Straight to the Heart
- Here I depart from strictly physical principles to editorialize on the application of concepts from the foregoing articles to clinical issues. This is a web page. You must consult the scientific literature devoted to these topics to evaluate and treat your patients. I hope these musings will enhance the framework for your interpretations; as always I don't have anything to impart here that hasn't been said better elsewhere and with factual substantiation. I'm not really worried about any physicians reading this, but there is a comparative medicine angle included that may be of interest.
- Technical issues / differences between methods
- Derivation of indices such the Gorlin area or stenosis "resistance" entails a reduction of data that may be ill-advised for clinical purposes. Consider that each of these 2 indices incorporate the exact same 2 measurements which are then distilled into a single value; only the derived index is different. Instead of computing an index, what if we simply plot data points of pressure gradient against the ejection flow rate for different patient groups. If the groups were separable using only these 2 measurements, then this would be apparent visually from the plot. For clinical purposes the argument for Gorlin versus resistance reduces to whether it's a straight line versus a parabola through the origin that best separates the patient groups (and so it is difficult to demonstrate clinical superiority of one over the other). If the patient groups occupy separable spaces on the plot (they don't), then we're done. - just draw a circle around the 2 (or more) data groupings and use the plot to see where your next patient lands! It's a logistic regression on raw or transformed data to determine the best statistically supportable separation criteria. We've thrown something away by reducing 2 dimensional data to 1 dimension. (NOTE: An appropriate logistic regression yields the best straight line, not necessarily through the origin, that separates the groups in this hypothetical 2D plot. Data transformation has the potential to help us find the best (nonlinear) curve which might be suggested by the distribution of plotted points.)
- While gradient and flow are continuous variables, the plot just described can be considered as dividing this 2 variable space into 4 quadrants: low versus high flow and low versus high gradient. This approach can be extended to encompass other methods, either in use or as yet undeveloped. To grade a stenosis we typically employ 2 measured quantities, a "flow variable" (e.g. thermodilution ejection flow rate, subvalvular LV VTI, pulmonic VTI, etc.) and a "severity variable" (e.g. various measures of pressure gradient, stenosis peak velocity or VTI). This principle depends of course on the existence of a definitive method ("gold" standard) for determination of the patient groups (or possibly a preconception of the result).
- A color M-mode of the outflow tract contains information about the functional (axial) configuration of the stenosis, limited by aliasing restrictions.
- Valvular AS in people is often seen in the elderly and may be accompanied by coronary disease and significant myocardial dysfunction. This causes difficulty in the quantification of the stenosis, potentially placing these individuals in the low-flow/low-gradient quadrant. ANY hemodynamically derived stenosis index may exhibit significant flow dependence in this regime. It is my view that clinical stenosis severity cannot be considered apart from the ventricle producing the flow. A simple cutoff value for "critical stenosis" (e.g. 0.5 cm2/m2) neglects the afterload-dependence of the ventricle which becomes increasingly significant with myocardial disease. "Critical stenosis" isn't a number, it's a function that depends on at least one other variable pertaining to ventricular function. The use of dobutamine in these patients as a diagnostic adjunct helps to identify those with myocardial reserve but has the potential to reveal flow dependence that may be both hemodynamic AND anatomic, i.e. increasing flow and/or distending pressure may yield a greater physical cross sectional area. In conjunction with the above, this thought segues into the concept of a multidimensional (multivariate) plot to separate the patient groups with one or more of them pertaining to the myocardium. There are, of course, published studies that have employed multivariate approaches. Newer imaging modalities (TEE, MRI, 3D echo) likely will improve anatomic quantification of the condition, particularly valuable where stratification by hemodynamic means is problematic.
- In veterinary medicine the principles described previously often are applied to the problem of detecting subaortic stenosis (SAS) in an attempt to permit exclusion of affected individuals from the breeding population. SAS is known to be a genetically transmitted, pseudo-dominant condition with variable penetrance (in studied breeds). The phenotype (if manifested to a detectable degree) appears at a variable time interval after birth and may result in lesions so mild as to be apparent only on post mortem exam (or some imaging modality as yet untested). Currently there is no genetic test for the condition (other than breeding/pedigree studies) and no definitive (gold) standard to define the presence of the disease. The current consensus standard consists of ranges for peak velocity values (\(V_p\)), determined from Doppler echocardiography, thought to roughly define normal, equivocal, and affected groups. While we know from the physics that \(V_p\) necessarily is flow dependent, phenotype identification problems arise with the (potentially) mildly affected individuals where ANY hemodynamic stenosis index may exhibit flow dependence. For more severely affected dogs, the problem of accurate prognosis is hampered by catastrophic impact of unpredictable events such as CHF following the development of mitral regurgitation or sudden death attendant to the exuberant pursuit of a squirrel.
- It's my opinion that clinical quantification of a stenosis based on either hemodynamics or anatomy alone will fall short of expectation (poor prediction of outcome) due to the lack of inclusion of the biological factors involved. These clearly include the severity of myocardial dysfunction and coronary artery disease (in people), but also many unseen factors may be imposed to a degree that depends on the biology (hypertrophy, ischemia, fibrosis, loss of compliance, autonomic imbalance, predisposition to arrhythmia, neurohumoral factors, and others that I don't care to list or don't know). This is a statistical problem, given the current level of understanding, and not a physical one alone. Flow-independence is an aspect of the problem to be aware of, not an end goal. Understanding the physics sets you free (to find the answers).