Cell-Proliferation Imaging for Monitoring Response to CDK4/6 Inhibition Combined with Endocrine-Therapy in Breast Cancer: Comparison of [¹⁸F]FLT and [¹⁸F]ISO-1 PET/CT

PET imaging of cell proliferation provides a noninvasive approach for assessing response to cancer therapeutics (1). This capability is especially well-matched to cell-cycle-specific therapeutics that are being increasingly used in the management of metastatic breast cancer. Although, endocrine therapy has been the mainstay of treatment in estrogen receptor (ER)-positive disease, acquired resistance is a major clinical challenge (2). One of the important mechanisms of such resistance involves alteration in cell-cycle regulators such as Rb and cyclin-D1 with disruption of tightly regulated cyclin-dependent kinase 4/6 (CDK4/6). Changes in any of these proteins or genes result in alteration of G₁–S checkpoint and subsequent dysregulated cell-cycle progression, contributing to loss of endocrine responsiveness (3, 4). One emerging strategy to overcome this resistance targets the cross-talk between the ER pathway and cell cycle by combining endocrine therapy with cell-cycle targeted agents such as CDK4/6 inhibitors. This approach utilizes CDK4/6 inhibition to arrest cancer cells at G₁, so that, when combined with ER modulators, they will synergistically inhibit cancer growth resulting in tumor cell senescence (5, 6). Palbociclib is an FDA approved oral inhibitor of CDK4/6 and is increasingly used as the first-line or second-line treatment of ER-positive metastatic breast cancer together with endocrine-based therapy in postmenopausal women (7, 8). Despite the efficacy of this combined approach and its documented impact on progression-free survival, not all patients benefit from the combination therapy. The results of the PALOMA-3 trial showed that 30 of 72 (42%) premenopausal patients with breast cancer treated with palbociclib and fulvestrant demonstrated disease progression or death (9). Additionally, although well tolerated, CDK4/6 inhibitors do carry some toxicity and may not be needed in women who would respond well to endocrine therapy alone. Biomarkers that can monitor changes in tumor proliferation early in the course of therapy to identify patients most likely to benefit from this combinational therapy versus endocrine therapy alone might therefore improve treatment selection for patients with ER-positive breast cancer. The preclinical study described herein was designed to test the feasibility of this approach.

Currently, Ki-67 expression in tissue is the clinical standard technique for evaluating cell proliferation, however this technique is invasive, requiring serial biopsies with potential sampling errors and underestimation of tumor heterogeneity (10, 11). Correlation of Ki-67 to cell proliferation markers such as 3′-deoxy-3′[¹⁸F]fluorothymidine ([¹⁸F]FLT) has been shown for several tumor types including breast cancer; however, the literature is controversial (12, 13).

Molecular imaging with PET utilizing radiotracers that are precursors for DNA synthesis such as [¹⁸F]FLT is an attractive alternative. [¹⁸F]FLT targets the activity of the thymidine salvage pathway of DNA synthesis and quantifies the S-phase fraction of cycling cells, reflecting tumor proliferation (14). [¹⁸F]FLT PET has been used for monitoring treatment response in various tumor types such as lung, head and neck, and breast cancer both in preclinical and clinical studies (15, 16). However, changes in [¹⁸F]FLT retention following treatment is variable depending on the tumors proliferation rate and treatment regimen (17). A multicenter phase II ACRIN trial demonstrated a positive correlation between early decrease in [¹⁸F]FLT after the first cycle of chemotherapy and pathologic complete response in locally advanced breast cancer (13).

However, pulse-label measurement of S-phase fraction of a tumor with [¹⁸F]FLT may be an inherent limitation of this tracer that underestimates tumor heterogeneity and asynchronous growth as it cannot differentiate between proliferating cells in G₁, G₂, and M phases versus quiescent cells at G₀, which is of importance, especially for cell-cycle targeted agents such as CDK 4/6 inhibitors. Mach and colleagues have demonstrated σ₂ receptors which are an alternative biomarker of cell proliferation in breast cancer and developed [¹⁸F]ISO-1, a σ₂-receptor ligand for PET imaging of proliferative tumors (18, 19). Given the high expression of σ₂ receptors in proliferating tumor cells compared with quiescent cells at G₀, this biomarker holds promise for providing a global representation of tumor proliferation by measuring the ratio of cells between proliferative and quiescent states. Preclinical studies have assessed [¹⁸F]ISO-1 PET as a biomarker of proliferation and showed significant correlation with tumor uptake and the proliferation to quiescent ratio. Dehdashti and colleagues reported the first human study to evaluate the safety and feasibility of [¹⁸F]ISO-1 PET in different cancers including breast cancer and found a significant correlation between Ki-67 and tumor-to-muscle uptake ratio (20). They suggested that [¹⁸F]ISO-1 uptake can be helpful in stratification of the tumors based on the proliferation status and this information can be potentially used in treatment planning.

In this study, we evaluated two PET tracers with differing mechanisms for measuring proliferation, [¹⁸F]FLT and [¹⁸F]ISO-1, for monitoring early responses from combination therapy with palbociclib and fulvestrant. As CDK4/6 inhibitors and endocrine therapy target the cell cycle at two different checkpoints, we hypothesized that changes in [¹⁸F]FLT and [¹⁸F]ISO-1 uptake correlate with S-phase depletion and G₀ arrest, respectively, as an indication of contribution of palbociclib and fulvestrant in treatment response (Fig. 1). To better understand how these radiotracers reflect tumor proliferation, we first explored the differences of the tracers in predicting changes in cell-cycle progression in response to CDK4/6 inhibitors, ER-targeted therapy, or both in a panel of breast cancer cell lines with different ER status and sensitivity to palbociclib, noting that CDK4/6 inhibitors have been applied to both ER-expressing and ER-negative breast cancer (21). Subsequently, we assessed in vivo performance of [¹⁸F]FLT and [¹⁸F]ISO-1 PET/CT using a xenograft model of ER-positive human breast cancer sensitive to both agents.

To study the correlation between changes in tumor proliferation in response to CDK4/6 inhibition and [¹⁸F]FLT and [¹⁸F]ISO-1 activity, six human breast cancer cell lines with different ER status and a range of reported sensitivities to palbociclib (22) were selected (Supplementary Table S1). All cell lines were obtained from ATCC in the beginning of the study period (August 2017) and cultured as recommended on the ATCC website. All experiments were performed within 6 months of cell-line purchase. Phenol red-free medium supplemented with 10% charcoal/dextran-treated FBS were used for endocrine-based therapies. Experiments were carried out with cells below 85% confluence for the experiments as the primary focus of this study was evaluation of tumor cell proliferation.

For each experiment, cells were plated at the desired concentration (depending on plating efficiency) and were allowed to grow for 24 hours prior to being exposed to treatment regimens. Estrogen receptor negative cells were plated and treated with 500 nmol/L palbociclib (PD0332991; SelleckChem), whereas ER-positive cells were treated with 500 nmol/L palbociclib, 100 nmol/L fulvestrant (ICI-182-780; Tocris-Bioscience), or a combination of both. Fulvestrant was included in the treatment regimen in ER-positive cells to evaluate the role of combination therapy (7). The treatment doses were selected based on the results of previous reports (23–25). Cells were treated with the indicated agents or DMSO controls for 1, 3, and 6 days and medium were replaced every other day.

For cell-proliferation studies, 10,000 to 15,000 cells were seeded in 12-well plates and treated as described. At each time-point, cells were collected and cell numbers counted using Automated Cell Counter (Countess; ThermoFisher Scientific). To confirm cell growth data and to relate findings to the standard proliferation marker in the clinical setting, we investigated Ki-67 expression using indirect immunofluorescence. Cells were initially cultured on slides, treated as above, fixed in 4% paraformaldehyde, and permeabilized with 0.5% triton X-100. They were incubated with mouse anti-human Ki-67 antibody (Invitrogen) followed by goat–anti-mouse Alexa Fluor-488 conjugated secondary antibody (Invitrogen).

Fluorescent sections were then imaged on a Zeiss Axio Observer (The Netherlands) and Zeiss Zen software was used for quantification of Ki-67 labeling index (percent of positively-labeled/negatively-labeled cells). An average of four regions was used for calculations.

Cell-cycle analysis was used in cell-line studies to track the different impact of palbociclib versus fulvestrant on the cell cycle, which is incompletely captured by the Ki-67 assay. Cell-cycle distribution was assessed using FACSCalibur flow cytometer (BD-Immunocytometry Systems). A multiplexed staining technique was used to enable a targeted assessment of cell-cycle changes to reflect the contribution of each drug in growth inhibition. We used EdU for S-phase assessment and Ki-67 to estimate G₀ versus G₁ accumulation for assessing quiescence (26). Expression of Ki-67 has been reported during active phases of cell cycle (G₁, S, G₂, and M phases), accordingly Ki-67 negative population in G₁–G₀ region can represent G₀ phase (27). Briefly, 20,000 to 40,000 cells were plated in six-well plates and after the treatment period, cells were pulse labeled with EdU (ThermoFisher Scientific) for 1-hour before harvesting. Harvested cells were fixed in 70% ethanol (1 × 10⁶ cells/mL) and stored at −20°C until analysis (28). Fixed cells were initially stained with Alexa Fluor 488 anti-human Ki-67 Antibody (Biolegend) according to the manufacture's protocol and subsequently stained with propidium iodide (Sigma-Aldrich). Click-iT EdU Alexa Fluor 647 Flow Cytometry Assay Kit (ThermoFisher Scientific) was used for accurate measurement of changes in S-phase. Data were analyzed with FlowJo software (Tree Star).

TMEM97 is a conserved integral membrane protein that was recently identified as containing the σ₂-receptor. It has been unofficially renamed as “TMEM97/σ₂-receptor,” and has been shown to have important role in regulating cell survival, morphology, and differentiation (29). Accordingly, protein expression levels of TMEM97 was evaluated as a biochemical surrogate assay for quantifying σ₂ receptors and correlated to [¹⁸F]ISO-1 ligand binding which is a measure of receptor density. For this purpose, cells were plated in 150-mm dishes and treated similar to prior experiments and were harvested and lysed in RIPA-lysis buffer. Protein content was separated by gel electrophoresis. TMEM97 expression was detected using rabbit anti-human TMEM97 antibody (Avivas System-Biology). The intensity of band was normalized using ratio relative to ß-actin as a housekeeping gene.

[¹⁸F]FLT was prepared as a service by our institution's cyclotron facility, as previously described (30). The radiochemical purity of [¹⁸F]FLT was >95% as analyzed with analytical HPLC. Specific activity of [¹⁸F]FLT ranged from (5,500 Ci/mmol) 203,500 GBq/mmol to (8,000 Ci/mmol) 296,000 Ci/mmol. To measure the impact of treatments on [¹⁸F]FLT trapping, we measured cell uptake as a function of tracer exposure time in culture. Cells were plated in 96-well plates at 10,000 to 25,000 cells per well and treated according to the described protocol (14). [¹⁸F]FLT (20 kBq/well) was added and incubated at 37°C for 5, 10, 15, 30, 45, and 60 minutes. Subsequently, radioactivity was assayed on Wizard automatic gamma-counter (PerkinElmer).

[¹²⁵I]RHM-4, chemical analogue of [¹⁸F]ISO-1, is a σ₂-receptor ligand that is used for in vitro pharmacologic assays (31). Given the similar binding properties of [¹⁸F]ISO-1 and [¹²⁵I]RHM-4 and their equivalence in prior in vitro studies (32) and longer half-life of ¹²⁵I (59-days vs. 109-minutes), [¹²⁵I]RHM-4 has been used as a practical analogue for in vitro binding studies. [¹²⁵I]RHM-4 was prepared by an iododestannylation reaction of the corresponding tributyltin precursor (33). Cells were cultured in 150-mm dishes (500,000–1.5 × 10⁶ cells per plate, depending on plating efficiency and growth rate of cell line) and treated according to the protocol. The cells were mechanically scraped, homogenized, followed by saturation binding assay. This approach is tailored to the properties of [¹²⁵I]RHM-4 (and [¹⁸F]ISO-1) as tracers of receptor-mediated binding and has shown to be an optimal evaluation of σ₂-receptor ligand uptake (34). Briefly, the cell membranes were incubated with [¹²⁵I]RHM-4 (0.02–9 nmol/L) for 90 minutes and the nonspecific binding was defined. Then membrane-associated receptor bound ligands were filtered, collected, and counted on a Perkin-Elmer Wizard gamma-counter. The equilibrium dissociation constant (K_d) and maximum number of binding sites (B_max) were determined by a nonlinear regression analysis using Prism GraphPad (version 6.0).

To evaluate the performance of [¹⁸F]FLT and [¹⁸F]ISO-1 PET in predicting response to cell-cycle targeted treatment in the preclinical setting, we selected the MCF7 xenograft model, which is an ER-positive cell line with high sensitivity to both fulvestrant and palbociclib, as confirmed by our in vitro studies. All animal experiments were approved by the Institutional Animal Care and Use Committee. Female SCID mice (6-week-old) were purchased from Charles River Laboratories (Wilmington, MA) and housed in laminar flow cabinets under specific pathogen-free conditions. The body weight and behaviors of mice were monitored on regular basis. Mouse xenograft models were prepared by implanting an estradiol pellet (0.72 mg, 60-day release; Innovative Research) 72 hours prior to the subcutaneous injection of 8 to 10 × 10⁶ MCF7 cells in 200 μL of DMEM:Matrigel (BD-Biosciences) in the left flank. All invasive procedures were performed under general anesthesia using isoflurane. Tumor growth was monitored with caliper measurement and volume was calculated as V = (length × width²)/2. Once the tumors reached an average volume of 100 mm³ (8 weeks), the tumor-bearing mice were randomized into four groups (n = 4) and treated with vehicle (normal saline), palbociclib, fulvestrant (Faslodex; AstraZeneca-Pharmaceuticals), or combination of both for 14 days (23, 24). Palbociclib was administered daily via oral gavage (75 mg/kg). Although this differs from the cyclical use of CDK4/6 inhibitors in patients, prior publication in mouse model support the efficacy of this approach (24, 35). Fulvestrant was injected intramuscularly (5 mg/mice) weekly (35, 36). For the combination-treatment group, mice received both agents. The design of the study was longitudinal and [¹⁸F]FLT and [¹⁸]ISO-1 PET/CT scans were performed in the same animals at baseline, then repeated on days 3 and 14 of treatment.

[¹⁸F]ISO-1 and [¹⁸F]FLT were prepared at our institution's cyclotron facility, as previously described (37). Each mouse underwent [¹⁸F]FLT scan followed by [¹⁸F]ISO-1 on consecutive days. The animals were injected with 7 to 8 MBq of the radiotracer intravenously 45 minutes before micro-PET/CT imaging (MoleCubes; Ghent Belgium). The duration of [¹⁸F]ISO-1 and [¹⁸F]FLT PET imaging was 15 minutes for each mouse. During the micro-PET/CT imaging, the mice were kept under isoflurane anesthesia. Biodistribution assessment was not the focus of this study and as these data were previously published for both tracers, we decided to utilize this available data (20, 38, 39). The PMOD software (PMOD-Technologies Ltd.) was used for co-registration of PET and CT images. Circular-3D region of interest (ROI) were drawn manually to assess the maximum tracer accumulation in the tumor to calculate maximum standardized uptake value (SUV_max), which is commonly used in both clinical and pre-clinical studies (23) and have been use for both tracers in human breast cancer studies (20, 38) and supported by mouse model data on σ₂-receptor expression (39). Tumor SUV_max was corrected based on paraspinal muscle uptake and expressed as tumor-to-muscle ratio (T/M) to correct for differences in tracer delivery with the tail vein injections (20). Attenuation correction was not performed as prior studies have shown that the amount of attenuation of signal in mice is not significant (40).

To confirm PET/CT imaging findings and to better quantify in vivo tracer uptake changes in response to therapy, ex vivo autoradiography was performed on the excised tumors after the last imaging session. Animals were euthanized and the tumors and contralateral quadriceps muscle were excised. Subsequently, the tumors and muscles were frozen, cryo-sectioned (10-μm thick), and exposed to a phosphor imaging plate to assess the distribution of radioactivity. Consecutive tissue slices were also prepared for immunofluorescent study as described below. Digital autoradiography of tissue sections was performed after overnight exposure (Typhoon FLA7000; GE-Healthcare Bio-Sciences). The tumor-to-muscle ratio was calculated for each radiotracer (41). These analyses were performed on four to six tumor sections for each animal.

Immunofluorescent staining: For tissue analysis of markers of proliferation, we stained frozen tumor sections for Ki-67 and TMEM97. Frozen sections were fixed in 4% paraformaldehyde and after permeabilization, the cells were incubated with mouse anti-human Ki-67 antibody (Santa-Cruz Biotechnology) and rabbit anti-human TMEM97 antibody (Avivas System Biology). Alexa Fluor 568 conjugated secondary antibody (Invitrogen) were then applied. Ki-67 and TMEM97 labeling index was calculated as above.

Flow-cytometric analyses: Portion of freshly excised tumors were minced and dissociated with an enzyme cocktail consisting of 0.04% collagenase (Gibco) and 0.05%DNAase I (Worthington Biochemical) followed by digestion with 0.25% trypsin/EDTA (Gibco). The suspension was then filtered through 40 μm cell strainer to obtain single cell suspension. Flow cytometric analysis of DNA content and Ki-67 was performed based on simultaneous analysis of Ki-67 and propidium iodide (27). The percentage of cells in G₀- and S-phase were calculated and correlated to changes in [¹⁸F]ISO-1 and [¹⁸F]FLT PET/CT, respectively.

Data were expressed as mean ± SE. For in vitro phase, Mann–Whitney U test was used to determine statistical significance between experimental and control groups. Longitudinal [¹⁸F]FLT and [¹⁸F]ISO-1 uptake parameters were compared within and between cohorts using the nonparametric Wilcoxon-rank and Mann–Whitney U tests, respectively. Correlations were calculated using linear regression, and expression as Pearson correlation coefficients. Statistical analysis was performed using SPSS statistic software package. For all experiments, P < 0.05 was considered significant.

Combination treatment with palbociclib and fulvestrant in MCF7 and MDA-MB134 resulted in significant inhibition of proliferation on day 1 (Fig. 2A). On day 3, treatment with palbociclib demonstrated a significant growth inhibition in MCF7 and MDA-MB134. Inhibition of ER enhanced the antiproliferative effect of palbociclib (P = 0.009 for MDA-MB134 and P = 0.012 for MCF7), indicating additive effects compared with individual agents. For example, only the combination of palbociclib and fulvestrant showed significant growth inhibition at day 1, and combined therapy showed a greater reduction in Ki-67 on day 6 compared with either agents alone. There were no significant differences in Ki-67 labeling index on day 1 of treatment. However, changes in Ki-67 correlated with cell-proliferation assay on day 3 and 6 (r = 0.78 and 0.89 for MCF7, respectively). There was significant reduction in Ki-67 expression on day 6 of combinational therapy in MCF7 and MDA-MB134 when compared with palbociclib alone (Fig. 2B), confirming the enhanced growth inhibitory effect of combination-therapy (P = 0.040 for MDA-MB134 and P = 0.020 for MCF7). ER-negative cell lines revealed significant growth inhibition in HCC38 and MDA-MB231 after 1 day of treatment (P = 0.006 and 0.008, respectively; Supplementary Fig. S1A). In agreement with findings of Finn and colleagues (22), treatment with 500 nmol/L palbociclib did not have significant effects on proliferation or Ki-67 expression of HCC1806 and UACC732 (Fig. 2; Supplementary Fig. S1).

Early and persistent G₁ arrest with associated S-phase depletion was observed in the ER-positive cell lines, MCF7 and MDA-MB134, after palbociclib and to lesser extent fulvestrant on day 1 (Fig. 3). There was a time-dependent increase in number of cells in G₀ in all treated groups. After 3 days of treatment with fulvestrant, a notable increase in the fraction of cells in G₀ was observed. Addition of palbociclib to fulvestrant led to further increase in percentage of cells accumulating in G₀. Combination therapy resulted in significantly higher G₀ accumulation in MCF7 cells when compared with palbociclib or fulvestrant alone on day 6 (34% vs. 11%, P = 0.001 for palbociclib and 34% vs. 22%, P = 0.016 for fulvestrant). Palbociclib resulted in early arrest at G₁ to S transition resulting in S-phase depletion in HCC38 and MDA-MB231 on day 1 (P < 0.0001). There was subsequent increase in the percentage of cells in G₀, after 3 days of treatment (HCC38, 10% vs. 3%; P = 0.005 and MDA-MB231, 11% vs. 3.4%; P = 0.001), which was further augmented on day 6 (Supplementary Fig. S2). The cell-cycle distribution of the palbociclib-resistant UACC732 was not affected by treatment with palbociclib; fulvestrant resulted in small but significant S-phase reduction (P = 0.012 on day-1). Similarly, treatment with palbociclib did not have significant effects on cell cycle of HCC1806.

Incubation with palbociclib resulted in marked decrease in [¹⁸F]FLT uptake in the sensitive cell lines as early as day 1 (P < 0.0001; Fig. 4A). In ER-positive cell lines, fulvestrant induced a moderate uptake reduction (−42% for MCF7 and −52% for MDA-MB134, on day 1). The combination therapy groups showed enhanced reduction, irrespective of incubation time (−95% for MCF7 and −93% for MDA-MB134, on day 1; Supplementary Table S2). However, differences in uptake when comparing palbociclib with combination therapy group were not significant. The decline in [¹⁸F]FLT uptake correlated significantly with flow cytometric S-phase depletion (e.g., r = 0.98, P < 0.0001 for MCF7 on day 3). The results of [¹⁸F]FLT uptake also correlated with Ki-67 expression on day 3 and 6 of treatment (Supplementary Table S3; r = 0.74 and 0.69, P = 0.009 and P = 0.01 for MCF7 on day 3 and 6, respectively). Similarly, treatment with palbociclib lead to marked decrease in [¹⁸F]FLT uptake in the sensitive ER-negative cell lines (P < 0.0001; Supplementary Fig. S3, panel I).

It is notable that the magnitude of changes in [¹⁸F]FLT uptake and S-phase were different from changes in Ki-67 expression. For example, 3 days of combination therapy in MCF7 cells resulted in 37% reduction in Ki-67 expression whereas there was 94% and 97% decrease in [¹⁸F]FLT uptake and S-phase, respectively. This was expected given that palbociclib induces G₁ arrest, which would diminish S-phase, and thus [¹⁸F]FLT uptake, but not Ki-67, which is positive in all phases of the cell-cycle besides G₀ (27).

We evaluated saturation binding of [¹²⁵I]RHM-4 on day 3 and 6 of treatment in sensitive cell lines. On day 3, we did not identify significant differences in K_d and B_max values between treated cells and controls. However, there was significant reduction in B_max after 6 days of treatment (Fig. 4B; Supplementary Fig. S3, panel II, P = 0.002 and P = 0.01; combination therapy vs. controls of MDA-MB134 and MCF7, respectively), indicating a reduction in σ₂ expression. Accordingly, in MCF7 and MDA-MB134 cells, 6 days of treatment with combination therapy resulted in significant reduction in TMEM97 expression, in line with the results of [¹²⁵I]RHM-4 binding (Fig. 4B). The results of [¹²⁵I]RHM-4 binding also correlated with Ki-67 expression on day 6 (r = 0.68, P = 0.01 for MCF7). Furthermore, HCC38 and MDA-MB231 treated with palbociclib for 6 days resulted in a significant decrease in TMEM97 expression. Palbociclib-resistant cells, HCC1806 and UACC732, did not show a significant difference in TMEM97 expression compared with controls.

There was a significant increase in tumor size only in control mice on day 14 (P = 0.02; Supplementary Fig. S4). Mean baseline [¹⁸F]FLT T/M ratio was 1.98 ± 0.37 with a significant reduction in treated tumors, on day 3 (42% reduction in FLT T/M for palbociclib, P = 0.009, 25% reduction for fulvestrant, P = 0.01, and 44% reduction for combination group, P = 0.004; Fig. 5, panel I A and C). Follow-up [¹⁸F]FLT PET/CT on day 14 revealed a persistent reduction in T/M ratio in all the treated mice, when compared with controls (40% decrease in uptake in combination therapy group).

Baseline [¹⁸F]ISO-1 T/M ratio was 2.64 ± 0.69. Three days of treatment with palbociclib, fulvestrant, or combination therapy did not result in significant changes in [¹⁸F]ISO-1 T/M (P = 0.99, 0.45, and 0.34, respectively; Fig. 5, panel IIA and C). On day 14, combination therapy caused significant reduction in uptake to 1.8|7 \pm $|0.12 (27% mean reduction; P = 0.021). On day 14, T/M ratio in palbociclib and fulvestrant monotherapy groups was 2.24 ± 0.89 and 2.1 ± 0.32, respectively, which was only slightly lower compared with baseline (7% and 13% reduction, respectively), and the difference was not significant (P = 0.086 and 0.051, respectively). We note that modest T/M ratio in for [¹⁸F]ISO-1 in the MCF-7 xenograft model limits contrast recovery from microPET images. This effect is better demonstrated in the tumor autoradiographic data described below.

Two tumors from each group were assessed by ex vivo autoradiography of [¹⁸F]FLT and [¹⁸F]ISO-1. Tumor-to-muscle ratios quantified from ex vivo analysis supported PET/CT findings. [¹⁸F]ISO-1 uptake revealed a decreasing trend in all treated groups with significantly lower uptake in fulvestrant and combination therapy mice (Fig. 6A). A significantly lower [¹⁸F]FLT uptake was observed in all treated mice without any appreciable difference across the treatment groups (Fig. 6B). Treatment with palbociclib or fulvestrant led to a significant decrease in Ki-67 and TMEM97 on day 14 (Fig. 6D) with pronounced reduction in the combination therapy mice. To further evaluate whether the changes in [¹⁸F]FLT and [¹⁸F]ISO-1 corresponded to the markers of proliferation, we correlated the Ki-67 and TMEM97 scoring with autoradiographic T/M ratio of both tracers (Fig. 6C). We detected a strong correlation between Ki-67 and [¹⁸F]FLT uptake (r = 0.87), whereas TMEM97 and Ki-67 demonstrated a moderate correlation with [¹⁸F]ISO-1 uptake (r = 0.57 and 61, respectively; Supplementary Fig. S5B and S5C). Cell-cycle distribution of the tumors was affected by all three treatment regimens, demonstrating significant S-phase depletion in all treatment groups when compared with controls (Fig. 5, panel IB). Palbociclib-treated tumor cells experienced a mild but significant arrest in G₀ (P = 0.022 vs. control). Treatment with fulvestrant resulted in significant arrest in G₀ (P = 0.008 vs. control). The combination therapy led to enhancement of the G₀ quiescence (34% in combination vs. 19% in palbociclib group; P = 0.031 and 23% in fulvestrant group; P = 0.014; Fig 5, panel IIB). The changes in [¹⁸F]FLT uptake correlated with S-phase depletion on flow cytometry (r = 0.78), in line with our in vitro results. We also noticed a reverse correlation between G₀ arrest and [¹⁸F]ISO-1 uptake (r = 0.57).

In this preclinical study, we explored a potential strategy for predicting response to combination therapy with palbociclib and fulvestrant in breast cancer. We assessed two proliferation radiotracers, [¹⁸F]FLT and [¹⁸F]ISO-1, with the hypothesis that the combination of imaging biomarkers might be able to discern distinct patterns of response for the individual agents and their combined use. Several studies have evaluated the performance of the individual tracers (15, 17, 19) but this study is the first to evaluate both tracers side-by-side. We analyzed the antiproliferative effect of combined CDK4/6 inhibition and ER modulation on a panel of breast cancer cell lines and found that both [¹⁸F]FLT and [¹⁸F]ISO-1 measure the changes in cell cycle, which correspond to cell-line drug sensitivity, but with different magnitude and time course. We noted that while [¹⁸F]FLT provides a larger indication of response to the individual agents, [¹⁸F]ISO-1 had superior performance in identifying the added effect of combining an ER-targeted agent with CDK4/6 inhibitor. Next, a MCF7 xenograft breast cancer model was used for in vivo verification demonstrating time-dependent changes in tracer uptake after treatment that were temporally distinct for each radiotracer, supporting the in vitro findings. [¹⁸F]FLT PET correlated with S-phase and detected early changes of tumor proliferation in response to both agents, especially palbociclib, whereas [¹⁸F]ISO-1 PET provided a better measure of the impact of more prolonged therapy of the combination of agents. These results provide new insight into imaging strategies for monitoring response to both early and sustained response to palbociclib–fulvestrant combination therapy.

Breast cancer is a heterogeneous disease with subtypes that have distinct biologic features resulting in various treatment response. Based on this consideration, we chose to perform the in vitro phase using two luminal-A cell-lines (MCF7 and MDA-MB134), one luminal-B cell line (UACC732), and three triple-negative cell-lines (HCC38, MDA-MB231, and HCC1806) to provide a range of potential treatment response. This approach allowed a comprehensive assessment of the biological processes engaged in response to this regimen across different spectrums of breast cancer biology to test the potentials of [¹⁸F]FLT and [¹⁸F]ISO-1 tracers for predicting response. Mechanistically, palbociclib results in the maintenance of Rb tumor suppressor in a hypophosphorylated state leading to G₁ arrest (42). However, fulvestrant with the withdrawal of estrogen effect results in G₀ arrest among other responses (43). Quiescent G₀-phase denotes the nonproliferating state of cells that have withdrawn from the active cell cycle and can shift toward senescence and cell death, which is deemed a clinically favorable response to chemotherapy given the potential for stimulating tumor death or clearance by the immune system (44). As culmination of the ER-mediated reactions is in the control of CDK activity (45), the combination of the two drugs has been proven to be synergistic in inhibition of cell-cycle progression (22). The result of our study is in line with previous reports (22). The early effect of palbociclib in cells was arrest of G₁ to S progression with drastic S-phase depletion and G₁ accumulation on day 1 in all palbociclib-sensitive cell lines, regardless of ER status.

In agreement with these findings, [¹⁸F]FLT uptake in cells reduced after short-term exposure to palbociclib and/or fulvestrant in palbociclib-sensitive and/or ER-positive cell lines, respectively. Addition of fulvestrant in ER-positive cells led to gradual G₀ accumulation that corresponded with time-dependent arrest at G₀–G₁ starting after 3 days of treatment, as indicated by cell-cycle assays in the cell-line studies. When treatment was continued for 6 days, substantial progression of cells to G₀ was achieved in the ER-positive luminal-A cells, especially for combined therapy, whereas maintaining S-phase depletion. Although, the changes in [¹⁸F]FLT uptake mirrored S-phase changes, it failed to predict the subsequent changes due to delayed G₀ arrest. Accordingly, for combined therapy, [¹⁸F]FLT predominantly reflected the contribution of palbociclib whereas changes due to fulvestrant remains undervalued. To address this shortcoming, we interrogated the associated cell-cycle changes and their influence on σ₂-receptor radiotracers to assess whether this technique can predict transition of cells from proliferating status to quiescence. In our study, changes in [¹²⁵I]RHM-4 binding was time dependent and a significant reduction was noted only on day 6 in ER-positive luminal cells treated with combinational therapy. These findings demonstrated a significant reversed correlation of [¹²⁵I]RHM-4 binding with accumulation of cells in G₀ on day 6, indicating induction of quiescence mainly due to the contribution of fulvestrant and its synergy with palbociclib.

We selected MCF7 cell line, for in vivo testing in xenograft models as the best available cell-line representative of the most common form of breast cancer in the clinical practice, the luminal-A subtype (46). We performed PET/CT at baseline and after 3 and 14 days of treatment to reflect the immediate and delayed posttreatment changes that were noted in vitro. Significant reduction in [¹⁸F]FLT T/M uptake was noticed in all the treated mice both on day 3 and 14 of treatment (approximately 18%–50% reduction), correlating to S-phase reduction, corresponding to in vitro findings, supporting the role of [¹⁸F]FLT as an early indicator of therapeutic efficacy. Despite the robust posttreatment changes in [¹⁸F]FLT, we did not detect any significant differences between our treatment groups and also different timelines, indicating a limitation of [¹⁸F]FLT as a stand-alone biomarker to identify the contribution of each component of the combinational therapy. Our results revealed no significant change in [¹⁸F]ISO-1 uptake after 3 days of treatment, despite the dramatic [¹⁸F]FLT changes. The simplest explanation for this observation is that the short-term treatment leads to early changes in S-phase fraction and arrest at G₁. Continued exposure leads to transition to an increase in the fraction of cells in the quiescent (G₀) phase, not well reflected by [¹⁸F]FLT. However, the increase in quiescence was well measured by [¹⁸F]ISO-1 PET imaging, showing a significant decline in uptake after 14 days of combinational therapy correlated with G₀ arrest on tissue analysis.

Histologic assessments of proliferation markers validated the strong correlation between [¹⁸F]FLT uptake and Ki-67 expression after 14 days of treatment, in line with previous reports (47). Expression of TMEM97 also paralleled the uptake of [¹⁸F]ISO-1, which confirmed the PET results by indicating that the decrease in tracer uptake was due to a real change in expression of proliferation marker, σ₂-receptor, and not a consequence of decreased delivery of radiotracer. However, the correlation between TMEM97 labeling index and [¹⁸F]ISO-1 T/M ratio was only moderate that might be due to the fact that we used a tissue-based technique by immunofluorescence for quantification of TMEM97 and future studies using more direct quantification technique are required before drawing a conclusion.

Our study has several limitations. The animal model in this study was used as a proof of concept and only a single cell line was used. We note that MCF-7 cells are somewhat modestly proliferative and larger changes may be noted in more proliferative human tumors. We also note that the daily administration of palbociclib used in this study differs from cyclic use in patients (9) and may need to be considered in translation of the results to patient studies. We also noted modest T/M ratios for [¹⁸F]ISO-1 in our MCF-7 xenograft, which might also be increased in human breast cancer (20). Additionally, optimizing the timing of sequential PET imaging was challenging to ensure timely imaging to reflect posttreatment changes at the similar interval for both radiotracers and at the same time avoid contamination of signals. In order to minimize the effect of first radiotracer on the subsequent set of imaging we allowed 18 to 20 hours interval to allow decay of the tracer. These issues will need to be considered if the approach is tested in future patient studies. It is notable that the proliferation tracers used in this study are not specific to BC or the mentioned combinational treatment and the xenograft model is a proof of concept for validating the in vitro behavior of each tracer. This concept can be applied in the setting of other cell-cycle targeted treatment regimens in oncology.

[¹⁸F]FLT and [¹⁸F]ISO-1 PET probes can evaluate the time-course of changes that occurs in response to cell-cycle targeted therapy targeting CDK4/6, endocrine agents, and their combination in a preclinical model of breast cancer. Comparison of the two radiotracers revealed that [¹⁸F]FLT measures immediate changes in S-phase as a predominate effect of targeting CDK4/6, providing a very early prediction of tumor response, whereas [¹⁸F]ISO-1 can assess relatively delayed changes reflecting cell-cycle arrest and transition to quiescence. Future studies are needed to determine the timing of [¹⁸F]ISO-1 and [¹⁸F]FLT PET/CT imaging for prediction and monitoring tumor-proliferation changes in the setting of cell-cycle targeted therapy; however, these early results indicate the potential of this approach to guide combined CDK4/6 inhibitors and anti-ER therapy in breast cancer and merit further preclinical studies and early-phase clinical studies.

No potential conflicts of interest were disclosed.

Conception and design: A. Elmi, A.S. Clark, R.H. Mach, D.A. Mankoff

Development of methodology: A. Elmi, R.H. Mach,

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Elmi, C.-C. Weng, C. Hou,

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Elmi, C.-C. Weng, A.S. Clark, D.A. Mankoff

Writing, review, and/or revision of the manuscript: A. Elmi, M. Makvandi, C.-C. Weng, A.S. Clark, R.H. Mach, D.A. Mankoff

Study supervision: R.H. Mach, D.A. Mankoff,

The authors thank Kuiying Xu, PhD, at Dr. Mach's radiochemistry lab for assistance with radiochemistry; the University of Pennsylvania Small Animal Imaging Facility staff for technical assistance; and the Cyclotron staff for production of radiopharmaceuticals. They also appreciate the excellent support of Dr. Eric Brown and the technical staff at the Brown Lab, in particular Dr. Veena Jagannathan for facilitating flow-cytometry study. This work was supported by Radiological Society of North America (RSNA) Research & Education Foundation resident grant (to A. Elmi), Susan G. Komen Foundation grant (to D. Mankoff SAC130060 and A. Clark CCR16376222), and NIH Grant 5T32EB004311 for radiology research track residency.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.