Aging Accelerates While Multiparity Delays Tumorigenesis in Mouse Models of High-grade Serous Carcinoma

Abstract

Objectives:

The “incessant ovulation” hypothesis links increased risk for tubo-ovarian high-grade serous carcinoma (HGSC) due to more ovulations and reduced risk conferred by pre-menopausal exposures like oral contraceptive use, multiparity, and breastfeeding. However, most women diagnosed with HGSC are postmenopausal, implying age is a major risk factor for HGSC. Our mouse model for HGSC, based on tamoxifen (TAM)-induced somatic inactivation of the Brca1, Trp53, Rb1, and Nf1 (BPRN) tumor suppressor genes in oviductal epithelium, recapitulates key genetic, histopathologic, and biological features of human HGSCs. We aimed to credential the model for future efforts to define biological and risk modification factors in HGSC pathogenesis.

Methods:

BPRN mice were treated with TAM to induce tumors at defined ages and parity status.

Results:

BPRN mice aged 9-months prior to tumor induction had markedly shorter survival than 6–8 week old mice induced to form tumors (median 46.5 weeks versus 61.5 weeks, log-rank test P=0.0006). No significant differences in cancer phenotypes were observed between multiparous versus nulliparous BPRN mice. However, using a modified tumor model with one wild-type Nf1 allele (BPRN^fl/+), nulliparous mice had more advanced tumors than multiparous mice (Mantel-Haenszel Chi-square test of association, P=0.01).

Conclusions:

Our findings show aging is associated with significantly shortened survival post tumor induction in the BRPN model and multiparity delays development and/or progression of HGSC in certain genetic contexts. The findings support relevance of our mouse model to gain mechanistic insights into how known factors exert their protective effects and to test novel approaches for HGSC prevention.

Introduction

In the United States, ovarian cancer is the fifth leading cause of female cancer deaths [1]. High-grade serous carcinoma (HGSC) accounts for up to 75% of cases and is the leading cause of ovarian cancer-related deaths, in part because early-stage tumors are usually asymptomatic. Most women with HGSC are diagnosed with widespread disease and cannot be cured with available therapies. Effective prevention strategies would bring new hope for reducing mortality from ovarian cancer, particularly for those women at increased risk due to inherited predisposing genetic variants. Risk-reducing salpingo-oophorectomy (RRSO) can effectively prevent HGSC development in such women [2], but given the adverse consequences of premature menopause associated with RRSO in young women, non-invasive approaches for prevention would be much preferred to reduce morbidity and mortality from the disease.

The “incessant ovulation” hypothesis first proposed by Fathalla in 1971 was based on epidemiological studies and animal model experiments, and suggested that repetitive ovulations could be a causative factor for ovarian carcinoma [3]. Further evidence supporting this hypothesis has been obtained. For example, administration of oral contraceptives (OC) provides long-lasting effects on reducing ovarian cancer risk [4,5]. The synthetic hormones in OCs block ovulation by suppressing natural hormone secretion. Likewise, ovulation ceases during pregnancy and for variable periods of time post-partum, and multiparity also has a strong protective effect on ovarian cancer development [6,7]. Early menopause and late menarche similarly decrease ovarian cancer risk [6,7]. In contrast, nulliparity, early age at menarche, and late age at natural menopause increase risk of developing ovarian cancer; all are associated with increased ovulations [7]. The underlying biological mechanisms of how incessant ovulation contributes to ovarian cancer pathogenesis remain under debate, particularly given data indicating that many, if not most, HGSCs arise from epithelium in the fallopian tube rather than on the ovarian surface [8,9]. During ovulation, the ovarian surface ruptures under the stimulus of local inflammatory reactions driven by ovulatory gonadotropin surge. Ovulation is accompanied by extrusion of follicular fluid, which is comprised of an enriched pool of hormones, growth factors, and potentially injurious components, such as reactive oxygen species (ROS) and pro-inflammatory factors [10,11]. The repetitive rupture-repair cycles affecting the ovarian surface and periodic exposure of local tissues to follicular fluid could cooperate in promoting tubo-ovarian carcinogenesis [11,12].

Most women diagnosed with HGSC are postmenopausal and nearly half are 65 years of age or older [13,14]. Overall survival of women with ovarian cancer decreases strongly with age. While there have been recent improvements in outcome for younger patients, this is not the case for older women [13]. Importantly, the “incessant ovulation” hypothesis and most non-inherited genetic factors known to influence ovarian cancer risk (e.g., parity and OC use) are manifest in the premenopausal years, whereas most women are diagnosed with HGSC following menopause and often many years after the last ovulatory cycle. This apparent paradox remains unresolved.

Genetically engineered mouse models (GEMMs) of cancer have the potential to provide tractable and relatively rapid systems with which to test cancer prevention strategies and inform cancer prevention trials in humans. The lengthy timeframe makes it challenging to conduct such studies in humans. We developed Ovgp1-iCreER^T2 mice in which the Ovgp1 promoter controls expression of tamoxifen (TAM)-regulated Cre recombinase in secretory cells of the oviductal epithelium – the murine equivalent of human FTE [15]. Our HGSC GEMM employs Ovgp1-iCreER^T2 mice to inactivate murine homologs of tumor suppressor genes (TSGs) that are frequently mutated in human HGSCs, specifically Brca1, Tp53, Rb1 and Nf1. These mice, hereafter referred to as BPRN mice, develop oviductal tumors closely resembling human HGSCs [15]. Precursor lesions known as serous tubal intraepithelial carcinomas (STICs) arise several months after transient treatment with TAM, and progress to locally invasive and then metastatic disease. The murine HGSCs acquire pervasive copy number alterations akin to those observed in human HGSCs, and their transcriptomes are also similar to their human tumor counterparts, particularly the immunoreactive and mesenchymal subsets of human HGSCs as defined by the TCGA [16]. If HGSC GEMMs are to be used to explore mechanisms that underlie the protective or tumor-promoting effects of factors such as pregnancy and aging, respectively, we must first show these factors have comparable effects on HGSC development and progression in the mouse as they do in humans. Herein, we employed our BPRN model to test effects of aging and parity on HGSC pathogenesis to determine if the model system supports prior claims drawn from epidemiological studies in humans.

Materials and Methods

Genetically engineered mice and animal care

Generation and characterization of Ovgp1-iCreER^T2 transgenic mice with various combinations of TSG alterations, including BPRN mice, have been described previously in detail [15–17]. Genotypes of individual mice were confirmed by polymerase chain reaction analysis, as previously described. Mice were housed in standard housing and fed the same chow diet. All procedures for the research described herein have been approved by the University of Michigan’s Institutional Animal Care and Use Committee (PRO00010212).

In vivo induction of oviductal tumors

Female Ovgp1-iCreER^T2 mice carrying floxed Brca1, Trp53, Rb1 and Nf1 alleles were intraperitoneally injected with TAM (0.2g/kg body weight, on day 1 and day 3) to induce Cre-mediated inactivation of the targeted genes in the oviductal epithelium. To characterize the time-dependent phenotypic features of oviductal tumorigenesis in mice treated with TAM at 6–8 weeks of age, cohorts of mice were euthanized at 2, 4, 6, 8, 10 and 12-months after the second TAM injection. To model effects of aging prior to TAM treatment in the BPRN model, TAM treatment in the aged cohort was delayed until mice were 9 months old and compared to a control cohort treated with TAM at 6–8 weeks of age. Mice in these two cohorts were sacrificed upon reaching humane endpoints. To test effects of multiparity, control (nulliparous) and test (multiparous) cohorts were treated with TAM at 6–8 weeks of age. Mice in the control cohort were caged with other female mice for the duration of the study, while mice in the test cohort were continuously caged with a breeding male mouse for up to 9 pregnancies. Control and test cohorts of BPRN mice were euthanized 60 weeks post-TAM, or earlier if humane endpoints were reached. Comparable numbers of mice in the control and test cohorts of BPRN^fl/+ mice were euthanized at either 52 or 60 weeks post-TAM. To test effects of multiparity in the context of aging, cohorts of nulliparous and multiparous BPRN mice were injected with TAM at 9 months of age and then sacrificed 46 weeks post-TAM or earlier if humane endpoints were reached.

Tissue sampling, histopathological and immunohistochemical analysis

Mice sacrificed at study endpoints were necropsied to determine the presence and extent of oviductal HGSCs as previously described [15–17]. Formalin-fixed and paraffin-embedded (FFPE) sections of each tissue were stained with hematoxylin and eosin (H&E) and examined by light microscopy. Oviducts without gross lesions were exhaustively sectioned and sections at regular intervals were microscopically examined by three individuals (XH, YZ, and KRC). Selected cases were further characterized with immunohistochemical analysis for PAX8 and CK8 expression as previously described [15]. The primary antibodies used were rabbit anti-PAX8 (Proteintech, Chicago, IL, USA) and rat anti-cytokeratin 8 (CK8, Developmental Studies Hybridoma Bank, University of Iowa). To evaluate infiltration of tumors by immune cells, the following primary antibodies were employed: rabbit anti-CD4, anti-CD8, anti-Foxp3, anti-CD68, anti-CD163 (Abcam, Cambridge, UK) and rat anti-CD45R (eBioscience, San Diego, CA, USA). Stained sections were digitally imaged by the Tissue and Molecular Pathology shared resource of the University of Michigan’s Rogel Cancer Center. Immune infiltrates in designated areas of tumor on 20x whole slide images were quantified using open source software QuPath [18] per the developer’s instructions, and reported as positive cell percentage and positive cell number per mm² tissue.

DNA isolation and targeted sequencing

Multiplexed barcoded libraries were made from 20 ng DNA isolated from FFPE mouse tumor tissues using a commercial DNA extraction kit (Qiagen, Germantown, MD, USA). Sequencing was performed on a custom-targeted panel created through the Life Technologies Ampliseq service on an Ion Torrent S5 Prime sequencer. The panel consisted of 4262 amplicons spanning 128 gene targets selected because they are commonly altered oncogenes and tumor suppressor genes in multiple cancer types based on the TCGA. Sequencing was performed following the manufacturer’s instructions, and each sample had an average of 2,569,828 mapped reads and 662x coverage (range 182x – 1327x). Sequence data was aligned to mm10 and read counts quantitated using Ion Torrent Suite’s TMAP custom alignment and quantitation algorithm [19] as previously described [16].

Statistical analyses

Survival rates of tumor-bearing aged and matched control mice were estimated by Kaplan-Meier analysis using GraphPad Prism software version 9. The Mantel-Haenszel Chi-square test of association was used to assess effects of aging and/or multiparity on disease severity at each study endpoint. Differences at P<0.05 were considered statistically significant.

Results

Characterization of the timing of oviductal tumor development and progression in BPRN mice following tumor induction at 6–8 weeks of age.

We previously showed tumors arising in BPRN mice closely recapitulate the morphology, genetic alterations, gene expression profiles, and biological behavior of their human HGSC counterparts [15,16]. We did not however, systematically characterize the time-dependence of tumor development in the BPRN model system. Before testing effects of multiparity or aging on HGSC pathogenesis, we wished to determine if BPRN mice transiently treated with TAM at 6–8 weeks of age consistently developed oviductal tumors following Cre-mediated inactivation of the targeted TSGs, with a similar latency period from mouse to mouse. Cohorts (n = 6–8 each) of TAM-treated BPRN mice were sacrificed and necropsied at 2, 4, 6, 8, 10 and 12 months post-TAM (Figure 1A). None of the mice evaluated at 6 months or earlier post-TAM showed detectable oviductal lesions (Table 1 and Figure 1B). However, at 8 months post-TAM, all mice analyzed had at least one oviduct with a STIC or early HGSC (eHGSC, locally invasive but still confined to oviduct), and half of the mice had bilateral oviductal lesions. All mice evaluated at 10 or 12 months post-TAM had at least one oviduct with a STIC or eHGSC (13 of 15 had bilateral lesions), and at 12 months post-TAM, more than half of the mice had full blown HGSC or carcinosarcoma – an HGSC variant also known as malignant mixed Mullerian tumor (MMMT) invading beyond the oviduct into adjacent tissues, such as the ovary, ovarian bursa or fat pad (Table1, Figure 1B). Representative photomicrographs of neoplastic lesions arising in BPRN mice are provided in Supplemental Figure 1. The findings indicate that i) there is a long (> 6 months) latency between Cre-mediated TSG inactivation and formation of recognizable oviductal lesions; ii) the tumorigenic phenotype is completely penetrant; and iii) once established, progression from STIC to invasive disease is relatively rapid. We have previously shown the majority of tumor-bearing BPRN mice followed to humane endpoints have peritoneal metastasis and/or ascites [16].

Figure 1. Timing of oviductal tumor development and progression in TAM-treated BPRN mice.

A. Schematic showing the experimental design. B. Summary of oviductal tumor phenotype in TAM-treated BPRN mice. Months post-TAM indicates the time points at which mice were sacrificed. L, left; R, right oviduct..

Table 1.

Time dependence of oviductal HGSC development in TAM-treated BPRN mice

Months post-TAM	Total number of oviducts	Oviductal morphology n (%)				Ovarian extension
		Normal	STIC	eHGSC	HGSC/MMMT
2	12	12 (100.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
4	12	12 (100.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
6	12	12 (100.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)
8	12	3 (25.0)	5 (41.67)	4 (33.33)	0 (0.0)	0 (0.0)
10	16	1 (6.25)	8 (50.0)	6 (37.50)	1 (6.25)	1 (6.25)
12	14	1 (7.14)	3 (21.43)	4 (28.57)	6 (42.86)	3 (21.43)

Tumor induction in aged mice is associated with shorter survival

Most women who develop ovarian cancer are diagnosed after menopause [13,14]. While it is challenging to assess age in isolation as a contributing factor in humans, this is more easily done in mice, for which genetic and environmental factors such as housing and diet can be controlled. Laboratory mice typically reach puberty around 6 weeks after birth and approach the endocrine equivalent of human perimenopause by ~8–9 months of age [20]. We wished to compare the post-TAM survival of BPRN mice in which tumor formation was induced during perimenopause compared to early in sexual maturity. Briefly, mice modeling the “perimenopausal” population were injected with TAM at 9-months of age (aged cohort), while control mice were exposed to TAM at 6–8 weeks of age. TAM-treated mice in both cohorts were monitored, then euthanized upon reaching humane endpoints (Figure 2A). At necropsy, the presence and extent of oviductal tumors were evaluated by gross examination and light microscopic evaluation of tissue sections. The median post-TAM survival for mice in the aged versus control cohorts was 46.5 versus 61.5 weeks, respectively. Kaplan-Meier analysis indicated that the aged cohort had significantly shortened survival post-TAM (Log-rank test, P=0.0006) (Figure 2B). Aging was associated with a hazard ratio of 2.70 (95% CI of 1.30–5.63 evaluated by Cox proportional hazard models), supporting an adverse effect of age on outcome in tumor-bearing BPRN mice. Although aged mice had shorter post-TAM survival than the younger control group, there was no significant difference in the presence or extent of disease in the two groups (Mantel-Haenszel Chi-square test of association, P=0.47) (Figure 2C). Because all mice were followed until they reached humane endpoints and most of the mice in both cohorts had advanced (metastatic) disease at the time of euthanasia, we were unable to determine if older age at the time of tumor induction shortens the latency of tumor development, accelerates tumor progression, or both, in BPRN mice. To test whether advanced tumors in aged mice had differences in immune cell infiltration compared to tumors in younger control mice, immunohistochemistry was used to identify selected immune cells in tumor tissues (n=7 in each group) using antibodies targeting CD45R, CD4, CD8, Foxp3, CD68, and CD163 (Supplemental Figures 2 and 3). Quantification of immune cells in digital whole slide images of tumor tissues showed slightly more immune cell infiltration of tumors in aged mice, but none of the immune cell types assessed showed significant differences between the two groups of tumors (Supplemental Table 1). Of the cell types analyzed, macrophages were most abundant in the tumors. Two macrophage markers, CD68 and CD163 were used to quantify the total and M2-polarized macrophage populations, respectively. The average CD68-positive cell count per square millimeter of tumor tissue was only slightly increased in the aged compared to control groups (fold change=1.09). However, the average CD163-positive cell count was more substantially increased in the aged group relative to control (fold change = 1.67) suggesting an enhanced pro-tumorigenic M2-polarization in the tumor microenvironment associated with aging.

Figure 2. Effect on survival of mouse age at the time of BPRN tumor initiation.

A. Schematic summarizing the experimental design. B. Kaplan-Meier analysis of mice euthanized upon reaching humane endpoints. Aged mice showed significantly reduced overall survival time post TAM-injection (log-rank test, P=0.0006). C. Summary of oviductal tumor phenotypes. The time points at which mice were sacrificed is defined as weeks post-TAM. L, left; R, right. Other non-oviductal tumors were found in a subset of mice at necropsy, including lung adenomas (LA), lymphomas (Ly), thyroid tumors (Th), and salivary gland tumors (Sg).

Genomic instability is one of the hallmarks of aging [21]. Genetic defects resulting from extrinsic or intrinsic damage accumulate with age in somatic cells, including chromosomal gains and losses [21]. We employed targeted next-generation sequencing to assess copy number alterations (CNAs) in tumors from aged (n=18) and control mice (n=13). Copy number estimates (log-transformed) for the 128 sequenced genes in these tumors are shown in Supplemental Figure 4, in which the tumor samples are grouped by cohort and then hierarchically clustered based on CNAs. Consistent with our previous findings, tumors arising in BPRN mice demonstrated widespread CNAs distributed across all 19 autosomes sampled by the targeted panel. Overall, there were no significant differences in the prevalence of copy number gains or losses in tumors from aged vs. control mice. Notably, as in our previous study [16], one tumor acquired high copy amplification of Myc, a common form of CNA in human HGSCs based on data from the TCGA [22].

Multiparous BPRN^fl/+ but not BPRN mice show fewer and less advanced tumors than nulliparous control mice

Parity is one of the most well-documented factors associated with reduced risk of ovarian cancer [6,7]. We previously discovered (unpublished data) that TAM-treated BPRN female mice remain fertile and can produce litters of viable pups (litter size ranging from 4 to 10 pups) from as many as 9 sequential pregnancies. Our initial experiments to test effects of multiparity on HGSC pathogenesis focused on BPRN mice with two floxed alleles of each of the four targeted TSGs. Individual female mice treated with TAM at 6–8 weeks of age (n=15) were caged with a single male for monogamous breeding. Multiparous mice were matched with a comparable number (n=14) of TAM-treated control littermates that remained nulliparous (Figure 3A). The number of litters born to multiparous mice ranged from 5 to 9. BPRN mice in both cohorts were euthanized 60 weeks post-TAM, except for one mouse in the nulliparous cohort that reached humane endpoints and was euthanized at 56 weeks post-TAM. While STIC or more advanced lesions were present in all oviducts from the nulliparous group, 4 of 30 oviducts in multiparous mice had no detectable lesions at study endpoint. Overall, however, there was no significant difference in the presence or extent of disease in the two BPRN cohorts (P=0.65, Mantel-Haenszel Chi-square test of association) (Figure 4B).

Figure 3. Effect of multiparity on oviductal tumor development and progression in TAM-treated BPRN and BPRN^fl/+ mice.

A. Schematic summarizing experimental design. B. Summary of oviductal tumor phenotypes in nulliparous vs. multiparous BPRN and BPRN^fl/+ mice. Weeks post-TAM indicate the time points at which mice were sacrificed. L, left; R, right. Other tumor row indicates mice in which lymphoma (Ly) was found at necropsy. Mouse genotypes are indicated in the color key at right.

Figure 4. Effect of parity on oviductal tumor development and progression in aged TAM-treated BPRN mice.

A. Schematic summarizing experimental design. B. Summary of oviductal tumor phenotypes in aged nulliparous vs multiparous mice. Weeks post-TAM indicate the time points at which mice were sacrificed. L, left; R, right. Other non-oviductal tumors found at necropsy included lung adenomas (LA) and lymphomas (Ly).

The long latency required for tumor development in BPRN mice (Figure 1) likely reflects a requirement for accumulation of additional somatic alterations beyond inactivation of the targeted TSGs. Indeed, in our prior work and confirmed here, we have shown that advanced BPRN tumors acquire abundant copy number alterations and somatic mutations in the mouse homologs of other genes that are recurrently mutated in human HGSCs [16]. Interestingly, TAM-treated BPRN mice in which one or more of the four targeted TSGs are heterozygous for the floxed allele, also develop HGSCs, but tend to have longer overall survival [16]. We wished to test whether effects of parity might be more easily detected in mice with fewer initial targeted gene mutations. Specifically, we employed BPRN^fl/+ mice that have one wild-type Nf1 allele, to test effects of multiparity on tumorigenesis. One cohort (n=8) of BPRN^fl/+ mice was injected with TAM at 6–8 weeks of age, then bred continuously for up to 9 pregnancies per mouse and compared to a control cohort (n=8) of littermate BPRN^fl/+ mice that remained nulliparous. Because euthanasia at an earlier time point might reveal effects of multiparity on tumor initiation (i.e., duration of latency), half of the mice in each cohort were sacrificed at 52 weeks post-TAM, while the other half were monitored until 60 weeks post-TAM. In contrast to BPRN mice, multiparity in BPRN^fl/+ mice was significantly associated with less advanced disease at study endpoints (Mantel-Haenszel Chi-square test of association, P=0.01). In the multiparous BPRN^fl/+ mice, 13 of 16 oviducts showed only STIC or no lesion, while the same was true for only 3 of 16 oviducts from the nulliparous mice (P=0.001, two-tailed Fisher’s exact test). If only mice euthanized at 60 weeks post-TAM are included in the analysis, 6 of 8 oviducts from multiparous mice showed only STIC or no lesion, while all 8 oviducts from mice in the nulliparous group showed eHGSC or more advanced tumors (P=0.007, two-tailed Fisher’s exact test) (Figure 3B). These findings suggest that multiparity delays tumor initiation but do not exclude an effect on the rate of tumor progression as well.

Finally, we wished to test whether multiparity might mitigate the adverse effects of aging in our BPRN model system. For this experiment, a cohort of BPRN mice (n=14) were continuously bred until 9 months of age, and then injected with TAM to induce oviductal tumor formation. These mice were compared to a cohort (n=16) of BPRN littermate controls that remained nulliparous before being injected with TAM at 9 months of age (Figure 4A). Mice were euthanized and necropsied 46 weeks post-TAM, or earlier if they reached humane endpoints (two mice in each cohort). We found no significant difference between the two groups with respect to the presence or extent of disease at study endpoint (Mantel-Haenszel Chi-square test of association, P=0.61), suggesting multiparity does not significantly impact effects of aging, at least in the BPRN model with biallelic targeting of all four TSGs (Figure 4B).

Discussion

Several studies indicate that STICs can be considered direct HGSC precursors [23–26]. Detailed molecular analyses of multiple independent tumor samples from a given individual have led to the conclusion that the interval between development of STIC and formation of frank HGSC is approximately 6–7 years, followed by rapid progression to metastatic disease [23,27]. Because the mean age for HGSC diagnosis is 61 [28], it is likely that most STICs, especially in women without genetic predisposition, do not arise until women are in their 50’s. Unfortunately, reliable methods with which to screen for or detect early tubal precursor lesions in humans are not yet available. Robust preclinical model systems may therefore prove to be particularly useful for studying how STICs develop and progress, and how progression to frankly invasive HGSC can be prevented or at least significantly delayed. As shown here, tumor induction in young BPRN mice does not result in detectable STICs until mice are in the equivalent of human peri- or early menopause. This aspect of the model, like the morphologic, molecular, and biological features shown earlier, closely recapitulates human HGSC pathogenesis.

Several studies have shown that older age at ovarian cancer diagnosis is associated with poorer outcome [29–31]. Aged patients are less likely to complete standard therapy, perhaps due to poorer tolerance for initial cytoreductive surgery and/or adjuvant chemotherapy [32]. In a recent retrospective study by Mallen and colleagues [33], patients over age 70 had significantly worse progression-free and overall survival compared to younger patients. The aged women were less likely to complete the adjuvant chemotherapy, presented more often with decreased platinum sensitivity, and had a lower clinical trial participation rate than younger women. Interestingly, although the older women had more severe co-morbidities than younger patients, their increased likelihood for incomplete adjuvant chemotherapy had no association with co-morbidities. A population-based study from Switzerland demonstrated that ovarian cancer patients over 70 years old tended to be diagnosed at more advanced stages compared to younger women [34]. Even after adjusting for confounding factors, including patients’ clinicopathological characteristics and treatment delivered, the risk of death was still roughly 2-fold higher in the older cohort compared to the younger. We confirmed the adverse effect of age in our model system by showing that aged BPRN mice have significantly shorter survival post-TAM compared to their younger counterparts. As reviewed by Chon et al. [35], the potential underlying mechanisms connecting aging and the prognosis of epithelial ovarian cancer may involve immune function, inflammatory responses, metabolic disorders, tumor microenvironment, DNA damage and/or epigenetic factors. While we have not yet elucidated the mechanisms underlying the effect of age on outcome, aging is associated with organ degeneration and tissue damage largely attributed to the continuous accumulation of deleterious molecular events involving genomic instability as well as epigenetic alterations [21]. The accumulation of additional genetic alterations presumably required for neoplastic transformation in our model system may occur more rapidly in old compared to young mice. In this scenario, shorter post-TAM survival in older mice might be associated with more rapid development of STICs post-TAM. Additional studies in which cohorts of mice are euthanized at specific intervals post-TAM would be required to test this hypothesis. While our targeted genomic sequencing analysis of advanced tumors in control vs. aged mice did not identify significant differences in the prevalence of CNAs in tumors from the two groups at study endpoint, this does not exclude the possibility that aged mice acquire these alterations more quickly than young mice. Shorter post-TAM survival of aged mice might also perhaps be explained by immune system compromise associated with aging [21]. In this scenario, anti-tumor immune surveillance would be less effective in aged mice, possibly leading to more rapid development of STIC and/or more rapid progression to invasive and metastatic disease. A previous study using ID8 murine ovarian cancer cells allografted into syngeneic mice reported that aged hosts are more susceptible to metastasis. Metastatic tumor burden increased with advancing age and tumors in the aged hosts had more CD45R+ tumor infiltrating lymphocytes [36]. We did not detect significant differences in immune cell infiltrates (tumor-infiltrating lymphocytes [TILs] or tumor-associated macrophages [TAMs]) in oviductal tumors from aged versus control mice, though our cohort sizes may have been too small to detect subtle differences between the two groups.

Multiparity, breastfeeding and combined oral contraceptive use are the most well-established protective factors related to reduced risk of ovarian cancer [7,37–39]. Based on a large study of European women, parous women had a significant 29% lower risk of ovarian cancer than nulliparous women, with each additional term pregnancy lowering risk by up to 8% [40]. We confirmed a protective effect of multiparity, but only in certain genetic contexts. Specifically, we found fewer and less advanced tumors in the oviducts of multiparous compared to nulliparous post-TAM BPRN^fl/+ mice, but not in similar groups of BPRN mice carrying two floxed alleles of all four targeted TSGs. If this difference is attributable to effects of parity on the necessity for more somatic “hits” in BPRN^fl/+ compared to BPRN mice, we could test if effects of parity are even more pronounced in mice carrying only one floxed copy of each of the four targeted TSGs. These latter mice also develop oviductal lesions with complete penetrance post-TAM, but take nearly twice as long to reach humane endpoints compared to BPRN mice with two floxed alleles of all four TSGs [16].

Although we showed that tumor-bearing aged BPRN mice have significantly shorter post-TAM survival than young mice, the effect was not mitigated by multiparity. This result contrasts with a previous study that compared the survival of aged multiparous versus comparably aged nulliparous syngeneic mice bearing tumors induced by injection of spontaneously transformed mouse ovarian surface epithelial (MOSE) cells. In that study, while median survival times of multiparous and nulliparous cohorts were 98 versus 87 days, respectively, comparison of intraperitoneal tumor burden and volume of ascites measured at death showed no statistical difference between the two groups [41]. Importantly, the influence of parity on factors that impact the propensity of normal oviductal epithelium to undergo neoplastic transformation cannot be assessed when testing effects of parity on fully transformed MOSE cells.

The more closely a given GEMM recapitulates its human tumor counterpart, the more useful it is likely to be for studying tumor biology and testing novel strategies for early detection, treatment, and prevention of cancer. While the utility of HGSC GEMMs to test approaches for translational clinical trials remains unproven, the work described herein helps credential our BPRN models for use in future studies to gain mechanistic insights into how known factors exert their protective effects and to test novel approaches for HGSC prevention and prioritize them for trials in humans.

Highlights:

• Aging accelerates, while multiparity delays tumorigenesis in mouse models of high-grade serous carcinoma

• Mouse models can be used to test how known deleterious and protective factors exert their effects on HGSC pathogenesis

• Genetically engineered mouse models may prove useful in testing novel strategies for HGSC prevention