Deep molecular tracking over the 12-yr development of endometrial cancer from hyperplasia in a single patient

Abstract

Although the progressive histologic steps leading to endometrial cancer (EndoCA), the most common female reproductive tract malignancy, from endometrial hyperplasia are well-established, the molecular changes accompanying this malignant transformation in a single patient have never been described. We had the unique opportunity to investigate the paired histologic and molecular features associated with the 12-yr development of EndoCA in a postmenopausal female who could not undergo hysterectomy and instead underwent progesterone treatment. Using a specially designed 58-gene next-generation sequencing panel, we analyzed a total of 10 sequential biopsy samples collected over this time frame. A total of eight pathogenic/likely pathogenic mutations in seven genes, APC, ARID1A, CTNNB1, CDKN2A, KRAS, PTEN, and TP53, were identified. A PTEN nonsense mutation p.W111* was present in all samples analyzed except histologically normal endometrium. Apart from this PTEN mutation, the only other recurrent mutation was KRAS G12D, which was present in six biopsy samplings, including histologically normal tissue obtained at the patient's first visit but not detectable in the cancer. The PTEN p.W111* mutant allele fractions were lowest in benign, inactive endometrial glands (0.7%), highest in adenocarcinoma (36.9%), and, notably, were always markedly reduced following progesterone treatment. To our knowledge, this report provides the first molecular characterization of EndoCA development in a single patient. A single PTEN mutation was present throughout the 12 years of cancer development. Importantly, and with potential significance toward medical and nonsurgical management of EndoCA, progesterone treatments were consistently noted to markedly decrease PTEN mutant allele fractions to precancerous levels.

INTRODUCTION

Endometrial cancer (EndoCA) is the most common cancer of the female reproductive tract. Historically, EndoCA is divided into two types, type I endometrioid and type II nonendometrioid, both differing in etiology and prognosis (Bokhman 1983; Rajmohan et al. 2014). The Cancer Genome Atlas (TCGA) Consortium has reported a new classification system of EndoCA based on molecular characteristics, which is beginning to be introduced into clinical practice as it shows high prognostic relevance (Levine and Cancer Genome Atlas Research Network 2013; Talhouk et al. 2015). Endometrioid tumors, the most common EndoCA, result from unopposed estrogen stimulation, are low-grade, and overall have a good prognosis attributable to their usual early-stage diagnosis. Nonendometrioid tumors are generally rarer, are high-grade, have a poor prognosis, and have a less well-defined etiology (Bokhman 1983; Felix et al. 2010). Incidence of endometrioid endometrial carcinoma (EEC) is on an unprecedented rise attributed to the obesity epidemic and prescribed estrogen drugs (Constantine et al. 2019). Development of EEC is indicated by a defined precursor lesion that arises from endometrial hyperplasia. Endometrial hyperplasia is an estrogen-driven, aberrant proliferation of endometrial epithelial glandular cells that results in an abnormal thickening of the endometrium. This leads to morphological changes in gland architecture defined by an overall low gland-to-stroma ratio (Ring et al. 2022). The glandular architectural changes are defined as simple or complex and may present with or without cellular atypia. As of 2014, the World Health Organization (WHO) has defined the malignant potential of endometrial hyperplasia by the cellular atypia status alone (Kurman et al. 2014). Simple and complex hyperplasia without atypia are considered benign lesions, whereas any hyperplasia with atypia can be defined as an endometrial intraepithelial neoplasm (EIN), atypical hyperplasia (AH), or an EIN/AH and is considered premalignant (Baak et al. 2005; Kurman et al. 2014). Although endometrial thickness can be detected through imaging, such as transvaginal sonography, histological phenotypes are definitively diagnosed by a surgical biopsy. If cellular atypia is detected, the curative intervention is hysterectomy owing to the increased cancer risk. Notably, in excess of 40% of hysterectomies performed because of the presence of an EIN/AH lesion are found to harbor an occult endometrial carcinoma (Trimble et al. 2006). Progesterone, a natural antagonist to estrogen, is a nonsurgical treatment option for women who wish to preserve fertility or for those patients who are surgically ineligible for hysterectomy (Kim and Chapman-Davis 2010; ACOG 2015). Although surgical intervention has become standard of care in EIN/AH diagnoses, progesterone therapy is a conservative treatment approach for both benign, AH, and grade 1 EEC (Pal et al. 2018; Ring et al. 2022).

Although atypia status alone indicates cancer risk, there is currently no way to predict EEC risk from analysis of the benign or normal endometrium. To gain a better understanding of the EEC cancer initiation and progression, EEC and EIN/AH genomes have been molecularly analyzed (Mutter et al. 2000; Levine and Cancer Genome Atlas Research Network 2013; Russo et al. 2017; van der Putten et al. 2017). EIN/AH lesions are highly like endometrial carcinomas as they share mutational similarities, most notably somatic mutations in PTEN, KRAS, PI3KCA, and CTNNB1 (Sun et al. 2001; Yeramian et al. 2013) and have been shown to share identical mutations with neighboring EECs when present (Russo et al. 2017). A pilot study found identical mutations present in benign endometrium that later presented in subsequent endometrial carcinomas in 50% of patients studied (Lupini et al. 2019). Although methylation status and exome sequencing are beginning to resolve the molecular transition from benign to malignant histology (Multinu et al. 2020; Russo et al. 2020), there remains a need to further investigate molecular attributes in benign hyperplasia and report follow-up information regarding EndoCA status of the patient.

Presumed cancer driver mutations in benign hyperplastic tissue may represent the normal background mutations of the mutationally dynamic endometrium (Nair et al. 2016; Moore et al. 2020), or they may signify the earliest malignant events that will drive neoplastic growth. Discerning which somatic mutations in hyperplasia will ultimately drive EEC from the mutations that are predicted to be oncogenic but remain inert and pose little cancer risk may lead to a better understanding of neoplastic progression and inform the list of candidate biomarkers that can be trained as molecular screening tools.

There are few, if any, opportunities to longitudinally track the mutational events underlying the discrete histology-defined steps from benign hyperplasia to EIN/AH development through to subsequent adenocarcinoma formation. Practically, one of the limitations to this possibility is that surgical intervention is used to remove the uterus when atypia is detected. Here, we present a unique longitudinal investigation in a single patient who initially presented with benign hyperplasia which progressed into an endometrial carcinoma over a 12-yr period. The patient was medically managed with progesterone treatment because of her surgical ineligibility. An endometrial biopsy was taken nearly every year from her first biopsy until her cancer diagnosis 12 yr later, followed by an additional benign biopsy 2 yr after that cancer diagnosis. We retrospectively performed targeted exome sequencing on each biopsy using a 58-gene oncology panel. Our primary aim was to investigate the sequence of mutational events associated with malignant transformation occurring over time.

RESULTS

Case Description

The case of a 57-yr-old postmenopausal woman, para 4, with multiple coexistent comorbidities including morbid obesity, meningioma, diabetes mellitus type II, heart disease, and a history of opioid abuse is described. She initially presented to our gynecology service with postmenopausal bleeding in 2005 and underwent hysteroscopy and a dilation and curettage. She was diagnosed with endometrial benign hyperplasia (Fig. 1A) and discharged with follow-up plans. The patient returned a year later in 2006 with recurrent abnormal uterine bleeding. Her biopsy at that time revealed complex hyperplasia with squamoid morules, glandular structures associated with a premalignant phenotype (Fig. 1B; Lin et al. 2009). Importantly, as it relates to the possibility of this longitudinal study, the patient was not considered a surgical candidate for hysterectomy because of her comorbidities. She was instead prescribed progesterone therapy in the form of progestin megestrol acetate 80 mg taken orally twice a day for 6 mo. The following year she presented with benign simple hyperplasia (Fig. 1C).

Figure 1.

Representative longitudinal biopsy findings over the 12-yr period leading to cancer diagnosis. Each panel shown is a hematoxylin and eosin (H&E)-stained biopsy slide from a specific year (magnification, 100×). (A) 2005 benign endometrial hyperplasia. (B) 2006 benign endometrial hyperplasia. (C) 2007 benign simple endometrial hyperplasia. (D) 2009 atypical complex hyperplasia. (E) 2010 benign simple hyperplasia. (F) 2014 proliferative endometrium. (G) 2017 atypical complex hyperplasia with adenocarcinoma. (H) 2017 complex hyperplasia. Laser-capture microdissection (LCM) was performed on glands taken from slides A, B, D, G, and H and 6-μm curls were cut and used for molecular analysis obtained from the blocks represented in panels C, E, and F.

Because of intercurrent medical complications, the patient only reported intermittently for gynecologic evaluations over the next several years. Table 1 summarizes the pathologic diagnoses of biopsy samples obtained over the 14-yr course of the patient's evaluation, diagnoses, treatment, and follow-up. Figure 1 shows the hematoxylin and eosin (H&E)-stained biopsy slides of her 12-yr progression to carcinoma. During this time, endometrial biopsies were performed with final pathology-confirmed diagnoses ranging from non-AH to AH. In 2009, based on the presence of complex hyperplasia and/or hyperplasia with atypia (Fig. 1D), megestrol acetate was again prescribed. The patient returned for gynecologic evaluations in 2010 and 2014, at which time biopsies were performed. These biopsies revealed areas of hyperplasia but with no evidence of atypia (Fig. 1E,F).

Table 1.

Endometrial biopsy results

Year	Final pathologic diagnosis	Pathology of biopsy samples sequenced	Hormone therapy
2005	Endometrial hyperplasia ranging from simple to complex without atypia	(1) Nonatypical complex hyperplasia; (2) normal endometrium	No
2006	Endometrial complex glandular hyperplasia with squamoid morules	Nonatypical complex hyperplasia	Megestrol acetate (80 mg 2× daily)
2007	Endometrial nonatypical simple hyperplasia	Nonatypical simple hyperplasia	No
2009	Endometrial hyperplasia ranging from simple to complex with focal atypia	Atypical complex hyperplasia	Megestrol acetate (80 mg 2× daily)
2010	Endometrial nonatypical simple hyperplasia	Nonatypical simple hyperplasia	No
2014	Proliferative endometrium with menstrual-like breakdown	Proliferative endometrium	No
2017	Endometrial complex hyperplasia with and without atypia; adenocarcinoma	(1) Complex hyperplasia without atypia; (2) adenocarcinoma	Megestrol acetate (80 mg 2× daily)
2019	Benign inactive glands indicative of hormone treatment	Benign endometrium	No

Pathologic diagnosis based on each biopsy obtained is provided along with year of the sampling and description of the pathology present on the sample obtained for genomic analysis and hormone therapy at those visits.

In 2017, the patient presented with abnormal uterine bleeding, and an endometrial biopsy was again performed. At that time, the patient was diagnosed with endometrial endometrioid adenocarcinoma (Fig. 1G) with adjacent complex hyperplasia (Fig. 1H). Because the patient was still not a surgical candidate, she was restarted on progesterone therapy. The patient returned for reevaluation in 2019 and a follow-up biopsy was performed. At this time, the biopsy results documented benign endometrium and without evidence of malignancy.

Targeted Sequencing of Endometrial Cancer–Linked Genes Identifies a Consistently Mutated Gene

To investigate the potential driver-gene mutations and selections occurring at each step of the malignant transformation pathway from hyperplasia to adenocarcinoma formation, we performed next-generation sequencing (NGS) using a specifically designed panel of 58 genes frequently mutated in cancer (Swift Biosciences). The pathogenic/likely pathogenic mutations identified in each of the samples are shown in Table 2. A complete list of all mutations, including pathogenic, likely pathogenic, and nonsynonymous mutations of uncertain significance, and their associated allele frequencies, are presented in Supplemental Table 1.

Table 2.

All pathogenic and likely pathogenic mutations in each biopsy sample

Sample	Gene	Position change	Nucleotide mutation	Protein mutation	Variant type	Allele freq uency (%)	Predicted effect	COSMIC ID	dbSNP	Genotype
2005 Histologically normal  endometrium	KRAS	Chr 12:25398284_C > T	c.35G > A@	p.G12D	Missense	9.2	Pathogenic	COSM1135366	121913529	+/−
2005 Nonatypical complex  hyperplasia	PTEN	Chr 10:89692848_G > A	c.333G > A	p.W111*	Stop gain	14.2	Pathogenic	COSM1166809	-	+/−
2006 Nonatypical complex  hyperplasia	PTEN	Chr 10:89692848_G > A	c.333G > A	p.W111*	Stop gain	23.3	Pathogenic	COSM1166809	-	+/−
2007 Nonatypical simple  hyperplasia	PTEN	Chr 10:89692848_G > A	c.333G > A	p.W111*	Stop gain	1.1	Pathogenic	COSM1166809	-	+/−
	KRAS	Chr 12:25398284_C > T	c.35G > A	p.G12D	Missense	5.9	Pathogenic	COSS1135366	121913529	+/−
2009 Complex hyperplasia  with focal atypia	APC	Chr5:112175936_C > T	c.4645C > T@	p.Q1549*	Stop gain	31.6	Pathogenic	COSM19414	863225357	+/−
	PTEN	Chr 10:89692848_G > A	c.333G > A	p.W111*	Stop gain	26.1	Pathogenic	COSM1166809	-	+/−
	KRAS	Chr 12:25398284_C > T	c.35G > A	p.G12D	Missense	19.0	Pathogenic	COSM1135366	121913529	+/−
	PTEN	Chr 10:89717729_G > A	Unknown@	p.D55N	Missense	17.5	Likely pathogenic	COSM6849498	-	+/−
	CDKN2A	Chr 9:21971183_C > T	c.175G > A@	p.V59M	Missense	16.2	Likely pathogenic	COSM12485	-	+/−
	TP53	Chr 17:7577153_C > A	c.785G > T@	p.G262V	Missense	15.4	Pathogenic	COSM11198	1131691025	+/−
	CTNNB1	Chr 3:41266124_A > G	c.121A > G@	p.T41A	Missense	9.9	Pathogenic	COSM5664	121913412	+/−
2010 Nonatypical simple  hyperplasia	PTEN	Chr 10:89692848_G > A	c.333G > A	p.W111*	Stop gain	19.1	Pathogenic	COSM1166809	-	+/−
	KRAS	Chr 12:25398284_C > T	c.35G > A	p.G12D	Missense	3.6	Pathogenic	COSS1135366	121913529	+/−
2014 Proliferative  endometrium	KRAS	Chr 12:25398284_C > T	c.35G > A	p.G12D	Missense	5.5	Pathogenic	COSS1135366	121913529	+/−
	PTEN	Chr 10:89692848_G > A	c.333G > A	p.W111*	Stop gain	3.3	Pathogenic	COSM1166809	-	+/−
2017 Nonatypical complex  hyperplasia	PTEN	Chr 10:89692848_G > A	c.333G > A	p.W111*	Stop gain	31.9	Pathogenic	COSM1166809	-	+/−
	KRAS	Chr 12:25398284_C > T	c.35G > A	p.G12D	Missense	7.1	Pathogenic	COSS1135366	121913529	+/−
2017 Adenocarcinoma	PTEN	Chr 10:89692848_G > A	c.333G > A	p.W111*	Stop gain	36.9	Pathogenic	COSM1166809	-	+/−
	ARID1A	Chr 1:27106823..27106825 _AGA > -	Unknown@	p.K1929del	In-frame	8.1	Likely pathogenic	-	-	+/−
2019 Benign inactive  glands	PTEN	Chr 10:89692848_G > A	c.333G > A	p.W111*	Stop gain	0.7	Pathogenic	COSM1166809	-	+/−

All pathogenic and likely pathogenic mutations in each biopsy sample identified using the 58-gene panel that had an allele frequency >5%, unless they were present in the patient's malignant and premalignant biopsies—namely, the 2017 tumor, 2017 adjacent hyperplasia, or the premalignant 2009 complex hyperplasia.

For mutations found in the 2017 and 2009 biopsies, we searched for mutation down to 1% allele frequency (AF) and then we in all other samples and if they were found, they were included in this table as they represent molecular links to malignancy. All mutations in which there was enough DNA were validated using an orthogonal sequencing technology; only seven samples did not have enough DNA for secondary sequencing validation, and they are highlighted with a superscripted “@” in the nucleotide change column. Heterozygous (+/−).

Quite intriguingly, at the time of the patient's first presentation in 2005, a high allele frequency KRAS G12D mutation was identified in histologically normal endometrium. This sample was obtained adjacent to a second and independent sample obtained at the same time, but which was histologically defined as nonatypical complex hyperplasia. This sample contained a high allele frequency PTEN stop mutation, W111*. Both mutations would be identified again throughout the course of the patient's care as described below and as shown in Table 2; however, only the PTEN W111* mutation would be present in all future samples.

The sample containing the most mutations was the complex hyperplasia with focal atypia sample obtained in 2009 and which predated the diagnosis of cancer by 8 yr. This 2009 ACH biopsy had the most pathogenic mutations (APC Q1549*, PTEN W111*, KRAS G12D, TP53 G262V, and CTNNB1 T41A) and likely pathogenic mutations (PTEN D55N and CDKN2A V59M) of all the samples sequenced and each mutation was associated with high allele frequencies (range: 9.9–31.6; Table 2). Intriguingly, this relatively high mutational burden was not present in her adenocarcinoma, and only one of the mutations identified in this sample was present in the cancer sample—namely, PTEN W111*.

In the 2017 cancer sample, only two mutations were identified: PTEN W111* and ARID1A K1929del (Table 2). The truncating PTEN W111* mutation present in the tumor sample (allele frequency [AF] 36.9%) was present in every biopsy sample analyzed throughout the course of the patient's medical journey, including the 2009 ACH sample, but with the sole exception being that of the histologically normal endometrial biopsy in 2005. This finding is consistent with the knowledge that PTEN is the most mutated gene in EEC (Levine and Cancer Genome Atlas Research Network 2013) and is also found to play a role in early malignant transformation events in EEC (Levine et al. 1998; Mutter et al. 2000). The second tumor-linked mutation identified was a likely pathogenic, in-frame deletion ARID1A K1929del (AF 8.1%). This ARID1A mutation was not identified in any of the patient's biopsy samples including the 2009 ACH sample. In general, ARID1A-inactivating mutations are common in EEC (Guan et al. 2011; Levine and Cancer Genome Atlas Research Network 2013) and ARID1A protein expression has been found to decrease in the progression to EEC from AH (Mao et al. 2013).

Apart from the PTEN W111* mutation, the only other recurrent mutation identified in the samples was the pathogenic, missense KRAS G12D mutation (Table 2). This mutation was present in a total of six samples. As noted above, the KRAS G12D mutation was present in the histologically normal endometrium sample obtained at the time of the patient's first visit in 2005. Although we did not detect this KRAS mutation in our patient's cancer sample using our next-generation sequencing panel, we were curious to use an even more targeted and sensitive technology. We, therefore, performed droplet digital polymerase chain reaction (ddPCR). The KRAS mutation was detected but at a very low allele frequency of 0.3%; a level below our initially defined cutoff threshold (Supplemental Table 2). Although KRAS is in general found to be mutated in many different cancer types, the exact role KRAS mutations play in EndoCA is still unknown (Sideris et al. 2019).

PTEN Mutant Expression Levels Increase over Time and Decrease following Progesterone Therapy

Interestingly, the identified pathogenic, nonsense mutation in the PTEN gene was present in the patient's 2005 nonatypical complex hyperplasia biopsy and persisted in each of her subsequent biopsies. Furthermore, the allele fraction of this PTEN mutation reached the highest allele frequency in her 2017 adenocarcinoma and adjacent complex AH samples (AF 36.9% and 31.9%, respectively), suggesting its role in clonal expansion (Fig. 2). Overall, the allele fractions of this PTEN mutation were commensurate with histological changes (Table 2; Fig. 2).

Figure 2.

Allele frequencies from next-generation sequencing results of the PTEN pW111* mutation over time. The patient was prescribed megestrol acetate during the highlighted times. Higher PTEN allele fractions are associated with a more severe pathological phenotype and present at lower frequency in benign phenotypes. Megestrol acetate plays a therapeutic role in mitigating hyperplasia. The chart depicts the pathology that underwent sequencing. Each biopsy was independently validated by droplet digital polymerase chain reaction (ddPCR) except for the 2019 sample, which represents ddPCR for PTEN mutation only. Tic marks between years 2007 and 2009, 2010 and 2014, and 2014s and 2017 represent the intervals the patient was lost to follow-up.

This W111* mutation is predicted to result in a loss of function due to a premature termination in exon 5. PTEN functions as a tumor-suppressor gene by down-regulating the PI3K–AKT pathway (Carracedo and Pandolfi 2008). A loss of function mutation would, therefore, promote progression into the cell cycle via AKT–mTORC signaling cascade (Nussinov et al. 2021). Allele fractions for the PTEN mutation were comparable between NGS and ddPCR in all samples tested apart from the 2006 biopsy results in which the AF measured by ddPCR was 6.3% AF but 23.3% by NGS (Supplemental Table 2). Allele fractions from NGS >15% were independently validated using Sanger sequencing—namely, the 2010 simple hyperplasia biopsy (AF 19.1%) (Supplemental Fig. 1A), the 2017 complex hyperplasia biopsy (AF 31.9%) (Supplemental Fig. 1B), and the adenocarcinoma biopsy (AF 36.9%) (Supplemental Fig. 1C).

As shown in Figure 2, the most consistent drops in PTEN allele fraction levels were noted following the three intervals in which the patient was prescribed megestrol acetate: 2006, 2009, and 2017 (Fig. 2). As noted, levels decreased from 23.3% in 2006 to 1.1% in 2007, 26.1% in 2009 to 19.1% in 2010, and then finally from 36.9% in 2017 to 0.7% when again measured in a biopsy sample obtained in 2019.

DISCUSSION

Here, we report the first study linking histologic development of EndoCA from benign hyperplasia in a single patient with the somatic, driver mutations accompanying that transformation and the effects of medical treatment on those mutations. Studies that longitudinally track benign hyperplasia into cancer are generally lacking because most cases are believed to regress without medical/surgical intervention. Similarly, the transformation of complex hyperplasia with atypia into cancer is not amenable to studying because once these preneoplastic lesions are identified they are surgically excised. As our patient was not deemed a surgical candidate, we had a rare opportunity to study the 12-yr histologic/molecular evolution of her cancer and the histologic and molecular effect of treatment with progesterone.

Identifying those actual mutations linked to driving cancer development, and not just associated passenger mutations, is a critical goal in understanding endometrial carcinogenesis and, simultaneously, a necessity in developing sensitive and specific molecular diagnostic and screening tools for EndoCA. This n-of-1 study allowed us the unique opportunity to follow the real-time development of EndoCA, describe the specific molecular events associated with each histologic transformational step, and, quite importantly, describe the effect of medical management on those molecular features. In doing so, we were able to track the rise, fall, and disappearance of many EndoCA-linked mutations and the persistence of one specific PTEN mutation linking its presence to the development of EndoCA. Although we focused on only those pathogenic or likely pathogenic mutations, a full documentation of all mutations identified and associated allele fractions are reported (Supplemental Table 1).

A discrete number of pathologic and likely pathologic mutations in known cancer driver genes were identified in the different histologically defined samples across the transformation timeline. At the time of her presentation in 2005, and in histologically normal tissue, one cancer driver mutation, KRAS G12D, was identified. The relatively high expression of this mutation in histologically normal endometrial tissue, an AF of 9.2%, but lack of expression in other samples and eventual very low expression, 0.3%, in the EndoCA sample suggest that this mutation was not a driver of cancer development in this patient. Although the presence of a cancer driver mutation in histologically normal tissue may have until recently been considered surprising, this is not unexpected. Specifically, we had hypothesized that normal endometrium harbors a previously unrecognized and intense landscape of somatic mutations (Nair et al. 2016). This was based on our original demonstration of the presence of bona fide mutations in uterine lavage fluid from women with and without EndoCA. Indeed, soon after our report based on liquid biopsy–based studies, we and others confirmed these findings by direct examination of endometrial tissue (Suda et al. 2018; Lac et al. 2019; Moore et al. 2020; D Pandya, S Tomita, MM Padron-Rhenals, et al., submitted).

The most complex set of mutations in our patient were identified nearly 8 yr prior to her cancer diagnosis. Specifically, a total of seven high allele fraction mutations were identified in six genes associated with the development of complex hyperplasia with focal atypia, a known preneoplastic lesion (Table 2). All but two of these, KRAS G12D and PTEN W111*, would be single-occurring events. Only one of these mutations, PTEN W111*, would be detectable in the patient's future EndoCA sample. The second mutation present in the patient's tumor, ARID1A K1929del, was not present in the 2009 preneoplastic lesion nor was it detected in any other sample in this study. Its allele fraction was markedly lower than that of the PTEN mutation (8.1% vs. 36.9%) suggesting a secondary role when compared to PTEN in this patient's tumor development. It is noted that the 2009 sample had the highest number of mutations identified in any of the biopsies. Because all samples were obtained by the same surgical technique (i.e., endometrial biopsy), it would be expected that the amount of tissue biopsied would be relatively the same for each sampling. Therefore, the number of mutations found was not based on the amount of tissue sampled but could, however, be determined by the location of a deeply heterogenic tissue sample. As with any biopsy of the uterus, concurrent histologies and mutations may be missed by chance or mutations may disappear because of the dynamic mutational nature of the endometrial lining.

PTEN is a known critical tumor-suppressor gene and is frequently mutated in endometrioid EndoCA. Mutations in PTEN have also been reported in both benign and atypical endometrial hyperplasia, suggesting mutations in PTEN may be one of the earliest molecular events contributing to EndoCA carcinogenesis (Mutter et al. 2000; Sun et al. 2001; Konopka et al. 2002; Gbelcová et al. 2015). The specific PTEN mutation in our patient, p.W111*, has been previously identified in EndoCA (Nair et al. 2016; Zehir et al. 2017), breast cancer, gliosarcoma, and lung adenocarcinoma (Reis et al. 2000; Kan et al. 2010; Imielinski et al. 2012).

The identification of and increases in the allele fraction of this PTEN p.W111* mutation correlated with the malignant transition of benign hyperplasia to adenocarcinoma. Interestingly, decreases in the PTEN p.W111* allele fractions were consistently noted following progesterone therapy (Fig. 2). Progesterone acts as an estrogen antagonist resulting in the prevention of uncontrolled cell proliferation in estrogen-dependent endometrioid EndoCAs (Kim and Chapman-Davis 2010). In 2006, the PTEN p.W111* allele fraction prior to the initiation of medical treatment was 23.3%. The allele fraction was dramatically reduced to 1.1% after the patient was treated with megestrol acetate. Similarly, 2 yr later in 2009, the PTEN p.W111* allele fraction reached 26.1% in the complex atypia sample. That expression was reduced to 19.1% at the end of the year's treatment and then was measured at 3.0% in a sampling of proliferative endometrium several years later (2014). As noted above, the highest PTEN allele fraction, 36.9%, was detected in the 2017 tumor sample. Intriguingly, and although only detectable by targeted ddPCR, secondary to its low level of expression of 0.7%, the PTEN p.W111* mutation was still detectable in 2019 in a benign inactive gland sampling of tissue. Although we cannot definitively conclude that progesterone therapy directly caused the associated drops in the PTEN W111* expression, a strong association is suggested. Better understanding of the timeline and clonal behavior of driver mutations in combination with presenting histology could provide an interesting possible real-time view on the effect of hormone treatment antagonizing molecular drivers and its action on delaying tumor development.

Given that the PTEN W111* mutation was still detectable 2 yr after the diagnosis of EndoCA, we will continue to monitor for the changes in expression level when biopsies are available for molecular analysis. A previous study in patients who had not undergone hysterectomy following the diagnosis of non-AH found a lead time of >9 yr to the diagnosis of carcinoma and just >4 yr for diagnosis of AH to cancer (Kurman et al. 1985). Taken together with our findings, we believe the lead-up time to the development of EndoCA presents a clinically ideal scenario for the development and use of screening/monitoring tests so that nonsurgical options can be offered to patients.

We previously had the opportunity to track the development of EndoCA over time in an asymptomatic 67-yr-old female without histopathologic evidence of premalignant lesions or EndoCA patient who was found to express two recognized oncogenic PTEN mutations in a uterine lavage fluid biopsy obtained as part of a study (Martignetti et al. 2018). Nearly 1 yr later, the patient returned with postmenopausal bleeding. A single microscopic focus of EndoCA was detected at that time, and DNA was isolated and sequenced from laser-capture microdissected tumor tissue. Analysis revealed the same two PTEN mutations identified nearly 1 yr earlier. The likelihood that these two mutations would occur by chance alone was essentially nil (P < 3 × 10⁻⁷). Thus, when viewed together, these two patients suggest that future, tumor-specific mutations can be identified in asymptomatic individuals and potentially used as screening tools in certain subsets of patients. However, if broad, large-scale screening approaches toward EndoCA are to be developed, we believe this needs to be balanced against other findings.

Specifically, we previously demonstrated that the presence of bona fide mutations in recognized cancer driver genes is not sufficient for diagnosing cancer given that the normal endometrium harbors a previously unrecognized and intense landscape of somatic mutations (Nair et al. 2016; Martignetti et al. 2018). Indeed, soon after our studies based on liquid biopsies, others confirmed our findings by direct examination of endometrial tissue (Suda et al. 2018; Lac et al. 2019; Moore et al. 2020). Thus, we suggest being wary of an overly simplified dependence based solely on mutation detection, or allele fractions, as our understanding between truly pathogenic and passenger is still evolving. More reasonable diagnostic and screening may, therefore, include a combination of DNA and proteomic-based (Wang et al. 2018) or completely proteomic-driven (B Reva, D Rykunov, P Dottino, et al., in prep) tests. Today, there are no screening tests for EndoCA. Ultimately, detecting early-stage, localized EndoCA or even premalignant lesions that can be medically managed means a potential for cure.

METHODS

Study Design and Sample Collection

This retrospective case study was conducted at the Icahn School of Medicine at Mount Sinai (ISMMS), New York, New York. Patient informed consent was obtained in accord with the Institutional Review Board ISMMS by the patient's attending physician. Formalin-fixed, paraffin-embedded (FFPE) samples were obtained from the Mount Sinai Hospital Pathology Department. FFPE samples were obtained from the patient's endometrial biopsies, starting with her initial visit in 2005, subsequent visits in 2006, 2007, 2009, 2010, and 2014, her diagnosed EIN/AH and adenocarcinoma in 2017, to her last follow-up in 2019. Ten milliliters of blood was collected into a BD Vacutainer K3 EDTA tube (Fisher Scientific) in 2018 for defining germline DNA status.

FFPE Tissue Processing

The slides from each FFPE block were again reviewed at the beginning of this study by an American Medical Board–approved Mount Sinai gynecological oncological pathologist, T.K., to confirm each previously reported pathology. FFPE blocks containing a single defined histology that represented >70% of the specimen were cut into 6-μm tissue sections and stored until DNA extraction at −80°C. FFPE samples where tissue sections were obtained were from years 2007, 2010, 2014, and 2019. In biopsies with more than one histology, 4-μm sections were cut and transferred to unfrosted glass slides and laser-capture microdissection (LCM) was performed (LCM-7000; Leica Systems). LCM was performed on samples from 2005, 2006, 2009, and 2017. The accompanying H&E slide was used to map the different histologies of each biopsy.

DNA Extraction and Quantification

For germline DNA extraction, whole blood was centrifuged at 2500 rpm for 10 min to separate out plasma. Remaining red blood cells were resuspended in 5 mL of RBC lysis buffer (155 nM ammonium chloride, 10 mM sodium bicarbonate, and 0.1 mM EDTA) and spun at 1200 rpm for 10 min. The resulting PBMC pellet was suspended in cell lysis solution (QIAGEN) and mixed with Protein Precipitation Solution (Thermo Fisher) and spun at 1200 rpm for 10 min. Supernatant was then transferred into 300 µL of 100% isopropanol and centrifuged at 13,000 rpm for 1 min. The DNA pellet was resuspended in 100% ethanol to wash, spun at 13,000 rpm, and rehydrated in molecular-grade water.

DNA from samples from which 6-μm curls were cut was purified using QIAamp DNA FFPE Tissue Handbook (QIAGEN) according to the manufacturer's instructions. DNA from LCM tissue was purified using QIAamp DNA Micro Kit (QIAGEN) according to the manufacturer's instructions. DNA quantification for all samples was determined by QuBit fluorometry (Thermo Fisher Scientific).

Next-Generation Panel–Based Sequencing

Next-generation sequencing was performed using a custom-designed Swift Biosciences 58 oncology gene panel: ABL1, AKT1, ALK, APC, ATM, ARID1A, BRAF, CDH1, CDKN2A, CSF1R, CTNNB1, DDR2, DNMT3A, EGFR, ERBB2, ERBB4, EZH2, PIK3R, FBXW7, FGFR1, FGFR2, FGFR3, FLT3, FOXL2, GNA11, GNAQ, GNAS, HNF1A, HRAS, IDH1, IDH2, JAK2, JAK3, KDR, KIT, KRAS, MAP2K1, MET, MLH1, MPL, MSH6, NOTCH1, NPM1, NRAS, PDGFRA, PIK3CA, PIK3R1, PTEN, PTPN11, RB1, RET, STK11, SMAD4, SMARCB1, SMO, SRC, TP53, TSC1, VHL, ARID1A, and PIK3R. Libraries from 10 FFPE samples and a germline sample from whole blood were created using the multiple targeted amplicon technology provided by Swift Biosciences (custom 58G Oncology Panel Kit, Swift Biosciences AL-56248). All samples were sequenced on an Illumina Hiseq 4000 at ultrahigh depth of 1500×. Variant calling on each sample's FASTQ files was performed using LoFreq and the Genome Analysis Toolkit HaplotypeCaller (GATKhc). The human genome Hg19 (GRCh37.p5) was used as the reference genome. All variants were further filtered using QIAGEN's CLC Genomics Workbench 12.0.2 (https://digitalinsights.qiagen.com). Somatic variants were produced by removing variants found in the germline sample. Somatic filtered variants from all 10 samples were then processed through the Qiagen Clinical Insight (QCI) tool to remove common variants (minor allele frequency [MAF] > 1%) and assign functional impact https://digitalinsights.qiagen.com). All variants were defined as pathogenic, likely pathogenic, uncertain significance, likely benign or benign by QCI based on the American College of Medical Genetics Guidelines, and further using SIFT (2016-02-22), PolyPhen-2 (v2.2.2) algorithms. Additional annotation with CADD score and COSMIC status were also produced for each variant.

Sanger Sequencing Validation

Sanger sequencing was performed for the PTEN p.W111* mutation that had a >15% allele fraction from the NGS results. The following primers were designed flanking the detected PTEN mutation: F: 5′-ACCTGTTAAGTTTTATGCAACATTTC-3′ R: 5′-AGAAATCTAGGGCCTCTTCTGC-3′. PCR was performed using GoTaq Colorless Master Mix to amplify this genomic region. An amount of 25-µL reactions were set up using 10 ng of DNA template with the following cycling protocol: 1 cycle of 95°C for 2 min; 40 cycles of 95°C for 1 min; 60°C for 30 sec; 72°C for 1 min; and 1 cycle of 72°C for 7 min. PCR products were purified using the QIAquick PCR Purification Kit (QIAGEN) according to the manufacturer's protocol.

Droplet Digital PCR Validation

Mutation validation was performed via ddPCR using RainDance technologies when AF were <15% or insufficient DNA was available for PCR. Probes were designed using a web-based design tool (Life Technologies). VIC and FAM probes were designed and used flanking the PTEN and KRAS chromosomal mutations Chr 10:89692848_G > A and Chr 12:25398284_C > T, respectively, along with wild-type sequences.

ADDITIONAL INFORMATION

Data Deposition and Access

The variants described in Table 2 have been submitted to ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/) and can be found under accession numbers SCV004035019, SCV004034087, SCV004034088, SCV004034089, SCV004034999, SCV004035001, SCV004035002, and SCV004035006.

Ethics Statement

This study was approved by the Icahn School of Medicine at Mount Sinai Institutional Review Board under the GCO# 10-1166. Written informed consent was obtained from the patient.

Acknowledgments

The authors thank the patients for their participation in this study. We also thank Dr. Jean-Noel Billaud's expert opinion and guidance, which significantly enhanced the quality of our research.

Author Contributions

K.R., O.C.-V., P.D., and J.A.M. conceptualized the study. Writing of the original draft was done by K.R., A.-M.B., and J.A.M., and review/editing was done by K.R., O.C.-V., D.P., S.C.C., M.M.P.-R. R.F.Q., T.K., A.-M.B., P.D., and J.A.M. Next-generation sequencing analysis performed by D.P. Molecular methods were performed by K.R., O.C.-V., R.F.Q., and S.C.C. T.K. and P.D. confirmed the histologic analysis of the biopsy slides.

Funding

The Laboratory received generous funding support from the Gordon family, the Ruttenberg family, the Mannheimer family, and the Goldstone family. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing Interest Statement

Drs. Dottino and Martignetti are cofounders and equity owners of MDDx, Inc., a private company that is developing new diagnostic tests for early detection of benign and malignant gynecologic conditions. They are also named inventors of intellectual property filed through the Icahn School of Medicine at Mount Sinai, which is currently licensed to MDDx, Inc.