Introduction
Adjuvant endocrine therapy improves outcomes for estrogen-receptor positive (ER+) breast cancer [
1‐
3]. However, 25–50% of women with early stage breast cancer (stages I and II) will experience tumor recurrence [
4]. Pre-operative or neoadjuvant 'window' studies provide short exposures to systemic therapy between cancer diagnosis and surgery, potentially providing early insight into tumor sensitivity and resistance [
5‐
7]. Recent and ongoing trials use an early biopsy strategy to determine whether alternative treatment (such as chemotherapy) is indicated [
8,
9]. Serial biopsy studies have shown a decrease in proliferative index (Ki-67) following as little as 2 weeks of successful neoadjuvant endocrine therapy [
10]; in the IMPACT study, proliferation dropped at 2 weeks and remained low for the subsequent 10 weeks in the majority of patients [
11]. Post-therapy Ki-67 levels following 2 weeks of neoadjuvant endocrine therapy have been shown to predict progression-free survival [
12], but the requirement for biopsy, and often serial biopsies, results in limited clinical use.
As more data emerge that endocrine therapy alone is sufficient for some patients [
13], tools are needed to measure tumor response to determine which patients benefit from chemotherapy or molecularly targeted therapies [
13,
14]. Oncotype Dx is a tissue-based genomic assay that, obtained prior to therapy, is widely used to assign individual treatment options [
13‐
15]. The ability to measure the impact of endocrine therapy could add value beyond pre-therapy predictions of response. PET imaging biomarkers offer a distinct and complementary approach to tissue sampling for evaluating early treatment response. Unlike genomic assays which rely on core-biopsy, PET has the potential to avoid sampling error, and to noninvasively assesses the entire tumor burden in vivo, allowing for serial studies.
18F-Fluorodeoxyglucose (FDG), the most commonly used PET imaging biomarker, measures glucose metabolism. FDG-PET has been shown to correlate with tumor proliferation in some studies [
16], but is also associated with other processes such as inflammation, cellular repair, and apoptosis.
18F-Fluorothymidine (FLT) is an investigational imaging probe of tumor proliferation [
17] shown to correlate with Ki-67 in breast, lung, and brain cancer [
18,
19]. Both imaging agents have potential to identify endocrine sensitive tumors early in treatment and may identify patients who could avoid cytotoxic therapy and/or benefit from combination endocrine therapy [
20] or endocrine therapy plus molecularly targeted agents [
21].
Early changes in FDG-PET measure response to chemotherapy, endocrine therapy, and targeted therapy in breast and other tumors [
22,
23]. FLT-PET imaging has demonstrated ability to measure early response to systemic endocrine, chemotherapy, radiotherapy, and combined chemoradiotherapy in multiple tumor types [
21,
24]. FLT-PET is correlated with changes in tumor proliferation early after initiating second-line docetaxel chemotherapy and following completion of variable neoadjuvant chemotherapy regimens in breast cancer [
19,
25]. Pre-clinical studies suggest that FLT-PET may be useful for indicating the need for combined endocrine therapy and cell-cycle targeted drugs (CDK4/6) [
26].
We prospectively evaluated early response to neoadjuvant aromatase inhibitor (AI) therapy using baseline and pre-surgery FDG- and FLT-PET imaging in two different protocols, in conjunction with tissue Ki-67 assay in early-stage ER+ tumors under the hypothesis that one or both tracer imaging approaches would produce results similar to Ki-67 biopsy levels. Our goals were to use PET imaging to evaluate feasibility of the window-of-opportunity approach to assess endocrine therapy early response in breast cancer, and to measure early tumor response in order to improve treatment selection for early stage breast cancer that would provide insight into potential mechanisms of resistance to therapy using the Ki-67 assay, an established predictor of endocrine responsiveness [
8,
12] as the reference standard.
Discussion
This study shows that it is feasible to monitor patients and measure change in tumor metabolic activity with serial PET imaging during neoadjuvant endocrine therapy to assess for early response in vivo. The majority of tumors manifest a decline in uptake beyond what would be expected for the established reproducibility of the imaging test; specifically, 50% (95% CI 31–69%) of patients studied by FDG-PET and 70% (95% CI 52–84%) of patients studied by FLT-PET. We noted a statistically significant association between both PET imaging measures and Ki-67 values both pre- and post-therapy (Fig.
5), noting that Ki-67 after a short exposure to endocrine therapy has been shown to have predictive value for long-term response [
12]. Taken together, these data indicate promise for both PET tracers as imaging biomarkers of the impact of endocrine therapy on tumor proliferation, with a narrower confidence interval for FLT, as expected by the tighter correlation to proliferation.
In both studies, we found a correlation between baseline and pre-operative uptake and proliferation measures in tissue. Recent studies of serial FLT in breast cancer patients undergoing neoadjuvant chemotherapy showed good correspondence between post-therapy uptake and Ki-67 and were predictive of response [
19,
38]. The value of post-therapy FDG-PET has also been shown for PET imaging studies of breast cancer [
16,
23,
39,
40].
Our hypothesis that change in FDG or FLT uptake between baseline and pre-surgery would predict endocrine response based on post-therapy Ki-67 values was not confirmed (Fig.
6). Response as defined by post-therapy Ki-67 does not show perfect concordance with imaging response, as it did for our pilot study with FDG-PET imaging in patients with advanced disease [
23], but is similar to observations of endocrine therapy impact by others [
21]. One reason for the discrepant results could be the small tumor size in this study. However, our analysis with partial volume correction also did not reveal an association (Additional file
1: Fig. S5). Another possibility is that indolent tumors with low pre-therapy uptake of tracer might respond differently to therapy [
41]. However, tumors with Ki-67 response that lacked response by imaging included both indolent and metabolically active tumors (Fig.
6). We also used a two-tissue compartment model to test dynamic measures of tracer flux, which has been shown to provide greater sensitivity to uptake changes in response to therapy [
41]. The dynamic measures did not reveal the expected association between imaging changes and post-therapy Ki-67, which suggests changes in these radiotracers’ uptake in tumors may be measuring changes in different underlying biological mechanisms than those assessed by changes in Ki-67 values (Additional file
1: Fig. S9).
We prospectively defined imaging response as a 20% decline in FDG-PET [
35] and 15% for FLT based on prior published test/re-test data for these tracers [
36]. Low pre-therapy tracer uptake as well as low Ki-67, present in many patients in this study, likely impacted our findings, especially the ability to measure changes in uptake by SUV [
41]. Within the FDG cohort, 8 patients were classified as metabolic non-responders but had a Ki-67% < 10% at surgery. All of these patients had baseline and pre-surgery SUVmax values of ≤ 3.
Limitations of our studies include a relatively small number of patients and variability in duration of AI therapy; this was due to patient convenience sampling. We obtained the tissue and imaging around the time of the surgical resection. It is possible that treatment lasting longer than the planned 2-week window confounded comparison between PET measures and Ki-67 tissue assays; however, the intervals we encountered are typical of clinical practice and the duration of therapy did not appear to influence the magnitude of change in PET measures or Ki-67 (Additional file
1: Figs. S1–S2). Moreover, in the IMPACT trial, the Ki-67 drop noted at two weeks persisted at 12, suggesting that the decline in tumor proliferation endured at a similar level in patients remaining on therapy [
11]. Both cohorts contained samples of ductal and lobular lesions. Patients with lobular cancers responded well to endocrine therapy, as one would anticipate with endocrine sensitive tumors. There were too few patients with lobular disease, however, to draw any conclusions, although these tumors did not appear to differ significantly from ductal tumors (Figs.
2,
3,
5 and Additional file
1: Figs. S4, S5, S7, and S8). Both pre- and post-menopausal patients were included in the study. While this makes the group potentially more heterogeneous, the therapy was identical, and favorable baseline characteristics similar. Enrolled patients all had operable tumors, and baseline Ki-67 did not differ by menopausal status (
p = 0.3, Wilcoxon rank-sum test). Our findings suggest that both pre- and postmenopausal women’s tumor response can be successfully measured with FDG- and FLT-PET.
Endocrine therapy is powerful treatment for ER+ breast cancer, alone or in synergy with other therapies. In clinical practice, genomic assays routinely determine which patients merit chemotherapy in addition to endocrine therapy [
42]. FDG- and FLT-PET are complementary tools to tissue assays that hold promise to measure early tumor changes to indicate sensitivity in vivo. Several recent and ongoing studies are looking at Ki-67 to stratify which patients require chemotherapy. An in vivo marker of similar response could avert the need for biopsy and allow whole tumor measures of response. FDG and FLT-PET are promising to detect changes in tumor biology early, prior to shrinkage of tumor, and could be used to measure the impact of CDK4/6 inhibitors, increasingly used with endocrine therapy in ER+ breast cancer [
25], or other molecularly targeted agents. FDG-PET is commonly used clinical practice and has a favorable biodistribution for both primary tumors and metastases, is Food and Drug Administration (FDA) approved, and routinely available, but does not directly measure cellular proliferation. FLT, on the other hand, is a validated tracer of cellular proliferation, but is investigational, and while it is well visualized in breast and regional nodal lesions, it has high liver and bone marrow uptake, making its application to metastatic breast cancer more challenging. As other novel tracers are in development, PET imaging can help to evaluate molecularly targeted agents and allow patients to remain on neoadjuvant treatment safely for a longer duration to then achieve a measurable pathologic response at surgery, determine which patients could avoid chemotherapy, and/or which patients will benefit from endocrine therapy alone [
26,
43].
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.