitaxel ahead of solvent-based paclitaxel. ETNA found the two indistinguishable. GeparSepto enrolled every subtype, including HER2-positive disease treated with dual antibody blockade, and scored pathological complete response as ypT0 ypN0. ETNA enrolled HER2-negative disease only and scored ypT0/is ypN0, a definition that counts residual in situ carcinoma as a response. The nab-paclitaxel dose differed between them, and GeparSepto reduced its own dose partway through accrual6,7. Population, response definition and dose all differ. The drug does not.
The repair-deficiency rows behave as one finding. INFORM substituted cisplatin for doxorubicin and cyclophosphamide in germline BRCA carriers8. GeparOLA substituted olaparib for carboplatin alongside paclitaxel in tumours selected for homologous recombination deficiency9. Neither substitution met the endpoint it was built to test. In GeparOLA the long-term invasive disease-free and overall survival estimates ran against the experimental arm rather than with it10. Both trials were majority triple-negative. Neither carries a receptor-positive result beyond a subgroup analysis.
That leaves four rows whose chemotherapy arm was given to a receptor-restricted, luminal population. Those control arms are the honest estimate of what cytotoxic therapy achieves at surgery in this cell, and they are the column to read first. I-SPY 2 is the one row that connects the endpoint to an outcome within each molecular subtype. The association was similar across subtypes, although tumours of this subtype with a low 70-gene score were excluded11.
The harder version of the question is whether response can be used to withhold chemotherapy altogether. Two families of trial have tried and they did not reach the same standard of proof. Response-guided omission has a noninferiority result behind it, from a programme that allocated by genomic score and by endocrine response rather than by randomisation12. Its window is preoperative and the therapy it decides is adjuvant, so it does not test preoperative treatment at all. Substituting a CDK4/6 inhibitor for chemotherapy has mostly similar estimates behind it, read on endpoints scored at surgery. Similar estimates are not noninferiority, and none of those trials carried a noninferiority margin Designing de-escalation trials and the non-inferiority margin problem. NEOLBC is the one substitution trial powered against chemotherapy on a prespecified difference. It randomised ribociclib with letrozole against chemotherapy in the tumours still proliferating after a two-week letrozole run-in. It was designed to show that the combination doubles the rate of complete cell-cycle arrest in the surgical specimen, and it did not13. The tumours it randomised, those with a Ki67 of 1% or more on letrozole, are the group least likely to produce such a difference.
What the table cannot show is the comparison that matters most. No row randomised chemotherapy against no chemotherapy in this subtype. No column reports the distribution of residual disease rather than its absence, and residual cancer burden is continuous and remains prognostic here with a shallower gradient than in triple-negative disease14,15. The index and its measurement caveats are set out in Residual cancer burden and continuous response measures. A report giving only the presence or absence of a complete response has discarded the part of the measurement that still carries information. No column records whether the operation changed, which is the reason chemotherapy is given before surgery here in the first place.
Caution
A low complete response rate is routinely read as chemoresistance, which is a different claim. Adjuvant chemotherapy reduces recurrence in selected hormone receptor-positive disease, in populations where almost nobody would have achieved a complete response had they been treated preoperatively Chemotherapy selection when it is indicated. Sensitivity measured by eradication and benefit measured by recurrence are different quantities. The first systematically understates the second whenever the tumour grows slowly.
Neoadjuvant trials in this subtype carrying a chemotherapy arm.
Read the control arms first, because they are where the response rate of chemotherapy alone in this cell is recorded.
ribociclib plus an aromatase inhibitor, with a gonadotropin-releasing hormone agonist in premenopausal women, vs chemotherapy followed by endocrine therapy
invasive disease-free survival
not yet reported; local against central agreement for post-endocrine Ki67 at or below 10% was 80.5%, 95% CI 79.0 to 82.1
pembrolizumab plus neoadjuvant paclitaxel then anthracycline-cyclophosphamide vs placebo plus the same chemotherapy, each followed by adjuvant pembrolizumab or placebo and endocrine therapy
pathological complete response and event-free survival
pCR 24.3% vs 15.6%, difference 8.5 percentage points, 95% CI 4.2 to 12.8, P = 0.00005; event-free survival immature at this analysis
Paclitaxel 80 mg/m2 weekly plus olaparib 100 mg twice daily for 12 weeks vs paclitaxel plus carboplatin AUC 2 weekly for 12 weeks, both followed by epirubicin and cyclophosphamide.
Pathological complete response, ypT0/is ypN0, tested by a two-sided one-group chi-square test planned to exclude a pCR rate of 55% or lower in the paclitaxel plus olaparib arm.
pCR 55.1%, 90% CI 44.5 to 65.3, with paclitaxel plus olaparib against 48.6%, 90% CI 34.3 to 63.2, with paclitaxel plus carboplatin. The trial could not exclude a pCR rate of 55% or lower in the olaparib arm, so the primary endpoint was not met.
Preoperative cisplatin 75 mg/m2 every 3 weeks for four doses versus doxorubicin 60 mg/m2 with cyclophosphamide 600 mg/m2 every 2 to 3 weeks for four doses, each followed by surgery.
Pathological complete response (ypT0/is, N0) confirmed by central review, the trial asking whether pathological complete response was at least 20% higher with cisplatin than with doxorubicin-cyclophosphamide.
Pathological complete response was 18% with cisplatin and 26% with doxorubicin-cyclophosphamide, risk ratio 0.70 (90% CI 0.39 to 1.2). Residual cancer burden 0 or 1 was 33% with cisplatin and 46% with doxorubicin-cyclophosphamide, risk ratio 0.73 (90% CI 0.50 to 1.1).
pembrolizumab added to weekly paclitaxel then doxorubicin and cyclophosphamide vs the same chemotherapy, adaptively randomised against a shared control
pathological complete response, with Bayesian graduation
estimated pCR 60% with pembrolizumab vs 22% in triple-negative disease, 44% vs 17% across HER2-negative disease
Weekly nab-paclitaxel versus weekly solvent-based paclitaxel, each followed by epirubicin plus cyclophosphamide.
Pathological complete response, defined as ypT0 ypN0.
Pathological complete response occurred in 38% of patients given nab-paclitaxel (233 of 606) versus 29% given solvent-based paclitaxel (174 of 600), odds ratio 1.53, 95% CI 1.20 to 1.95, p=0.00065.
Preoperative AC then surgery versus preoperative AC then docetaxel then surgery versus preoperative AC then surgery then docetaxel.
Disease-free survival and overall survival, with pathologic complete response.
Adding docetaxel to preoperative doxorubicin and cyclophosphamide did not significantly change disease-free survival or overall survival, although preoperative docetaxel doubled the pathologic complete response rate and reduced local recurrence as a first event (P=.0034).
Four cycles of doxorubicin and cyclophosphamide before surgery versus the same chemotherapy after surgery.
Disease-free survival and overall survival.
Disease-free survival and survival did not differ between preoperative and postoperative chemotherapy (P=.99 and P=.83 at five years; 55 versus 53 percent disease-free survival at nine years, P=.50).
3 · Neoadjuvant endocrine therapy and who it suits
The preoperative window as a measurement and as a treatment, continued
Does a better receptor-directed drug improve the tumour's early biological response more than an aromatase inhibitor?
Rationale
The first preoperative comparisons put an aromatase inhibitor against tamoxifen, in postmenopausal women and, with ovarian suppression, in premenopausal women. They are what established preoperative endocrine therapy as something worth doing at all. Changing between the aromatase inhibitors does not push the receptor axis any further, because they behave alike. Degrading the receptor rather than starving it should shut down more signalling. It should also keep working where the receptor is mutated.
Experiment
Short preoperative windows of two or four weeks, each ending in a central Ki67 score. One randomised an oral receptor degrader against anastrozole before putting both groups on the same combination. One randomised two doses of a second oral degrader against letrozole. One was a single arm read against an arrest threshold. Against these sit the older tamoxifen comparisons, which scored tumour size rather than proliferation, and a full-length preoperative phase III trial of an injectable degrader given alone and added to an aromatase inhibitor, with a prognostic composite scored at surgery.
Finding
The older comparisons did not all run one way. The aromatase inhibitor gave more tumour response than tamoxifen in the trial that established the approach, and again in premenopausal women treated with ovarian suppression. In two other preoperative comparisons the response rates were close and the difference was not significant. Among the newer agents, one oral degrader suppressed proliferation more than anastrozole over two weeks. The other matched letrozole without beating it. The single-arm study arrested the cycle in a minority of the tumours it treated. Over a full preoperative course the injectable degrader did not improve on an aromatase inhibitor on the trial's own prognostic endpoint, given alone or added to it.
Limitation
A window study is small because a short exposure is all its design asks for. Its endpoint is a ratio of two Ki67 scores. That ratio measures what the drug does to proliferation. It has never been shown to buy a lower recurrence rate. A better ratio is a better ratio and nothing more. The comparisons are also not on one footing. The older agent trials scored tumour response by callipers and by ultrasound. The two methods gave different response rates inside the same trial. One of those trials permitted concurrent chemotherapy, which contaminates the endocrine comparison. One window study stopped early for slow accrual. In another the primary comparison rests on the subset with paired central scores rather than on everyone randomised. The phase III trial enrolled a broader population and read a different endpoint entirely. It does not settle the window question. Neither do the window results overturn it.
Next question
If the ceiling is not set by which receptor drug is given, is it set by the tumour's histology instead?
P024IMPACTPROACTSTAGEcoopERA Breast CancerAMEERA-4ELIPSEALTERNATE
Does invasive lobular carcinoma need its own preoperative endocrine experiment?
Rationale
Lobular tumours lose E-cadherin and infiltrate in single files rather than forming a discrete mass. They are almost always receptor-positive and low grade. They respond poorly to cytotoxic treatment. Both halves of a neoadjuvant argument therefore change. Endocrine therapy becomes the plausible option rather than one of two. A tumour with no discrete mass cannot be measured by callipers or by imaging the way a ductal tumour can.
Experiment
A randomised short preoperative exposure to tamoxifen, to an aromatase inhibitor or to fulvestrant in lobular disease, read out as target engagement rather than as tumour response. Separately, a randomised preoperative trial of adding a CDK4/6 inhibitor to endocrine therapy that enrolled receptor-positive disease of any histology, so that lobular and ductal tumours could be compared inside one trial.
Finding
Neither has reported the comparison it was built to make. The short-exposure study has conference and registry material only. Nothing on its record may be read as a published outcome. The only publication on the randomised trial's record is its axillary analysis, which found that clearing the axilla added nothing over sentinel node biopsy in node-positive disease after preoperative endocrine therapy.
Limitation
The short-exposure study is a phase zero design. It measures whether a drug reaches and engages its target. It cannot rank endocrine agents for efficacy. Reading it that way is the error its design invites. The randomised trial is named for lobular cancer but enrolled any histology. Lobular tumours were a minority of the patients in its published analysis. That report is a comparison inside a mixed population rather than a lobular trial. The endpoints of that publication are ones the paper itself calls prespecified exploratory. They say nothing about the trial's systemic question. No other trial at this position enrolled a lobular-only population. The histology argument rests on these two and on subgroups borrowed from elsewhere.
Next question
If neither escalating the receptor drug nor separating the histology has raised the response, can a different mechanism raise it?
TBCRC 037PELOPS
Neoadjuvant endocrine therapy is the one preoperative strategy in this subtype whose rationale never depended on a pathological endpoint.
The six randomised comparisons agree on two things. Three to six months of endocrine therapy produces measurable regression in a substantial minority of receptor-rich tumours, enough to convert mastectomy candidates to conservation. And within the aromatase inhibitor class the choice of drug is immaterial. ACOSOG Z1031 randomised exemestane, letrozole and anastrozole against each other and separated them on nothing but the clinical response rate used to pick two agents forward22.
Duration and menopausal status have been varied outside the comparative trials. Six months of exemestane shrank strongly ER-positive tumours further than three months without significant side effects, and improved the feasibility of breast conservation by almost 10%23. In premenopausal women with locally advanced disease, letrozole with a GnRH analogue was given alongside preoperative chemotherapy and after surgery, in a retrospective series24. After neoadjuvant aromatase inhibition in Z1031, five-year locoregional recurrence after surgery was 1.5%, which supports conservation after this route25. The 21-gene score predicted clinical response to neoadjuvant letrozole26, and in a national cohort it was associated with complete response to neoadjuvant chemotherapy27.
The disagreement is entirely between aromatase inhibitor and tamoxifen, and it is a disagreement about method. P024 and STAGE found the aromatase inhibitor clearly ahead. P024's letrozole arm also suppressed proliferation more than tamoxifen, measured by Ki6728. IMPACT found no significant difference between its arms. PROACT returned response rates that sat close together on both of the instruments it used. Four features separate the two pairs and none of them is pharmacological. P024 ran four months in women who were not candidates for conservation, and scored response by clinical palpation29. STAGE ran twenty-four weeks in premenopausal women on goserelin, with matching placebos, and scored by callipers30. IMPACT ran three months and scored by callipers and by ultrasound31. PROACT ran twelve weeks and scored by both instruments as well. Its two instruments returned different response rates inside the one trial. PROACT also permitted concurrent chemotherapy, which contaminates the endocrine comparison outright32. Duration, menopausal status, measuring instrument and contamination all vary across these four. Read the endpoint column before the result column.
The decision rests on P024. It is a randomised double-blind comparison against the drug the clinic would otherwise have given. It is also the one that moved a surgical endpoint rather than a biomarker. STAGE is what extends the class below the menopause, because it holds ovarian suppression constant in both arms and varies only the partner drug.
Escalating within the receptor axis has now been tested twice and has not moved the ceiling. Z1031 found nothing between the three aromatase inhibitors. ALTERNATE gave a full six-month preoperative course and found that fulvestrant, alone or added to anastrozole, did not improve the proportion of tumours reaching the endocrine-sensitive state33. The ceiling is low and it is set by the tumour rather than by the drug.
The oral receptor degraders are absent from this table by design, and the reason is the endpoint rather than the result. coopERA Breast Cancer, AMEERA-4 and ELIPSE ran two-week and four-week exposures whose read-out is a ratio of two Ki67 scores34,35,36. Giredestrant suppressed proliferation more than anastrozole over that window. Amcenestrant matched letrozole without beating it, in a trial stopped early for slow accrual. Elacestrant was single-arm and read against an arrest threshold. None of the three scored tumour response, so none of them is commensurate with the six trials in the table Window-of-opportunity studies and short-term biologic endpoints.
Who it suits follows from the biology rather than from the trials. Strongly oestrogen receptor-positive, low-grade tumours are the ones that respond. Older or less fit patients are the group in whom the toxicity comparison against chemotherapy is most one-sided.
Lobular carcinoma has the strongest prior claim on this strategy and the weakest evidence behind it. Lobular tumours lose E-cadherin and infiltrate in single files rather than forming a discrete mass. They are almost always receptor-positive, low grade and poorly responsive to cytotoxic treatment. Both halves of the neoadjuvant argument therefore change at once, and the instruments used to score response in the trials above stop working on a tumour with no discrete mass. TBCRC 037 is the only histology-specific design, and it is a phase 1 target-engagement study that cannot rank endocrine agents for efficacy however it reads out. PELOPS is named for lobular disease and enrolled any histology. The only publication on its record is an axillary analysis whose endpoints the paper itself calls prespecified exploratory37. The histology argument rests on biology and on subgroups borrowed from elsewhere.
Duration is where practice most often goes wrong, because response accrues over months rather than weeks. The trials here that state a duration ran three to six months. A course cut to a few weeks has no downstaging evidence behind it at all. Shortened courses are the commonest avoidable cause of a disappointing result.
What the table cannot show is any survival column, because none of these trials had one. Nobody randomised preoperative endocrine therapy against preoperative chemotherapy in a trial powered to compare them. No trial in this table enrolled a lobular-only population.
In practice
Fix the goal before the first prescription. Surgical downstaging and measurement of endocrine sensitivity are different objectives with different stopping rules.
Plan at least four months, reassessing by examination and imaging at intervals rather than waiting for a fixed date.
Take a baseline Ki67 and a repeat at two to four weeks only if the result will change a decision. If it will not, the biopsy is not justified by the information.
Treat progression on neoadjuvant endocrine therapy as a significant event rather than an ordinary one. It is uncommon, and it identifies a tumour unlikely to be controlled by the same drug class given for the next five years.
The randomised comparisons between endocrine agents given before surgery.
Compare how each trial scored response before comparing the rates.
postmenopausal, oestrogen receptor-rich (Allred 6-8 or above 66%), clinical stage II-III; 1298 of 1362 evaluable; week 4 or week 12 Ki67 above 10% triggered a switch to chemotherapy or immediate surgery
neoadjuvant fulvestrant, or anastrozole plus fulvestrant, vs anastrozole
endocrine-sensitive disease rate at 6 months
endocrine-sensitive disease rate 22.8% with fulvestrant, 20.5% with the combination and 18.7% with anastrozole; neither fulvestrant-containing arm improved on anastrozole
Premenopausal women with oestrogen receptor positive, HER2 negative, operable breast cancer and WHO performance status 2 or lower.
Goserelin plus anastrozole versus goserelin plus tamoxifen, each with matching placebo, for 24 weeks before surgery.
Best overall tumour response, complete or partial response by callipers, during the 24 week neoadjuvant period.
Complete or partial response occurred in 70.4 percent of 98 patients given anastrozole and 50.5 percent of 99 given tamoxifen, a difference of 19.9 percent, 95 percent CI 6.5 to 33.3, p=0.004.
neoadjuvant exemestane vs letrozole vs anastrozole
clinical response
letrozole and anastrozole taken forward on clinical response; no difference between agents in Ki67 suppression or PEPI score; breast-conserving surgery in 51% of patients presenting as mastectomy-only candidates
Postmenopausal women with untreated primary breast cancer positive for estrogen or progesterone receptors and not candidates for breast-conserving surgery.
Letrozole 2.5 mg daily for four months versus tamoxifen 20 mg daily for four months.
Overall objective response by clinical palpation.
Objective response by clinical palpation was 55 percent with letrozole versus 36 percent with tamoxifen (P<0.001).
4 · Ki67, PEPI and response measures for a setting where pCR fails
The preoperative window as a measurement and as a treatment, continued
Does the proliferation left after a short preoperative exposure measure how much a tumour depends on oestrogen?
Rationale
Oestrogen drives the cell cycle in this cell. The fraction of cells still dividing after a short exposure is therefore the most direct read-out of whether the drug is working. A pretreatment biopsy says how fast the tumour was growing. A biopsy taken on treatment says how much of that growth oestrogen was responsible for. CDK4/6 sits immediately downstream of the receptor. A tumour still proliferating on an aromatase inhibitor is the obvious place to put a cell-cycle drug.
Experiment
Paired biopsies taken before and during preoperative treatment, scored centrally for Ki67. The largest trial randomised a short perioperative course of an aromatase inhibitor against no perioperative treatment. Its primary endpoint was time to recurrence. The paired Ki67 was only a secondary objective. Another trial randomised the three aromatase inhibitors against each other. Smaller trials added a CDK4/6 inhibitor to an aromatase inhibitor, either as a randomised arm or as a within-patient step after an endocrine-only run-in. One ongoing trial profiles serial biopsies at single-cell resolution.
Finding
Endocrine therapy alone leaves a proliferating fraction in most tumours. How large that fraction is after the short exposure separates patients across a wide span of recurrence risk. The exposure itself changed no outcome. The three aromatase inhibitors are indistinguishable on this read-out. It reports the class and not the molecule. Adding a CDK4/6 inhibitor drives most tumours below the arrest threshold, far more of them than an aromatase inhibitor does alone. The arrest reverses once the drug is withdrawn.
Limitation
Complete cell cycle arrest is an assay threshold, not a biological state. The cut point was calibrated on tumours given endocrine therapy alone. It was built to grade endocrine dependence. A drug that stops the cycle whether or not the tumour depends on oestrogen crosses that line in nearly everyone, which is precisely what destroys its power to select. The thresholds in use are not even the same number from trial to trial. The within-patient designs also confound the added drug with longer time on treatment, because the second biopsy is both on more drug and later.
Next question
If proliferation can be shut down almost completely, does that deeper suppression produce more of the tumour response a surgeon and a patient can see?
If the binary endpoint carries little information, something has to replace it. The four trials here agree on the biology and disagree on almost everything about the measurement.
The agreement is that endocrine therapy alone leaves a proliferating fraction in most tumours. How large that fraction is after a short exposure grades how far the tumour depends on oestrogen. POETIC is where that becomes a clinical statement rather than a pharmacodynamic one. Among treated HER2-negative women, five-year recurrence risk ran from 4.3% where Ki67 was low before and after treatment to 21.5% where it was high at both points38. The trial is load-bearing for an unusual reason. Its intervention changed no outcome, so the prognostic signal cannot be an effect of the two weeks of treatment. The measurement carries the information and the drug does not.
The divergence is exposure. POETIC read fourteen perioperative days. Z1031 and neoMONARCH read weeks. ALTERNATE read six months, with a switch to chemotherapy triggered at week four or week twelve. A threshold calibrated on one of these exposures does not transfer to another, which is why these four cannot be pooled on a cut-point. neoMONARCH sits in this table because Ki67 is its endpoint rather than because it asks the same question, and what it does to the threshold is argued in Neoadjuvant CDK4/6 inhibition and window-of-opportunity designs.
Two conventions coexist and they are not interchangeable. Complete cell-cycle arrest is set at a Ki67 of 2.7% or below, which is a pharmacodynamic threshold used in window studies. Endocrine response is set at 10% or below, which is an allocation threshold used to direct treatment. Neither boundary was placed where biology changes. Both were placed where observers could agree.
The measurement side deserves as much attention as the biology. Ki67 scoring varies between laboratories and observers even under a common protocol39, which is why successive working group statements concentrate on method rather than on cut-point40,41. The largest contemporary test of that comes from the trial now using post-endocrine Ki67 to allocate treatment. Local and central scoring agreed on the binary endocrine-response call in 80.5% of specimens16. Where the two readings disagreed, the patient would have been sent the other way.
The preoperative endocrine prognostic index answers a different question, which is what to do after surgery. It combines pathological tumour size, nodal status, Ki67 and oestrogen receptor status measured in the surgical specimen. It was derived in P024 and validated in IMPACT. A score of zero identified a group at very low risk of relapse on endocrine therapy alone42. That is a prognostic statement and not a predictive one. It does not establish that chemotherapy would have added nothing, because nobody in either cohort was randomised to find out. Z1031 later scored a preoperative endocrine prognostic index of zero more often in luminal A tumours than in luminal B ones22.
What the table cannot show is who read the slide. There is no column for central against local scoring, for antibody clone or for scoring protocol, and the preceding paragraph is the reason that matters. Nor has any trial here randomised the decision rule itself. WSG ADAPTcycle applies that rule to select patients and then randomises the drug, which is a different experiment.
Caution
A fall in Ki67 is a change in a proliferation marker rather than a change in outcome. The single trial large enough to separate the two found a measurement that predicted recurrence and a treatment that prevented none of it38. Reading a suppressed Ki67 as evidence that the drug is helping that patient runs the two together.
Trials whose read-out was proliferation rather than tumour size.
Note how far apart their exposures are, from fourteen days to six months.
neoadjuvant fulvestrant, or anastrozole plus fulvestrant, vs anastrozole
endocrine-sensitive disease rate at 6 months
endocrine-sensitive disease rate 22.8% with fulvestrant, 20.5% with the combination and 18.7% with anastrozole; neither fulvestrant-containing arm improved on anastrozole
2-week lead-in of abemaciclib plus anastrozole, or abemaciclib alone, vs anastrozole alone, then 14 weeks of the combination in all groups
change in Ki67 from baseline to 2 weeks
complete cell-cycle arrest in 68% with the combination and 58% with abemaciclib vs 14% with anastrozole, P < 0.001; pCR in 4% of the 190 patients operated
neoadjuvant exemestane vs letrozole vs anastrozole
clinical response
letrozole and anastrozole taken forward on clinical response; no difference between agents in Ki67 suppression or PEPI score; breast-conserving surgery in 51% of patients presenting as mastectomy-only candidates
5 · Neoadjuvant CDK4/6 inhibition and window-of-opportunity designs
The preoperative window as a measurement and as a treatment, continued
Does suppressing proliferation more deeply produce a better tumour response at surgery?
Rationale
Ki67 was adopted before surgery because it is quick, cheap and readable in a core biopsy. The whole case for using it rests on the assumption that it stands in for something the patient experiences. A drug that suppresses proliferation far more should therefore shrink tumours more, or leave less disease behind at operation. If it does neither, the surrogate and the outcome have come apart. The surrogate is then the thing to distrust.
Experiment
Trials that reported a proliferation endpoint and a clinical or pathological endpoint out of the same randomisation. One added a CDK4/6 inhibitor to letrozole and reported the change in Ki67 beside the clinical response rate. Another gave a short lead-in of the combination against an aromatase inhibitor alone, then operated on everyone who completed the preoperative course. Two more carried the combination through a full preoperative course against chemotherapy. One of those read residual cancer burden, pathological complete response and a composite prognostic index scored on the surgical specimen, the other a molecular risk score read on the same specimen. One trial asked directly whether early proliferation suppression converted into that composite index.
Finding
The proliferation endpoints separated the arms decisively. The clinical and pathological endpoints did not follow. Where both came out of one randomisation, a large difference in Ki67 sat beside a clinical response rate that was no different. In the comparison against chemotherapy the two arms changed places depending on which endpoint was read, chemotherapy ahead on residual disease and the combination ahead on the composite index. The trial that read a molecular risk score instead found the two arms level. Where early suppression was tested directly as a predictor of the composite index, it did not deliver one.
Limitation
Pathological complete response sits near its floor in this cell. A trial powered on it has almost no room to move. A null result there means very little. The composite index is not independent of the surrogate, because proliferation is one of its own components. A drug that suppresses proliferation therefore improves the index partly by construction. The molecular risk score is a reclassification of the tumour rather than a measure of what the patient gained at surgery. The trial that tested the translation most directly has conference and registry material behind it and no peer-reviewed report. Its negative result cannot be weighed as a published trial result.
Next question
If pathological response cannot carry the argument in this cell, can the response measured during treatment still be used to decide who needs chemotherapy at all?
PALLETneoMONARCHFELINENeoPALCORALLEEN
This is the cleanest demonstration in the book that arresting proliferation and eradicating a tumour are different events.
Four of the five rows have reported. Two of them set a CDK4/6 inhibitor added to an aromatase inhibitor against endocrine therapy alone, and both found proliferation suppressed well beyond what the endocrine arm achieved. PALLET and neoMONARCH show that in the same direction, over fourteen weeks and over two44,43. PALLET gave no clinical response advantage over letrozole alone. neoMONARCH put every group on the combination after its two-week lead-in, so it has no clinical comparison to offer. The pathological complete response rate neoMONARCH reports is a single-group figure.
The table looks like five trials of one strategy. It is two questions with different control arms. PALLET and neoMONARCH compared the combination against endocrine therapy alone and read proliferation. NeoPAL and CORALLEEN compared it against chemotherapy and read the surgical specimen. A reader who lines up the result column across all five is comparing answers to different questions.
That first combination is what a cytostatic agent should produce, because blocking the G1 restriction point stops cells dividing Cell cycle regulation. It does not kill them, so the tumour bed is still cellular at surgery. A trial judging such a drug by a pathological complete response will call an active agent inactive.
NeoPAL makes the point sharpest by disagreeing with itself. Judged on residual cancer burden, the chemotherapy arm did better. Judged on the preoperative endocrine prognostic index, the combination arm did better21. One randomisation of one set of patients produces opposite rankings depending on which post-treatment measurement is chosen. CORALLEEN read a third measure, a molecular risk score on the surgical specimen, and found the two arms level19. NeoPAL was designed as a non-comparative trial. Neither trial carried a noninferiority margin.
The composite index deserves a further warning, because proliferation is one of its own components. A drug that suppresses proliferation improves that index partly by construction. Reading such an improvement as evidence of deeper tumour control is circular. The molecular risk score has the opposite problem. It reclassifies the tumour rather than measuring what the patient gained at surgery.
There is a second reason to distrust the arrest threshold once a cell-cycle inhibitor is in the regimen. Complete cell-cycle arrest was calibrated on tumours given endocrine therapy alone. It was built to grade endocrine dependence. A drug that stops the cycle whether or not the tumour depends on oestrogen crosses that line in nearly everyone. Crossing it then says nothing about the receptor, which is exactly what destroys its power to select.
Suppression also requires continuing exposure. Both neoMONARCH and the within-patient NeoPalAna design recorded Ki67 recovering once the inhibitor was withdrawn before surgery, and in NeoPalAna a further cycle given immediately beforehand suppressed that recovery again45. NeoPalAna is single-arm and each patient is her own control, so its comparison confounds the added drug with longer time on treatment. The rebound observation survives that objection because it runs the other way. It makes the discontinuation and adherence questions in Toxicity, discontinuation, and the cost of two or three years questions about efficacy rather than about administration.
NEOLETRIB is the ongoing row. It profiles serial biopsies at single-cell resolution through six months of letrozole and ribociclib. It is single-arm, so it will describe what months of exposure do to a human tumour and will settle nothing about whether that helps anyone. FELINE asked the translation question directly, whether early proliferation suppression converts into a better prognostic index. It has conference and registry material only. Its reported negative result cannot be weighed as a published trial result.
What the table cannot show is any survival column. No trial here was powered to test whether a CDK4/6 inhibitor can replace chemotherapy. Pathological complete response also sits so near its floor in this cell that a null result on it means very little. The window logic, and the reason a pharmacodynamic effect does not license an outcome claim, is worked out in Window-of-opportunity studies and short-term biologic endpoints using POETIC.
Preoperative phase II trials adding a CDK4/6 inhibitor to endocrine therapy in this subtype.
Four different endpoints appear in five trials, which is the argument of this section.
2-week lead-in of abemaciclib plus anastrozole, or abemaciclib alone, vs anastrozole alone, then 14 weeks of the combination in all groups
change in Ki67 from baseline to 2 weeks
complete cell-cycle arrest in 68% with the combination and 58% with abemaciclib vs 14% with anastrozole, P < 0.001; pCR in 4% of the 190 patients operated
letrozole plus palbociclib for 19 weeks vs FEC100 followed by docetaxel
residual cancer burden class 0 or I rate
RCB 0-I in 7.7% vs 15.7%; pCR 3.8% vs 5.9%; PEPI score 0 in 17.6% vs 8.0%; 2 vs 17 serious adverse events
6 · Immunotherapy in hormone receptor-positive early disease
The preoperative window as a measurement and as a treatment, continued
Can checkpoint blockade or an antibody-drug conjugate raise response where endocrine escalation has not?
Rationale
Luminal tumours are usually immunologically quiet. The high-grade, receptor-low end of this cell is not. An adaptive platform trial had already reported a pembrolizumab signal across HER2-negative disease rather than in triple-negative disease alone. Separately, most receptor-positive tumours express HER2 below the amplification threshold, which an antibody-drug conjugate can use as an address rather than as a driver.
Experiment
Two phase III trials added a checkpoint inhibitor to anthracycline and taxane-based preoperative chemotherapy in luminal disease, with pathological complete response as a primary endpoint and event-free survival to follow. Alongside them sit a randomised trial of preoperative radiation dose as an immune primer before checkpoint blockade, and a preoperative trial of an antibody-drug conjugate given with and without an aromatase inhibitor in receptor-positive, HER2-low disease rather than HER2-amplified disease. A single-arm study profiled what a CDK4/6 inhibitor on an endocrine backbone does to the immune and stromal compartments over a preoperative course.
Finding
Both phase III trials raised pathological complete response. In the one whose record reports the subgroup, the increase concentrated where PD-L1 was expressed. The event-free survival that decides whether the increase matters is immature where it has been reported at all. The profiling study found that a CDK4/6 inhibitor on an endocrine backbone diminished adaptive anti-tumour immunity, which is a reason to doubt that the two escalation strategies combine. The radiation-priming trial and the antibody-drug conjugate trial have conference and registry material only. Neither yet has a result to read.
Limitation
Pathological complete response is not the surrogate here that it is in triple-negative disease. A gain in it does not convert into an expectation about recurrence. One of the two phase III trials selected explicitly on grade, on low oestrogen receptor expression and on PD-L1, which is the end of this cell that behaves least like the rest of it. Its result does not generalise to the ordinary luminal tumour. The PD-L1 result there is scored with a single antibody clone against a single cut-off. A group defined by a marker is not a randomised comparison. The profiling study is single-arm. Its immune observation is a reason for doubt rather than a test of whether the strategies combine. The radiation-priming trial holds its receptor-positive and triple-negative cohorts separately. Nothing there may be read as a pooled population.
Next question
Does a gain in pathological response, in a cell where pathological response is rare, convert into fewer distant recurrences?
All three rows agree on one finding. Adding a checkpoint inhibitor to anthracycline and taxane-based preoperative chemotherapy raises the pathological complete response rate in this subtype.
They are not three equivalent estimates of it. KEYNOTE-756 and CheckMate 7FL are double-blind, placebo-controlled phase 3 trials with a concurrent control arm18,17. The I-SPY2 pembrolizumab arm row is an adaptively randomised platform arm whose rates are Bayesian estimates read against a control shared with other arms and accrued over a different period20. The drugs differ too. The estimation method differs more, and it is what stops the three response rates being read off against each other.
Whom they enrolled matters more than what they showed, because both phase 3 trials restricted entry by grade. KEYNOTE-756 admitted grade 3 tumours only18. CheckMate 7FL admitted grade 2 tumours only where oestrogen receptor expression fell between 1% and 10%17. These are the tumours that behave least like typical luminal disease. Transferring either result to a grade 1, strongly receptor-positive tumour is not supported by either trial.
KEYNOTE-756 is the load-bearing row. It is the largest of the three. It is also placebo-controlled, and it carries event-free survival as a dual primary endpoint rather than as an afterthought. That is the endpoint the decision will eventually rest on. It was not mature at the analysis that reported the response result.
The effect concentrates further inside those populations. In CheckMate 7FL the response gain was larger among PD-L1-positive tumours than across the trial as a whole. That is a group defined by a marker rather than a randomised comparison. The marker was scored with one antibody clone against one cut-off. Statistical and clinical significance are different claims sets out what such an observation can and cannot carry.
The endpoint problem now arrives at its sharpest point. A gain in pathological complete response has been demonstrated in the one subtype where that endpoint is least prognostic1,2. The gain may well translate into fewer recurrences. Nothing published so far shows that it has.
A practical conflict is already visible, because adjuvant abemaciclib and adjuvant ribociclib are options for the same high-risk population. Neither is given alongside a checkpoint inhibitor. A patient eligible for both escalations can have only one of them, and no trial compares the two. There is also a mechanistic reason to doubt that the two would combine. NeoRHEA, a single-arm profiling study, found that a CDK4/6 inhibitor on an endocrine backbone diminished adaptive anti-tumour immunity47. That is a reason for doubt rather than a test of the combination.
What the table cannot show is the column that decides the question. There is no event-free survival and no overall survival in any of these rows. There is no PD-L1 column, so the subgroup that drives the enthusiasm is invisible where the trials are enumerated. And nobody has randomised a preoperative checkpoint inhibitor against adjuvant CDK4/6 inhibition, which is the choice the clinic actually faces. The other mechanisms being tried here, an antibody-drug conjugate and radiation given as an immune primer, have no result to read at all Unresolved questions and what reads out next.
Caution
The intuitive reading is that a higher complete response rate in a poor-prognosis population must translate into fewer recurrences. In triple-negative disease that reading has been supported by subsequent survival data. Here the same inference is being drawn in the subtype where the trial-level link between response and outcome is weakest. Late recurrence also means the follow-up needed to test it is measured in decades rather than years Late recurrence risk and surveillance implications.
Preoperative checkpoint inhibitor trials in this subtype.
Read the eligibility column before the result column, because all three selected for the end of this cell that behaves least like the rest of it.
untreated grade 3 oestrogen receptor-positive HER2-negative, T1c-2 with cN1-2 or T3-4 with cN0-2
pembrolizumab plus neoadjuvant paclitaxel then anthracycline-cyclophosphamide vs placebo plus the same chemotherapy, each followed by adjuvant pembrolizumab or placebo and endocrine therapy
pCR 24.3% vs 15.6%, difference 8.5 percentage points, 95% CI 4.2 to 12.8, P = 0.00005; event-free survival immature at this analysis
high-risk stage II-III HER2-negative disease; 29 of 69 in the experimental arm triple-negative
pembrolizumab added to weekly paclitaxel then doxorubicin and cyclophosphamide vs the same chemotherapy, adaptively randomised against a shared control
estimated pCR 60% with pembrolizumab vs 22% in triple-negative disease, 44% vs 17% across HER2-negative disease
7 · Unresolved questions and what reads out next
Six questions are open. The five trials below work on four of them. Two have nothing in the table at all.
The first is whether a response gain translates. Event-free survival from KEYNOTE-756 and CheckMate 7FL is the readout that will settle whether the argument in Immunotherapy in hormone receptor-positive early disease was pessimism or realism. A positive result would make pathological complete response useful in this subtype for the first time. A null result would confirm that the endpoint measured the wrong thing. No trial in the table below asks this question, which is worth noticing.
The second is whether endocrine or cell-cycle response can take the place of chemotherapy in intermediate-risk disease. WSG ADAPTcycle allocates on a genomic score together with the post-endocrine Ki67 described in Ki67, PEPI and response measures for a setting where pCR fails, then randomises ribociclib with endocrine therapy against chemotherapy, with invasive disease-free survival as the endpoint. Its randomisation is not purely adjuvant, because its chemotherapy arm may be given before surgery16. That design makes a null result ambiguous, because the allocation rule and the drug would fail together. RIBOLARIS asks a narrower version. It gives everyone the same preoperative ribociclib and letrozole, routes on a molecular risk score read at surgery, and carries five-year distant metastasis-free survival in the low-risk cohort as its endpoint48. That is the right endpoint and a long wait. It is also single-arm through the preoperative phase, so the low-risk cohort is compared against an external benchmark rather than against a randomised control.
The third is what a CDK4/6 inhibitor does to a human tumour over months rather than weeks. NEOLETRIB samples serially through six months of exposure with single-cell sequencing46. Its value will be mechanistic rather than clinical.
The fourth is whether a different mechanism can raise response where endocrine escalation has not. TRIO-US B-12 TALENT moves an antibody-drug conjugate before surgery in untreated hormone receptor-positive, HER2-low disease, given alone and with an aromatase inhibitor. It uses HER2 expression below the amplification threshold as an address rather than as a driver. P-RAD / TBCRC 053 randomises preoperative radiation dose as an immune primer before checkpoint blockade and chemotherapy, holding its hormone receptor-positive and triple-negative cohorts separate. Neither has a result.
The fifth is whether lobular histology needs a preoperative experiment of its own. TBCRC 037 is the histology-specific row, a randomised short preoperative exposure in lobular disease read out as target engagement. It will say whether each drug reaches its target in that histology. It cannot rank the drugs on efficacy.
The sixth question has no trial behind it. Nobody is testing whether preoperative treatment of any kind improves outcome over upfront surgery in this subtype. Nobody has prospectively validated a preoperative endocrine prognostic index of zero as a gate for omitting adjuvant chemotherapy. Both would change practice. Neither is being studied.
The table below prints what each open trial asks rather than what it found, because three of the six have no primary publication at all. Treating a conference abstract as a trial result is the specific error this table is built to prevent.
Preoperative trials in this subtype that have not reported a primary result.
No result column is printed, because for three of these six there is no publication to take one from.
Trial
Population
Experimental vs control
P-RAD / TBCRC 0532 · awaited
untreated early breast cancer, enrolled into separate hormone receptor-positive and triple-negative cohorts
randomised preoperative radiation dose levels before pembrolizumab and chemotherapy; the dose levels are not stated in the material available
TBCRC 0371 · awaited
Patients with invasive lobular breast cancer.
Randomised short preoperative exposure to tamoxifen, anastrozole or fulvestrant.
TRIO-US B-12 TALENT2 · awaited
Patients with untreated hormone receptor-positive, HER2-low early breast cancer.
Neoadjuvant trastuzumab deruxtecan alone versus trastuzumab deruxtecan plus anastrozole.
Patients with stage II, grade 2 or 3 HR-positive, HER2-negative early breast cancer defined as high risk by a Ki-67 index of 20% or more and/or high genomic risk on a validated gene expression signature.
Single arm through the neoadjuvant phase; after surgery patients with a ROR-low score or a pathological complete response continue adjuvant ribociclib and endocrine therapy, whereas patients with ROR-medium or ROR-high receive standard chemotherapy first.
intermediate-risk HR-positive, HER2-negative early breast cancer selected by a 21-gene recurrence score and by endocrine response measured as post-endocrine Ki-67; 5292 screened and 1684 randomised at 85 German sites
ribociclib plus an aromatase inhibitor, with a gonadotropin-releasing hormone agonist in premenopausal women, vs chemotherapy followed by endocrine therapy
oestrogen receptor-positive HER2-negative luminal A or B disease
letrozole plus ribociclib for 6 months, single group, with serial biopsy and single-cell RNA sequencing
Interplay
Part VI argues that a tumour is a population rather than a value, and this chapter is where that argument has a bedside consequence for the commonest subtype. Every response measure used here averages over that population. A complete response asks whether the whole population was eradicated, which in slowly proliferating disease is almost never true. Residual cancer burden keeps the variance and is therefore the more informative instrument14. A two-week Ki67 measures the fraction of the population still cycling under oestrogen deprivation, which is a statement about composition rather than about size38. Read this way, the low complete response rate in hormone receptor-positive disease is not a failure of the drugs. It is what averaging over a heterogeneous, slowly dividing population looks like when the measurement is forced into two categories.
Cortazar P, Zhang L, Untch M, et al. Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis. Lancet 2014 384:164-172. PMID 24529560
von Minckwitz G, Untch M, Blohmer JU, et al. Definition and impact of pathologic complete response on prognosis after neoadjuvant chemotherapy in various intrinsic breast cancer subtypes. J Clin Oncol 2012 30:1796-1804. PMID 22508812
Fisher B, et al. Effect of preoperative chemotherapy on the outcome of women with operable breast cancer. J Clin Oncol 1998 16:2672-85. PMID 9704717
Wolmark N, et al. Preoperative chemotherapy in patients with operable breast cancer: nine-year results from National Surgical Adjuvant Breast and Bowel Project B-18. J Natl Cancer Inst Monogr 2001. PMID 11773300
Bear HD, et al. Sequential preoperative or postoperative docetaxel added to preoperative doxorubicin plus cyclophosphamide for operable breast cancer:National Surgical Adjuvant Breast and Bowel Project Protocol B-27. J Clin Oncol 2006 24:2019-27. PMID 16606972
Untch M, et al. Nab-paclitaxel versus solvent-based paclitaxel in neoadjuvant chemotherapy for early breast cancer (GeparSepto-GBG 69): a randomised, phase 3 trial. Lancet Oncol 2016 17:345-356. PMID 26869049
Gianni L, et al. Comparing Neoadjuvant Nab-paclitaxel vs Paclitaxel Both Followed by Anthracycline Regimens in Women With ERBB2/HER2-Negative Breast Cancer-The Evaluating Treatment With Neoadjuvant Abraxane (ETNA) Trial: A Randomized Phase 3 Clinical Trial. JAMA Oncol 2018 4:302-308. PMID 29327055
Tung N, Arun B, Hacker MR, et al. TBCRC 031: Randomized Phase II Study of Neoadjuvant Cisplatin Versus Doxorubicin-Cyclophosphamide in Germline BRCA Carriers With HER2-Negative Breast Cancer (the INFORM trial). J Clin Oncol 2020 38:1539-1548. PMID 32097092
Fasching PA, Link T, Hauke J, et al. Neoadjuvant paclitaxel/olaparib in comparison to paclitaxel/carboplatinum in patients with HER2-negative breast cancer and homologous recombination deficiency (GeparOLA study). Ann Oncol 2020 32:49-57. PMID 33098995
Fasching PA, Schmatloch S, Hauke J, et al. Neoadjuvant Paclitaxel/Olaparib in Comparison to Paclitaxel/Carboplatin in Patients with HER2-Negative Breast Cancer and HRD-Long-term Survival of the GeparOLA Study. Clin Cancer Res 2025 31:1596-1604. PMID 39998850
I-SPY2 Trial Consortium, Yee D, DeMichele AM, et al. Association of event-free and distant recurrence-free survival with individual-level pathologic complete response in neoadjuvant treatment of stages 2 and 3 breast cancer: three-year follow-up analysis for the I-SPY2 adaptively randomized clinical trial. JAMA Oncol 2020 6:1355-1362. PMID 32701140
Nitz UA, et al. Endocrine Therapy Response and 21-Gene Expression Assay for Therapy Guidance in HR+/HER2- Early Breast Cancer. J Clin Oncol 2022 40:2557-2567. PMID 35404683
de Groot AF, Cohen D, Heijns JB, et al. Ribociclib-Letrozole Combination as an Alternative for Neoadjuvant Chemotherapy in Selected Postmenopausal Patients with Luminal Breast Cancer (BOOG 2017-01). Clin Cancer Res 2025 31:3175-3181. PMID 40495424
Symmans WF, Wei C, Gould R, et al. Long-term prognostic risk after neoadjuvant chemotherapy associated with residual cancer burden and breast cancer subtype. J Clin Oncol 2017 35:1049-1060. PMID 28135148
Yau C, Osdoit M, van der Noordaa M, et al. Residual cancer burden after neoadjuvant chemotherapy and long-term survival outcomes in breast cancer: a multicentre pooled analysis of 5161 patients. Lancet Oncol 2022 23:149-160. PMID 34902335
Hamann M, Christgen M, Gluz O, et al. Endocrine response assessment in HR-positive HER2-negative early breast cancer: concordance of local versus central Ki67 measurements in the WSG ADAPTcycle trial (n = 5292). ESMO Open 2025 10:105552. PMID 40815984
Loi S, Salgado R, Curigliano G, et al. Neoadjuvant nivolumab and chemotherapy in early estrogen receptor-positive breast cancer: a randomized phase 3 trial. Nat Med 2025 31:433-441. PMID 39838118
Cardoso F, O'Shaughnessy J, Liu Z, et al. Pembrolizumab and chemotherapy in high-risk, early-stage, ER+/HER2- breast cancer: a randomized phase 3 trial. Nat Med 2025 31:442-448. PMID 39838117
Prat A, Saura C, Pascual T, et al. Ribociclib plus letrozole versus chemotherapy for postmenopausal women with hormone receptor-positive, HER2-negative, luminal B breast cancer (CORALLEEN): an open-label, multicentre, randomised, phase 2 trial. Lancet Oncol 2020 21:33-43. PMID 31838010
Nanda R, Liu MC, Yau C, et al. Effect of pembrolizumab plus neoadjuvant chemotherapy on pathologic complete response in women with early-stage breast cancer - an analysis of the ongoing phase 2 adaptively randomized I-SPY2 trial. JAMA Oncol 2020 6:676-684. PMID 32053137
Cottu P, D'Hondt V, Dureau S, et al. Letrozole and palbociclib versus chemotherapy as neoadjuvant therapy of high-risk luminal breast cancer. Ann Oncol 2018 29:2334-2340. PMID 30307466
Ellis MJ, Suman VJ, Hoog J, et al. Randomized phase II neoadjuvant comparison between letrozole, anastrozole, and exemestane for postmenopausal women with estrogen receptor-rich stage 2 to 3 breast cancer: clinical and biomarker outcomes and predictive value of the baseline PAM50-based intrinsic subtype - ACOSOG Z1031. J Clin Oncol 2011 29:2342-2349. PMID 21555689
Fontein DB, Charehbili A, Nortier JW, et al. Efficacy of six month neoadjuvant endocrine therapy in postmenopausal, hormone receptor-positive breast cancer patients--a phase II trial. Eur J Cancer 2014 50:2190-200. PMID 24970786
Torrisi R, Bagnardi V, Rotmensz N, et al. Letrozole plus GnRH analogue as preoperative and adjuvant therapy in premenopausal women with ER positive locally advanced breast cancer. Breast Cancer Res Treat 2011 126:431-41. PMID 21221766
Hunt KK, Suman VJ, Wingate HF, et al. Local-Regional Recurrence After Neoadjuvant Endocrine Therapy: Data from ACOSOG Z1031 (Alliance), a Randomized Phase 2 Neoadjuvant Comparison Between Letrozole, Anastrozole, and Exemestane for Postmenopausal Women with Estrogen Receptor-Positive Clinical Stage 2 or 3 Breast Cancer. Ann Surg Oncol 2023 30:2111-2118. PMID 36653664
Iwata H, Masuda N, Yamamoto Y, et al. Validation of the 21-gene test as a predictor of clinical response to neoadjuvant hormonal therapy for ER+, HER2-negative breast cancer: the TransNEOS study. Breast Cancer Res Treat 2019 173:123-133. PMID 30242578
Pease AM, Riba LA, Gruner RA, Tung NM, James TA. Oncotype DXRecurrence Score as a Predictor of Response to Neoadjuvant Chemotherapy. Ann Surg Oncol 2019 26:366-371. PMID 30542840
Ellis MJ, Ma C. Letrozole in the neoadjuvant setting: the P024 trial. Breast Cancer Res Treat 2007 105 Suppl 1:33-43. PMID 17912634
Eiermann W, et al. Preoperative treatment of postmenopausal breast cancer patients with letrozole: A randomized double-blind multicenter study. Ann Oncol 2001 12:1527-32. PMID 11822750
Masuda N, et al. Neoadjuvant anastrozole versus tamoxifen in patients receiving goserelin for premenopausal breast cancer (STAGE): a double-blind, randomised phase 3 trial. Lancet Oncol 2012 13:345-52. PMID 22265697
Smith IE, et al. Neoadjuvant treatment of postmenopausal breast cancer with anastrozole, tamoxifen, or both in combination: the Immediate Preoperative Anastrozole, Tamoxifen, or Combined with Tamoxifen (IMPACT) multicenter double-blind randomized trial. J Clin Oncol 2005 23:5108-16. PMID 15998903
Cataliotti L, et al. Comparison of anastrozole versus tamoxifen as preoperative therapy in postmenopausal women with hormone receptor-positive breast cancer: the Pre-Operative "Arimidex" Compared to Tamoxifen (PROACT) trial. Cancer 2006 106:2095-103. PMID 16598749
Ma CX, Suman VJ, Sanati S, et al. Endocrine-sensitive disease rate in postmenopausal patients with estrogen receptor-rich/ERBB2-negative breast cancer receiving neoadjuvant anastrozole, fulvestrant, or their combination: a phase 3 randomized clinical trial. JAMA Oncol 2024 10:362-371. PMID 38236590
Hurvitz SA, Bardia A, Quiroga V, et al. Neoadjuvant palbociclib plus either giredestrant or anastrozole in oestrogen receptor-positive, HER2-negative, early breast cancer (coopERA Breast Cancer): an open-label, randomised, controlled, phase 2 study. Lancet Oncol 2023 24:1029-1041. PMID 37657462
Campone M, Bidard FC, Neven P, et al. AMEERA-4: a randomized, preoperative window-of-opportunity study of amcenestrant versus letrozole in early breast cancer. Breast Cancer Res 2023 25:141. PMID 37950338
Vidal M, Falato C, Pascual T, et al. Elacestrant in Women with Estrogen Receptor-Positive and HER2-Negative Early Breast Cancer: Results from the Preoperative Window-of-Opportunity ELIPSE Trial. Clin Cancer Res 2025 31:1223-1232. PMID 39820652
Weiss A, Jin Q, Tayob N, et al. Axillary Management and Outcomes After Neoadjuvant Endocrine Therapy in the Randomized PELOPS Trial. Ann Surg Oncol 2025 33:2275-2283. PMID 41276768
Smith I, Robertson J, Kilburn L, et al. Long-term outcome and prognostic value of Ki67 after perioperative endocrine therapy in postmenopausal women with hormone-sensitive early breast cancer (POETIC): an open-label, multicentre, parallel-group, randomised, phase 3 trial. Lancet Oncol 2020 21:1443-1454. PMID 33152284
Polley MY, Leung SC, McShane LM, et al. An international Ki67 reproducibility study. J Natl Cancer Inst 2013 105:1897-1906. PMID 24203987
Dowsett M, Nielsen TO, A'Hern R, et al. Assessment of Ki67 in breast cancer: recommendations from the International Ki67 in Breast Cancer working group. J Natl Cancer Inst 2011 103:1656-1664. PMID 21960707
Nielsen TO, Leung SCY, Rimm DL, et al. Assessment of Ki67 in breast cancer: updated recommendations from the International Ki67 in Breast Cancer Working Group. J Natl Cancer Inst 2021 113:808-819. PMID 33369635
Ellis MJ, Tao Y, Luo J, et al. Outcome prediction for estrogen receptor-positive breast cancer based on postneoadjuvant endocrine therapy tumor characteristics. J Natl Cancer Inst 2008 100:1380-1388. PMID 18812550
Hurvitz SA, Martin M, Press MF, et al. Potent cell-cycle inhibition and upregulation of immune response with abemaciclib and anastrozole in neoMONARCH, phase II neoadjuvant study in HR+/HER2- breast cancer. Clin Cancer Res 2020 26:566-580. PMID 31615937
Johnston S, Puhalla S, Wheatley D, et al. Randomized phase II study evaluating palbociclib in addition to letrozole as neoadjuvant therapy in estrogen receptor-positive early breast cancer: PALLET trial. J Clin Oncol 2019 37:178-189. PMID 30523750
Ma CX, Gao F, Luo J, et al. NeoPalAna: Neoadjuvant Palbociclib, a Cyclin-Dependent Kinase 4/6 Inhibitor, and Anastrozole for Clinical Stage 2 or 3 Estrogen Receptor-Positive Breast Cancer. Clin Cancer Res 2017 23:4055-4065. PMID 28270497
Fjermeros K, Ghannoum S, Geisler SB, et al. The NEOLETRIB trial: neoadjuvant treatment with letrozole and ribociclib in ER-positive, HER2-negative breast cancer. Future Oncol 2024 20:2457-2466. PMID 39073142
Papagiannis A, Majjaj S, Duhoux FP, et al. Palbociclib and endocrine therapy diminish adaptive anti-tumor immunity in early breast cancer: The NeoRHEA phase 2 study. Nat Commun 2025 16:11659. PMID 41298425
Cottu P, Pascual T, Lemonnier J, et al. Rationale and trial design of RIBOLARIS: A phase II trial of neoadjuvant and adjuvant ribociclib plus endocrine therapy for clinically high-risk HR+/HER2-negative breast cancer. Ther Adv Med Oncol 2026 18:17588359261467380. PMID 42602220