| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Cardiac Imaging |
1 Department of Radiology and Radiological Sciences, Vanderbilt University Medical Center, MCN R-1322, 21st Ave S, Nashville, TN 37232-2675 (M.A.B.)
2 Department of Radiological Sciences, University of California–Irvine Medical Center, Orange (E.N.C.M.)
3 Department of Radiology, University of Iowa College of Medicine, Iowa City (W.S.)
4 Department of Radiology, Memorial Medical Center of Long Beach, Calif (C.W.S.).
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
MATERIALS AND METHODS: Nine readers graded a set of 25 chest radiographs (15 cases of aortic valve disease [AVD], 10 control cases without AVD) for blood volume and ventricular size. Blood volume was graded on a scale of -3 (severe hyperemia) to 0 (normovolemia) to +3 (severe oligemia). Ventricular size was graded on a scale of 0 (normal) to 3 (massively enlarged). The oligemia and ventricular size grades were added to yield the radiographic severity index. Pulmonary capillary wedge pressure, pulmonary arterial pressure, stroke volume, and cardiac output were measured at the time of catheterization.
RESULTS: The five more experienced readers achieved good nonchance agreement (
= 0.48; P < .001). They were unanimous in scoring 12 cases as oligemic; variations occurred only in severity assessments. Oligemia was due to emphysema in one case and to AVD in 11. In oligemic cases, radiographic severity correlated significantly with wedge pressure (r = 0.93, P < .001) and pulmonary arterial pressure (r = 0.93, P < .002).
CONCLUSION: Many cases of AVD show oligemia. The severity of oligemia correlates well with hemodynamic abnormality. Oligemia may be caused by atrial-pulmonary-vascular reflex vasoconstriction, low right ventricular output, and possibly high levels of atrial natriuretic factor.
Index terms: Aorta, flow dynamics, 535.91 • Aortic valve, 535.83, 535.84 • Lung, perfusion, 60.91 • Pulmonary arteries, flow dynamics, 564.91
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
One of the authors (E.N.C.M.) has observed that many cases of aortic valve disease (AVD), including aortic stenosis (AS), aortic incompetence (AI), and mixed AS and AI, also show oligemia (reduced pulmonary BV) where one might have expected congestion. In AVD, however, this oligemia is generalized, not confined to the bases (Fig 1) (7). We have not been able to find any previous reports in the radiology literature of oligemia associated with AVD; however, it has been conclusively documented in both pathology studies and physiologic studies (8–11) that AVD is frequently accompanied by a reduction in pulmonary BV. This work will be discussed more fully later.
|
|
|
|
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
We received 25 posteroanterior chest radiographs that represented two populations of patients. The cases consisted of (a) eight cases of AS and seven of mixed AS and AI (nine men and six women aged 23–57 years [mean age, 39 years]) and (b) 10 age- and sex-matched control cases without AVD that included seven healthy subjects and three patients with chronic obstructive pulmonary disease (eight men and two women aged 20–67 years [mean age, 43 years]).
The radiographs were randomized and labeled only with a code number. Nine readers from several hospitals were asked to assess each case for pulmonary oligemia (defined as less than normal pulmonary BV), normovolemia (normal pulmonary BV), or hyperemia (increased pulmonary BV). (Assessment of pulmonary BV is covered further in reference 7.) They were also asked to grade oligemia by assigning a score of 3 for severe oligemia, 2 for moderate oligemia, and 1 for minor oligemia. Normovolemia was assigned a score of 0. Hyperemia was graded with a score of -1 for minor hyperemia, -2 for moderate hyperemia, and -3 for severe hyperemia (Fig 3). (No AVD cases were classified as hyperemic; therefore, in the results there are only four grades of vascularity.) In addition, the readers were asked to grade right and left ventricular size: Normal size was assigned a grade of 0, mild enlargement, a grade of 1; moderate enlargement, a grade of 2; and massive enlargement, a grade of 3. Because the accuracy of separate assessment of right and left ventricle size on a frontal view alone is in some doubt, particularly in cases of AS, where concentric left ventricular hypertrophy can elevate the cardiac apex and mimic right ventricular enlargement, these assessments were added together to yield a single cardiac size index (Table).
|
|
The image readers spanned a wide range of levels of training and experience and included three chest radiologists, two experienced general radiologists in community hospital practice, a radiology fellow, and three residents in radiology (one at the 3rd-year postgraduate level, two at the 5th-year postgraduate level). Each reader in turn received the study images (including those of control cases), example radiographs, and the reading table. The analyses were carried out in the readers' own offices, and the readers used their normal reading conditions. No clinical or hemodynamic information was supplied, and the readers were led to believe they were simply evaluating the reader's ability to quantify pulmonary BV on the basis of the plain radiograph. They were unaware that the results of this study (if successful) were also to be used to test our observation that many patients with AVD have pulmonary oligemia that can be recognized on chest radiographs.
When all nine readers had returned their results, the code numbers were matched to the patient information. Interobserver reliability for the detection and quantification of oligemia was determined by using a analysis (16–21) (see Appendix). Complete agreement on detection and grading of severity was required. The readers' grading of pulmonary vascularity was then added to their grades for cardiac enlargement to provide an index of overall severity. For example, a case with severe oligemia (grade 3) and a cardiac index value of 4 would have an overall severity index of 7. The overall severity grades were then correlated with the catheterization and clinical data.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
value was 0.15 (standard error, 0.09), which demonstrated a significant (P < .01) degree of nonchance agreement, with a probability of less than 10% that the agreement observed could have occurred by chance (see the Discussion).
When the analysis was limited to the responses from the five more experienced readers (with the results from the fellow and the three radiology residents omitted), the value was 0.48 (standard error, 0.08), which indicated a higher level of nonchance agreement. This result was significantly different from chance agreement (P < .001) and could be classified as good agreement (on a scale from slight to perfect agreement).
Radiographic-Hemodynamic Correlation
Only the data from the five experienced radiologists were used for the radiographic-hemodynamic correlations. Of the 25 cases analyzed, 12 were classified as oligemic and 13 as normovolemic. In one of the 12 oligemic cases, there was no history of cardiac disease, but classic emphysema was present. AVD was present in the remaining 11 cases (eight cases of AS, three of mixed AS and AI). A less expected finding was a second population of four proved cases of AVD that also were normovolemic. Three of these four were cases of mixed AS and AI, and one was a case of AS.
When all 15 cases of AVD were rank ordered in terms of overall radiographic severity, it appeared that the cases fell into two separate groups both radiographically and clinically (Table). Group 1 consisted of four normovolemic cases in which the mean pulmonary capillary wedge pressure (± standard error) was 14.3 mm Hg ± 2.5 (n = 4), the mean pulmonary arterial pressure was 25.2 mm Hg ± 3.9 (n = 4), the mean stroke volume was 88.0 mL ± 9.5 (n = 3), and the mean cardiac output was 6.0 L/min ± 0.3 (n = 4) (Table). Note that there are only four cases in this normovolemic group—too few to derive statistical correlations. However, it is evident from the catheterization data that, in our series, the normovolemic cases exhibited little hemodynamic abnormality (Figs 4, 5). This finding has been confirmed with results from much larger studies (2,11) in which nonradiographic techniques were used.
|
|
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
value, we chose to use conservative formulas, which would be expected to overestimate interobserver error (see the Appendix). Our results with all nine observers showed only slight nonchance agreement, but when the effect of experience was assessed by omitting the results from the four less experienced readers from the panel, the test of significance became overwhelmingly positive (P < .001). There was no interobserver disagreement between the more experienced observers about the presence of oligemia—only about the severity grades. It should be pointed out that the value of 0.15 was for the entire group (inexperienced and experienced readers together), and it is highly likely that if we removed the experienced observers' results, the value would decrease to a nonsignificant level, which would indicate that there was no agreement in the less experienced group of observers with regard to the presence of oligemia. Our data strongly suggest, however, that the ability to recognize oligemia can be acquired with experience and/or instruction, particularly if the image reader becomes aware that oligemia does occur in patients with AVD (21). Because we did not perform any radioisotope measurements of pulmonary BV, the question might be raised: What evidence is there that the "appearance" of oligemia was not caused by some other pathologic condition or by an artifact? However, if one wishes to argue that the radiographic appearance of oligemia may not indicate a reduction in pulmonary BV, one must offer viable alternatives that cause the same radiographic appearance, and we submit that there are no such alternatives. It has been argued that an increase in the quantity of air in the lungs without a decrease the pulmonary BV will increase the air-to-blood ratio and cause the lung to appear oligemic, but it is difficult to think of a mechanism by means of which this could occur. For example, deep inspiration (without a Valsalva maneuver) will certainly increase the air in the lungs, but it also increases the pulmonary BV (7,22). Deep inspiration with a Valsalva maneuver causes a reduction in pulmonary BV, as may positive pressure ventilation and other forms of air trapping, whether functional or organic (4,7). The most common cause of an artifactual appearance of generalized oligemia is overpenetration of the film, which is easily corrected by means of bright lighting. We would submit, therefore, that there should be no artifact simulating oligemia that a trained reader would not recognize.
None of our cases of AVD had a hyperemic appearance, and radioisotope measurements of pulmonary BV in patients with AVD confirm that hyperemia (commonly called congestion) does not frequently occur in AVD, even in patients with congestive heart failure (10,11,23,24). Our findings of two groups of cases—a smaller, normovolemic group with nearly normal hemodynamics and an oligemic group with abnormal hemodynamics—have been duplicated by other investigators (8–11) who were operating independently and who used absolute measurements of pulmonary BV. It is statistically nearly impossible that this could have occurred by chance, which emphasizes the finding that our readers were accurate in their detection and grading of oligemia and normovolemia. We shall now discuss the nonradiographic evidence for diminution of pulmonary BV in AVD.
Physiologic Evidence
Balbarini et al (10) examined 34 patients: 20 with AI, 10 with mixed AS and AI, and four with AS. The mean pulmonary BV in these patients was 299 mL/m2 (normal pulmonary BV, 300–600 mL/m2). This mean value, however, was derived from two separable groups of patients. The first group had a mean pulmonary BV of 329 mL/m2, which is normovolemic, and only slightly elevated pulmonary arterial pressure with a mean pulmonary vascular resistance of less than 2.5 mm Hg/L/min/m2. In the second group, the mean pulmonary BV was 270 mL/m2, which is oligemic; the pulmonary arterial pressure was elevated, and pulmonary vascular resistance was greater than 2.5 mm Hg/L/min/m2.
The results of the study by Balbarini et al (10) confirm our observation that, in the oligemic subgroup, the left ventricle functioned less efficiently than normal. The mean stroke volume in the patients (both groups) in the study by Balbarini et al was 67 mL as compared with a normal volume of 90.9 mL. This correlates well with our own catheterization data (Table, Fig 4), which showed a marked reduction in stroke volume in the oligemic group (range, 31–55 mL) as compared with a range of 86–90 mL in the normovolemic and hypervolemic groups, (mean of both groups, 64 mL). By design, our 15 cases consisted of patients with AS and patients with mixed AS and AI, whereas Balbarini et al, in addition to four cases of AS and 10 of mixed AS and AI, studied 20 cases of AI alone. Their findings indicate that oligemia in AVD is not confined to patients with AS or mixed AS and AI but also occurs in pure AI. A reduction in pulmonary BV (oligemia) and low stroke volume in a subgroup of cases of AVD (including predominant AS and predominant AI) has also been described by Yu (11).
Pathologic Evidence
In 1960, James et al (8) were trying to determine what changes in the pulmonary vascular bed might lead to right ventricular hypertrophy in certain cases of cardiac and respiratory disease. By serendipity, the patients they examined included four young patients with AVD (three with AS and one with mixed AS and AI) who had died in accidents. The results of James et al are best delivered in their own words:
"[T]he first surprise in this work came from the results in the four cases of aortic valve disease. . . . [T]he figures for these cases indicated a considerable reduction in the peripheral arterial bed. . . . [O]ur findings suggest very strongly that there is a reduction in the small arterial bed of the lung in aortic stenosis even before the advent of congestive cardiac failure, and this is supported by the histologic findings."
Their histologic analysis demonstrated considerable hyperplasia of small (50–200-µm) arteries (8). These results were confirmed independently by Pirincci et al (9) in 1961, who demonstrated severe reactive pulmonary hypertension secondary to AS.
There is, therefore, a strong body of evidence—physiologic, pathologic, and now radiographic—that pulmonary BV is reduced in a proportion of patients with AVD (oligemia was present in 73% of cases in our series and in 71% of cases in the series of Balbarini et al). With this result established, it becomes pertinent to try to determine why oligemia should be present, why it should occur in one group and not in another, and what radiographic value these changes may have. That is, can we determine from their presence the hemodynamic status of the patient?
Possible Causes of Oligemia
Organic changes in the microvasculature.—James et al (8) suggested that the reduction in pulmonary BV in AVD is caused by endarteritis in the small pulmonary vessels. If the reduction in pulmonary BV were caused simply by organic changes in the small pulmonary vessels, however, one would anticipate finding "pruning" and tortuosity of the segmental vessels, with enlargement of the central pulmonary arteries and considerable enlargement of the right ventricle. We and others (25) have not observed any such changes in the pulmonary vessels, and no significant correlation between oligemia and right ventricular size has been found. In fact, severe pulmonary arterial hypertension can be present in AVD with little evidence of its presence on the chest radiograph. For these reasons, it seems that organic changes in the microvasculature are not the cause of oligemia but occur secondary to chronically elevated left atrial pressure, as in mitral valve stenosis (as will be discussed later in this article) (5).
Neuronal effects.—Long-standing elevation of left atrial pressure causes narrowing of the basal lung vessels (arterial and venous) of all sizes (ie, not confined to the microcirculation) (6,7,13). We have previously discussed the reasons why this vascular narrowing and basal oligemia cannot be caused by pulmonary edema (7), as has frequently been suggested (26,27). We believe it is caused by an atrial-pulmonary-vascular reflex (7). There are certainly pressure receptors in the left side of the heart (26), but why should elevated pressure in the left atrium cause pulmonary vasoconstriction?
Our hypothesis to explain this is as follows: Because there are no valves or sphincters in the pulmonary veins, free reflux occurs normally from the left atrium into the pulmonary veins during each atrial systole (atriovenous reflux; Fig 5) (7,27–30). As left atrial pressure rises due to outflow resistance, this atriovenous reflux increases progressively, which reduces left atrial outflow into the left ventricle (Fig 5) (7,28). To prevent this reflux and maintain left atrial output, the elevated left atrial pressure initially causes increased tonus of the left atrial wall, so that the atrium does not enlarge, even when left atrial pressure is high, and can continue to pump blood into the left ventricle (7). This fits well with the absence of atrial enlargement and the lack of flow inversion in acute failure of the left side of the heart, even when left atrial pressure is very high. Eventually the left atrial strength diminishes, and a different mechanism must take over to protect left atrial function. The left atrium enlarges, and reflex venospasm narrows the lower lobe vessels, which decreases their capacitance and markedly increases their resistance to flow, thereby reducing atriovenous reflux and improving flow into the left ventricle. With constriction of only the lower lobe vessels, there is little mechanical increase in pulmonary vascular resistance, whereas if the upper lobe vessels were also to narrow, pulmonary vascular resistance would greatly increase, which would cause pulmonary arterial hypertension and right ventricular failure (7). This hypothesis also explains why flow inversion is never seen in the pulmonary edema of renal failure, overhydration, or low oncotic pressure, even when edema is severe.
Wedge pressure also is chronically elevated in AVD (1), and the severity of oligemia was correlated with wedge pressure (Fig 4), which strongly suggests that the same type of relationship between the elevation of left atrial pressure and vasospasm found in mitral valve stenosis and chronic failure of the left side of the heart also exists in AVD. If the mechanism is the same as that in mitral valve disease, however, why should the upper lobe vessels also be narrowed in AS? We believe that in AVD there may be additional mechanisms that are different, either in kind or in strength, from the mechanisms in mitral valve disease. We have previously observed (4,7) that, in combined AVD and mitral valve disease, flow inversion is usually absent, and generalized pulmonary oligemia is usually present; that is, the factors caused by the AVD appear to predominate.
Circulating atrial natriuretic factor.—In AVD, the obstruction to outflow from the left side of the heart is further "downstream" than in mitral valve stenosis or failure of the left side of the heart, resulting in left ventricular hypertrophy. It has been shown (31–33) that the level of circulating atrial natriuretic factor is markedly increased in patients with ventricular hypertrophy (mean ± standard error, 120 pg/mL ± 59 [120 ng/L ± 59]) as compared with the level in patients with a normal heart (20 ng/L ± 12). A strong relationship has been reported (34) between the degree of ventricular hypertrophy (right, left, and septal) and the level of circulating atrial natriuretic factor. One of the effects of increasing atrial natriuretic factor appears to be a reduction in circulating blood volume (29,30). Balbarini et al (10) have shown in AVD patients that as pulmonary BV decreases, total BV decreases, which would explain why the vascular pedicle is usually somewhat narrow in patients with AVD (Fig 1) (35–37).
Why should some cases of AVD not demonstrate oligemia? It is possible that, in these cases, ventricular hypertrophy may not yet be sufficiently advanced to produce a high level of atrial natriuretic factor. In support of this possibility, we should point out that those cases in which there is little reduction in cardiac output or stroke volume do not show oligemia (25).
Right ventricular output.—Right ventricular output and pulmonary BV are usually closely related, and part of our hypothesis is that low right ventricular output secondary to the low left ventricular output (vis a tergo [effect of a force acting from behind]) (30) contributes to the oligemia in AS. We (7,13) and others (25) have found an excellent correlation between pulmonary BV and stroke volume that fits this hypothesis.
Pulmonary Arterial Hypertension
In a small group of AVD cases, the mean pulmonary arterial pressure is too high to be caused simply by a passive increase secondary to elevated left atrial pressure. This finding confirms that there must be altered compliance (confirmed by Balbarini et al [10]) due to vasospasm, endarteritis, or both (8,9,37).
Clinical Relevance of Radiographic Findings
It is important to reiterate that although some degree of pulmonary venous and arterial hypertension is always present in cases of AVD, this may not produce any of the radiographic manifestations usually seen in pulmonary venous and arterial hypertension. Therefore, once we have established the diagnosis of AVD, criteria different from those for ischemic failure of the left side of the heart or mitral valve disease must be used to assess the severity of the disease. By using the correlations discussed earlier between radiographic severity, wedge pressure, pulmonary BV, and stroke volume, we may be able to perform clinically useful assessments of the hemodynamic severity of the disease. For example, if a patient with AVD has oligemic lungs and cardiac enlargement, he or she is likely to have a markedly diminished stroke volume and cardiac output, with moderate to severe pulmonary venous and arterial hypertension (Fig 6b).
|
|
Generalized pulmonary oligemia in AVD should now be added to the known spectrum of pulmonary vascular changes initiated by cardiac dysfunction, including the "bat's wing" edema of acute left atrial pressure elevation, the basal oligemia (flow inversion) of chronic left atrial pressure elevation, and the decreased compliance of the pulmonary vasculature that occurs in failure of the left side of the heart (5,7,22).
|
APPENDIX |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
analysis is a statistical method to help determine interobserver reliability by estimating the amount of chance-corrected nominal scale agreement between readers (16–20). The method has been developed particularly for treatment of problems that concern the reliability or reproducibility of diagnostic categorization and is quite stringent in that the proportion of agreement that can be attributed to chance is determined and subtracted out. Thus, a negative value is rendered when the level of agreement is less than that expected by chance; zero, when the observed agreement can be exactly accounted for by chance; and unity, when there is complete agreement (18). Compelling arguments have been proposed (16–20) that is the best of such measures.
There are many established methods for the calculation of the coefficient. We used the most stringent form of , as described originally by Fleiss (19), without recourse to formulas that allow "partial credit" for partial agreement (16) or to attempts to increase the value of by adjusting for observer-bias "base rates" (17). Although statistically valid, such adjustments to the outcome of would tend to make the results seem to be more significant (17,18,21). We preferred to use the older, more stringent formulas, which do not adjust for "negative bias" and yield a more conservative test of significance.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Abbreviations: AI = aortic incompetence AS = aortic stenosis AVD = aortic valve disease BV = blood volume
Author contributions: Guarantor of integrity of entire study, E.N.C.M.; study concepts and design, E.N.C.M., M.A.B.; definition of intellectual content, E.N.C.M.; literature research, E.N.C.M., M.A.B.; clinical studies, W.S., C.W.S.; data acquisition and analysis, E.N.C.M., M.A.B.; statistical analysis, E.N.C.M., M.A.B.; manuscript preparation, editing, and review E.N.C.M.
Received December 8, 1997;
revision requested March 11, 1998; revision received June 12, 1998;
accepted August 24, 1998.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
) in which the mean stroke volume was 88 mL, and the mean cardiac output was 6.0 L/min. Group 2 (radiographic severity grade, 2–8) consisted of oligemic cases (•) in which the mean stroke volume was 40 mL, and the mean cardiac output was 4.6 L/min. The vertical line indicates the junction between groups. (a) Correlation between radiographic severity and wedge pressure (in millimeters of mercury). For group 1 (n = 4), there was no correlation between wedge pressure and radiographic severity. Significant correlations were found for groups 1 and 2 together (r = 0.81, P = .002; n = 12, with cases 10, 16, and 24 excluded, because no wedge pressure was recorded) and for group 2 alone (r = 0.93, P < .001; n = 8). (b) Correlation between radiographic severity and pulmonary arterial (P.A.) pressure (in millimeters of mercury). For group 1 (n = 4), there was no correlation between pulmonary arterial pressure and radiographic severity. Significant correlations were found for groups 1 and 2 together (r = 0.64, P = .044; n = 10, with cases 10, 16, 17, 20, and 24 excluded, because no pulmonary arterial pressure was recorded) and for group 2 alone (r = 0.93, P = .002; n = 6).
