Pulmonary Vascular Steal in Chronic Thromboembolic Pulmonary Hypertension*

Transcription

1 Pulmonary Vascular Steal in Chronic Thromboembolic Pulmonary Hypertension* Mitchell A. Olman, M.D.; William R. Auger, M.D., F.C.C.P.;t Peter F. Fedullo, M.D., F.C.C.P.; and Kenneth M. Moser, M.D., F.C.C.P. After pulmonary thromboendarterectomy, performed for relief of chronic thromboembolic pulmonary hypertension, perfusion lung scans have frequently disclosed new perfusion defects in segments served by undissected pulmonary arteries. Our hypotheses were that these new postoperative defects occurred with great frequency and did not represent postoperative vessel occlusion. We retrospectively reviewed the preoperative and postoperative perfusion scans of 33 consecutive patients undergoing pulmonary thromboendarterectomy. New postoperative perfusion defects were noted in 23 of 33 patients. The incidence of new defects was increased tenfold in segments that had (1) normal preoperative angiographic Gndings, (2) normal preoperative radionuclide perfusion, and (3) not been entered at the time of surgery. Postoperative angiograms, available in 15 of 33 patients, documented the nonembolic, nonocclusive nature of the new perfusion scan defects. The most plausible alternate explanation for this previously undescribed Goding is a redistribution of pulmonary arterial resistance induced by the thromboendarterectomy, namely, a pulmonary vascular "steal." (Chest 1990; 98: ) everal groups have now shown that chronic, large S vessel thromboembolic pulmonary hypertension is potentially correctable by surgical thromboendarterectomy. l-4 However, the postoperative course of these patients is complex, presenting management problems, including reperfusion edema, 5 persistent hypoxemia, pericardial effusion, and psychiatric disturbances.6 One unusual postoperative observation, seen in a significant number of our patients, has been the appearance of new perfusion defects in the postoperative lung scan. 7 To our knowledge, this phenomenon has not been reported previously. These new defects appeared to occur, most commonly, in areas of the lung served by segmental and lobar arteries that had not been entered at surgical thromboendarterectomy. The analysis that follows was undertaken to determine the incidence of this phenomenon and its potential basis. The findings suggest that these new perfusion scan defects are due to postoperative redistribution of regional pulmonary vascular resistance, a phenomenon we have labeled pulmonary blood flow steal. underwent pulmonary thromboendarterectomy at the University of California at San Diego Medical Center (UCSD) for chronic thromboembolic pulmonary hypertension between 1985 and 1988 and who met the following criteria were reviewed: {1) a standard six-view {anterior, posterior, right posterior oblique, left posterior oblique, right lateral, and left lateral) lung perfusion scan (3 mci of technetium-labeled MAA, 500,000 counts per image) and a xenon ventilation scan had been obtained preoperatively and before hospital discharge {two to three weeks after surgery); 2) right heart catheterization data and a pulmonary angiogram had been obtained at UCSD preoperatively; {3) complete hemodynamic data were obtained during postoperative monitoring; the values provided are those obtained on the third or fourth postoperative day; {4) one of the authors was in the operating room at the time of surgery, reviewed the thrombus specimens and, with the surgeon, recorded the vessels entered at thromboendarterectomy. The lung perfusion scans were analyzed in the following manner. Using an anatomically derived template {Fig 1), the overall perfusion image in each projection was divided according to the template bronchopulmonary segmental boundaries. The perfusion to each anatomic segment was visually scored as normal {2.0), decreased {0.5 to 1.5), or absent {0). Only those segments with normal ventilation were ranked. Steal was noted to have occurred when a lung segment score decreased by 1.0 or more between the METHODS The medical records of 33 consecutive patients who successfully *From the Division of Pulmonary and Critical Care, Department of Medicine, University of California at San Diego. Presented in part at the annual meeting, American Federation for Clinical Research, Carmel, CA, February 6-9, Supported in part by NHLBI Institutional Research Training Award {HL-07022) and the UCSD-NHLBI SCOR in Acute Respiratory Failure {HL-23584). twill Rogers Institute Research Fellow. Manuscript received April 30; accepted May 3. Reprint requests: Dr. Moser; UCSD Medical Center (H-772), 225 Dickinson Street, San Diego, CA Posterior view FIGURE l. Anterior (left) and posterior (right) views of perfusion scan template that demonstrates the normal bronchopulmonary segmental anatomy. Similar diagrams were used to assess the lateral and oblique perfusion scan images Pulmonary Vascular Steal (0/man et a/)

2 Table!-Preoperative and Postoperative Hemodynamic Data PAP, Hg 41.7 ± 12 Cardiac output, Umin 3.85 ± 1.1 PVR, dyne-s/em Preoperative Postoperative Hemodynamics Hemodynamics p Value 24.2±6 5.9± ± 100 *All values are reported as the mean± I SD; p values reflect the result of a two-tailed t test for paired values. preoperative vs the predischarge scan. The scans were scored by two of the authors (K.M.M., M.A.O.). There was interobserver agreement on the occurrence of steal in more than 90 percent of the segments. When disagreement existed, the segment was scored as showing no steal. Those scoring the scans were blinded as to the patient's name, the hemodynamic and angiographic data, and the findings at surgery. The preoperative pulmonary arteriogram was similarly scored. Lobar and segmental pulmonary arteries were judged to be normal (2.0), partially obstructed (1.0), or completely obstructed (0). In 15 patients, postoperative pulmonary arteriography was perfonned at 10 to 19 months after the operation and reviewed as above. 3.14). We also noted an increased incidence of steal in undissected relative to dissected segments (p<o.oool, odds ratio: 3.40). These two factors were synergistic in that a given segment with a normal preoperative arteriogram and scan that was not dissected was ten times more likely to demonstrate steal than a segment with an abnormal preoperative arteriogram and scan that was endarterectomized. Follow-up pulmonary angiograms were available for review in 10 of the 23 patients who demonstrated steal. In nine of them, the arterial segments with steal were normal preoperatively and remained so on the follow-up arteriogram. An example is shown in Figure St6tistical Analysis The preoperative and postoperative hemodynamic variables in all 33 patients were analyzed using the paired t test. 11 The Fisher's exact test" was used to assess the possible association of steal with the preoperative scan, the preoperative angiographic findings, and with whether the segmental arteries had been entered as a consequence of the endarterectomy procedure. The analysis was supplemented by review of the postoperative angiograms that were available. Values are reported as the mean± 1 SD. Statistical significance was accepted at the p<0.05level for all analyses. RESULTS Patients ranged in age from 20 to 70 years (45 ± 15 years) with a male to female ratio of 1.1:1. The preoperative and postoperative hemodynamic data in the 33 patients are presented in Table l. Significant declines in pulmonary artery mean pressure and pulmonary vascular resistance, as well as a rise in cardiac output, were noted postoperatively. All patients underwent a pulmonary thromboendarterectomy. However, in 27 patients, some segmental arteries (4.6 ± 2 segments) were not dissected at surgery. Twenty-one (78 percent) of these patients demonstrated steal in one or more (2.5±2 segments) of these untouched segments. In six patients, all pulmonary arterial segments were endarterectomized. Two patients in this group demonstrated a steal phenomenon. Thus, steal occurred in 23 (69 percent) of the 33 cases reviewed. The most common site of steal occurrence was the left upper lobe. In most instances, the perfusion decrements were visually striking (Fig 2). Further analysis of possible predictive factors disclosed a strong association between postoperative steal and a normal preoperative angiogram and perfusion scan in a given segment (Fisher's exact test, p<0.0005, odds ratio: FIGURE 2a (upper). 'Ibp row: Preoperative anterior view perfusion scan (left) and arteriogram (right) of the lea: lung. The upper lobe pulmonary artery is unobstructed on the arteriogram and perfusion is nonnal on the scan image. FIGURE 2b (lower). Bottom row: Postoperative anterior view on predischarge perfusion scan (left) and follow-up arteriogram (right) of the lea: lung. Please note the decrement in predischarge perfusion of the upper lobe, demonstrating upper lobe vascular steal and the nonnal follow-up angiographic appearance of the upper lobe polmonary artery. CHEST I 98 I 6 I DECEMBER,

3 2. In one patient, the site of steal was abnormal on the preoperative angiogram and remained so at followup. The postoperative arteriograms were also reviewed in five of the ten cases without postoperative perfusion scan steal. We did not detect any new vascular abnormality on the postoperative study in any case. DISCUSSION The major finding in our study is the frequent occurrence of new postoperative, segmental, perfusion scan abnormalities in lung zones served by angiographically patent, undissected pulmonary arteries. The appearance of "spurious" recurrence of embolism on perfusion scan has been reported previously during the course of resolution of acute pulmonary embolism.' In that study, new scan defects appeared as resolution proceeded; and it was shown angiographically that these defects reflected differential rates of embolic dissolution in different vessels. Lung scan perfusion defects resolved distal to those vessels in which rapid resolution of obstruction had occurred, while new defects appeared distal to vessels that retained significant, although partial, obstruction. These spurious recurrences were postulated to reflect a changed pattern of blood How as a consequence of the redistribution of pulmonary vascular resistance. In the present series, the differential resolution rates of acute emboli played no role in the "steal syndrome" because acute emboli were not demonstrated at surgery. Yet the "steal" phenomenon again appears to reflect a redistribution of flow-in this instance following surgical thromboendarterectomy. What, then, are the potential mechanisms responsible? The answers must remain speculative at this juncture, but certain insights are available. Our initial concern when this phenomenon was first seen was that in situ thrombosis or embolic recurrence might have occurred perioperatively. Our analysis and other considerations make these possibilities unlikely. First, these patients are routinely protected by preoperative or intraoperative placement of a caval Greenfield filter, and lower extremity pneumatic compression plus heparin prophylaxis are begun in the immediate postoperative period. These measures should render perioperative embolism or rethrombosis low in probability-less than 4 percent overall for the caval filter alone Second, if a perfusion scan is obtained within 18 hours after surgery, a policy we have recently adopted, the steal phenomenon is already present. Indeed, the immediate postoperative scan is often a mirror image of the preoperative scan (Fig 3). Thirdly, if thrombosis in situ had occurred, we would expect it to occur in dissected segments in light of the thrombogenic potential of vessels whose endo thelium has been perturbed H In fact, we found that steal occurred far more frequently in segments untouched at surgery. Finally, the postoperative pulmonary arteriograms in this series have not demonstrated new occlusion of vessels that were patent on the preoperative angiogram. Absolute assurrance of vascular patency postoperatively in lung zones with steal would require an early postoperative pulmonary arteriogram. We have chosen not to carry out this procedure for ethical reasons. Segmental atelectasis and postoperative reperfusion edema could be causes of new perfusion scan defects in our patient population. Both pulmonary parenchymal abnormalities have been reported to induce regional decrements in How However, the steal Centimeters I I It,,,, I I It,, I I,, It,,,, I I I I I I I FtGURE 3. 1bp panel: Preoperative portable perfusion scan, anterior view. Excellent perfusion to right upper lobe; by angiogram, no thrombi were present. Middle panel: Surgical specimen. Chronic thrombus was removed from the right middle and lower lobes, as wed as all segments on the left. No thrombus was present in the right upper lobe. Bottom panel: Anterior view, portable perfusion scan, performed 12 hours after surgery (view is slightly tilted). There is striking steal of the previously well-perfused right upper lobe that persisted to the time of hospital discharge. Pulmonary Vascular Steal (0/man et al)

4 zones have been consistently free of parenchymal infiltrates or areas of collapse on the postoperative chest roentgenograms; the postoperative ventilation scans have shown no abnormality in these segments; and, as previously reported, 4 reperfusion edema is limited to zones oflung served by endarterectomized vessels, whereas steal occurs dominantly in untouched segments. What alternative possibilities exist? A brief note regarding the approach to endarterectomy may provide some insight. Arteries are not endarterectomized under two circumstances: when no chronic thrombus is visible in the lobar or segmental artery at surgery; or when thrombus is seen, but is too thin to allow dissection to the segmental level. In addition, despite careful attempts at endarterectomy, the specimen may break distally, leaving behind obstructing organized thrombus in some segmental or subsegmental vessels. Given these considerations, we can put forth two different explanations for the postoperative regional decrement in flow responsible for the steal. First, in some cases, partially occluding thrombus may remain after surgery either because it was too distal to allow removal or, as noted above, "broke off' at endarterectomy. However, the presence of partially occlusive residual distal thrombi-as seen on the preoperative arteriogram-appears to explain the steal phenomenon in a small minority of patients. A more comprehensive explanation can be supported by the following two observations: (1) the normal preoperative scan and angiogram in most patients with "steal" and the normal follow-up angiograms available in some, and (2) pathologic examination of lung biopsy specimens in a number of these patients that have shown extensive pulmonary hypertensive changes in the small distal arteries changes not detectable by routine angiography. We speculate that these changes, like those in patients with congenital right-to-left shunts, are induced by the high flows and pressures to which the "open" vascular bed is exposed for months to years in patients with chronic thromboembolic pulmonary hypertension (C T-E PH). We further speculate that such hypertensive changes in the vascular bed that was "open" preoperatively may reduce flow to these zones postoperatively as pulmonary flow is diverted to the relatively normal microcirculation distal to endarterectomized segmental arteries. A final potential explanation might exist, namely: that surgical disruption of the pulmonary vascular endothelium may induce a state of nonresponsive vasodilatation in the endarterectomized areas. Evidence for this hypothesis is primarily circumstantial. The vascular endothelium has been shown to secrete vasodilatory and vasoconstrictor substances. 25-llll In the setting of experimental endothelial stripping or carotid endarterectomy, the vascular endothelium has been shown to regenerate, forming a coherent layer in seven to ten days. However, morphologic and functional alterations in the neoendothelium have been shown to exist as late as four weeks postoperatively It is possible, therefore, that the neoendothelium in the operated-on segments in patients with thromboembolic pulmonary hypertension may create an abnormally low resistance segment in endarterectomized lung zones by its effect on vascular smooth muscle, thus shunting flow toward zones that were endarterectomized. Additional biopsy specimens, arteriographic, and physiologic observations should clarify the relative contributions of the various mechanisms postulated. Whatever its bases, however, the most important clinically relevant observations regarding steal are that it occurs rather commonly and that, if it is encountered after a successful thromboendarterectomy in which the involved vessels were not entered, rethrombosis or embolism is an unlikely cause. ACKNOWLEDGMENTS: The authors wish to express their appreciation to many individuals, i n c Drs. l u Kirk ~ Peterson, Maurice Buchbinder, Pat Daily, Walter Demoitsky, Joe Utley, and William Ashburn for their contributions to the evaluation and treatment of patients included in this study. REFERENCES 1 Moser K, Spragg R, Utley J, Daily P. Chronic thrombotic obstruction of major pulmonary arteries: results of thromboendarterectomy of 15 patients. Ann Intern Med 1983; 99: Moser KM, Daily P, Peterson K, Dembitsky W, Vapnek J, Shure D, et al. Thromboendarterectomy for chronic, rnajol'"vessel thromboembolic pulmonary hypertension. Ann Intern Med 1987; 107:56() Daily PO, Dembitsky W, Peterson KL, Moser KM. Modifications of techniques and early results of pulmonary thromboendarterectomy for chronic pulmonary embolism. J Thorac Cardiovasc Surg 1987; 93: Rich S, Levitsky S, Brnndage B. Pulmonary hypertension from chronic pulmonary thromboembolism. Ann Intern Med 1988; 108: Levinson RM, Shure D, Moser KM. Reperfusion pulmonary edema after pulmonary artery thromboendarterectomy. Am Rev Respir Dis 1986; 134: Wragg R, Dimsdale J, Moser K, Archibald C. Operative predictors of delirium following pulmonary thromboendarterectomy: a model for post-cardiotomy delirium? J Thorac Cardiovasc Surg 1988; 96: Olman M, Auger W, Ashburn W, Hartman M, Moser K. Pulmonary blood Row 'steal' after surgery for chronic thromboembolic pulmonary hypertension. Clin Res 1989; 37:146A 8 Wallace JM, Moser KM, Hartman MT, Ashburn W. Patterns of pulmonary perfusion scans in normal subjects. Am Rev Respir Dis 1981; 124: Nicod P, Peterson K, Levine M, Dittrich H, Buchbinder M, Chappuis F, et al. Pulmonary angiography in severe chronic pulmonary hypertension. Ann Intern Med 1987; 107: Blaufox M, Freeman L. The pulmonary profile: normal perfusion and ventilation lung scintigraphies: anatomic mrrelations. Medical Multimedia Corporation Press; 1982:1 11 Gray H. The lungs. In: Goss C, ed. Anatomy of the human CHEST I 98 I 6 I DECEMBER,

5 body. 29th ed. Philadelphia: Lea & Febiger, 1973; Wallenstein S, Zucker C, Fliess J. Some statistical methods useful in circulation research. Circ Res 1980; 47: Glantz S. Primer of biostatistics. New York: McGraw-Hill Books, 1981; Moser KM, Longo AM, Ashburn WL, Guisan M. Spurious scintiphotographic recurrence of pulmonary emboli. Am J Med 1973; 55: Greenfield LJ. Current indications for and results of Greenfield filter placement. J Vase Surg 1984; 1: Kanter B, Moser KM. The Greenfield vena cava 6lter. Chest 1988; 93:17Q Mason R, Mohammed S, Chuang H, Richardson P. The adhesion of platelets to subendothelium, collagen and artificial surfaces. Semin Thromb Hemost 1976; 3: Tschopp T, Baumgartner H, Siberbauer K, Sinzinger H. Platelet adhesion and platelet thrombus formation on subendothelium of human arteries and veins exposed to Rowing blood in vitro: a comparison with rabbit aorta. Haemostasis 1979; 8: Poulose K, Reba R, Wagner H. Characterization of shape of perfusion defects in certain pulmonary diseases. N Eng! J Med 1968; 279: Alderson P, Rujanavech N, Walker R, McKnight R. The role of 133 Xe ventilation studies in the scintigraphic detection of pulmonary embolism. Radiology 1976; 120: Anderson E, Simon G, Reid L. Primary and thromboembolic pulmonary hypertension: a quantitative pathological study. J Pathol1973; 110: Moser K, Daily P, Peterson K. The management of chronic unresolved large vessel embolism in the pulmonary circulation in health and disease. Ed. Will J, Dawson C, Weir K, Buckner C, eds. New York: Harcourt Brace Jovanovich, Rabinovich M. Mechanisms of pulmonary hypertension in chronic high How states. In: Weir K, Reesves J, eds. Pulmonary vascular physiology and pathophysiology, vol 23, Lung biology in health and disease. New York: Dekker, 1988; Heath D, Edwards J. The pathology of hypertensive pulmonary vascular disease: a description of six grades of structural changes in the pulmonary arteries with special reference to congenital cardiac septal defects. Circulation 1958; 18: Griffith T, Edwards D, Lewis M, Newby A, Henderson A. The nature of endothelium-derived vascular relaxant factor. Nature 1984; 308: Furchgott R, Vanhoutte P. Endothelium derived relaxing and contracting factors. FASEB J 1989; 3: Yanagisawa M, Kurihara H, Kimura Y, Tomobe Y, Kobayashi M, Mitsui Y, et al. A novel potent vasoconstrictor peptide produced by vascular endothelial cells. Nature 1988; 332: O'Brien R, Robbins R, McMurty I. Endothelial cells in culture produce a vasoconstrictor substance. J Cell Physiol 1987; 132: Cross K, EI-Sanadiki M, Murray J, Mikat E, McCann R, Hagan P. Functional abnormalities of experimental autogenous graft neoendothelium. Ann Surg 1988; 208: Skimokawa H, Aarhus L, Vanhoutte P. Porcine coronary arteries with regenerated endothelium have a reduced endotheliumdependent responsiveness to aggregating platelets and serotonin. Circ Res 1987; 61: Plan to Attend ACcP s 57th Annual Scientific Assembly San Francisco ~ November 4-8, 1991 nr-:: Pulmonary Vascular Slaal (Oiman et 81)