Many breast cancers, such as inflammatory breast cancer (IBC) and triple-negative breast cancer (TNBC), do not respond adequately to currently offered therapies. TNBC lacks the targeted therapy options available for hormone receptor–positive and HER2-positive breast cancers, so treatment relies on cytotoxic chemotherapy. A promising development for TNBC (and possibly for IBC) is availability of immune checkpoint therapies such as those targeting PD-L1, although the response rate with such therapies is low and the duration of response is short [1].
Many breast cancer patients receive neoadjuvant therapies prior to surgery, but those who do not have a pathologic complete response have a higher risk of relapse than those who do, because minimal residual disease (MRD) that persists after surgery (and other therapies, including radiation therapy and adjuvant chemotherapy) is often resistant to further treatment [2–4]. Some patients with MRD may have lengthy relapse-free survival, while others suffer early relapse. Most clinical trials in breast cancer aim to prolong survival after relapse, when the disease is at an advanced stage, a challenging task. To improve outcomes in patients with early relapse (among whom patients with IBC are overrepresented), we are investigating therapies that can reduce the likelihood of relapse. Intervention before relapse occurs will likely result in much better outcomes.
To accomplish this goal, we have developed an experimental approach involving the modeling of poor-prognosis MRD, which is likely responsible for early relapse, in a cell culture system; this model system can then be used for testing therapies. Cancer evolution is governed by an interplay between the genotypes of cancer cells and many selection pressures in the body, including those imposed by the immune system [5–9]. A major limitation of most cell culture systems is that although they can model a desirable genetic change, they lack the body-like evolutionary selection pressures that maintain the tumor adaptability that drives cancer evolution and therapy resistance. To address this limitation, we previously applied a severe body-like selection pressure, i.e., a prolonged lack of glutamine, in a cell culture of the triple-negative IBC (TN-IBC) cell line SUM149, which eliminated 99.99% of the cells [10]. Interestingly, the remaining cells survived in quiescence for several weeks and then proliferated indefinitely, indicating that they might have characteristics similar to those of poor-prognosis MRD. This tumor adaptability phenotype, i.e., an opportunistic switching by cancer cells between quiescence and proliferation, can be used to define poor-prognosis MRD in a way that is quantifiable in cell culture. In other words, if MRD mostly stays dormant in the body for years and decades, it might not cause relapse. However, if it constantly switches between quiescence and proliferation depending upon the selection pressures applied, it would lead to early relapse. The cell culture model we are developing can model this process. In support of the validity of cell culture-based approach for modeling deep intrinsic resistance, we have found that IBC-derived TNBC cell lines (SUM149 and FC-IBC02) are a much better source of highly adaptable cancer cells than non-IBC-derived TNBC cell lines (MDA-231 and its metastatic variants cultured from bone metastases in nude mice). This is evident from the observations that adaptable cells derived from IBC cell lines proliferate indefinitely in a glutamine-deficient medium, while adaptable cells derived from non-IBC cell lines fail to do so [10, 11]. This result in cell culture mirrors a higher adaptability of IBC than non-IBC in vivo.
We refer to the adaptable cancer cells selected in this manner as metabolically adaptable (MA) cells. SUM149-MA cells can also survive other metabolic challenges, such as a lack of glucose or oxygen, better than the parental cell line [10, 11]. They are resistant to the chemotherapeutic drugs paclitaxel and doxorubicin and to several targeted therapies that inhibit cell proliferation [12]. Importantly, SUM149-MA cells are highly tumorigenic in nude mice, efficiently metastasizing to multiple organs such as skin, lungs, and brain [10]. Molecular characterization has shown that they have activation of pathways that promote epithelial to mesenchymal transition (EMT) (e.g., high ZEB1, high SNAIL1, low GRHL2) and numerous changes affecting the epigenome (e.g., low TET2), alternative RNA splicing (e.g., low ESRP1, ESRP2), and RNA base modifications (e.g., high FTO) [12–15]. These suggest that body-like selection pressures select highly adaptable cancer cells that may drive therapy resistance. In the context of our cell culture model, genetic changes present in MA cells may cooperate with alterations in the epigenome, transcriptome, and proteome, to provide different selection advantages over time and to confer deep intrinsic resistance.
Although SUM149-MA cells exhibit some useful features of poor-prognosis MRD, the problems inherent to cell culture still exist. The selection pressures imposed by the body occur one after another, a sequence that is not feasible to emulate in cell culture. Once a given challenge has been overcome in cell culture, proliferative cells, irrespective of their inherent adaptability, will have an advantage. To address this issue at a practical level, we rely on another evolution-related concept that is well accepted in cancer research: the inverse correlation between cell proliferation and cell adaptability in biological systems. This concept also fits well with the cancer stem cell concept, wherein a small subpopulation of progenitor-like cancer cells (residing mostly in quiescence) drives the disease. Therefore, when we assay the efficacy of therapeutic drugs in cell culture, we assume that highly proliferative cancer cells are the first to be eliminated. Thus, by eliminating the vast majority of less-resistant cancer cells we can reveal the resistant cells, which survive by switching to quiescence (a measurable phenotype and a characteristic of poor-prognosis MRD).
To provide proof of concept for the usefulness of our cell culture model for evaluating cancer therapies, in the present study, we evaluated a well-known BET bromodomain inhibitor, JQ1, which has been shown to modulate the cancer epigenome favorably to overcome therapy resistance in many leukemias and solid cancers [16–21]. Preclinical studies suggest that JQ1 may overcome resistance to other therapies in estrogen receptor-positive and triple-negative breast cancers [22, 23]. The differentiation-state plasticity that often drives therapy resistance can be targeted by JQ1, which prevents changes to open chromatin architecture in basal-like breast cancer [24]. Interestingly, JQ1 treatment may also enhance antitumor immunity, which involves changes in expression of PD-L1 and PD-1 in tumor cells and immune cells [25, 26]. Thus, the purpose of this study was to use our novel cell culture model of resistant MRD to assess the efficacy of JQ1 against metabolically adaptable TN-IBC cells and its immune-modulatory effects in these cells via PD-L1.