Fifty-five patients who were scheduled to have a total of fifty-nine elective total knee arthroplasties with cement consented to participate in this study. The study was approved by the Human Research and Ethics Committee. The twenty-two men and thirty-three women ranged in age from thirty-two to eighty-three years (mean, 65.5 years). Thirty-one patients were obese, twenty-nine had a history of hypertension, seven had chronic atrial fibrillation, three had end-stage renal disease that was being treated with hemodialysis, four had a history of deep venous thrombosis, and two had a history of pulmonary emboli. One of the two patients who had had pulmonary emboli had had a Greenfield filter in place at the time.
Forty-seven of the arthroplasties were primary and twelve were revision procedures. A central venous catheter was inserted for intraoperative aspiration of blood during all of the primary procedures. Two patients had a bilateral procedure performed during the same anesthesia session, but hemodynamic data collected during the second procedure was not included in the analysis because of the absence of suitable hemodynamic baseline values. One patient had a staged bilateral procedure with a week between the operations, and a fourth patient had a revision total knee arthroplasty because of loosening and then had a second revision two months later because of recurrent dislocation. Each of these procedures was considered a separate operative event.
All patients had general anesthesia with endotracheal intubation. Patients who had a history of esophageal abnormality or rheumatoid arthritis of the cervical spine or who chose regional anesthesia were excluded from the study. Four staff surgeons performed the procedures. For each operation, the involved limb was exsanguinated with elevation and an Esmarch bandage, after which a tourniquet was applied and inflated to 350 millimeters of mercury (46.66 kilopascals). The prosthesis was chosen by the surgeon. All tibial and femoral components were inserted with use of an intramedullary guide system. The tibial and femoral canals were irrigated with jet lavage before insertion of the cement. All implants were cemented with standard techniques; the procedures did not include venting of the femoral or tibial shaft or overdrilling of the pilot hole for the intramedullary guide. Before the tourniquet was deflated, each patient received one unit of autologous blood to provide adequate preload according to departmental protocol.
After induction of the anesthesia, a five-megahertz single-plane transesophageal echocardiography probe (Hewlett-Packard, Wayne, Pennsylvania) was inserted to image the right atrium and the right ventricle. Transesophageal echocardiography images were recorded for two minutes after induction of the anesthesia, inflation of the tourniquet, and insertion of the cement. Continuous recordings of the right atrium and the right ventricle began just before the tourniquet was released and were continued for fifteen minutes.
Two echocardiologists (R. S. and K. C.) from the Division of Cardiology, Department of Medicine, who were blinded to all information regarding patient demographics and intraoperative events at the time of imaging, independently reviewed the transesophageal echocardiography images postoperatively. In the event of disagreement between these observers, a third echocardiologist reviewed the images. All echocardiograms were evaluated with use of a modification of the grading system of the Mayo Clinic for echogenic emboli5 (Table I). Each image was assigned a score of 1, 2, or 3 points for the amount of filling of the right atrium, the duration of echogenesis, and the diameter of the largest embolic particles, for a possible total score of 3 to 9 points. The charts were reviewed to document the presence or absence of risk factors associated with the development of deep venous thrombosis or pulmonary embolism; these included an age of more than forty years, a history of deep venous thrombosis or pulmonary embolism, obesity, a malignant lesion, prolonged immobilization, and use of an estrogen preparation.
All patients had insertion of a radial artery catheter in addition to standard anesthetic monitors. Echogenic patterns were determined with transesophageal echocardiography and arterial catheters in the initial twenty-one patients (twenty-four knees). In an effort to identify the composition of the emboli, a number-7.5 French catheter was inserted in the right internal jugular vein in the last ten patients managed in this series. Ten milliliters of heparinized saline solution (one unit per milliliter) was placed in all catheters to prevent intracatheter thrombosis, and all catheters were tested for patency before the tourniquet was deflated. At the time that the tourniquet was released, the anesthesiologist aspirated sixty milliliters of blood from the catheters. A pathologist then sectioned and stained macroscopic material from each sample with hematoxylin and eosin, noting the presence or absence of methylmethacrylate, thrombus, and fat. If histological examination suggested fat, Sudan and oil-red-O stains were used to determine its presence.
Because the need for additional hemodynamic data became evident as the study progressed, an oximetry pulmonary artery catheter was also inserted in the remaining thirty-four patients (thirty-five knees). The thirty-five echo cardiograms for these patients were assigned to one of two groups on the basis of the maximum diameter of the echogenic material that had been identified. The images in Group I showed echogenic material that was less than 0.5 centimeter in maximum diameter, and those in Group II showed large discrete particles that were 0.5 centimeter or larger in maximum diameter. These larger particles were sometimes superimposed on a shower of small emboli.
Hemodynamic parameters, pulse, mixed venous oxygen saturation, and exhaled gas tensions were measured with the oximetry pulmonary artery catheters after induction of the anesthesia and before inflation of the tourniquet (baseline); at the time that the tourniquet was inflated; after the femoral and tibial components were cemented; and one, three, five, ten, and fifteen minutes after the tourniquet was deflated. The pulmonary vascular resistance index and the systemic vascular resistance index were calculated as: pulmonary vascular resistance = (mean pulmonary arterial pressure - pulmonary artery wedge pressure)/cardiac index, and systemic vascular resistance = (mean arterial pressure - central venous pressure)/cardiac index.
The last ten of the thirty-four patients who had an oximetry pulmonary artery catheter also had insertion of a number-8.5 French femoral vein sheath in the treated limb to assess embolic events immediately at the time that the tourniquet was released. These catheters were flushed with heparinized saline solution to prevent clotting and then aspirated when the tourniquet was deflated. Specimens were analyzed as previously described.
Electrocardiograms and radiographs of the chest were made preoperatively and postoperatively for all fifty-five patients. All patients received low-dose warfarin therapy (2.5 milligrams) beginning on the evening of the operation. Complications during the postoperative period were documented for each patient. When clinically indicated, venous Doppler ultrasound, ventilation-perfusion scanning, and pulmonary angiography was performed.
Statistical Analysis
The statistical analysis involved the eight measured hemodynamic variables (mean arterial pressure, heart rate, central venous pressure, mean pulmonary arterial pressure, pulmonary arterial systolic pressure, pulmonary arterial diastolic pressure, pulmonary arterial occlusion pressure, and cardiac index), the two calculated hemodynamic variables (systemic vascular resistance and pulmonary vascular resistance), and the four other measured variables (arterial oxygen saturation determined by pulse oximetry, mixed venous oxygen saturation, end tidal carbon dioxide, and end tidal nitrogen). Two-way repeated-measures analysis of variance determined the presence or absence of significant effects of the eight measurement periods, of Group I or Group II, and of their interaction. Whenever a variable was found to be significant with the original two-way repeated-measures analysis, that variable was subjected to repeated-measures analysis of variance separately for Groups I and II to determine which group demonstrated the significant repeated-measure effect. The Tukey test was used for post hoc comparisons. For all determinations, a p value of less than 0.05 was considered significant.
Echocardiography Results
Transesophageal echocardiography revealed showers of echogenic material traversing the right atrium, right ventricle, and pulmonary artery outflow tract in all patients after the tourniquet was deflated (Figs. 1, 2, and 3). The echogenic material first appeared in the right atrium and right ventricle ten to fifteen seconds after the tourniquet was released. The showers peaked in intensity within thirty seconds (range, twenty-four to forty-five seconds) and lasted longer than three minutes. Several patients had continuous echogenic showers after collection of the data ceased (at fifteen minutes after the tourniquet was released). Intraoperative movement of the knee joint by the surgeon increased the intensity of the echogenic showers.
Analysis of the relationship between the embolic scores5 and the risk factors associated with thrombogenesis revealed no significant association, with the numbers available, between the magnitude of the echogenic phenomenon and an age of more than forty years or previous deep venous thrombosis or pulmonary embolism. We also detected no significant association with secondary risk factors (obesity, a malignant lesion, prolonged immobilization, or use of an estrogen preparation), gender, or the duration for which the tourniquet had been inflated.
Thirty-five echocardiograms were also evaluated solely for the presence or absence of large echogenic particles. Nine images (26 per cent) exhibited only a so-called snowstorm pattern, in which echogenic particles of less than 0.5 centimeter in diameter obliterated the right atrium, right ventricle, and pulmonary artery (Group I). The remaining twenty-six images (74 per cent) exhibited distinct particles of at least 0.5 centimeter in diameter (Group II); those were often superimposed on a snowstorm of smaller particles. There was no association between the duration of tourniquet use and the presence of particles that were at least 0.5 centimeter in diameter.
Hemodynamic Results
Immediately after the tourniquet was released, the mean arterial pressure (and standard error of the mean) decreased, from the baseline value of 85 ± 2.8 millimeters of mercury (11.33 ± 0.37 kilopascals), to 79 ± 2.7 millimeters of mercury (10.53 ± 0.36 kilopascals). The mean mixed venous oxygen saturation also decreased, from 83 ± 0.9 to 72 ± 1.5 per cent. However, the mean pulmonary arterial systolic pressure increased, from 20 ± 1.0 millimeters of mercury (2.67 ± 0.13 kilopascals) to 24 ± 0.98 millimeters of mercury (3.20 ± 0.13 kilopascals). These changes were significant (p < 0.05) throughout the study population. However, with the numbers available, no significant association could be detected between the embolic scores5 and the intraoperative hemodynamic changes.
In both Group I and Group II, the mean pulmonary arterial pressure three minutes after the tourniquet was released (27 ± 1.0 millimeters of mercury [3.60 ± 0.13 kilopascals]) was increased compared with the baseline value (p < 0.05). In Group II only, the mean pulmonary vascular resistance increased, from 205 ± 16 dyne·s·cm-5·m2 to a maximum of 328 ± 29 dyne·s·cm-5·m2, beginning five minutes after the tourniquet was deflated (p < 0.05). There were no notable changes in the heart rate or cardiac index during the study.
In both groups, the mean mixed venous oxygen saturation decreased sharply one minute after the tourniquet was deflated (from a baseline value of 83 ± 0.9 per cent to 72 ± 1.5 per cent); a return of deoxygenated blood from the ischemic limb with release of the tourniquet most likely contributed to this response. The mean mixed venous oxygen saturation returned to the baseline value by five minutes after the tourniquet was released, whereas the mean pulmonary vascular resistance remained elevated in Group II. End tidal carbon dioxide increased five minutes after the tourniquet was deflated in both groups and returned to the baseline within ten minutes. These increases were not sufficient to account for the increased pulmonary vascular resistance seen in Group II. End tidal nitrogen did not vary intraoperatively.
Embolic Composition
Blood sampled from the pulmonary artery catheter at the onset of the embolic showers in ten patients did not demonstrate evidence of fat bodies. The blood aspirated from a distal port of the pulmonary artery catheter of one patient demonstrated fresh thrombus. In ten patients, a sample of blood was withdrawn from a number-8.5 French femoral vein sheath in the treated limb at the time that the tourniquet was released. Histological samples, which were stained and reviewed by the Department of Pathology, did not reveal the presence of fat. Assays for polymethylmethacrylate and its breakdown products revealed no important levels in any of the specimens. Blood aspirated from five of the ten femoral catheters demonstrated fresh venous thrombus (Fig. 4).
Four of the fifty-five patients had embolic material that was less than 0.5 centimeter in diameter coursing from the right side of the heart to the left through a patent foramen ovale. None of these patients were known to have a patent foramen ovale before the operation; the diagnosis was made intraoperatively on the basis of real-time echocardiograms. A documented increase in pulmonary vascular resistance secondary to the observed embolic events may have caused opening of a closed foramen ovale, predisposing the patient to neurological sequelae. However, none of these four patients who had so-called paradoxical emboli demonstrated clinical evidence of neurological sequelae.
The two patients who had had previous pulmonary emboli had a Greenfield filter in place at the time of the total knee arthroplasty. The filter was inserted in one patient immediately before the operation; this patient had relatively little echogenic material, and no large particles were evident on the echocardiograms (Group I). The filter in the second patient had been inserted eight months before the operation; this patient had moderate-to-heavy echogenic emboli at the time that the tourniquet was released (Group II). Clinical pulmonary embolus later developed in this patient, suggesting the development of collateral circulation after the filter was inserted.
No major changes were evident on postoperative electrocardiograms, measurement of arterial blood gases, or radiographs of the chest, compared with preoperative studies, in fifty-two of the fifty-five patients. Three patients had acute respiratory distress postoperatively, and ventilation-perfusion scans were positive for pulmonary embolism. All three of these patients had had echogenic particles that were at least 0.5 centimeter in diameter on images made at the time that the tourniquet was released. One of these patients had a bilateral total knee arthroplasty performed during the same anesthesia session, and the pulmonary vascular resistance doubled compared with the baseline values on release of the second tourniquet; postoperatively, this patient demonstrated new perfusion defects consistent with multiple pulmonary emboli.
There were no intraoperative cardiac arrests, no perioperative myocardial infarctions, and no perioperative deaths.
Despite the use of prophylaxis to decrease the risk of thromboembolic events, deep venous thrombosis and pulmonary embolism remain major causes of perioperative morbidity and mortality in association with total knee arthroplasty. In one series, 57 per cent (362) of 638 total knee arthroplasties were associated with deep venous thrombosis, with pulmonary embolism occurring in 8 per cent (thirty-nine) of the 517 patients26.
Intraoperative transesophageal echocardiography has provided reproducible evidence of an echogenic embolic phenomenon associated with total joint arthroplasty and the use of a tourniquet. With total knee arthroplasty, embolic showers are observed after the tourniquet is released18, suggesting that massive pulmonary emboli may occur at that time. With the use of intraoperative transesophageal echocardiography, pulmonary emboli can be diagnosed rapidly. In our opinion, it is not necessary to perform a ventilation-perfusion scan or angiography before initiating treatment13.
To quantify the emboli at the time that the tourniquet was released, we initially used a modification of the grading system of the Mayo Clinic, which assesses the degree of involvement of the right atrium, the duration of the echogenesis, and the diameter of the largest embolic particles5. With the numbers available, we were unable to detect a significant association between the intensity of the echogenic showers, as scored with this system, and the intraoperative hemodynamic changes. Similarly, Ereth et al. found no association between the embolic score and hemodynamic changes in a study comparing total hip arthroplasties performed with cement with those performed without it. In response to these initial findings, we began evaluating images solely on the basis of whether they showed embolic particles that were at least 0.5 centimeter in maximum diameter. We then identified a significant association between emboli of this size and increases in pulmonary vascular resistance, thereby implicating the size of the particles as a critical variable. Interestingly, there was no association between the duration of tourniquet use and the intensity of the embolic showers; this suggested a so-called all-or-none effect at the time of tourniquet release.
The effect of release of the tourniquet on pulmonary hemodynamics is illustrated by the relationship between larger embolic particles and increases in pulmonary vascular resistance. Intraoperatively, the mixed venous oxygen saturation returned to the baseline level by five minutes after the release, while pulmonary vascular resistance remained elevated in Group II (in which the particles were at least 0.5 centimeter in diameter); thus, a metabolic etiology for the increased pulmonary vascular resistance in Group II is unlikely. The fact that the pulmonary vascular resistance was increased in Group II but not in Group I (in which the particles were less than 0.5 centimeter in diameter) again implicates vascular occlusion rather than metabolic sequelae in the pathway to pulmonary dysfunction.
Theories regarding the pathophysiology of this embolic phenomenon have often focused on fat emboli as a suspected etiology. Penetration into the medullary canal during endoprosthetic procedures and intramedullary stabilization of fractures of the long bones with subsequent release of bone marrow into the circulatory system have been well documented13,17,21. The pathologists in the present study, however, identified thrombus, not fat, in macroscopic emboli found in the central circulation after release of the tourniquet. This finding is consistent with that of Healy et al., who observed little fat and no polymethylmethacrylate monomer in samples of systemic blood from twenty-five patients who had a total knee arthroplasty. Such findings may reflect a complex pathophysiology in which the release of free fatty acids from marrow debris into the central circulation is implicated in the activation of the clotting system6,22,25. In combination with venous stasis and intimal damage, this hypercoaguable state completes the Virchow triad of thrombus formation. Hofmann et al. proposed a pathophysiological model that links the release of fat emboli and the formation of venous thrombus to echogenic emboli as seen on transesophageal echocardiography.
Inadequate exsanguination of a limb coupled with tourniquet-related stasis and cooling contribute to the formation of fresh thrombus16. However, the patients in the present study had echogenic emboli even though exsanguination had been clinically adequate. Activation of the coagulation cascade by bone cement may contribute to the formation of fresh clot in such patients23,27. Most probably, the formation of venous clot associated with total knee arthroplasty with cement is attributable to many factors, such as tourniquet stasis; increases in intramedullary pressure; fat; activation of the coagulation cascade by bone cement; and other, unidentified variables7,20.
There are numerous concerns regarding the relationship between echogenic emboli and operative morbidity14. No permanent lung injury was identified in the present study, although patients with particles of at least 0.5 centimeter in diameter had increased pulmonary vascular resistance that did not return to baseline levels during the intraoperative observation period. Hemodynamic recordings and echocardiography were discontinued fifteen minutes after the tourniquet was released and were not performed postoperatively in our patients. All three patients in whom a clinical pulmonary embolism developed had been found to have particles of at least 0.5 centimeter in diameter on the intraoperative echocardiograms, and the patient who had a functioning Greenfield filter (inserted just before the operation) had little showering. Although anecdotal, these observations warrant additional investigation into the postoperative implications of these embolic events.
We suspect a relationship between the size of echogenic emboli and perioperative morbidity and mortality. The present study documented an increase in pulmonary vascular resistance in the twenty-six patients (74 per cent of the thirty-five included in the analysis) who had emboli of at least 0.5 centimeter in diameter at the time that the tourniquet was deflated. Hofmann et al. suggested that this physiological response can be lethal for patients who have massive release of bone marrow or an underlying compromise of cardiorespiratory reserve. Notably, similar increases in pulmonary vascular resistance and decreases in oxygen saturation are seen in patients who have clinical pulmonary embolism11.
The period after release of the tourniquet during total knee arthroplasty with cement represents a critical time of potential hemodynamic instability. The consistent release of fresh venous thrombus into the circulation at the time that the tourniquet is deflated was confirmed histologically in the present study2,14. Surgeons should consider acute pulmonary embolism when evaluating a patient who has intraoperative hemodynamic collapse during total knee arthroplasty with cement. Future investigation should focus on the identification of patients who are at risk for hemodynamic collapse from intraoperative embolic insult and on the use of antithrombotic agents to prevent the formation of thrombus when a tourniquet is employed.