Pulmonary injury may occur after a fracture of a long bone, in association with intravasation of marrow fat into the systemic circulation and a systemic inflammatory response to trauma13,36,54. The effects of fat embolism vary from an absence of symptoms to the manifestations of fat-embolism syndrome9,36, with hypoxemia, cerebral dysfunction, and development of fever within twenty-four to forty-eight hours after the onset of the injury43. The clinical prevalence of fat-embolism syndrome in patients who have fractures has been reported to range from 0.25 per cent (of 4530 patients) to 1.2 per cent (of 7701 patients)36. Multiply injured patients who have a fracture of a long bone are at risk for subsequent development of fat-embolism syndrome and adult respiratory distress syndrome3,18,19,31,33,39. The rate of mortality for patients who have adult respiratory distress syndrome associated with fat-embolism syndrome has been reported to be 10 per cent55.
Current reports in the literature have suggested that fixation of a fracture is best done early2,3,19,21,31,32,47,49. Retrospective studies2,19 have shown that intramedullary nailing within twenty-four hours after a femoral fracture reduced the prevalence of fat-embolism syndrome, adult respiratory distress syndrome, and pneumonia. Similarly, Bone et al. showed, in a prospective study, that fixation within twenty-four hours after a femoral fracture in multiply injured patients decreased the prevalence of pulmonary complications and the duration of hospitalization compared with those following fixation more than forty-eight hours after the fracture3. However, the definition of early fixation in those studies ranged from zero to forty-eight hours after the onset of the injury. Moreover, the benefit of early fixation of a femoral fracture by intramedullary nailing is questionable because of the effects of reaming26,33-35,37,38.
Intramedullary reaming causes an increase in intramedullary pressure and intravasation of fat emboli from the bone marrow6,15,20,22,24,56. Previous clinical studies have shown that nailing with reaming performed within twenty-four hours after an injury exacerbated hypoxemia33. In an experimental study of sheep, Wenda et al. revealed that an intramedullary pressure of fifty millimeters of mercury (6.67 kilopascals) in the femoral canal produced small amounts of fat emboli from bone marrow, as detected with echocardiography51. A study of intramedullary nailing of non-fractured femora of sheep56 showed that the procedure of reaming elevated the intramedullary pressure to 753 millimeters of mercury (100.37 kilopascals) (range, 310 to 1126 millimeters of mercury [41.32 to 150.10 kilopascals]). The first two passes of the reamer were responsible for most of the increase in intramedullary pressure. Interestingly, although reaming produced embolic showering as seen on an echocardiogram, the actual insertion of the nail led to the greatest amount of fat embolization. However, insertion of the nail increased intramedullary pressures by only sixty-nine millimeters of mercury (9.20 kilopascals). In patients who have pulmonary dysfunction, intramedullary nailing with reaming may cause additional pulmonary damage and may trigger adult respiratory distress syndrome4,32,33. Insertion of a femoral nail without reaming in multiply injured patients has been found to have fewer deleterious pulmonary effects than nailing with reaming33.
The purpose of the present study was to investigate the fate of fat emboli after they had been created and to examine the pulmonary effects of the timing and method of fixation of a fracture in association with underlying pulmonary fat embolism. We used a canine model of fat embolism created by pressurization of the medullary canal. The results may have implications for optimizing the management of patients who have pulmonary dysfunction and a fracture of a long bone.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. Funds were received in total or partial support of the research or clinical study presented in this article. The funding source was Grant MA-11292 of the Medical Research Council of Canada.
†Division of Orthopaedic Surgery, Department of Surgery, St. Michael's Hospital, 55 Queen Street East, Suite 800, Toronto, Ontario M5C IR6, Canada.
‡Department of Pathology, Mount Sinai Hospital, 600 University Avenue, Toronto, Ontario M5G 1X5, Canada.
§Musculoskeletal Research Laboratory, Division of Orthopaedic Surgery (G. I. A.), and Department of Anaesthesia (R. J. B.), St. Michael's Hospital, 30 Bond Street, Toronto, Ontario M5B 1 W8, Canada.
Induction of Fat Embolism
Fat embolism was created in forty-two skeletally mature, conditioned dogs of mixed breed that weighed from twenty-two to thirty-five kilograms. One dog died of ventricular tachycardia during induction of fat embolism. Therefore, data on forty-one dogs were available for analysis.
After sedation of the animals by subcutaneous injection of 0.05 milligram of acepromazine per kilogram of body weight and 0.6 milligram of atropine per kilogram of body weight, anesthesia was induced with intravenous administration of sodium pentothal. Endotracheal intubation was performed, and general anesthesia was maintained with a mixture of 1.5 per cent halothane, 65.5 per cent oxygen, and 33 per cent nitrous oxide. Fluid requirements were met by intravenous administration of thirty milliliters of lactated Ringer solution per kilogram of body weight per hour. The left hind extremity, right side of the groin, and left half of the anterior aspect of the neck were shaved and cleaned with soap (70 per cent isopropyl alcohol and povidone-iodine [Betadine]). One gram of cefazolin (Ancef) and one milligram of oxymorphone were given intravenously. An 18-gauge femoral arterial catheter (Arrow, Mississauga, Ontario, Canada) was inserted into the femoral artery through a cut-down. An 8.0-French (2.64-millimeter-diameter) Cordis catheter (percutaneous sheath introducer set; Arrow) was inserted in the left external jugular vein. A 7.5-French (2.5-millimeter-diameter) Swan-Ganz catheter (Baxter-Edwards, Mississauga, Ontario, Canada) was placed through the introducer and positioned in the pulmonary artery with use of fluoroscopy. The femoral and pulmonary arterial catheters were connected to calibrated transducers (Cobe Medical Products, Scarborough, Ontario, Canada), which provided input to a recording device (model TA2000; Gould Instruments, Cleveland, Ohio) for blood pressure and pulmonary artery pressure-readings. An analog-to-digital converter (Perimed, Piscataway, New Jersey), connecting the Gould recorder to a portable computer (286V Ultralite; NEC, Tokyo, Japan), allowed continuous graphic monitoring of the blood pressure and the pulmonary arterial pressure with use of a software program (Perisoft; Perimed).
A lateral parapatellar approach to the left knee joint was made, the knee was flexed, and the patella was displaced medially. Retrograde reaming of the femoral canal was performed through the intercondylar notch in one-millimeter increments, beginning at six millimeters (six-millimeter reamer; Howmedica, Pfizer Hospital Products, Guelph, Ontario, Canada) and progressing to nine millimeters (seven to nine-millimeter reamers; Synthes, Mississauga, Ontario, Canada). An 8.25-millimeter polyethylene cement-restrictor (DePuy, Warsaw, Indiana) was inserted into the proximal aspect of the femur. The tibial canal was perforated through the tibial plateau 1.5 centimeters posterior to the anterior aspect of the tibial tuberosity. A six-millimeter flexible reamer (Howmedica, Pfizer Hospital Products) was used to ream the tibia in an antegrade manner.
Semiliquid polymethylmethacrylate bone-cement (Simplex P; Howmedica, East Rutherford, New Jersey) was introduced into the femoral and tibial canals after reaming. The cement was pressurized by inserting precontoured Steinmann pins, measuring 5.0 by 150.0 millimeters, into the canals. Previous studies of arthroplasty20,30,42,56 and work in our own laboratory1 have shown that insertion of a femoral prosthesis with pressurization of the cement elevates intramedullary pressure to levels ranging from 290 to 900 millimeters of mercury (38.66 to 120.00 kilopascals).
After irrigation of the knee joint with sterile saline solution, the patella was reduced. The incision was closed in layers, and the left hind extremity was bandaged. The arterial line was removed, the right femoral artery was ligated proximal to the insertion point of the catheter, and the incision in the groin was closed. The Swan-Ganz catheter was removed from the animals that were to survive for twenty-four hours. However, the Cordis catheter remained in place and covered with padded dressings. For the animals that were to survive for four hours, the Swan-Ganz catheter was disconnected but not removed from the pressure transducers. The animals were transferred to the recovery room. Unrestricted weight-bearing was permitted. Postoperatively, one gram of cefazolin as well as one milligram of buprenorphine were administered intravenously for analgesia. Buprenorphine tends to produce less respiratory depression than morphine, but it provides sufficient analgesia to allow the animals to walk.
Control Groups
The eight dogs in the control group were divided into two subgroups, consisting of four dogs that were killed four hours after pressurization of the medullary canal and four that were killed twenty-four hours after pressurization of the medullary canal. The dogs were killed with an intravenous injection of pentobarbital sodium (340 milligrams per milliliter of solution). Immediately before the dogs were killed, an arterial blood sample was taken for analysis of blood gas and the pulmonary arterial pressure was measured.
Experimental Groups
After pressurization of the medullary canal in the left hind extremity, a fracture was created in the contralateral femur and then fixed in thirty-three dogs. These dogs were randomly divided into two groups: fifteen dogs had fixation of the fracture four hours after embolic showering and eighteen dogs had fixation twenty-four hours after the procedure. After induction of general anesthesia, the right thigh was shaved and was cleaned with povidone-iodine (Betadine). The Swan-Ganz catheter was reconnected to pressure transducers in the animals that had fixation after four hours as well as in the animals that had fixation after twenty-four hours (the catheter was reinserted in those dogs). A lateral approach was made to the middle part of the femur, and an oscillating saw (Synthes) was used to notch the lateral cortex. A three-point bending jig applied to the lateral aspect of the femur created a standard transverse fracture of the middle of the femoral shaft.
The two temporal groups were subdivided into three subgroups according to the method of fixation of the fracture (Figs. 1-A, 1-B, and 1-C). Eleven dogs (five in the four-hour group and six in the twenty-four-hour group) had application of an eight-hole 3.5-millimeter dynamic compression plate (Synthes) on the lateral aspect of the femur. Eleven dogs (five in the four-hour group and six in the twenty-four-hour group) had a stainless-steel forearm nail (Biomet, Warsaw, Indiana) that was 4.5 millimeters in diameter and 150.0 millimeters long inserted in the intramedullary canal without reaming, and eleven dogs (five in the four-hour-group and six in the twenty-four-hour group) had a stainless-steel humeral nail (Biomet) that was 8.0 millimeters in diameter and 150.0 millimeters long inserted in the intramedullary canal after reaming. The intramedullary nails were inserted in an antegrade manner through the piriformis fossa. In the animals that had a nail inserted without reaming, a 2.0-millimeter Kirschner wire was passed just medial to the greater trochanter to locate the medullary canal. With use of a 5.5-millimeter cannulated drill-bit, an entry hole was made for the nail. The 4.5-millimeter-diameter nail then was placed in the femoral canal. For the animals that had insertion of the nail with reaming, the femur was reamed in an antegrade manner from six to eight millimeters. One hour after fixation of the fracture, the animals were killed by administration of an overdose of pentobarbital sodium (340 milligrams per milliliter solution).
Collection of Data
Readings of pulmonary artery pressure were recorded for one minute at several time-intervals. A baseline reading was recorded immediately after induction of anesthesia and again before reaming to induce pulmonary embolism. Readings were then recorded during reaming of the femur at six millimeters and at nine millimeters; during reaming of the tibia at six millimeters; during pressurization of the cement; and at one, five, fifteen, thirty, and fifty-five minutes after use of the cement. Pulmonary artery pressure was recorded before the fracture was created (at four or twenty-four hours after induction of the pulmonary fat embolism), at the time that the fracture was created, immediately after fixation of the fracture, and one hour after fixation. A software program (Perisoft; Perimed) was used to compute the mean pressure of the pulmonary artery during one respiratory cycle.
Baseline measurement of arterial blood gas was performed on samples taken just after induction of anesthesia with the dog breathing room air. Measurements were also made before the long bones were reamed to induce pulmonary embolism; immediately after reaming of both the femur and the tibia; during pressurization of the cement; and at one, five, fifteen, thirty, and fifty-five minutes after the use of cement. Measurements of blood gas were repeated at four or twenty-four hours after fat embolization and again one hour after fixation of the fracture. Samples of arterial blood gas were analyzed with a blood-gas analyzer (model 178P8; Corning, Medfield, Massachusetts), and the pH, PCO2, PO2, HCO3-, and oxygen saturation were recorded. The alveolar-arterial PO2 gradient was calculated as the difference between the calculated alveolar oxygen tension and the measured arterial oxygen tension (PO2)28,52. Alveolar oxygen tension was determined according to the formula: FiO2 X (atmospheric pressure - vapor pressure of water) - PCO2/R, where FiO2 is the fractional inspired oxygen concentration and R is the respiratory quotient with a value of 0.8. The alveolar-arterial PO2 gradient is a measure of the amount of ventilation-perfusion inequality in a lung28,52. The alveolar-arterial gradient or alveolar-arterial PO2 difference is calculated by subtracting the arterial PO2 from the ideal alveolar PO2. The latter measurement is the PO2 that the lung would have if there were no ventilation-perfusion inequality and if it was exchanging gas at the same respiratory exchange ratio as the real lung52. An increased alveolar-arterial gradient usually is caused by an abnormally high ventilation-perfusion ratio within the lung52. An abnormally high ratio is related to the presence of an increased number of unperfused but ventilated alveoli28,52. This mismatch of ventilation and perfusion, or ventilation-perfusion inequality, causes relative hypoxemia and hypercapnia52.
Measurement of the Serum Level of Methylmethacrylate
Samples of pulmonary arterial blood were taken immediately after pressurization of the cement to measure the serum level of methylmethacrylate. The blood samples were stored frozen and thawed for the assays. Equal parts serum and acetonitrile were vortexed and centrifuged. The supernatant then was subjected to high-pressure liquid chromatography with use of a C18 column to detect methylmethacrylate monomer; detection was at 254 nanometers. No detectable methylmethacrylate monomer (limit of detection, 0.7 microgram per 100 milliliters) was found in any of the samples of pulmonary arterial blood.
Preparation and Analysis of Histological Specimens
After the dogs were killed, the lungs were removed with the heart en bloc and the major vessels were ligated. The lungs then were inflated with 100 per cent oxygen to a pressure of fifteen centimeters of water. To maintain inflation, an umbilical clamp was applied to the trachea. The lungs, brain, and kidneys were immersed in 10 per cent neutral buffered formalin for at least one week before histological processing.
At the time of processing, the gross specimens were washed with running water for twelve hours. Ten one-centimeter-thick sections were cut from the periphery of the lungs, with the assumption that the periphery probably would have the greatest concentration of fat emboli44, and from the brain and the kidneys. Two specimens each were taken from the right superior lobe, the right middle lobe, the right inferior lobe, the left superior lobe, and the left inferior lobe of the lung. Three specimens were taken from the frontal lobe of the brain; three, from the occipital lobe; and two, from both the basal ganglia and the brain stem. Ten specimens were taken from both the right and the left kidney. The heart of each animal was examined to exclude the presence of a patent foramen ovale or any right-to-left intracardiac shunt. No patent foramen ovale was found in any heart at the postmortem examination.
The specimens were placed in osmium tetroxide (BCN Chemicals, Beaconsfield, Quebec, Canada) for staining of fat8,12 and were shielded from light for at least two weeks. The specimens were processed, with use of a tissue-processor (Miles Canada, Etobicoke, Ontario, Canada), in sequential solutions, beginning with 50 per cent ethanol and ending with the tissue embedded in paraffin (Paraplast; Fisher Scientific, Unionville, Ontario, Canada). The specimens were stained with hematoxylin and eosin, cut with a microtome to five-micrometer sections, and mounted on slides (Permount; Fisher Scientific).
In order to rule out artefacts from the preparation, equivalent specimens were obtained from four animals that had been anesthetized and had had another operative procedure that did not involve pressurization of the medullary canal. Organs were removed immediately after the animal was killed. Specimens were prepared in the same manner as those from the experimental animals. No intravascular fat was seen in any specimen obtained from the four animals that did not have pressurization of the medullary canal. Intravascular fat was seen in all of the specimens obtained from the lungs, kidneys, and brain of the animals that had had reaming and pressurization of a long bone with cement.
Histological analysis was performed with use of a semiautomated image analyzer (model 2001; Leco, Mississauga, Ontario, Canada) on a personal computer system (model 486; Leco). We examined two fields per section of lung at a magnification of 100 times (Fig. 2-A). To account for differences in inflation of the lungs, the mean ratio of the percentage of the area of the field occupied by fat and the percentage of the area occupied by lung tissue was calculated. A similar procedure was performed on the specimens from the brain, except that four high-power fields per section were examined at a magnification of 200 times (Fig. 2-B). A previous investigator reported that fat emboli tend to be deposited in the capillary loops of renal glomeruli50. Each kidney was analyzed, at a magnification of 200 times, with respect to the amount of fat occupying a glomerulus (Fig. 2-C). Ten random glomeruli were examined in each of the ten sections from the two kidneys, for a total of 200 glomeruli. The total area of fat seen in the 200 glomeruli was divided by the total area of the glomeruli, thus quantifying the amount of glomerular fat.
Two fields per pulmonary section were examined at a magnification of 200 times for the presence of pulmonary edema. To quantify pulmonary edema, we used the scoring system of Turchin et al.48. A field with no fluid in any alveoli received a score of 0 points; a field with fluid in less than one-half of the alveoli, 1 point; and a field with fluid in more than one-half of the alveoli, 2 points. Therefore, the maximum possible score for pulmonary edema was 40 points (two times twenty fields per pulmonary specimen). Two fields per pulmonary section were examined at a magnification of 500 times for the presence of neutrophils.
Statistical Analysis
Data are reported as mean values and the standard error of the mean. Statistical analysis was performed with use of SAS software (version 5; Statistical Analysis System, Cary, North Carolina) on a personal computer (MAG 1450 Plus; Magnum International, Markham, Ontario, Canada). Analysis of variance with a split-plot design16,27 was used to compare data obtained one hour after fixation in the three subgroups that had fixation of the fracture at each of the two time-intervals after induction of fat embolism. Paired t tests were used to detect differences between data obtained after fixation and baseline data as well as data obtained after embolization within each subgroup. Moreover, the multiple-range tests of Tukey and Bonferroni were used to compare the data obtained after the different methods of fixation. The use of multiple comparison tests still is valid even when the F test for equality of different treatment effects reveals no significant difference17. The more powerful, sequentially rejective Bonferroni procedure also was used to determine the presence of significant differences between the various methods of fracture fixation at each of the two time-intervals17. The level of significance was a = 0.05.
Induction of Fat Embolism
In the control animals, pressurization of the medullary canal caused a significant increase (p < 0.03) in pulmonary artery pressure that persisted for at least fifty-five minutes. In the thirty-three experimental animals, pressurization of the medullary canal also produced a significant increase in the mean pulmonary arterial pressure, from 22.0 ± 0.9 millimeters of mercury (2.93 ± 0.12 kilopascals) to 29.5 ± 1.5 millimeters of mercury (3.93 ± 0.20 kilopascals) (p < 0.001). Pulmonary hypertension after pressurization of the canal persisted for at least fifty-five minutes, with a mean pulmonary arterial pressure of 28.0 ± 0.9 millimeters of mercury (3.73 ± 0.12 kilopascals) at the four or twenty-four-hour time-interval for all subgroups that had fixation of the fracture (Figs. 3 and 4 ). No significant difference was found between the control animals and the subgroups of experimental animals with respect to the baseline pulmonary artery pressure or the pressure at four hours or twenty-four hours after induction of fat embolism. Over-all, pulmonary artery pressure was greater at four hours than at twenty-four hours after embolization (p < 0.03) (Table I).
The alveolar-arterial PO2 gradient measured four hours after embolization was decreased compared with the baseline value in each subgroup to be treated with fixation of the fracture (p < 0.001) (Fig. 5). The animals that were assessed at four hours had remained under general anesthesia for approximately two hours after intramedullary pressurization and the induction of fat embolism to allow for measurement of pulmonary artery pressure and the alveolar-arterial PO2 gradient fifty-five minutes after embolization. Afterward, the femoral arterial catheter was removed, the femoral artery was ligated, and the left parapatellar incision was closed. As the animals were to be returned to the operating room within two hours, they were taken to the recovery room instead of their cages. Supplemental oxygen was provided during the recovery phase. The amount and effect of the supplemental oxygen, which was not measured until the standardized measurements were made at four hours, resulted in a spurious decrease in the alveolar-arterial PO2 gradients at that time. The animals that were assesed at twenty-four hours had been returned to their cages overnight after intramedullary pressurization and the induction of fat embolism. The alveolar-arterial PO2 gradients calculated twenty-four hours after embolic showering for the control and experimental animals were not significantly different from the baseline values, with the numbers available (Table II).
Fixation of the Fracture
Four-Hour Group
Pulmonary artery pressures measured before the creation and fixation of the fracture remained elevated compared with the baseline values (p < 0.0006) (Fig. 3). However, none of the methods of fracture fixation affected pulmonary artery pressure. Also, no difference was seen with respect to the pulmonary artery pressure between the control group and the experimental subgroups at the time that the animals were killed (Table I). The alveolar-arterial PO2 gradient measured one hour after fixation with a plate was not significantly different, with the numbers available, from the baseline value or the value determined four hours after embolization (Fig. 5). Nailing without reaming led to a higher alveolar-arterial PO2 gradient compared with the baseline value (p = 0.048) and with that calculated four hours after embolic showering (p = 0.03) (Fig. 5). Similarly, the alveolar-arterial PO2 gradient determined one hour after nailing with reaming was elevated compared with the baseline measurement (p = 0.02) and the value calculated four hours after embolization (p = 0.01) (Table II).
We compared the effect of the type of fixation on the alveolar-arterial PO2 gradient. One hour after nailing with reaming and nailing without reaming, the alveolar-arterial PO2 gradients were approximately four (p < 0.05) and 3.5 times higher, respectively, than that measured after fixation with a plate. No significant difference was found, with the numbers available, between the alveolar-arterial PO2 gradients calculated after nailing with reaming and after nailing without reaming. The alveolar-arterial PO2 gradient one hour after fixation with either of those methods was higher when the fixation had been performed at four hours than when it had been performed at twenty-four hours (p < 0.01).
One hour after fixation of the fracture, no difference was seen with respect to the amount of embolic fat relative to lung tissue among the animals in the control group and those that had application of a plate, nailing with reaming, or nailing without reaming. The scores for pulmonary edema were not significantly different among the subgroups that had fixation of a fracture, but they were greater than that in the control group (p < 0.01). Moreover, a maximum of only two neutrophils was seen in the twenty fields from each pulmonary specimen from the animals in the control group and from those in the subgroups that had fixation of the fracture. No hyaline membranes were seen at any time-period in any subgroup.
The amounts of embolic fat relative to glomerular area were similar among the control group and the subgroups that had fixation of the fracture (Table III). The quantities of embolic fat relative to brain tissue also were similar among the groups. No signs of ischemia or inflammation were seen in either the cerebral or the renal specimens.
The diameters of the pulmonary vessels occluded by fat varied greatly, but the vessels occluded by fat in the specimens from the kidneys and the brain had a more consistent diameter. Those vessels also had a smaller diameter than those in the lungs (p < 0.0001). The mean diameter of the glomerular vessels occluded by fat was 7.8 ± 0.4 micrometers (range, 4.9 to 13.5 micrometers), the mean diameter of the cerebral vessels was 4.2 ± 0.2 micrometers (range, 1.6 to 7.8 micrometers), and the mean diameter of the pulmonary vessels was 26.3 ± 1.6 micrometers (range, 6.0 to 80.4 micrometers).
Twenty-four-Hour Group
Twenty-four hours after induction of fat embolism, the pulmonary artery pressure was not significantly different from the baseline value, with the numbers available (Fig. 4). None of the methods of fracture fixation substantially increased the pulmonary artery pressure from the baseline value. Also, the pulmonary artery pressures measured at the time that the animals were killed did not differ substantially between the control group and the subgroups that had fixation of a fracture. The alveolar-arterial PO2 gradients measured twenty-four hours after induction of fat embolism were not significantly greater than the baseline values, with the numbers available. None of the methods of fracture fixation affected the alveolar-arterial PO2 gradient compared with the baseline value or with the value determined twenty-four hours after embolization. The alveolar-arterial PO2 gradients measured when the animals were killed were similar among the control group and the subgroups that had fixation (Fig. 6 and Table II).
After the animals were killed, no difference was found with respect to the amounts of embolic fat detected in the lungs relative to lung tissue among the control group and the subgroups that had fixation of the fracture. As in the four-hour group, the quantification of embolic fat in the kidneys and the brain did not show any differences among the various subgroups (Table III).
The scores for pulmonary edema were not significantly different, with the numbers available, among the subgroups that had application of a plate, nailing without reaming, and nailing with reaming (Table III). However, the scores for pulmonary edema in these subgroups were higher than that in the control group (p < 0.01). Scores for pulmonary edema over-all were less in the four-hour group than in the twenty-four-hour group (p < 0.03). No differences in neutrophil count were seen among the subgroups that had fixation of the fracture or between the findings in the four-hour group or the twenty-four-hour group.
The model used in the present study produces pulmonary fat embolism, pulmonary hypertension, hypoxemia, and impaired pulmonary gas exchange. The findings from this model cannot be applied to the development of adult respiratory distress syndrome in humans, but they can be used to study various interventions for the treatment of fractures associated with injury of the lungs secondary to fat embolism.
Reports in the literature on the fixation of fractures in multiply injured patients2,3,19,27,47,49 have indicated that early fixation of a fracture is important for reducing the prevalence of pulmonary complications. In our investigation, fixation of the fractures was performed early—that is, within twenty-four hours. Our results show that, in the setting of induced fat embolism, the method of fixation plays only a small role in the development of additional compromise of pulmonary function. The greater impairment of pulmonary gas exchange seen when fixation was performed four hours rather than twenty-four hours after induction of fat embolism does not imply that early fixation of a fracture is injurious. The effects of delaying fixation of a fracture have been shown to be detrimental to pulmonary function by causing additional embolization of marrow fat51. In the present investigation, the femora were not fractured at the time of induction of fat embolism. Therefore, the effect of continued motion at the site of the fracture with additional embolization of marrow fat was not seen. In addition, the effect of prolonged recumbency (which is necessary for fractures of the femoral shaft that are not stabilized) on pulmonary dysfunction did not occur in this model of fat embolism.
In the present study, application of a plate four hours after induction of increased pulmonary artery pressure had a less deleterious effect on the alveolar-arterial gradient than did nailing with reaming after the same time-interval. Other authors also have shown an early measurable effect on pulmonary function after insertion of an intramedullary device with reaming10. This finding was not demonstrated when fixation was performed twenty-four hours after the increase in pulmonary artery pressure. Despite these early differences in the alveolar-arterial gradient between the subgroups that had fixation four hours after the increase in pulmonary artery pressure, no differences were noted among these groups with respect to the pulmonary artery pressure, the score for pulmonary edema, or the amount of fat in the lungs, brain, and kidneys. Thus, there is little evidence over-all to suggest that clinical pulmonary decompensation occurs less often with application of a plate than with intramedullary nailing. Given the short (one-hour) duration of follow-up after fixation of the fracture, the alteration in the alveolar-arterial oxygen gradient may not have clinical relevance.
A previous study showed that insertion of an intramedullary nail without reaming in intact sheep femora increased the pressure in the canal by seventy millimeters of mercury (9.33 kilopascals) and did not cause substantial embolic showering51. Moreover, in a prospective, non-randomized study of multiply injured patients who had fracture fixation within twenty-four hours after the injury33, nailing without reaming produced fewer deleterious effects on pulmonary function than did nailing with reaming. In contrast, our investigation demonstrated that nailing without reaming affected pulmonary gas exchange to the same extent as did nailing with reaming performed four or twenty-four hours after the onset of pulmonary fat embolism.
The ability of the model to produce an elevation in pulmonary artery pressure reliably is important, as elevation of pulmonary artery pressure has been found to be a feature of acute respiratory failure, even in the absence of systemic hypoxemia58. Furthermore, an elevation of pulmonary artery pressure within two hours after multiple injuries was found to be a predictor of death45. Our investigation showed that, four hours after induction of pulmonary fat embolism, there was persistent pulmonary hypertension, which was not affected by any method of fixation. After fixation of the fracture, the pulmonary artery pressure decreased in all three subgroups. Twenty-four hours after induction of pulmonary fat embolism, the pulmonary artery pressure returned to the baseline value and again was not affected by any method of fixation of the fracture. The exact timing of the operation, four or twenty-four hours after the induction of fat embolism, is not an issue as fractures in multiply injured patients should be fixed as soon as possible after injury. The four and twenty-four-hour intervals served as points for differentiating the effects of fat emboli on the pulmonary vascular tree with time.
The embolization of marrow fat to major organs was quantified by histological examination of specimens from the lungs, brain, and kidneys. The lungs are the first organs to receive emboli from marrow fat released into the venous circulation. Fat emboli can deform and pass through the lungs over time5,41,48. They fragment and travel through the larger pulmonary vessels into the smaller ones40,41,48. The distribution of fat emboli as they travel through the lungs and into the systemic circulation depends on the distribution of cardiac output41, making the brain and kidneys more likely to receive emboli25,41. Our investigation did not reveal any significant differences with respect to the content of embolic fat in the specimens from the lungs, brain, and kidneys obtained one hour after fixation among the control group and the experimental groups, which had different timings and methods of fixation of the fracture. This finding shows that the initial embolic showering produced by the induction of fat embolism had a greater impact than did any additional embolization produced by fixation of the fracture. Moreover, creation of the femoral fracture required exposure of the lateral aspect of the femur to allow application of the bending jig. Therefore, open—not closed—intramedullary nailing was performed. This may have decompressed the femoral canal during nailing, thus diminishing the increase in intramedullary pressure and the subsequent embolization of marrow fat associated with nailing. Furthermore, the scores for pulmonary edema were similar between the four-hour and twenty-four-hour groups and among the subgroups that had different methods of fracture fixation. Very few neutrophils were found in the specimens from the lungs in any of the groups. The absence of both a significant pulmonary inflammatory response and quantitative differences in embolic fat content in the organs studied may have been the result of the relatively short (one-hour) duration of follow-up after fixation of the fracture. The inflammatory process may require a greater amount of time before it is manifested histologically.
The presence of fat emboli alone does not produce fat-embolism syndrome and acute pulmonary injury, but it can lead to hypoxemia and impaired pulmonary gas exchange. The quantity of embolic fat is not necessarily associated with the amount of decline in the function of the end organ14. Cofactors, such as the release of mediators as well as the presence of other injuries and hypotension, potentiate the development of adult respiratory distress syndrome13,19,31,57. Injury, especially multiple trauma, activates a systemic inflammatory response that produces elevated levels of mediators such as fibrinogen13,43, tissue thromboplastin11, prostacyclins53, cytokines23,29,46, and elastase29. The presence of cofactors may be more important than the presence of fat or the method of fixation of the fracture in the development of adult respiratory distress syndrome10. The release of these cofactors also may account for some of the deleterious effects on the alveolar-arterial gradient seen in the current study. The role of fixation of the fracture is controversial, as the findings of clinical studies have both supported and rebutted the claim that there is an increase in adult respiratory distress syndrome after intramedullary nailing of femoral fractures in patients who have concomitant pulmonary injuries7,38,49. We found no difference in the scores for pulmonary edema among the subgroups that had fixation of the fracture, and the substantial pulmonary inflammatory response typically seen with adult respiratory distress syndrome was absent. These findings suggest that the method of fixation of a fracture plays a minor role in the development of pulmonary dysfunction and cannot be related to the development of adult respiratory distress syndrome.
In conclusion, pulmonary fat embolism in a canine model can be used reliably to examine the pulmonary effects of fixation of a fracture and to study fat embolism pathobiologically. In the present study, fixation of a fracture produced no substantial evidence of acute inflammation of the end organ and had no effect on pulmonary artery pressure. The different methods of fixation had a similar effect on pulmonary function after treatment of the fracture.
NOTE: The authors thank Nancy Podworny for technical assistance.