CONTROVERSY WITH CIRCULATION

To begin with, reviewer #1 wrote: "... the authors do not appear to have understood the centerline method." This affirmation could be true, but he does not explain how our suggested inability would have influenced an angiographic system made by General Electric and operated by radiologists to produce insignificant diagnoses. He accuses us of having made unjustified changes, but our angiographic system carried certified centerline software utilized by technicians: how could we have changed the software of a complex X-ray system, used by others?

 

Reviewer' sentence: " The authors fail to justify why they are reverting to a method of classifying regional wall motion developed in the 1970's" is not only inaccurate (because CASS re-proposed this classification in the 1980's [Circulation, Vol 63, Suppl I, June 1981]), but it is also ridiculous, considering the centerline method refers to pressure/volume diagrams: an antiquated, hundred years old,  misinterpretation of the cardiac cycle! That some cardiologists have little knowledge of physics is not surprising, but to insist on an error once discovered is staggering.

 

The Reviewer finds fault with our positioning of the proximal end of the longitudinal axis, stating that "this axis is constructed from some arbitrary point located an unknown distance below the junction of the aortic and mitral valve planes." This one, among all objections, convinced us of the futility of discussing to get our paper published in Circulation, because it suggests how long way from problems about the longitudinal axis the reviewer is. However, here I provide explanations which, in my opinion, should be evident to any specialist performing and reading left ventriculography,  but evidently not so clear to specialists in reading reviews.

 

The starting point is that angiography projects a mass of radio-opaque blood changing volume onto a film plane (or screen), transforming it into a sequence of images. This process results in loss of a dimension and creates a movie representing the simplification of a complex three-dimensional phenomenon taking place over time. Angiographers draw diagnostic conclusions on ventricular wall motion running film back and forth. This modality is "subjective"  depending on single-reader interpretation (variable among different readers) and is verbal (difficult to compare with). Since the 1970's, the availability of computers has suggested the possibility of setting up "objective" interpretive systems founded on mathematical calculations. All proposed methods use only two silhouettes: the largest one (end-diastolic - ED) and the smallest one (end-systolic -ES), and all renounce the observation of movement. It should be clear that these two methods differ as they analyze different phenomena; only the reproducibility of diagnoses ensures the correctness of a computer-assisted process, so far subjective reading has been the only reference system available. Because none of the available techniques proposed until the mid-1980's was able to produce diagnoses comparable whit the accepted albeit subjective norms, the suspicion arose that some error in conception end/or calculation had occurred.

The longitudinal axis was the element most suspected for two reasons:

 

1 ) while in ED angiographic apex coincides with the anatomical apex of the ventricle, in ES this is no longer valid. As we know, after studies with metal markers implanted in the myocardium, it was shown that the cardiac apex remains stationary throughout the cycle, while the angiographic top shortens.

 

2 ) during systole the heart twists due to the arrangement of some muscular layers: this means that points of the wall defining the outline of a particular portion of the silhouette in diastole, are no longer in the same position during systole. Moreover, the angiographic apex elevates in systole.

 

Therefore, we do not construct longitudinal axes of the ventricle in diastole and systole, but the axes of two shadows of contrast medium. If the walls obliterate part of the apical shell in systole, of this portion of the ventricle, we will have no impressions, becoming invisible.

 

The last point to be discussed regards the term "axis." In my mind, this explanation seems more suitable for middle school students rather than graduates in medicine. In any case, from a technical point of view, the axis of a cylinder-conical volume (which represents the content of the left ventricle), is a segment starting from the distal apex of the cone, around which it is possible to rotate the cylinder uniformly. If the proximal face of the cylinder is smooth (that is a circle), then axis will pass through its center, but if the proximal face is uneven (the whole of aorta and mitral valve planes do not have a circular shape, and their morphology is anything but constant), then the axis proximal end will reach the proximal face casually, being important the equidistance between walls and central axis.

ED and ES silhouettes are plane projections (prolate ellipsoids) of two cylinder-conical volumes. The axes of these figures must meet the same criteria set for the masses from which they derive. The distal ends of these axes are always the apexes of these figures, and their purpose is still to divide the silhouettes into two parts as symmetrical as possible. Where the proximal ends of the two longitudinal axes reach the aortic-mitral border does not matter, because neither the aorta nor the mitral border has relevance in the wall motion assessment.

Although these statements can be considered correct even only on a rational basis, the various possibilities for tracking of the longitudinal axes reported in the literature (and never justified), were subjected to algorithmic tests that showed the correctness of the foregoing, i.e., the task of the longitudinal axes is split into two parts ED and ES silhouettes as symmetrical as possible, is totally irrelevant to the placement of their proximal ends.

 

The superiority of Klinger method over all other methods (centerline and endo-myocardial landmarks mainly) does not lie in using areas instead of cords, but having considered that the heart performs a complex motion, in agreement with and taking advantage of dynamics laws. This result was accomplished using algorithmic tests, a search method apparently not known by Circulation's reviewers.

Anyone using the centerline method may take as many chords he wants, and be looking for hundreds of standard cases, but he will never get the correct diagnosis about wall motion using a method set up to evaluate arterial stenoses and senselessly adapted to a much more complicated function.

 

As far as a number of normal population is concerned, more once I have explained that after 12 cases, the SD of standard curve stopped decreasing, making it useless to look for more examples. Moreover, every single Center can construct his standard curve. The problem if ever, would be comparing average curves of different Centers, but this concerns the concept of normality, not the method of reading. Numerical values of SD of my normal population have been published [Cathet Cardiovasc Diagn: 18:50-59, 1989], ranging from 0% (apical sectors 10, 11) to 15% (last area of the posterior-basal sector), and show a mean value of 6,8%. These low values allow Klinger method a refined sensibility for hypokinesis. On the contrary, I never read numerical values of SD in papers about centerline method. These values are high looking through published curves, particularly at the apex. High amounts of SD of any curve means substantial overlapping between standard variants and pathological modifications, reducing its diagnostic ability, exactly is the problem of centerline method.

 

Last but not least, we have compared the diagnostic ability of two different computer-assisted methods over the same patient population: for each process, we used criteria set by their respective authors. It is my opinion that our statistical calculations are correct (even because a professor of statistics has established them). So I think that reviewer's knowledge of statistics does not exceed his understanding of physics.

 

Following information furnished by reviewer #1, I searched the website weber.u.washington.edu/-sheehan, to discover it no longer exists. Also the word "centerline" doesn't address to any site about wall motion analysis using chords system. Probably all was erased. On the contrary, I will publish on my website (certainly not in Circulation), my contraction curves, furnishing all information to allow any programmer to construct software able to elaborate diagnoses about wall motion analysis based on calculations, so objectively comparable. The word "semi-automated" refers to the manual drawing of silhouettes edges because up to now, and there is no software with knowledge and interpreting ability our cerebral cortex has.

 

To conclude with, the guide-ideas that brought me to elaborate the Klinger method aren't fictional, but have been selected using algorithmic tests during three years work (1985 to 1987), and are mainly two:

 

1) Cardiac activity can be correctly described only using laws of dynamic physics (1° motion continuity; 2° force = mass*acceleration; 3° action-reaction).

 

2) Cardiac walls become in contact at the apex during systole, obliterating part of the apical shell.

 

Quite similar ideas have been proposed at that time by different Authors in different contexts: 1) Robinson T.F., Factor S.M., and Sonnenblick E.H.: The Heart as a Suction Pump. Scientific American 6: 62-69, 1986.

 

2) Nogueira E.A., Carvalhal S.S., MacMillan R.M., and Maranhao V.: Analysis of the Position of the Left Ventricular Apex and Base During Systole. Cathet Cardiovasc Diagn 13:253-261, 1987.

 

All the remainder (areas rather than cords or segments, the proximal end of longitudinal axis positioning, number of SD to be selected as discriminating value, and so on), seems secondary and subject to the mere expediency of choices.

 

I agree with reviewer #1 the conclusion of our paper was (cautiously) "the Klinger method as an improvement in the semi-automated analysis of wall motion." However the judgment should have been: the volume/pressure diagram is an erroneous interpretation of the cardiac cycle, and all methods of analysis based on this idea of contraction are conceptually wrong, and in spite of complex calculations, they are unable producing clinically useful diagnoses. On the contrary, Klinger method produces numbers automatically translated into terminology used by angiographers, formulating same diagnoses as subjective ones. In conclusion, Klinger method represents, for now, the unique possibility of analysis of wall motion by calculation.

 

I have long been uncertain whether to express an opinion on the reviewer #1 who wrote the assessment at the beginning of this article. Indecision was about choice: reviewer #1 was incompetent or a person in bad faith? I have concluded that both qualifications are right.