Talk:Models of cardiac cell
The article is a very good summary but a bit light on new models published in the past five years which are notable in that they: (1) tend to use Markov models for more channels; (2) have been applied successfully to model effects of drugs and genetic mutations; (3) have been more closely coupled with more detailed models of interacting subsystems such as metabolism, Ca handling and signal transduction.
This is a clearly presented and practically useful compilation of computer models of cardiac electrical activity. I have only a few suggestions:
Somewhere near the beginning of the article, the authors should include a disclaimer that the article contains a sampling of available ionic models – not an exhaustive survey. In addition, the models shown are almost exclusively models of normal cardiac electrophysiology. As the authors know, several models designed to replicate certain pathologies, such as acute myocardial ischemia and hypertrophy, now exist. In addition, some studies have combined models to study impulse transmission across different cell types (e.g., from the sinus node to surrounding atrial myocardium and across Purkinje-muscle junctions).
Although the authors occasionally refer to the behavior of a given model in 2D or 3D, it would be useful to have a general statement indicating that good behavior of a model in 1D does not guarantee good behavior in higher dimensions.
The authors also might briefly indicate that all of the human ventricular models suffer from a limited source of actual voltage clamp measurements from normal human myocytes. In that regard, the current models must be regarded as something of a hybrid between normal and abnormal ionic behavior. In addition, when modeling human myocytes, the tacit assumption is made that human myocytes contain the same types of channels as dog or guinea pig. Given the known difference between ion channel types in dog and guinea pig (not to mention the much more significant differences between rodent and dog), and the lack of a comprehensive characterization of all possible ion channel types in normal human myocardium, it seems possible that this assumption may not be valid.
Reply to reviewers
We appreciate the reviewers’ comments and suggestions and their recognition of the hard work that went into researching and classifying the models presented here as well as coding, validating, and simulating a large number of them. In addition we have added two more models since last review.
Reviewer B: We disagree that the article is light on models published in the last five years. In fact, 21 of the 45 models included were published in the last eight years. Many of these models do include Markov descriptions of channels (such as the Iyer et al., which includes only a single Hodgkin-Huxley-style gating variable compared with 57 Markov state variables for the currents, as well as the Greenstein-Winslow, Mahajan-Shiferaw et al., Bondarenko et al.). A number of others include complex and detailed calcium handling, such as the Hund-Rudy, Greenstein-Winslow and related models, Shannon et al., and Mahajan-Shiferaw et al. We recognize that there is no straightforward manner to identify new models are they are published, and sometimes it takes time for them to become widely known. We hope that this Scholarpedia article in the future will assist in this effort, as authors can provide information about their models here. If the reviewer has any specific suggestions for additional models to include, we would be pleased to hear about them.
We have now clarified our intentions at the beginning of the article in terms of discussing modifications of models to represent the effects of drugs and diseases. Unless different parameter values or equations were included in the original article, we have chosen not to include minor modifications to channel conductance values and other small changes. As we discussed above, it is already challenging to keep up with the new models being published, let alone their modifications. Perhaps this may be a possible way to expand the article in the future, if authors are willing to provide information about the ways they use existing models.
Reviewer A: We have modified the beginning of the article to explain that the article does not necessarily contain every published ionic model, and that we concentrate on the original models and variations included in the original articles only, rather than subsequent modifications made to represent the effects of different diseases or drugs. We have also added a new section, “Future directions,” in which we indicate that models that behave well in single cells may not behave well in tissue.
Although originally all models were forced to draw from existing voltage clamp data, which may not have been available from the species or region of heart of interest, over time voltage clamp data has become available from an increasing number of species and heart regions. For example, the Iyer et al. human ventricular model uses cloned human channel data as a basis for their current descriptions. The mixing of ion channel data from different species is a well-known problem in model development, but we believe that classifying the origin of the data underlying each ion channel of every model is beyond the scope of this article and may not always be possible based on the data provided in the model publications. We have included a short paragraph in the Ventricular models section that mentions this problem.