This study involved two consecutive experiments. In experiment I, the pulmonary response to intramedullary fixation of femoral fractures with reaming was compared with the response to the procedure without reaming in sheep that had normal pulmonary function. A second type of reamer was also tested to determine if the design of the reamer had an effect on pulmonary response. The first reamer (AO Synthes, Monument, Colorado) has a solid shaft that is the same diameter as the head of the reamer. This has been thought to increase the pressure created by the reamer and, thus, the number of emboli. The second reamer (Küntscher Reamer; Biomet, Warsaw, Indiana) is a fine-wire reamer and the diameter of the shaft is smaller than that of the head. It is thought to generate less pressure and, therefore, fewer emboli. In experiment II, the pulmonary response in sheep with normal lungs was compared with the response in sheep with a contused lung. The magnitude and duration of transvenous particulate embolization was measured with intracardiac ultrasound.
Study Design
Experiment I
Twenty-three yearling mixed-breed sheep that weighed forty-five to sixty kilograms were randomly assigned to one of four groups. All of the animals were allowed one hour to recover from the induction of anesthesia and insertion of a line before initiation of a protocol that consisted of baseline measurements and an orthopaedic intervention followed by a three-hour observation period.
Five animals, the control group, had a skin incision made over the lateral aspect of the femur; five animals had a lateral skin incision and creation of a fracture of the middle of the femoral shaft; six animals had a lateral skin incision, a fracture of the middle of the femoral shaft, and insertion of a femoral nail without reaming; and seven animals had a skin incision, a fracture of the middle of the femoral shaft, nine passes of a flexible intramedullary reamer, and insertion of an intramedullary nail.
After the initial randomized experiment was completed, a separate group of five sheep was studied to see what effect the type of reamer had on pulmonary function. In these animals, a femoral fracture was created according to protocol and an intramedullary nail was inserted after the use of a modified (Küntscher) reamer.
Serial hemodynamic measurements were made at thirty, sixty, 120, and 180 minutes after the intervention was completed. Blood was collected for the determination of arterial and mixed venous blood gas with a Radiometer ABL-30 analyzer (Radiometer, Copenhagen, Denmark) and for the determination of hemoglobin during the baseline period and at the completion of the experiments. Lung tissue was excised after death to quantify the deposition of emboli.
Experiment II
Thirty yearling mixed-breed sheep that weighed forty-five to sixty kilograms were divided into two groups. The first fifteen sheep had normal lungs and were assumed to have normal pulmonary function. In the second fifteen sheep, a unilateral pulmonary contusion was created with a blow to the right side of the chest with a captive bolt. A three by three by 0.25-inch (7.62 by 7.62 by 0.64-centimeter) steel plate padded with a patch of tire rubber was used to transfer the kinetic energy delivered by the bolt to the chest wall. The plate was placed subcutaneously in the chest wall at the level of the posterior axillary line. A Schermer mechanical stunning apparatus (Koch, Kansas City, Missouri) with number-21 bull load was held against the plate and fired at full expiration. After creation of the contusion, the plate was removed and the incision was closed. At the completion of the experiment, the lungs were examined grossly through a thoracotomy. There was consistent evidence of discoloration on the surface of the lung from hemorrhaging, which, on sectioning, was found to extend to the hilum.
The following orthopaedic interventions were used for both the animals with normal lungs and those with contused lungs. Five animals had a lateral skin incision and creation of a fracture of the middle of the femoral shaft; five animals had a skin incision, a fracture of the middle of the femoral shaft, and insertion of a femoral nail without reaming; and five animals had a skin incision, a fracture of the middle of the femoral shaft, nine passes of a flexible intramedullary reamer, and insertion of an intramedullary nail.
Hemodynamic, blood-gas, and ultrasound data were recorded before the intervention and for three hours after it at regular intervals. Baseline data were recorded one hour after the induction of anesthesia and insertion of the vascular line. The pulmonary contusion was induced after the baseline data were recorded. Orthopaedic intervention was initiated one hour after the baseline data were recorded (for the normal lungs) or one hour after the contusion was created (for the contused lungs). Hemodynamic, blood-sampling, and ultrasound data were recorded every sixty minutes after the orthopaedic intervention. Intracardiac ultrasound data were measured continuously during contusion of the lung and the orthopaedic intervention and every sixty minutes thereafter.
Preparation of the Animals and Hemodynamic Monitoring during Orthopaedic Intervention in Experiments I and II
The sheep fasted overnight but had free access to water. Anesthesia was induced by intravenous injection of ketamine (0.3 milligram per kilogram of body weight) and Valium (diazepam, five milligrams per kilogram of body weight). After endotracheal intubation, anesthesia was maintained by inhalation of 2 per cent isoflurane and 98 per cent oxygen. The animals were ventilated with an Ohio Omeda volume ventilator (Airco, Madison, Wisconsin) with tidal volumes set to deliver fifteen milliliters per kilogram of body weight at a rate of twenty breaths per minute. Positive end expiratory pressure was maintained at five centimeters of water. After the induction of anesthesia, the animals were given an intravenous infusion of Ringer lactate solution (five milliliters per kilogram of body weight), which was begun as the isoflurane was administered. Throughout the experimental protocol, the animals received Ringer lactate solution intravenously at two milliliters per kilogram of body weight per hour.
Single-lumen catheters were inserted into the carotid artery and a jugular vein for infusions and collection of blood samples. A balloon-directed pulmonary artery catheter inserted through a jugular vein was floated into the pulmonary artery. Pulmonary artery pressures, pulmonary arterial wedge pressures, systolic arterial pressures, and central venous pressures were continuously recorded in millimeters of mercury on a Hewlett-Packard monitor (model 78342A; Palo Alto, California), with zero referenced to the level of the heart in animals that lay with the left side dependent. Cardiac output (in liters per minute) was determined with thermodilution techniques after injection of iced saline solution; at least three values were obtained at each measurement point, and a mean value was calculated. Pulmonary vascular resistance was calculated, in millimeters of mercury per milliliter per minute, by subtracting the pulmonary capillary wedge pressures from the pulmonary arterial pressure and dividing the remainder by the cardiac output15.
All sheep had a three-centimeter-long incision made in the skin over the piriformis fossa of the left femur. To create a fracture in the middle of the left femoral shaft, a one-centimeter incision was made laterally over the middle of the shaft, the outer cortex was drilled with an AO 3.5-millimeter drill-bit, an osteotome was introduced into the drill-holes and grasped with a vise grip, and torque was applied to create an oblique osteotomy of the middle of the shaft. In one subgroup, a nine-millimeter solid femoral nail was inserted, without reaming, through the piriformis fossa across the fracture for reduction and stabilization. In another subgroup, an eleven-millimeter slotted femoral nail was inserted across the fracture after reaming of the intramedullary canal with an AO Synthes reamer. The reaming was performed in one-half-millimeter increments, beginning at seven millimeters and progressing to eleven millimeters, for a total of nine passes of the reamer.
Intravascular Ultrasound in Experiment II
A transvenous echocardiogram catheter (12.5-megahertz transient frequency and eight francium transducer) (SSD 550, Intravascular System; Aloka, Tokyo, Japan) was inserted through a jugular vein and advanced into the distal aspect of the right atrium guided by echo signals. The intravascular ultrasound output was continuously recorded on video cassette during the orthopaedic interventions, and each step in the orthopaedic treatment was monitored. From the cassette, the amount of embolized marrow was measured quantitatively and depicted in graph form. Echogenic embolic particles that were visualized entering the right atrium after each maneuver were quantified by counting the maximum number of particles per video frame as well as measuring the maximum size of the particles and the duration of the appearance of the particles9,22.
Pulmonary Tissue Response
Experiment I: At the completion of the experiment, the anesthetized sheep were killed by intravenous injection of phenobarbital. Two grams of lung tissue was excised from the cephalad and caudad edges of the left lung and placed immediately into tared vials that were then sealed and reweighed. The excised lung tissue was dried in an oven until the dry weight was stable on consecutive days. The amount of water in the lung tissue was determined as the ratio of wet-to-dry tissue weight. Excised lung tissue was also submitted for histological examination. Sections from samples were placed in 10 per cent formaldehyde solution for complete fixation. Several thin sections were taken from each lung sample and dehydrated for embedding in paraffin. Five-micrometer sections of tissue were placed on glass slides and stained with hematoxylin and eosin. The slides were evaluated blindly by one investigator to determine the presence or absence of fat emboli in the tissue. Because of the dissolution of fat by alcohol in the dehydration process, a clear negative histological image in erythrocyte-dense pulmonary microvessels was considered adipocyte or lipid and evidence of fat embolism.
Experiment II: At the completion of the physiological measurements, the anesthetized sheep were killed by intravenous injection of phenobarbital and lung tissue was obtained for histological examination. Samples were not taken from the areas of pulmonary contusion. Sections from samples were immediately flash-frozen in sodium pentane. Five-micrometer frozen sections were placed on glass slides and stained with oil red O. The slides were analyzed blindly by one investigator to determine the presence or absence of fat emboli in the tissue. Bright red foci of stain within erythrocyte-dense pulmonary microvessels were considered adipocyte or lipid and evidence of fat embolism. In order for a sheep to be considered to have fat emboli, multiple slides from all tissue samples had to have foci of oil-red-O staining. Lungs with consistently positive findings had multiple foci, and lungs with negative findings had few or no foci of stain.
Statistical Analysis for Experiments I and II
Hemodynamic parameters and the results of the blood tests obtained serially over time in the four groups were analyzed with two-way analysis of variance. If differences were noted among interventions or times with the analysis of variance, a Dunnett multiple-comparison procedure was used to determine which orthopaedic interventions or time-periods were associated with values that differed from the baseline measurements. The ratios of wet-to-dry tissue weight were compared between groups with use of one-way analysis of variance. The intracardiac ultrasound data were evaluated with one-way analysis of variance, with the investigator comparing the embolic responses (the size, number, and duration of the embolic shower) between the different orthopaedic interventions. Statistical comparisons between the embolic responses associated with the different steps of the orthopaedic intervention (creation of the fracture, opening of the canal, reaming, and insertion of the nail) were also made with one-way analysis of variance. The embolic response to each pass of the reamer was evaluated with a Duncan multiple-comparison procedure. The presence or absence of fat emboli in the tissue samples was analyzed with the chi-square test. A significance of p < 0.05 was set for all tests at a statistical power of 50 per cent, given the small size of the samples. All of the statistical analyses were performed on SAS for Windows (version 6.08; Statistical Analysis System, Cary, North Carolina).
Experiment I
Hemodynamic Parameters
The only significant change in the hemodynamic parameters as a result of the orthopaedic interventions was a transient increase in the pulmonary vascular resistance at thirty minutes compared with the baseline value in the animals that had a skin incision, a fracture of the middle of the femoral shaft, and nailing with reaming (p = 0.045) (Fig. 1). With the numbers available, we could detect no significant differences among the groups at any time-period (Table I).
We also could not detect, with the numbers available, a significant difference in any group between the baseline arterial and mixed venous PaO2 and the values at the end of the experiment. Similarly, the final hemoglobin and total protein concentrations were not different among groups.
Histological Analysis of the Tissue
There was no significant difference, with the numbers available, among the groups with regard to the frequency of lungs that had fat emboli in the pulmonary microcirculation or in the water content of the lung tissue (Table II).
Group Treated with the Modified Reamer
With the numbers available, we could not detect a significant difference between the pulmonary response associated with use of the modified Biomet reamer and that associated with use of the AO reamer.
Experiment II
Hemodynamic Parameters
In the small sample that we studied, systemic blood pressure, diastolic blood pressure, cardiac output, pulmonary arterial pressure, and pulmonary arterial wedge pressure did not change significantly in any group during the time-interval of observation (Table III). Pulmonary vascular resistance showed a transient increase after reaming both in the animals that had normal lungs (p = 0.022) and in those that had a contused lung (p = 0.013), but this did not persist at three hours after intervention (Figs. 2 and 3).
With the numbers available, the hemoglobin and white blood-cell count did not significantly change over the course of the experiment in any group. Fibrinogen levels declined slightly after femoral nailing with reaming in the sheep that had normal lungs.
With the numbers available, there were no significant changes in arterial oxygenation, pulmonary arterial oxygenation, the arterial-to-alveolar ratio, or pulmonary shunt due to any treatment in either the sheep with normal lungs or those with a contused lung. Pulmonary shunt consistently increased after pulmonary contusion, but not significantly.
Histological Analysis of the Tissue
Histological analysis revealed a significant difference in the presence of fat emboli in the animals that had nailing, with or without reaming, compared with the control animals that had normal lungs (p = 0.003) but not compared with the control animals that had a contused lung (Table IV). (This lack of significance was possibly due to the small size of the sample.)
Intravascular Ultrasound
The intravascular ultrasound was used to monitor and quantitate9,22 the presence of embolic material through the right side of the heart during the orthopaedic intervention. A few small embolic particles were identified during the creation of the fracture. Opening of the intramedullary canal with the awl was associated with the largest number of emboli (Figs. 4-A, 4-B, and 5). This procedure produced embolic particles of the largest size and for the longest duration in the animals that had nailing, with or without reaming. The first two passes of the reamer produced more emboli and embolism of longer duration than did the later passes (Fig. 6). Insertion of the intramedullary nails resulted in embolization; however, nailing without reaming produced significantly more embolic material than nailing with reaming did (p = 0.01).
Fat embolism syndrome has been described after fractures of the long bones of the lower extremities10,11. Certain orthopaedic interventions during intramedullary nailing of fractures and during intramedullary reaming and broaching in total hip and knee arthroplasty may result in serious pulmonary injury or death12-14,19,21,22,24,25.
In this study, we consistently found that intramedullary reaming before nailing of femoral fractures caused a transient increase in pulmonary vascular resistance but little pulmonary dysfunction. Our femoral osteotomy model has a built-in pulmonary-injury-prevention factor in that the animals were well hydrated and ventilated. This is not seen in the clinical situation of fractures of long bones. Thus, the controlled environment in our study may have produced better results than those in an actual clinical situation. Insertion of a nail without reaming did not cause a significant change in pulmonary function. In addition, we demonstrated that there was no added adverse risk associated with the presence of a pulmonary contusion. The concept that a pulmonary injury primes the lung for the damaging effects of fat embolism was not supported by the results of our experiment.
We used a sheep model because the ovine pulmonary response to microembolization injury has been documented extensively, and the pathophysiology of adult respiratory distress syndrome in sheep has been established. An advantage of the sheep model of lung dysfunction is that the results of previous research on fat embolization after femoral reaming in sheep have been reported20 and any discrepancy between those results and those of the present study cannot be attributed to interspecies variation. Another consideration was that the sheep femur is comparable in size and structure to the human femur and permits intramedullary reaming and insertion of a nail1,2,15,23,25.
Our specific objectives were to monitor hemodynamic and embolic events sequentially to determine if pulmonary dysfunction was increased by embolic phenomena associated with orthopaedic interventions and if specific orthopaedic interventions were associated with embolic events. Pape et al. reported substantial sustained pulmonary dysfunction after intramedullary reaming18. The difference between their conclusions and ours regarding the pulmonary effects resulting from treatment of femoral fractures can be attributed to experimental design. Our study more closely simulates the clinical situation because a standardized fracture was created, even though the fracture was the result of a controlled osteotomy and not of blunt forces that injure bone and soft tissue. We recognize that intervention of the fractured bone is necessary to study a model that is clinically relevant. We did not find any embolic phenomenon to be associated with a controlled fracture of the femur. In their clinical and sheep studies, Pape et al. used an intact femur model in which the femoral canal had no vent through which the increased pressure and marrow contents could be released8. Clinical studies have confirmed that intact femora and fractured femora are different with regard to the potential for embolization of the marrow. Kerr et al. noted a cumulative detrimental effect in intact femora after bilateral closed femoral nailing for impending pathological fracture. However, catastrophic consequences may occur as a more isolated event. Several authors have reported adverse effects as a result of embolization of marrow after hip and knee arthroplasty12,22,27.
Some important observations made in the present study were the timing and magnitude of transvenous embolic events associated with orthopaedic manipulation of the fractures. The operative intervention associated with the greatest amount of emboli were opening of the intramedullary canal with the awl, the first two passes of the reamer, and insertion of the intramedullary nail. The variability in embolic material during the time-course of fracture manipulation suggests that interventions intended to ameliorate or neutralize fat embolism have to be carried out in the proper time-frame.
Pulmonary dysfunction in the fat-embolism syndrome may be related to a cumulative dose of embolic material or to the acute magnitude of a single dose. In either case, activation of inflammatory mediators is an important pathophysiological event15. The importance of this distinction may have implications with regard to how pulmonary injury can be minimized in patients who have a femoral fracture. Efforts to decompress bone-marrow embolization before intervention and reduction of the fracture, regardless of the type of treatment, may be prudent17,18. Factors that influence pulmonary dysfunction and need additional study include the methods of reaming, the design of the reamer and the intramedullary nail, the techniques for decreasing the viscosity and thermal necrosis of the intramedullary tissue, and the pharmacological interventions that block mediators of pulmonary microvascular injury. Insertion of the awl and reaming may have a more profound effect in a long bone with an intact diaphysis, such as is found with an impending pathological fracture, closed femoral shortening, or total joint arthroplasty.
The clinical implications of this study are that the pulmonary dysfunction created with the controlled osteotomy model in this study was minimum, reaming had a modest effect on pulmonary vascular resistance that was not observed in the animals that had intramedullary nailing without reaming (neither type of treatment had a clear advantage with regard to pulmonary function, however), the presence of a pulmonary contusion did not increase the risk of pulmonary dysfunction after treatment of the fracture, the effect of reaming on pulmonary vascular resistance seemed to be transient, intravascular ultrasound documented that certain orthopaedic interventions were associated with significantly more and larger emboli and a longer duration of the pulmonary embolic shower, insertion of the awl caused the greatest amount of embolic material, and embolization of marrow is a cumulative phenomenon that continues in substantial amounts with the first two passes of the reamer and with the actual insertion of any intramedullary device. Our study does not support the hypothesis regarding the advantages of so-called minimum reaming, as there was no embolic effect after the first two passes of the reamer. There was no distinct advantage to reaming or not reaming with regard to pulmonary function.
NOTE: The authors thank David Senft, John Ganser, M.D., Jonathan Ellichman, M.D., Barbara Kochevar, and Don Trunkey, M.D., for help and assistance in the preparation of this manuscript.