Abstract
Chemotherapy-related cognitive impairments (CRCIs) encompass cognitive deficits in memory, attention, and executive function that arise during and following chemotherapy. CRCI symptoms are predominantly reported by female cancer patients but also occur in males. These impairments may involve reduced estradiol levels, which then increases vulnerability to the impact of tumors and chemotherapy on cognition. This study utilized the MMTV-PyVT mouse model of breast cancer to test the hypothesis that impaired ovarian function and associated estradiol levels play a critical role in CRCI susceptibility. Mice were either ovariectomized (OVX) or underwent sham surgery. The OVX group then received supplemental estradiol (E2) ad libitum in the drinking water to maintain physiological hormone levels. After tumor development, mice were trained in the Morris water maze to assess spatial memory, and subsequently, they received weekly injections of either saline or a combination of cyclophosphamide (CYP; 66.7 mg/kg, i.v.) and doxorubicin (DOX; 6.7 mg/kg, i.v.) for 4 weeks. Spatial memory was reassessed 10 d and then 35 d, after the final injections. Results demonstrated a significant disruption of normal ovarian cycling in sham-operated mice treated with CYP + DOX, as well as significant spatial memory impairments when compared with OVX mice supplemented with E2. This study suggests that chemotherapy-induced ovarian damage and the consequent drop in circulating estrogens significantly contribute to vulnerability to CRCIs, underscoring the importance of estradiol in mitigating CRCI risks.
Significance Statement
This study demonstrates that impairment of ovarian function in sham mice contributes to the manifestation of chemotherapy-related cognitive impairments (CRCIs). The observation that cognitive impairments were not present in E2-supplemented ovariectomized (OVX) animals suggests that disruptions of circulating E2 by chemotherapy mediate vulnerability to CRCIs. Unfortunately, OVX mice receiving E2 supplementation exhibited increased tumor volume in comparison with sham mice receiving tap water, suggesting that maintaining E2 can interfere with treatment. Therefore, E2 supplementation is unlikely to be an acceptable therapeutic intervention for CRCIs in cancers with tumors that express estrogen receptors. However, targeting mechanisms downstream of E2, such as high-affinity choline uptake, may be effective interventions for CRCIs.
Introduction
Chemotherapy-related cognitive impairments (CRCIs) affect up to 75% of cancer patients receiving chemotherapy. CRCIs, which include deficits in spatial memory, attention, processing speed, and executive function, can persist for 20 years following treatment, significantly affecting patients’ quality of life (Koppelmans et al., 2013). Although CRCIs are observed in both men and women, most patients who report CRCIs are women (Janelsins et al., 2014), suggesting that women may be more vulnerable than men to the adverse cognitive effects of chemotherapy. Multiple mechanisms have been identified as contributors to CRCIs including direct central nervous system (CNS) damage from chemotherapeutic agents as well as indirect damage through induced neuroinflammation and oxidative stress (Fernandez et al., 2020). However, these mechanisms do not explain why women report CRCIs more frequently than men.
Many chemotherapeutic agents cause gonadal suppression or ablation, reducing levels of circulating sex hormones, and many women receiving chemotherapy exhibit amenorrhea and early menopause, resulting in decreased circulating estrogen levels (Walshe et al., 2006). It is well established that fluctuations in circulating estrogen levels affect the cognitive performance of females on a variety of behavioral tasks suggesting that damage to the ovarian follicles and resulting impairment in circulating estrogens may contribute to cognitive impairment (Rosenberg and Park, 2002). To this point, women experiencing amenorrhea following chemotherapy exhibit significant cognitive deficits when compared with chemotherapy patients who do not experience amenorrhea, implicating reductions of circulating estrogen in the manifestation of CRCIs (Walshe et al., 2006).
In support of findings from clinical studies, research in rodents has demonstrated that the chemotherapeutic agents cyclophosphamide (CYP; Cytoxan) and doxorubicin (DOX; Adriamycin) reduce the size and number of ovarian follicles (Mattison et al., 1981; Y. Wang et al., 2019). This is particularly relevant because CYP and DOX are two of the more commonly used chemotherapeutic agents and are often used in combination (AC chemotherapy) for the treatment of breast cancer. Weekly injections of subtherapeutic doses of CYP + DOX have been demonstrated to induce spatial memory deficits in female, but not male, Balb/C mice, suggesting greater vulnerability in females. Furthermore, research modeling CRCIs indicates that suppression of the estrus cycle in breast tumor-bearing MMTV-PyVT female mice using therapeutic doses of CYP + DOX is associated with poor spatial memory in the Morris water maze (MWM; Philpot et al., 2016).
The ovarian follicles are the primary source of circulating estrogens. Estrogens have several neuroprotective effects including anti-inflammatory actions in the CNS (Vegeto et al., 2008), inhibitory effects on microglia activation (Drew and Chavis, 2000; Vegeto et al., 2001), and inhibitory effects on the intracellular localization of NF-κB (Ghisletti et al., 2005), preventing downstream activation of inflammatory genes. Estrogen also increases glutamate uptake in astrocytes (Arevalo et al., 2010), preventing glutamate toxicity, and can act as an antioxidant (Razmara et al., 2007). Importantly, cancer patients exhibit increased levels of proinflammatory cytokines when compared with healthy controls (Rego et al., 2014), and several chemotherapeutic agents, including CYP and DOX, increase proinflammatory cytokines (Janelsins et al., 2012).
Thus, chemotherapy-induced ovarian suppression or ablation and the resulting reductions in circulating estradiol (E2) may render females uniquely vulnerable to chemotherapy-induced neurodegenerative processes that can promote neuronal injury (Ahles and Saykin, 2007).
The goal of the present study was to test the hypothesis that the adverse effects of CYP + DOX on spatial memory in female mice are dependent on vulnerability of the ovaries to ablation by chemotherapy. To this end, the spatial memory of ovariectomized (OVX) and sham-operated tumor-bearing female MMTV-PyVT mice was assessed prior to and following repeated CYP + DOX administration. Tumor-bearing mice were used because evidence suggests that the presence of tumors themselves may contribute to neurotoxicity and cognitive impairments (Denlinger et al., 2014; Olson and Marks, 2019). OVX mice were placed on E2-supplemented water to maintain physiological concentrations of E2 in the circulation. This manipulation allowed for systemic CYP + DOX to produce all adverse effects, known and unknown, with the exception of ovarian ablation and reducing circulating E2. The sham surgery group remained on tap water, rendering them vulnerable to chemotherapy-induced ovarian suppression or ablation and consequential reductions in circulating E2. Results indicate that spatial memory following the repeated administration of a therapeutic dose of CYP + DOX is protected when circulating E2 concentrations are maintained but is impaired when the ovaries are vulnerable to damage from chemotherapy, a consequence that reduces circulating E2.
Materials and Methods
All procedures involving animals were performed in accordance with the animal care committee's regulations. Breeders and their offspring were given ad libitum access to water and standard Teklad Global 18% protein rodent chow (TD.2018, Envigo). B6.FVB-Tg (MMTV-PyVT) 634Mul/LellJ mice (Jackson Laboratory) were bred in-house using hemizygous male mice and C57BL/6J female mice. This model was selected because it has a close similarity to human breast cancers requiring chemotherapeutic intervention (Guy et al., 1992; Maglione et al., 2001; Lin et al., 2003). Additionally, the B6.FVB-Tg strain is cross-bred with C57Bl/6J mice and was chosen over the FVB-TgN strain because the former exhibits slower tumor progression and lower metastasis rates than the latter, allowing for long-term studies of the effects of chemotherapy on cognitive function. Crucially, the B6.FVB-Tg strain is not vulnerable to the retinal degeneration and reduction of visual acuity that occurs in the FVB-TgN strain, making the former suitable for studies where intact vision is important.
Genotyping
On postnatal day (P)21, when in-house–bred MMTV-PyVT mice were weaned, a tissue sample from the tail of each female mouse pup was collected and genotyped (Transnetyx) as MMTV-PyVT: Tg negative or positive. MMTV-PyVT: Tg-positive female mice (N = 53) were subsequently housed 3–4 per cage. MMTV-PyVT: Tg-negative females were used for other studies. Male mice were killed via CO2 inhalation followed by cervical dislocation.
OVX
Female mice were randomly assigned to either OVX (N = 28) or sham surgery (N = 25) groups between P63 and P69 in young adulthood. Mice were then anesthetized with isoflurane (3–5%) in an induction chamber, removed, and placed on a rodent-specific nose cone apparatus with maintenance anesthesia (1–2%). Anesthetic depth was confirmed via toe pinch. Meloxicam (5 mg/kg, s.c.) and buprenorphine-sustained release (1.0 mg/kg, s.c.) were administered as multimodal analgesia at the time of surgery. Ancillary heat was provided using warm water circulation heating pads. The ovaries were removed via bilateral incisions (5 mm) midway between the base of the tail and the middle of the back. The location of the ovarian fat pad under the muscle was identified, and a small incision was made through the muscle. The ovaries were pulled through the incisions using blunt forceps by grasping the fat pad surrounding it, blood vessels were ligated, and the ovaries were removed. The muscle layer was then closed using absorbable sutures, with the skin closed via wound clips. For sham surgery mice, the same procedures were performed except for the ligation of blood vessels and removal of the ovaries. Following surgery, mice were individually housed to recover for 1 week. Meloxicam (5–10 mg/kg, s.c.) was given every 24 h for 48 h after surgery.
Estradiol
Following surgery, OVX mice began estrogen replacement by supplementing the drinking water with 17β-estradiol (440 ng/ml). This method has been used to effectively maintain circulating estrogen at physiological levels (Gordon et al., 1986; Groeneveld et al., 2004). Water was monitored daily and changed regularly to verify normal fluid intake.
Timeline, weights, tumor assessment, and measurement
Mice were weighed and monitored weekly for tumor manifestation via manual manipulation and external tumor measurement using calipers. When a tumor >50 mm3 (using the modified ellipsoidal formula: volume(tumor) = ½(length × width2) was detected, the first day of that week was designated as Day 1 for that mouse (Fig. 1). Mice remained in the study unless a tumor ulcerated and reached a diameter >15 mm or a total volume of 2cm3 or other complications occurred. Mice were evaluated weekly for clinical signs requiring killing per Institutional Animal Care and Use Committee guidelines: ruffled coat hair, hunched posture, impaired mobility, cachexia (weight loss >20% of body weight), dehydration, anorexia, or dyspnea.
Study design and timeline for evaluating the impact of OVX and chemotherapy on spatial memory in MMTV-PyVT mice. This study began with 53 adult female MMTV-PyVT mice. Between P63 and P69, the mice underwent either sham surgery or OVX. OVX mice received estrogen (E2)-supplemented water postsurgery, while sham-operated mice had regular tap water. Mice were assessed weekly for tumor development, with Day 1 marking when tumors reached 50 mm3. Orientation to the MWM occurred on Days 11 and 12. Mice were trained from Days 13 to 17 to locate a consistently assigned hidden platform in one of the six zones. Proestrus status was checked on Days 15–18, 43–46, and 71–74. A baseline spatial memory test, including a probe trial, was conducted on Day 18. Mice were injected weekly on Days 19, 26, 33, and 40 with either a chemotherapy regimen (CYP and DOX) or saline. Reassessments of spatial memory in the MWM were conducted on Days 46 and 74 to gauge chemotherapy effects. The study concluded with 36 mice.
Starting with a tumor volume of 50 mm3 helps to ensure that all mice have tumors at a comparable stage of development and that the tumors are sufficiently established to mimic early-stage breast cancer. Importantly using this small tumor volume as a starting point allows sufficient time for the study of tumor-bearing controls who do not receive chemotherapy.
Estrus cycle measurements
Assessment of proestrus status was conducted on Days 15–18, 43–46, and 71–74 using an Electronic Vaginal-Estrous Cycle-Monitor (Stoelting). Four-day windows for measurement were used because mice typically exhibit a 4 d cycle. The Electronic Vaginal-Estrous Cycle-Monitor is used to identify the different phases of the estrous cycle in rodents by measuring electrical resistance or impedance in the vagina. Elevated estrogen stimulates the proliferation of epithelial cells in the vagina, and impedance is higher (>3 kΩ) during proestrus, when the vaginal epithelium is thickest than during other phases (estrus, metestrus, diestrus). For the purposes of this study, a threshold of >4 kΩ was considered proestrus.
Behavior
MWM
The MWM was chosen to assess cognitive function based on studies in mice indicating that chemotherapeutic agents impair performance in this task and findings in humans that spatial memory is impaired by chemotherapy (Hurria et al., 2006; Winocur et al., 2012, 2006; Philpot et al., 2016; Rendeiro et al., 2016).
Apparatus. A circular pool measuring 130 cm in diameter was filled with water at a temperature of 22–24°C. A clear Plexiglas-tiered platform, with a bottom-tier diameter of 15 cm and a top-tier diameter of 9.5 cm, was positioned in the water during all nonprobe trials. To visually obscure the platform, the water was made opaque with white nontoxic tempera paint (Tri Art). The pool was surrounded by a white polyvinyl curtain to block visual cues that might orient the subjects toward the experimenter or other external stimuli. Inside the curtain, distinct visual stimuli were placed to serve as static cues for spatial orientation. All behaviors were recorded by a video camera installed above the pool.
Behavior quantification within the pool utilized the Noldus EthoVision tracking software, which was used to define six equal-sized “virtual” zones within the pool. The software measured various parameters for each trial, including the average distance from the pool's perimeter, average swim velocity (VEL), total distance moved, latency to each zone, entries to each zone, latency to the platform, and time spent in the platform zone. The percentage of entries into the platform zone and the percentage of total distance moved (TDM) within the platform zone were subsequently calculated.
Training and baseline assessment. Beginning on Day 11, mice received orientation training consisting of five trials/day for a period of 2 d. This procedure has been demonstrated to improve the performance of female animals (Perrot-Sinal et al., 1996). For each trial, mice were placed in the pool facing the wall and allowed to swim to a visible platform. The starting location of the mice and the location of the platform were varied on each orientation trial, and the presence of the platform was indicated during orientation training by a Plexiglas rod (15 cm long, 3.8 cm diameter) that extended vertically from the platform 14.5 cm above of the surface of water. At the end of each trial, mice were towel dried and placed in a heated holding cage for 2 min before being returned to their home cages.
Spatial memory training consisted of five trials/day on Days 13–17. The intertrial interval was ∼25 min for each animal. During each trial, the platform was placed a few centimeter below the surface of the water. The platform location was varied across all six zones but remained constant for individual mice, so they could learn its location relative to cues in the environment. On every training day, each mouse was introduced into the pool from each of the five starting zones that did not contain the platform. The order of placement in each of the five zones was randomized for each animal. Animals were placed in the water, facing the wall, to start the trial. Each trial continued until the animal mounted the platform with all four paws or until 60 s elapsed. If a mouse failed to locate the platform within 60 s, the animal was gently guided to the platform. All mice were left on the platform for a minimum of 15 s before being removed from the water. On Day 18, Trials 1 and 2 occurred as described, with Trial 3 serving as a probe trial to assess spatial memory. During the probe trial, the platform was removed, and mice were allowed to swim for a total of 60 s before being removed from the apparatus. Following the probe trial, two additional platform trials were performed (Trials 4 and 5) to prevent possible extinction of spatial memory.
Following training and baseline assessment in the MWM (Day 18), mice that demonstrated spatial memory of the platform location by devoting ≥20% of their probe trial exploration to the zone that contained the platform during training were randomly assigned to saline or chemotherapy groups.
Reassessment of spatial memory. On Days 46 and 74, spatial memory was reassessed in the MWM to determine the effect of chemotherapy on spatial memory and the persistence of these effects, respectively. Mice received five trials on reassessment days with the platform for each individual mouse in the same location as during training on Trials 1, 2, 4, and 5. Trials 1 and 2 served to refamiliarize the mice with the MWM after a 28 d absence. To assess spatial memory, on Trial 3 (probe trial), we removed the platform from the pool, and we allowed the mouse to search for 60 s after being placed in a random, nonplatform zone. Trials 4 and 5 served to minimize extinction effects associated with the probe trial. The intertrial interval was ∼25 min for each animal.
Chemicals and injections
Saline (0.9%) was made in-house using sterile ddH2O. CYP was obtained as a solid from Tocris Bioscience and dissolved in 0.9% saline to a concentration of 20 mg/ml. DOX (2 mg/ml) was obtained from Teva Parenteral Medicines. Based on group assignment at the baseline, mice received weekly injections of either CYP (66.7 mg/kg, i.v.) + DOX (6.7 mg/kg, i.v.) or equivalent volumes of saline on Days 19, 26, 33, and 40. To mimic clinical treatment, we administered CYP and DOX over four cycles, a common number of infusions for AC chemotherapy. The cumulative doses of CYP and DOX administered to mice correspond to 800 and 80 mg/m2, respectively, doses in the 4 week cumulative dose range for standard breast cancer chemotherapy in humans.
Attrition
A total of 10 mice, six sham surgery and four OVX mice, did not demonstrate spatial memory of the platform location during the baseline probe trial and were removed from further study. Additionally, two sham mice receiving saline, three OVX mice receiving saline, and one OVX mouse receiving chemotherapy were removed during the study due to the tumor size or ulceration, and their data were not included in any analysis. Lastly, one mouse receiving chemotherapy was removed due to weight loss. At the end of the study, there were 8–10 mice per group and a final N of 36 for statistical analysis.
Statistical analysis
Statistical analyses were performed using SPSS 27 (IBM). For all analyses, an a priori alpha of 0.05 was used to define statistical significance. In instances where using repeated-measure or mixed-factor ANOVA revealed a violation of sphericity, the Greenhouse–Geisser correction was used for all determinations of statistical significance. All post hoc tests were performed using Sidak's correction for multiple comparisons.
To assess whether surgery and chemotherapy altered the frequency of proestrus, χ2 analyses were performed on group proestrus frequencies obtained on Days 15–18, following four injections (Days 43–46), and at the end of the study (Days 71–74).
To assess weights, we performed a 2 (group) × 2 (treatment) × 10 (timepoint) mixed-factor ANOVA. To assess tumor number and tumor volume, we performed 2 (group) × 2 (treatment) × 12 (timepoint) mixed-factor ANOVA.
The VEL, latency to the platform (LTP), TDM, and platform zone entries (PZE) during the first platform trial on Day 18, after 4 weeks of chemotherapy (Day 46) and 5 weeks postchemotherapy (Day 74) were analyzed using 2 (group) × 2 (treatment) × 3 (timepoint) mixed-factor ANOVA.
To assess general activity, we analyzed TDM and VEL during the probe trial on Day 18, after 4 weeks of chemotherapy (Day 46) and 5 weeks postchemotherapy (Day 74) using 2 (group) × 2 (treatment) × 3 (timepoint) mixed-factor ANOVA.
To assess memory of the platform location, we analyzed the distance moved in the correct zone (DMCZ), entries into the correct zone (ECZ), time in the correct zone (TCZ), and latency to the correct zone (LCZ) during the probe trial on Day 18, after 4 weeks of chemotherapy (Day 46) and 5 weeks postchemotherapy (Day 74).
Partial correlations were performed to assess relationships between DMCZ, ECZ, TCZ, and LCZ to determine suitability for use in multivariate analysis (MANOVA) of spatial memory. Subsequently, a 2 (group) × 2 (treatment) × 3 (timepoint) MANOVA was conducted using moderately correlated measures (r = 0.2–0.7). According to Hair et al. (1998), moderate correlations (0.2 to 0.7) provide a good balance between having related variables that are not overly redundant, and MANOVA is most effective when the DVs are moderately correlated. This structure ensures that the covariance matrices are balanced and the assumption of homogeneity is more likely to be met.
To characterize MWM behavior on probe trials while controlling for individual differences in total activity, we analyzed the proportion of total movement devoted to each zone during the probe trials using a 2 (group) × 2 (treatment) × 3 (timepoint) × 6 (zone) mixed-factor ANOVA.
Results
Proestrus frequency analysis
The effect of surgery (sham/tap vs OVX/E2) and chemotherapy (saline vs CYP + DOX) on the frequency of proestrus on Days 15–18, 43–46, and 71–74 was assessed using χ2 analysis for independence. At baseline, following the OVX or sham surgeries, proestrus was observed in 89% of sham/tap mice and 91% of OVX/E2 mice and did not differ across groups. The observation of proestrus in OVX mice may be confusing; however, estrogens promote the proliferation of epithelial cells in the vagina and the Electronic Vaginal-Estrous Cycle-Monitor measures impedance associated with the thickness of the vaginal epithelium. Thus, this result indicates that the 17β-estradiol in the drinking water (440 ng/ml) of OVX/E2 mice maintained physiological levels of estrogen at concentrations sufficient to thicken the vaginal epithelium.
The χ2 analysis revealed large (ϕ = 0.92) and significant [X2 (3, N = 36) = 30.44; p < 0.05] changes in the frequency of proestrus following four weekly injections of saline or CYP + DOX. On Days 43–46, 89% of saline-injected sham/tap mice exhibited proestrus, consistent with baseline measurements. Similarly, 80% of saline-injected OVX/E2 mice and 78% of CYP + DOX-injected OVX/E2 mice exhibited proestrus. However, only 56% of CYP + DOX-injected sham/tap mice exhibited proestrus over 4 d of measurement. This frequency was significantly lower than all other groups suggesting that CYP + DOX impaired normal estrus cycling in mice with intact ovaries (Fig. 2). When proestrus frequency was reassessed on Days 71–74, large and significant differences between groups remained [ϕ = 0.92; X2 (3, N = 36) = 30.4; p < 0.05], with saline-injected sham/tap mice, saline-injected OVX/E2 mice, and CYP + DOX-injected OVX/E2 mice all exhibiting proestrus frequencies between 78 and 89%, while only 56% of CYP + DOX-injected sham/tap mice exhibited proestrus, indicating a persistent impairment of estrus cycling in these mice.
Changes in proestrus frequency across treatment phases. The percentage of sham-operated (sham/tap) and OVX/E2 mice exhibiting proestrus during three key phases: baseline (Days 15–18), after 4 weeks of chemotherapy (Days 43–46), and 5 weeks postchemotherapy (Days 71–74). Red shaded areas highlight the periods when mice received weekly injections of either chemotherapy agents (CYP and DOX) or saline. Significant within-group changes from the baseline are marked with arrows, and between-group differences at the same timepoints are indicated by asterisks. Statistical significance was determined by a p value of <0.05 following Sidak's correction for multiple comparisons.
Tumor and weight changes
Weights
Factorial ANOVA demonstrated a large (partial η2 = 0.125) and significant difference in group (sham/tap vs OVX/E2) weights over time (F(9,288) = 4.554; p < 0.05). There was no significant difference in weights between groups at the start of the study (Table 1). However, 3 weeks after final injections (Day 64) and until the end of the study (Day 78), Sham/Tap mouse weights consistently increased at a slower rate (0.41–0.56 g/week) than for OVX/E2 mice (0.94–1.21 g/week), with OVX/E2 mice weighing significantly more by the end of the study.
Weekly measurements of weights, tumor counts, and tumor volumes
Tumor number
Factorial ANOVA revealed large (partial η2 = 0.347) and significant differences in tumor number changes over time due to treatment (F(11,352) = 17.004; p < 0.05). From Day 8 to Day 43, there was a significant stepwise increase in the tumor number of saline-injected mice which subsequently plateaued. In contrast, mice in the CYP + DOX groups exhibited an increase in tumor number from Day 1 to 15 which then plateaued during chemotherapy from Days 22 to 43 and only increased significantly again on Day 78, demonstrating the effectiveness of CYP + DOX in suppressing tumor emergence.
Tumor volume
Factorial ANOVA revealed large (partial η2 = 0.11) and significant changes in tumor volume (Table 1) over time due to interactions with the group and treatment (F(11,352) = 3.954; p < 0.05). During the initial phase of the study, from Days 1 to 15 (the preinjection window), an increase in total tumor volume was exhibited in all animals, increasing by 344% (CI95 311%, 379%). From Days 15 to 43, the tumors of saline-injected mice exhibited an increase in tumor volume of 393% (CI95 369%, 419%). By contrast, from Days 15 to 43, the tumors of CYP + DOX-injected mice exhibited a 44% decrease (CI95 35%, 51%). This change was consistent regardless of the animals’ OVX status, underscoring the effectiveness of CYP + DOX in reducing tumor volume. From Days 43 to 78, tumor volume in the saline-injected mice increased by 304% (CI95 285%, 323%), while tumor volume changed by 687% (CI95 587%, 807%) among mice previously receiving CYP + DOX during the same time span. Tumor volumes in the CYP + DOX group remained relatively smaller than those in the saline group by the end of the study (Day 78).
MWM
Test day platform trials
Platform trials in the MWM should not be considered a measure of spatial memory specifically because the platform can be located without utilizing spatial cues if an effective search strategy is established (i.e., swimming in an expanding or reducing spiral or in concentric circles of increasing or decreasing diameters). However, performance on the first platform trial of each day can indicate group differences in retention of memories from prior sessions. Therefore, we examined group differences in VEL, LTP, TDM, and PZE on the first platform trial at the baseline (Day 18) and on reassessment Days 46 and 74 for indications of a disruption of long-term memory.
Analyses of Platform Trial 1 results on Days 18, 46, and 74 revealed a significant group by treatment interaction for LTP (partial η2 = 0.128; F(1,31) = 4.542; p < 0.05) and TDM (partial η2 = 0.132; F(1,31) = 4.72; p < 0.05) as well as a significant change across days of assessment for VEL (partial η2 = 0.404; F(2,62) = 20.99; p < 0.05), LTP (partial η2 = 0.135; F(2,62) = 4.83; p < 0.05), and TDM (partial η2 = 0.100; F(2,62) = 3.45; p < 0.05).
Subsequent post hoc analyses using Sidak's correction for family-wise error revealed that VEL was significantly lower on Day 74 than on Days 18 or 46 (Fig. 3A). Because this effect did not differ between OVX/E2 mice and sham/tap mice or between mice receiving saline or chemotherapy, the effect likely reflects a tendency toward reduced activity as the mice mature and have more experience in the MWM.
Additional post hoc analyses revealed that, when compared with sham/tap mice receiving saline injections, sham/tap mice receiving chemotherapy took significantly longer and traveled significantly further to reach the platform on Day 74 Platform Trial 1, suggesting a chemotherapy-induced impairment. For OVX/E2 mice, post hoc analyses revealed that the Platform Trial 1 LTP (Fig. 3B) and TDM (Fig. 3C) was significantly lower at the baseline (Day 18) than observed in sham/tap mice; however, these differences were not present on Days 46 or 74. Additionally, the Platform Trial 1 LTP and TDM of OVX/E2 mice increased significantly from Day 18 to Days 46 or 74, but did not differ between OVX/E2 mice receiving saline or chemotherapy, indicating that this effect was associated with OVX and/or E2 supplementation and not a chemotherapy-induced change.
Test day probe trials
General activity
Because the presence and development of tumors, surgical manipulation, and/or the administration of chemotherapeutics could alter activity in the mice and distort or confound the interpretation of spatial memory measures taken on the Day 18, 46, and 74 probe trials, we analyzed the TDM and VEL during the probe trials for differences between groups and changes across days of assessment (Table 2). Results of mixed-factor ANOVA for both TDM and VEL indicated a general tendency for reduced activity across days of assessment (partial η2 = 0.481; F(2,64) = 29.706; p < 0.05). Both TDM and VEL were significantly lower on Day 46 than Day 18 and significantly lower on Day 74 than on either Day 18 or Day 46. Because there were no significant effects of the group or treatment on activity, this effect like reflects a tendency toward reduced activity as the mice mature and have more experience in the MWM, as observed in the Platform Trial 1 data.
TDM and VEL during probe trials
Correlations
Prior to analysis of MWM probe trial data via MANOVA, Pearson's r partial correlations (Table 3) were calculated between DMCZ, ECZ, TCZ, and LCZ measurements from probe trials conducted on Days 18 (baseline), 46 (after four weekly injections), and 74 (5 weeks following the last injection). These behavioral measures were chosen because they are traditionally considered indicative of knowledge of the platform location during MWM probe trials. Not surprisingly, there were significant correlations between DMCZ, ECZ, TCZ, and LCZ at nearly all timepoints, demonstrating convergent validity and suggesting each measurement was capturing some aspect of a common construct, in this case, spatial memory. However, TCZ was correlated >0.9 with DMCZ indicating these variables did not provide unique information. Therefore, TCZ was not included in the subsequent MANOVA to eliminate the adverse impact of redundancy on the analysis. Additionally, LCZ was negatively correlated with all variables, i.e., poor spatial memory was associated with a longer latency to enter the correct zone for the first time. Therefore, LCZ values from each mouse were inverted by subtracting each latency from the total trial duration (60 s). The inverted values, now positively correlated with the other variables, were used for the subsequent MANOVA. The final values used in the MANOVA were moderately correlated, with Pearson's r values between 0.218 and 0.668, indicating that each measure captured unique and overlapping aspects of spatial memory in the MWM.
Correlation matrices for probe trial metrics at various timepoints
Spatial memory
Results of a MANOVA using DMCZ, ECZ, and inverted LCZ measurements from probe trials conducted on Days 18, 46, and 74 demonstrated significant interactions between the group and treatment (partial η2 = 0.028; F(3,94) = 3.161; p < 0.05) and treatment and timepoint (partial η2 = 0.065; F(6,188) = 2.168; p < 0.05) as well as a significant change across Timepoint (partial η2 = 0.177; F(6,188) = 7.535; p < 0.05) suggesting group differences in spatial memory as well as changes in spatial memory across assessment timepoints. Subsequent univariate analyses were conducted on each measure to isolate effects.
DMCZ, ECZ, and LCZ analyses (Fig. 4)
DMCZ. There were significant effects of timepoint (partial η2 = 0.142; F(2,96) = 7.915; p < 0.05) and an interaction of the group and treatment (partial η2 = 0.062; F(1,96) = 6.364; p < 0.05) on DMCZ.
Subsequent multiple comparisons using Sidak's correction for family-wise error revealed that sham-operated mice receiving CYP + DOX injections moved significantly less in the correct zone during Day 46 (after four weekly injections) and Day 74 (5 weeks following the final injection) probe trials when compared with sham-operated mice receiving saline injections. Additionally, the DMCZ of sham surgery mice receiving CYP + DOX injections decreased significantly from the baseline (Day 18) to 5 weeks postinjections (Day 74; Fig. 4A). These results indicate a CYP + DOX-induced impairment of spatial memory in sham surgery mice receiving chemotherapy. For OVX/E2 mice, there was a significant reduction in DMCZ from the baseline (Day 18) to 5 weeks postinjections (Day 74); however, this measure was not significantly lower in CYP + DOX-injected OVX/E2 mice than saline-injected OVX/E2 mice at any timepoint, indicating this change in the CYP + DOX OVX/E2 mice was not due to chemotherapy but a trend in OVX/ E2 mice regardless of treatment.
Behavioral metrics in Platform Trial 1. Panels A–D depict key performance metrics during the first platform trial in the MWM at three timepoints: baseline (Day 18), after 4 weeks of chemotherapy or saline injections (Day 46), and 5 weeks post-therapy (Day 74). The metrics include average VEL (A), latency to reach the platform (B), TDM(C), and the number of times mice entered the platform zone (D). Group means are presented with standard error of the mean (SEM). Symbols and color codes for different groups are detailed in the legend. Asterisks indicate statistically significant differences between groups, with significance set at a p < 0.05 following Sidak's correction for multiple comparisons.
ECZ (Fig. 4B). There were significant effects of timepoint (partial η2 = 0.126; F(2,96) = 6.929; p < 0.05), group and treatment (partial η2 = 0.069; F(1,96) = 7.073; p < 0.05), and treatment and timepoint (partial η2 = 0.078; F(2,96) = 4.057; p < 0.05) on ECZ.
Subsequent multiple comparisons using Sidak's correction for family-wise error revealed that sham-operated mice that received CYP + DOX injections entered the correct zone less frequently during Day 74 (5 weeks following the final injection) probe trial than sham-operated mice that received saline injections. Additionally, sham-operated mice that received CYP + DOX injections exhibited a significant decrease in ECZ from Day 18 to Day 74. These findings suggest an impairment of spatial memory in sham surgery mice that emerges several weeks following CYP + DOX exposure. While OVX/E2 mice receiving CYP + DOX injections also exhibited a significant decrease in correct zone entries from Day 18 to Day 74, the ECZ of OVX/ E2 mice receiving CYP + DOX did not differ significantly from saline-injected OVX/E2 mice on Day 74, suggesting the reduction in correct zone entries across weeks of assessment was not due to CYP + DOX administration.
LCZ (Fig. 4C). There were significant effects of timepoint (partial η2 = 0.333; F(2,96) = 23.988; p < 0.05) on LCZ.
Subsequent multiple comparisons using Sidak's correction for family-wise error revealed no significant differences between saline and CYP + DOX-injected sham–operated mice at any assessment timepoint. Similarly, for OVX/E2 mice, LCZ did not differ between saline and CYP + DOX-injected mice at any timepoint. For sham-operated mice, LCZ did increase significantly on Day 74 (5 weeks following the final injection) compared with either Day 18 (baseline) or Day 46 (following four weekly injections). Similarly, for OVX/E2 mice, LCZ increased significantly from Day 18 to Day 74. Taken together these results indicate a tendency for mice to take longer to find the correct zone across weeks of assessment regardless of surgery type or treatment.
Disproportionate zone exploration. To confirm the finding that spatial memory was impaired by CYP + DOX in sham surgery mice and that these results were not due to any differences in activity, the proportion of distance moved in each zone during the Day 18, Day 46, and Day 74 probe trials was analyzed (Fig. 5). At the baseline (Day 18), when the proportion of distance moved in each zone was compared with unbiased exploration (100%/6 zones = 16.7% per zone) all groups explored the correct zone (0°) significantly above the expected value (16.67%) for unbiased behavior. Additionally, OVX/E2 mice assigned to receive saline injections also explored the +60° zone more than expected for unbiased behavior. In contrast, all groups explored the +/−180° and the −120° zones less than expected for unbiased behavior and OVX/E2 mice assigned to receive CYP + DOX injections explored the +120° zone less than expected for unbiased behavior. Taken together, these data indicate that on Day 18 (baseline), all groups could effectively utilize spatial cues to navigate to the correct area of the MWM and were generally aware of areas of the MWM that were not near the correct zone.⇓
Spatial memory assessment in probe trials. Panels A–C depict the distance moved within the correct zone (A), number of entries to the correct zone (B), and latency to reach the correct zone (C) across three timepoints: baseline (Day 18), after 4 weeks of either chemotherapy or saline injections (Day 46), and 5 weeks following the conclusion of injections (Day 74). Group means are shown along with SEM. Colors and symbols used to differentiate groups are detailed in the legend. Asterisks highlight statistically significant differences between groups at each timepoint. Significance was determined using a p value of <0.05 after applying Sidak's correction for multiple comparisons
Spatial exploration preferences in MWM zones. The proportion of distance moved within each zone of the MWM for sham and OVX mice at three assessment points: baseline (Day 18), 4 weeks postinjections (Day 46), and 5 weeks after injections (Day 74). Bars denote group means with SEM; the dashed line indicates an unbiased exploration expectation (16.67%). The 0° marks the target zone from the training trials. Each row details one of the four groups defined by OVX status (sham or OVX) and treatment type (saline or CYP + DOX). Upward arrows signify more exploration than unbiased expectation, while downward arrows show less. Statistical significance, marked by Sidak's correction for multiple comparisons, is noted for p < 0.05.
On Day 46, following four weekly injections, saline-injected sham/tap mice explored the correct zone (0°) more, and the +120°, +/−180°, and −120° zones less, than expected if the behavior was unbiased. This indicates that the spatial memory of saline-injected sham/tap mice was not impaired by the passage of time or repeated injections. In contrast, for CYP + DOX-injected sham/tap mice, no zones were explored significantly more than expected due to chance during the Day 46 probe trial, indicating an inability to identify the correct zone following four weekly injections of CYP + DOX. These mice did explore the +120° zone significantly less than expected, suggesting some spatial memory; however, the overall distribution of behavior strongly indicates spatial memory impairment.
On Day 46, both saline- and CYP + DOX-injected OVX/E2 mice explored the correct zone (0°) more than expected if the behavior was unbiased. Saline-injected OVX/E2 mice also explored the +120° and +/−180° zones less than expected and CYP + DOX-injected OVX/E2 mice explored the −120° zone less than expected if the behavior were unbiased. Taken together, these data suggest that OVX/E2 mice retained spatial memory of the correct zone and that CYP + DOX did not impair spatial memory.
On Day 74, 35 d following the final injection, saline-injected sham/tap mice explored the correct zone (0°) and the −60° zone significantly greater than expected and explored the +120° and +/−180° zones less than expected if the behavior was unbiased. This indicates that saline-injected sham/tap mice retained a spatial memory of the platform location throughout the study. In contrast, CYP + DOX-injected sham/tap mice did not explore any zones more than expected if the behavior was unbiased and explored the −120° zone less than expected if the behavior was unbiased.
This strongly suggests that CYP + DOX injections impair the spatial memory of mice that received sham surgery.
On Day 74, both saline- and CYP + DOX-injected OVX/E2 mice explored the correct zone (0°) more than expected if the behavior was unbiased. Saline-injected OVX/E2 mice also explored the −60° zone more than expected. Additionally, both saline- and CYP + DOX-injected OVX/E2 mice explored the +120° and +/−180° zones less than expected if the behavior was unbiased. Therefore, there does not appear to be a CYP + DOX-induced impairment of spatial memory in OVX/E2 mice.
Discussion
The present study assessed spatial memory in OVX/E2 and sham/tap mice on three occasions: before treatment (Day 18), after four weekly injections (Day 46), and 35 d postfinal injection (Day 74) of either saline or CYP + DOX. The study aimed to test the hypothesis that the cognitive impairing effects of AC chemotherapy are dependent on vulnerability of the ovaries to ablation. The results indicate that spatial memory impairment following exposure to CYP + DOX is notably influenced by the status of ovarian function, which affects circulating estrogen levels. Sham-operated mice, experiencing natural ovarian suppression and E2 depletion due to chemotherapy, exhibited significant impairments of spatial memory when compared with their saline-injected counterparts and to CYP + DOX-injected OVX mice maintained on estrogen-supplemented water. This finding is important because it demonstrates that even though CYP + DOX was present in the system, unless the ovaries are vulnerable to chemotherapy-induced damage, cognitive deficits do not occur. This suggests that chemotherapy-induced ovarian suppression or ablation heightens vulnerability to CRCIs and underscores the protective role of E2 against the effects of chemotherapeutic agents that lead to cognitive decline. These findings contribute to the expanding body of literature on CRCIs, particularly in the context of hormonal influence on neurocognitive outcomes postchemotherapy.
One interpretation of these findings is that reductions in circulating E2 due to chemotherapy directly impair cognitive function. It is well established that circulating E2 is associated with cognitive performance; for example, E2 contributes to normal verbal and spatial memory in females (S. J. Lee and McEwen, 2001; Sherwin, 2003). Furthermore, lower concentrations of circulating E2 are linked to poorer performance on cognitive function tests. Specifically, natural reductions in E2 during the menstrual cycle are associated with decreased verbal fluency, working memory capacity, perceptual speed, and fine motor function (Hampson, 1990; Phillips and Sherwin, 1992; Rosenberg and Park, 2002). Therefore, it could be argued that the cognitive deficits observed in CRCIs are a direct consequence of lowered E2 levels (Sherwin, 2003; Eberling et al., 2004). However, this interpretation does not account for studies of CRCIs that have identified chemotherapy-induced changes in the structure and function of the CNS, suggesting various mechanisms through which CRCIs may manifest (Ahles and Saykin, 2007; Deprez et al., 2011; Simó et al., 2013)
Estrogens mediate vulnerability to chemotherapeutic agents
Several mechanisms leading to CRCIs have emerged as relevant to the present findings including cytokine-mediated inflammation, microglial activation, oxidative stress, and disruptions in acetylcholine (ACh) synthesis (Berry et al., 2014; Christenson et al., 2017; Fujii et al., 2017, 2015). Each mechanism provides a plausible explanation for how ovarian suppression or ablation, and the consequent reduction in circulating estrogens, may heighten susceptibility to cognitive impairments induced by AC chemotherapy.
Estrogens modulate the immune response within the CNS, particularly by inhibiting the activation of microglia—the primary immune cells in the brain. When microglia are activated, they can exacerbate neuroinflammation, leading to cognitive deficits. By maintaining estrogen levels, either through ovarian function or supplementation, the inflammatory responses triggered by chemotherapy can be significantly reduced, thus preserving cognitive function (Vegeto et al., 2001, 2008; Ghisletti et al., 2005).
Furthermore, estrogens possess antioxidant properties that help in scavenging the free radicals generated during chemotherapy, reducing oxidative stress which is a known contributor to cellular damage and cognitive decline. Estrogens also enhance the expression of antioxidant enzymes, further protecting neural tissues against the oxidative stress induced by both cancer and chemotherapeutic treatments (Razmara et al., 2007).
Studies suggest that E2 can modulate the expression and activity of cholinergic systems, which are vital for cognitive processes such as learning and memory (Luine, 2014; Barth et al., 2015). Estradiol has been demonstrated to regulate high-affinity choline uptake (HACU), which is the rate-limiting step in the synthesis of ACh. In mouse models, disruptions of HACU across multiple brain regions have been shown to impact memory and cognition (O’Malley et al., 1987; Ragozzino and Choi, 2004; Bennett et al., 2009; Muramatsu et al., 2016). Thus, chemotherapy-induced reductions in circulating E2 can result in CRCIs because of impaired ACh synthesis and cholinergic function. Furthermore, ACh plays a key role in regulating inflammatory responses through the cholinergic anti-inflammatory pathway. Specifically, when ACh binds to α7 nicotinic ACh receptors on macrophages and other immune cells, it leads to the suppression of proinflammatory cytokines by inhibiting the NF-κB pathway, which regulates proinflammatory cytokines such as IL-1β, TNF-α, and IL-6. Consequently, the reduction in HACU induced by chemotherapy, and the subsequent decrease in ACh production, can increase susceptibility to CNS damage caused by neuroinflammation.
Estrogens mediate sex differences in CRCI vulnerability
The present findings, along with existing literature, highlight the protective effects of circulating E2. The impact of chemotherapy on circulating E2 suggests why females may be more vulnerable than males to the adverse cognitive effects of chemotherapy. However, males produce E2 through the aromatization of testosterone from the testes, with circulating concentrations ranging from 10 to 40 pg/ml. For comparison, adult premenopausal females have circulating E2 concentrations from 15 to 350 pg/ml, while postmenopausal females exhibit circulating E2 concentrations ∼10 pg/ml [111]. Because chemotherapy can lead to gonadal suppression or ablation in both sexes, males could also see reductions in circulating E2 during and after treatment, potentially increasing their vulnerability to these agents.
One factor contributing to the greater vulnerability of females to the negative effects of chemotherapy is the rapid division of cells in the ovarian follicles, which house and support the oocytes and are the primary source of circulating E2. The follicular cells are highly susceptible to damage from chemotherapeutic agents, which preferentially target rapidly dividing cells. This damage can disrupt the support system for the oocytes, leading to their dysfunction or death. Importantly, since females are born with a finite number of oocytes, any damage to these cells is irreversible and has long-term consequences for hormonal balance.
In contrast, Leydig cells in the testes produce testosterone and do not divide rapidly. Thus, these cells are less vulnerable to chemotherapies targeting rapidly dividing cells. Additionally, Leydig cells are relatively isolated from direct contact with blood-borne chemotherapeutic agents compared with ovarian follicles. These cells also typically operate below their maximal synthetic capacity, which allows them to maintain sufficient testosterone production even if some cells are damaged. Moreover, the testes contain progenitor Leydig cells capable of differentiating and expanding in response to injury or loss. Although this regenerative potential is limited, it offers a means to restore function over time if the initial damage is not overwhelming.
Taken together, this evidence suggests that the testes would be less vulnerable to systemic chemotherapy than the ovaries, and therefore, circulating E2 in males is less likely to be impaired by chemotherapy than in females.
Estrogen supplementation
The current results suggest that maintaining physiological levels of estrogen postchemotherapy could be beneficial in preserving cognitive function, opening the door to targeted therapeutic strategies that consider hormonal status as a factor in treatment planning. However, this potential protection comes with a caveat: direct hormonal replacement therapy has proven to be problematic. ER-positive tumors are associated with as high as double the risk of recurrence and metastasis (Dorgan et al., 2010), especially following menopause. Estradiol can induce direct DNA changes complicating DNA repair (Haakensen et al., 2011; Li et al., 2018; J. J. K. Lee et al., 2023) and enhance cellular proliferation and tumor invasiveness in ER-positive breast cancer cells (Haakensen et al., 2011; T. Wang et al., 2021; Pacheco-Velázquez et al., 2022; Gnant et al., 2023) by activating pathways such as ERK/MAPK and Wnt/β-catenin, which are crucial for cell proliferation and differentiation in cancer cells. Therefore, estrogen supplementation during chemotherapy is not advisable for ER-positive tumors as E2 is known to induce metastasis, proliferation, differentiation, and recurrence of cancer cells (Chen et al., 2020). Indeed, the tumor volumes in our OVX/E2 mice were consistently larger than their sham/tap counterparts during the last few weeks of assessment.
Conclusions
This study effectively demonstrates that E2 acts as a protective agent against the cognitive impairments associated with AC chemotherapy, specifically in the context of ovarian function and its consequential impact on circulating estrogen. Sham-operated mice, which experienced natural ovarian suppression and subsequent declines in circulating E2, exhibited significant deficits in spatial memory following chemotherapy, while OVX mice maintained on E2-supplemented water did not exhibit chemotherapy-induced impairment of spatial memory. This protective effect highlights the critical role of E2 in buffering the neurotoxic effects of chemotherapeutic agents that are likely exacerbated by induced ovarian suppression. This underscores the importance of considering hormonal health in the treatment of female cancer patients and may advocate for strategies that support estrogen levels during chemotherapy to mitigate cognitive decline in the absence of ER-positive tumors.
Footnotes
The authors declare no competing financial interests.
We thank Paula Bickford, Cheryl Kirstein, and Ashleigh Lacey. This work was supported by Bankhead-Coley Florida Biomedical Research Program 21b07.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license, which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed.
References
In this issue
