Table 1.
Representative genes and their functions incorporated in the DEABM.
Figure 1.
Overall schematic of cell-types and their interactions involved in duct epithelial cell life cycle.
This figure depicts the minimally sufficient set of cell types and their interactions necessary to represent the growth and maintenance of the breast duct epithelial cell population. In particular note that given this representation ER+ cells do not have proliferative potential, a state that is maintained through the suppression of cMet by RUNX3.
Figure 2.
Set of included representative “genes” and their relationship to cellular behaviors and general functions within the DEABM.
As the representational focus of the DEABM is on characterizing the functional dynamics associated with oncogenesis, potentially detrimental “genes” have been included on their known influences on those functions that are plausibly involved and altered in the process of tumorigenesis. Additionally, the arrows are intended to represent known direct regulatory effects; it is expected that there are many second and third order effects that might lead a named gene to affect other downstream behaviors. *Note that the labeling of these “genes” is not intended to be a comprehensive description of all the known effects of the named genes, but rather to label certain putative cellular behaviors possibly involved in malignant transformation.
Figure 3.
Schematic of control logic concerning DNA damage, repair and functional consequences of unrepaired DNA damage within the DEABM.
A baseline premise of the DEABM is that DNA damage can occur during a luminal epithelial cell’s life-time, and that damage that remains unrepaired by the time the cell is to divide can be passed on as a mutation, a certain subset of which may affect a critical cellular function that may influence tumorigenesis. The DEABM incorporates abstract representations of DNA damage, damage repair, senescence, apoptosis and passage of mutations to subsequent cellular generations.
Table 2.
Agent classes and their associated variables.
Figure 4.
Post-calibration behavior of the DEABM reproducing baseline, normal breast epithelial dynamics.
These graphs demonstrate the ability of the DEABM to generate recognizable fluctuations in luminal cell mass during normal menses (Letter A), demonstrating the first stage of the face validity of the DEABM in being able to reproduce self-sustaining cellular population without evidence of unconstrained growth. Furthermore, the DEABM was also able to reproduce expected alterations in luminal cell population dynamics associated with pregnancy, initiation depicted by red arrow (Letter B). These data.
Figure 5.
Demonstration of the DEABM to reproduce expected patterns of luminal cell growth in response to estrogen and progesterone within a menstrual cycle.
Panel A depicts the output of the DEABM in terms of differentiated luminal cell population during the course of a single menstrual cycle period. The individual runs (n = 5) are depicted in light grey dashed plots and demonstrate the inter-run variance expected from the stochastic nature of the DEABM. The average of these runs is seen in the solid black line, and reproduces the expected increase in luminal cell mass seen during the luteal phase. The general trajectory of the DEABM seen in Panel A is noted to be similar to reference data sets present in the literature, as seen in Panel B (reproduced with under the Creative Commons License from Ref [42]) and Panel C (reproduced with permission from Ref [43]); both of which depict the degree of luminal cell proliferation during various phases of the menstrual cycle. Note in particular the wide variance in the sample points present in the reference data sets, which represent multiple samples obtained from multiple individuals.
Figure 6.
Comparison of cumulative cancer incidence generated in the DEABM for wild-type/sporadic, TP53 and Myc mutation populations, as compared to the sporadic cumulative cancer incidence at age 55 reported in the SEER review.
Cumulative cancer risk following ∼40 simulated years of menstrual cycles (15000 steps), n-individual simulations = 500 in each group (N-group = 3). The baseline sporadic cancer risk was ∼3.6 (range 2.3 to 4.9)%, similar to the cumulative incidence rate of 2.94% in 55 year-old women as reported in the SEER 2010 review [44]. Dysfunction of p53 resulted in a nearly 7-fold increase in cumulative risk of invasive breast cancer, 24.6 (range 19.8 to 29.4)%, and hyper-activity of the proto-oncogene Myc resulted in over 2-fold increase in cumulative risk to 8.6 (range 6.4 to 10.8)%. These data demonstrate the ability of the DEABM to generate recognizable and plausible increases in cumulative cancer risk associated with known oncogenic mutations.
Figure 7.
Demonstration of the ability of the DEABM to recapitulate cumulative and longitudinal cancer incidence in populations with BRCA1 mutations.
Panel A demonstrates that the DEABM generated an increased aggregate cancer incidence over 40 years of menstrual cycles of ∼31.6% (range from 30.6–33.6%, N-groups = 3) compared to both sporadic/wild-type simulations and the SEER review data. The simulated BRCA1 values all fall well within the range of cumulative cancer incidences reported in BRCA1 population studies between 17–58% [46]–[50]. Panel B demonstrates that in addition to reproducing plausible cumulative cancer incidences, there is similar matching between longitudinal incidence over this ∼40 year interval between the output of DEABM simulations and the aggregated data from the previously noted studies on BRCA1 [46]–[50]. Published BRCA1 population study plots are labeled by study author name, whereas the individual DEABM cohorts of n-individuals = 500 are labeled by N-group number.
Figure 8.
Reproduction of ER tumor status in both wild-type/sporadic and BRCA1 mutated populations of breast cancer.
These data demonstrate the similarity between DEABM simulation runs and data extracted from the literature concerning the percentage of ER+ tumors generated in both wild-type/sporadic and BRCA1-mutated populations [46]–[57]. Panel A depicts the ER+ percentage among wild-type/sporadic populations from both the literature, ∼68% (range 60–77%) of premenopausal breast tumors [51]–[54], and in simulated populations (n-individuals = 500, N-groups = 3) of ∼65% (range 59–71%) of the simulated breast cancers. Panel B demonstrates the same comparison of ER+ tumors in the BRCA1 mutant population, where the DEABM shows that only ∼38% (range 29–44%) of tumors generated were ER+ as compared to published incidences of ER+ BRCA1 mutant tumors of ∼36% (range 19–52%) [51], [55]–[57]. For both Panel A and B published cancer population data is denoted by the name of the study’s first author, whereas the DEABM runs are labeled with their N-group number. These findings indicate that the DEABM incorporates plausible mechanisms for ER+ tumorigenesis, suggesting a role of RUNX3 expression (or other genes performing a similar function) in the selectivity of ER+ breast cancer previously unknown.
Figure 9.
Relationship between RUNX3 expression and ER status in breast cancer.
Graphical representation of search results from the Cancer Genome Atlas (TCGA) and Oncomine (www.oncomine.org) to identify correlations between RUNX3 expression and both ER+ and ER− breast tumors. These results suggest a trend towards decreased expression of RUNX3 in ER+ tumors, a finding consistent with our hypothesis that loss of RUNX3 function may be one of a class of genetic abnormalities that results in the loss of suppression of the proliferative potential of ER+ cells, and that this may be a precondition in the development of ER+ tumors.