Abstract GS2-09: Tamoxifen instigates uterine cancer development by activating PI3K signaling and supersedes PIK3CA driver mutations

Tamoxifen is widely used in the adjuvant treatment of estrogen receptor–positive (ER+) breast cancer and is an important drug for pre-menopausal women and post-menopausal patients who cannot tolerate aromatase inhibitors. Despite the clear clinical benefit in improving relapse-free and overall survival in these patients, an adverse effect of tamoxifen is a 2- to 7-fold increased risk of uterine cancer (UC) after 2-5 years of treatment. To date, the mechanism of tamoxifen-driven tumorigenesis is not well understood, and preventive approaches are lacking. Here, to molecularly characterize tamoxifen-associated uterine cancers (TA-UCs) and gain insights into their unique evolution, we performed whole-exome sequencing of 21 TA-UCs (discovery cohort) and droplet digital PCR (ddPCR) of an additional 40 TA-UCs (validation cohort) obtained from the ‘Tamoxifen Associated Malignancies: Aspects of Risk’ (TAMARISK) study. In addition, we used in vivo mouse models to: (i) further investigate tamoxifen-activated molecular pathways that may be involved in TA-UC tumorigenesis; and (ii) offer mechanistic insights. Overall, we discovered that TA-UCs were genomically similar to non–TA-UCs from The Cancer Genome Atlas (TCGA) project, with one profound exception: TA-UCs are characterized by a lower-than-expected frequency of mutations in two highly prevalent UC driver genes in the PI3K pathway: PIK3CA (14% [3/21] vs 48% [265/554] in non–TA-UC; P=0.003, Fisher’s exact test; Q=0.02, Benjamini-Hochberg FDR) and PIK3R1 (0%, [0/21] vs 31% [174/554]; P=0.001; Q=0.01). We used ddPCR in the independent TA-UC validation cohort and confirmed the low frequency of mutations in PIK3CA (7.5% [3/40] vs 21% [144/685] in control UCs from the Dana-Farber contribution to the AACR GENIE project; P=0.04). We next performed mouse in vivo studies and demonstrated that tamoxifen activated the PI3K pathway and increased cell proliferation in normal mouse uterine tissue through paracrine and autocrine effects, both of which were abrogated by the PI3K inhibitor alpelisib. Taken together, we describe a distinct and novel pathway of carcinogenesis in which tamoxifen acts as a driver event in the uterus and promotes tumor development in a mutation-independent manner. Indeed, tamoxifen may increase the risk of UC by activating the PI3K pathway, which can substitute for the early acquisition of oncogenic PIK3CA or PIK3R1 mutations observed in non–TA-UC tumors. Furthermore, the ability of a PI3K inhibitor to reduce cell proliferation in our mouse model raises the possibility that downregulating the PI3K pathway may prevent or significantly reduce TA-UC development, offering a potential future therapeutic and prevention strategy for specific high-risk patients undergoing tamoxifen therapy.

Citation Format: Kirsten Kubler, Agostina Nardone, Shankara Anand, Daniel Gorvich, Marjolein Droog, Francisco Hermida-Prado, Tara Akshi, Avery S Feit, Gabriella Cohen, Gwen Dackus, Matthew Pun, Yanan Kuang, Justin Cha, Mendy Miller, William J Gibson, Cloud P Paweletz, Eliezer M Van Allen, Flora E van Leeuwen, Petra Nederlof, Harry Hollema, Quang-Dé Nguyen, Marian JE Mourits, Ignaty Leshchiner, Chip Stewart, Ursula A Matulonis, Wilbert Zwart, Yosef E Maruvka, Gad Getz, Rinath Jeselsohn. Tamoxifen instigates uterine cancer development by activating PI3K signaling and supersedes PIK3CA driver mutations [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr GS2-09.

Abstract 2727: The premalignant state captured in the landscape of somatic mutations can reveal the cancer cell-of-origin

Despite increasing knowledge of tumorigenesis, the identity of the cancer cell-of-origin, i.e. the normal cell type that acquired the cancer-initiating event, remains largely unknown. Our approach of identifying the cell-of-origin is based on two observations: (1) the chromatin structure is cell-specific; and (2) the density of somatic mutations along the genome is associated with the regional profile of chromatin modifications.

We have previously developed a method that quantifies the ability to predict the mutational distribution along the cancer genome from the profile of epigenetic modifications in different normal cell types. Here we present the largest application of our method using 2,550 whole genomes representing 32 distinct cancer types. To identify the cell-of-origin, we determined the correlation between the observed density of mutations along the genome and the predicted values based on chromatin modifications from 104 different normal tissue types. The normal cell type that showed the strongest correlation with a specific cancer mutational landscape was the candidate cell-of-origin.

We found that in almost all cancer types the cell-of-origin can be characterized solely from DNA sequences. Interestingly, we found that the fallopian tube was the best match for high-grade serous ovarian cancer, providing independent evidence that this is the cancer’s site of origin. For breast cancer we found that the four distinct subtypes best-matched cells from the luminal cell lineage: basal-like breast cancer likely originates from luminal progenitors, whereas all other subtypes from luminal mature cells. This association holds true even when accounting for different alterations in the homologous recombination repair pathway, suggesting that subtypes are more determined by the cell-of-origin than the specific DNA repair defect. In addition, we found that we could identify the cell-of-origin using metastatic samples – a finding that may help in difficult clinical diagnoses. Moreover, we demonstrate that cancer drivers, both germline risk alleles and somatically mutated drivers, reside in active chromatin regions in the respective cell-of-origin.

Taken together, our findings indicate that many of the somatic mutations accumulated while the cells maintained a chromatin structure similar to the cell-of-origin (likely occurring prior to transformation). Therefore, this historical record, captured in the DNA, can be used to identify, the often elusive, cancer cell-of-origin. Our approach can ultimately help better understand the potential of particular normal cell types to transform and initiate cancer, as well as the association of the cell-of-origin with tumor subtypes and sensitivity to treatment.

Citation Format: Kirsten Kubler, Rosa Karlic, Nicholas J. Haradhvala, Kyungsik Ha, Jaegil Kim, Maja Kuzman, Wei Jiao, Sitanshu Gakkhar, Kent W. Mouw, Lior Z. Braunstein, Olivier Elemento, Andrew V. Biankin, Ilse Rooman, Mendy Miller, Christopher D. Nogiec, Edward Curry, Mari Mino-Kenudson, Leif W. Ellisen, Robert Brown, Alexander Gusev, Cristian Tomasetti, Hong-Gee Kim, Hwajin Lee, Kristian Vlahovicek, Charles Sawyers, Katherine A. Hoadley, Edwin Cuppen, Amnon Koren, Peter F. Arndt, David N. Louis, Lincoln Stein, William D. Foulkes, Paz Polak, Gad Getz. The premalignant state captured in the landscape of somatic mutations can reveal the cancer cell-of-origin [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2727.

Abstract 3304: The mutation landscape of cancers serves as a record of early malignant transformation

How the cell lineage influences a tissue's susceptibility to malignant transformation is a fundamental question in cancer biology, which has been barely addressed in cancer genomics thus far. Cell properties are encoded in the cell type-specific chromatin structure and we previously demonstrated that the cell-of-origin (COO) chromatin organization is a key determinant of the landscape of somatic mutations, which accumulated over lifetime serving as a memory of the historical cell lineage (Polak et al, Nature , 2015). We now show that this principle is generalizable to common tumor types and offers insights into the molecular events of cancer initiation. We extended our work to 2,641 genomes from 30 cancer types and epigenetic modifications from 98 normal tissues. In 25 cancer types, the tumor originated from a cell type that was its direct cellular counterpart or a close proxy; in only two, there was no match or a close proxy; and in the remaining three (esophageal, pancreatic ductal and biliary adenocarcinoma) the best matched cell type suggested metaplasia to stomack mucosa like tissue.The cellular context of breast tumor formation was investigated in more detail, showing that the COO, and not the gene inactivation event, determines the subtype. Basal-like tumors appeared to arise from luminal progenitor cells, while all other subtypes arose from mature luminal cells. Furthermore, irrespective of the inactivation mechanism (pathogenic germline, somatic truncating or epigenetic silencing event), all BRCA1/2- and RAD51C-altered basal-like tumors best matched to luminal progenitors while BRCA1/2- and CHEK2 -mutated luminal A/B subtypes best matched mature luminal cells. Finally, we observed that tumor type-specific driver genes reside in genomic regions that are defined by a highly active chromatin environment in their COOs. This highlights their essential role in cell type differentiation and implies the acquisition of somatic mutations early, when the chromatin architecture still reflected the COO. Taken together, our findings shed light on the crucial role of the COO in shaping the mutational landscape and tumor evolution.

Citation Format: Paz Polak, Rosa Karlic, Kirsten Kubler, William D. Foulkes, Gad Getz. The mutation landscape of cancers serves as a record of early malignant transformation [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3304.