Coverage from the 28th annual meeting of the Society for Maternal-Fetal Medicine
Slide 1 & 2-Preterm delivery is the leading cause of infant mortality in the United States and in the world. Although infection is found in the majority of early preterm delivery (<32 weeks) the mechanisms involved are not clearly known.
The pathologies that lead to preterm delivery are thought be in action as early as the second trimester; however there are no reliable early diagnostic tests.
The women at risk for preterm delivery are a heterogeneous group with host (ethnicity, stress level, multiple pregnancy etc) and environmental factors (pollution, nutrition etc) in play. The treatments proposed to prevent preterm delivery, such as progesterone, are effective in a small portion of women who are not well defined.
Another layer of complexity is the difficulty of performing research in pregnant women due to the potential risks of exposing the fetus to chemicals in utero.
Slide 3-Computer modeling is used to study the behavior of objects or systems that cannot be easily or safely tested in real life. Similar to the internet, it has been first used in the military, for war-simulations. Quickly computer modeling gained acceptance in civilian life. Simulations are now used in the automobile industry (car design), toxicology (drug side effects), chemistry (molecular interactions, drug design), genetics (mutations, polymorphisms) and epidemiology (vaccine effectiveness) Duarte et al. PNAS 2007; 104:6.
For patients this means expedited better, safer treatments. Computer modeling also allows patient-specific-factors to be tested in silico; which allows personalized medicine development.
Slide 4-Similar to in vitro and in vivo experimentation, in silico proteomic and genomic data are fed into the model. The model can be developed to include the interactions between multiple systems such as (placenta, mother and fetus) in a dynamic fashion. For example the effect of addition of different concentrations of progesterone at different time points can be assessed to prevent parturition.
The model is built based on the assumption that there is redundancy in the system but there are key regulatory molecules; modulation of which may, singly or in combination, regulate the outcome. The complex signaling cascades may be represented with the endpoint, master-regulatory molecules (i.e. infection is represented with increased activated nuclear-NF-kB concentration).
Slide 5, 6, 7- By using this cutting edge technology the myometrial parturition model was built using bottom-up approach, including the nuclear, cytoplasmic, intercellular and cell membrane components in collaboration with Cellworks Group Inc. The biochemical interactions between the molecules were represented with mathematical differential equations. A proprietary Cellworks Group Inc. software program B-PACE was used.
Our goal is to better understand the pathogenesis of preterm delivery, identify diagnostic markers and the key regulatory molecules (nodes) that may be modulated for treatment.
Slide 8- We developed the model based on the in vivo estrogen, progesterone, cortisol, CRH levels observed in fullterm pregnant women and in vitro molecular human data. In the first simulations the labor as represented with an increase in progesterone receptor (PR) A expression relative to PRB expression (PRA/PRB ratio) was the end-point.
Slide 9-12-Since NF-kB activation (increased level of nuclear NF-kB) is the hallmark of infection, we induced NF-kB activation (at 0.2 ?M LPS) in 2nd trimester to assess the effect of infection on PRA/PRB ratio. This level of NF-kB activation is below that seen in labor and above that seen in labor, and represent subclinical infection (PMID: 11385114). We observed that NF-kB activation leads to increased PRA/PRB to labor levels.
Slide 13- In humans there is abundance of estrogen and progesterone during pregnancy and their action is regulated at their respective receptor level (PMID:12050275). We represented progesterone treatment with increased PRB concentration.
Slide 14-We observed that progesterone treatment was effective in preventing PRA/PRB increase in the absence of infection; which was independent of estrogen receptor (ER) level.
The model predicted that progesterone treatment is effective in preventing PRA/PRB increase in women without infection. (in silico data corroborated the in vivo observations in 463 women clinical trial- NEJM June 2003 PMDI# 12802023).
The progesterone treatment was not effective in preventing parturition in the presence of infection. (needs to be tested in a focused clinical trial; this observation may explain why progesterone treatment is effective only in a small percent of women at risk, and may be used to identify the women who may benefit from progesterone treatment).
Our model shows that the response to progesterone treatment to prevent preterm delivery may be influenced by the presence of subclinical infection; therefore the early diagnosis and treatment of infection may improve the progesterone effectiveness.
We will continue to develop the model to include the innate immune system, contraction and anti-inflammatory molecular pathways.
Computer modeling is the most recent discovery tool that may be used to:
- understand the biochemical relationships between molecules identified using proteomics and genomics
- understand the pathogenesis
- identify diagnostic markers
- identify the key regulatory molecules for treatment
- develop personalized medicine
- design the best clinical trial by identifying the group of patients that will best respond to the treatment
- identify the potential side effects and toxicities before the clinical trial is initiated