Pulmonary Embolism in Cemented Total Hip Arthroplasty
Michael Clarius, Christian Heisel, Steffen J. Breusch
Summary
Embolism is a well-known complication of cemented to- tal hip arthroplasty (THA). As a result of manipulations of the medullary cavity, the intramedullary pressure rises and fat, bone marrow and air embolises into the venous system and to the lung. Clinically, this is seen as acute hypotension, which can go as far as cardiac failure. Al- though a fatal outcome is rare, fat embolism is a serious complication. The most effective prophylactic measure to reduce the risk is a thorough lavage of the femoral cavity. The use of pulsatile jet-lavage can be regarded as an obligatory preparatory procedure before cement ap- plication. Fat and bone marrow are removed as potential embolic sources and, further, the cement-bone interface is enhanced.
Introduction
Deep vein thrombosis (DVT) and pulmonary embolism (PE) are feared and well known complications of THA [54]. Fat embolism comprises an entity amongst the group of PE and usually occurs in the early perioperative phase and is unrelated to DVT. The incidence of postoperative DVT as assessed by phlebography is reported in literature to be as high as 77% [68], although, in clinical practice, it is frequently underestimated and consequently not diag- nosed. An important and life-threatening complication of DVT is secondary PE with an incidence estimated to be between 6–33% [28, 43, 68] confirmed by lung perfusion- and ventilation scans. Although effective anti-embolic and DVT prophylactic measures have been able to reduce these figures significantly, DVT and PE are still the most common causes of death after THA [54, 93].
Consequences of Pulmonary Embolism
Pulmonary embolism leads to an increase in pulmonary vascular resistance. The blockage of the pulmonary vas- cular system results in an increased AV-shunt and in pulmonary hypoperfusion [49, 60, 95]. The acute pressure increase in the pulmonary arteries [16, 17] causes the functional and structural alteration of a cor pulmonale to occur resulting in left ventricular hypovolaemia and a reduced cardiac output [1, 91]. Clinically, this is seen as acute hypotension, which can go as far as cardiac failure.
Cardiopulmonary Complications and Death During Total Hip Arthroplasty
Soon after introduction of methylmethacrylate (MMA)
as bone cement towards the end of the 1960s, reports of
adverse intraoperative complications were published and
associated with the use of bone cement. Sir John Charnley
[18] himself, who inaugurated MMA in orthopaedic sur-
gery observed a drop in blood pressure immediately after
implantation of the endoprosthesis in many of his patients
for up to 5 minutes which was more pronounced during
implantation of the stem than the cup. Some authors re-
ported instances of intraoperative cardiac arrest [18, 25,
27, 66, 67, 69, 78, 88, 98], which were fortunately manage-
able with resuscitation. Such instances of so-called »car-
diac arrest syndrome« [88] as a reaction to implantations
of cemented implants was not, however, always reversible
and a number of intraoperative deaths were reported in
literature ( ⊡ Table 15.1 ), mostly in patients with hip frac-
ture. At autopsy 15/49 cases showed extensive pulmonary
fat embolisation. An exact microscopic examination of
lung tissue revealed an additional bone marrow emboli-
⊡
Table 15.1. Intraoperative cardiac arrests and deaths during total hip replacement reported in literatureAuthor Diagnosis Cardiac Arrest
Syndrome
Death Autopsy Cause of Death
Charnley 1970 no information 4 2 – –
Powell et al. 1970 hip fracture 3 – – –
Hyland, Robins 1970 hip fracture 1 1 1 air and fat embolism
Burgess 1970 hip fracture 1 1 1 fat embolism
Gresham et al. 1971 hip fracture 2 2 2 fat embolism
Schulitz et al. 1971 OA 3 2 1 fat embolism
Thomas et al. 1971 hip fracture 1 1 – –
Cohen, Smith 1971 Revision 1 1 1 fat and bone marrow embolism
Phillips et al. 1971 hip fracture 1 1 1 fat and bone marrow embolism
Dandy 1971 hip fracture 4 2 2 fat embolism
Michelinakis et al. 1971 hip fracture 2 – – –
Sevitt 1972 hip fracture 2 2 2 fat embolism
Kepes et al. 1972 hip fracture 2 2 1 bone marrow embolism
Peebles et al. 1972 hip fracture 1 1 – –
Newens, Volz 1972 hip fracture 1 – – –
De Angelis et al. 1973 Revision, OA 2 – – –
Nice 1973 hip fracture 1 1 – –
Milne 1973 OA 1 – – –
Tronzo et al. 1974 no information 1 1 –
Jones 1975 no information 1 1 – fat embolism
Hyderally, Miller 1976 path. hip fracture 1 1 – –
Beckenbaugh, Ilstrup 1978 no information 1 1 – –
Engsaeter 1984 no information 1 1 1 fat embolism
Zichner 1987 no information 10 3
Hochmeister et al. 1987 hip fracture 1 1 1 fat and bone marrow embolism
Maxeiner 1988 hip fracture 3 3 1 fat embolism
Egbert et al. 1989 path. hip fracture 1 1 1 fat and bone marrow embolism
Patterson et al. 1991 hip fracture 7 4 1 no evidence of embolism
Bogner, Landauer 1991 hip fracture 10 10 1 no evidence of embolism
Pietak et al. 1997 hip fracture 2 2 2 fat and bone marrow embolism
Tsujitou et al. 1998 no information 2 2 2 fat embolism
Parvizi et al. 1999 17 HF., 4 OA, 1 RA, 1 NU
23 23 13 11/13 bone marrow, 3/13 bone
cement
Ortega et al. 2000 no information 5 1 1 fat embolism
Fallon et al. 2001 path. hip fracture 1 1 1 fat embolism
Leidinger 2002 hip fracture 12 12 12 1x fat embolism, 11x right heart
failure
total 115 87 49 –
HF hip fracture, OA osteoarthritis, RA rheumatoid arthritis, NU non union.
sation alongside fat embolisms in 18 patients whereby the particles were found to even contain cancellous bone fragments.
Intraoperative Mortality
Intraoperative mortality during hip replacement was eval- uated by Parvizi [77] in a large retrospective study involv- ing 38,488 patients and found to be at around 0.06%. One particularly risk-laden group was the patient with fracture of the femoral neck with an intraoperative mortality of 0.18%, those with pertrochanteric fracture and cemented hip replacement suffered a 1.6% risk. Leidinger [61], though, published data in 2002 from 150 patients with fracture of the femoral neck for whom the intraoperative mortality was found to be much higher at 8%. Elective THA carries a risk well below 0.1%.
Diagnosis and Visualisation of Intraoperative Embolism
The intraoperative complications forced anaesthetists to monitor patients very carefully when hip replacement was undertaken. During intraoesophagal cardiac auscultation so-called »mill-wheel-murmurs« could be detected dur- ing prosthesis implantation [70]. These phenomena were accompanied by hypotensive episodes and blood aspira- tion via a Swan-Ganz catheter allowed Jones to establish the first diagnosis of fat embolism in 1975.
Visualisation of intraoperative embolism was first achieved using echocardiography [20, 22, 44, 92, 109].
The development of transoesophageal echocardiography (TEE) allowed a continuous high-quality monitoring of the patient intraoperatively without complicating influ- ences such as breathing movements, adipositas or emphy- sema, because it is benefited by the close proximity of the oesophagus to the target organ the heart ( ⊡ Fig. 15.1 ).
Continuous echocardiographic monitoring during cemented hip replacement operations allowed an assess- ment of the relative frequency of intraoperative embolic events [22, 41]. Therefore TEE is regarded as the most sensitive method to detect intraoperative embolism [39], although the less invasive transcranial Doppler has been advocated.
Pathomechanism Cement »Toxicity«
Many attempts were made to explain the implantation syndrome [90]. Initially, bone cement was seen as the prime culprit and cause of cardiovascular complications.
Frost [37] suspected the heat emission during polymerisa- tion as the primary cause of the observed losses in pres- sure. Other authors postulated a connection between the acute onset of hypotension and the release of monomeric MMA, which was assumed to cause peripheral vasodilata- tion and bradycardia accompanied by a negative inotropic effect when released during the hardening process [34, 53, 58, 62, 66, 69, 72, 79, 97, 104]. Radioactive marking of the monomer proved the influx of methylmethacrylate into the blood circulation, which permitted in vivo MMA measurements [50, 65].
Hypotension could be reproduced in animal models after intravenous injection of the monomer with a dose dependent relationship. Intraoperative measurements in patients showed in vivo concentrations from 0.3–5.9 mg/100 ml [50, 58, 65, 103, 114], but Bright [14] found that a measurable hypotensive effect was only attained at concentrations around the tenfold of those achieved in humans. Death occurred in the animal model after injection of doses correlating with the hundredfold MMA concentrations (125 mg/100 ml) [50]. However, clinically a statistical correlation between measured MMA concen- trations and pressure drop could not be demonstrated.
15
⊡ Fig. 15.1. Normal cross-sectional view of the heart as seen during
transoesophageal echocardiography
Not just the monomers, but also initiators of the dimethyl-p-toluidin type were accused of causing acute hypotensive crises. But Schlag [96] was able to demon- strate that initiators of this type were totally depleted dur- ing the polymerisation process.
It is therefore important to realise the cement itself is not the »toxic culprit«.
Intramedullary Pressure
Another line of argument in the explanation of the im- plantation syndrome was the pathogenetic intravasation of air, fat and bone marrow components and the autopsy findings mentioned above appeared to support this. Caus- al in this argument is the increase in intramedullary pres- sure during implantation of the prosthesis [19, 26, 40, 80, 94]. Experiments involving femora from deceased donors were able to show very high intramedullary pressure peaks of up to 1447 kPa as can be seen in ⊡ Table 15.2 . Values of up to 250 kPa were attained during intraoperative pressure measurement ( ⊡ Table 15.2 ).
The increase in intramedullary pressure during cement insertion and stem implantation causes an influx of air [2, 3, 5, 24], fat and bone marrow fragments via the linea aspera [31] into the venous system [45, 112]. A pro- portion of these particles embolise swiftly, the remainder adhere to vessel walls and initiate the development of a mixed thrombus [114]. It was Pelling and Butterworth [80], who finally proved the mechanism of fat displace- ment from the femur as the decisive mechanism. There was no difference in the extent of fat embolism between bone cement, bone wax and simple dough.
Irrespective of the operative approach chosen, the femoral vein is subject to a significant torsion as a result of flexion and rotation of the leg and this can lead to tem- porary total occlusion of the vessel [101]. Intraoperative phlebography has been able to confirm this phenomenon [87]. Such occlusion implies a complete cessation of blood flow, torsion of the femoral vein leads to damage of the endothelial wall, the trauma of the operation itself causes a higher propensity for blood to coagulate and thus the triad of factors postulated by Virchow in 1856 for the develop- ment of venous thrombus is fulfilled. This shows that the high incidence of postoperative deep venous thrombosis of the leg after hip replacement as well as the subsequent pulmonary embolic events can be explained to a high degree by the operative method itself. A further conse- quence of the operative method is that the act of relocation of the hip implies also the re-canalisation of the femoral vein from which the newly formed thrombi are released into the venous system [22]. The danger of cardiopulmo- nary complications rises and it becomes apparent why this moment can be regarded as particularly dangerous, even more so, as it closely follows the most critical moment, that of cement and stem implantation.
Multicentre Study
Our own investigations within the framework of a multi- centre study included 96 cemented THA patients, which were continuously monitored using TEE. We were able to identify several intraoperative steps that were followed by embolic events. ⊡ Figure 15.2 shows such a risk profile during a total hip replacement operation.
Our studies revealed that femoral stem implantation and relocation of the hip (i.e. unkinking of the femoral vein) were the most embolism-prone operative steps.
A so-called »snow flurry« ( ⊡ Fig. 15.3 ), which is the echocardiographic correlate for an air embolism, was seen in virtually all patients immediately after cement and stem implantation. Diagnostic signs such as these are then frequently followed by a manifest embolus ( ⊡ Fig. 15.4 ), which can then circulate in the right-side of the heart for up to 2 minutes before being swept out into the blood ves- sels of the lung ( ⊡ Fig. 15.5 and 15.6 ).
⊡
Table 15.2. Intramedullary pressure during implantation ofthe femoral prosthesis
Author Pressure (kPa) Method
Without Drill Hole
With Drill Hole
Ohnsorge 1971 455 262 cadaver femur
Phillips et al. 1973 250 intraoperative
Hallin 1974 89 intraoperative
Breed 1974 114 13 animal model
Tronzo et al. 1974 75 intraoperative Kallos et al. 1974 118 32 animal model von Issendorff, Rit-
ter 1977
504 96 cadaver femur
Indong et al. 1978 1282 cadaver femur
Drinker et al. 1981 1243 animal model
Engsaeter et al.
1984
76 4 intraoperative
Orsini et al. 1987 223 animal model
Wenda et al. 1988 131 intraoperative
Song et al. 1994 488 cadaver femur
Yee et al. 1999 1052 cadaver femur
McCaskie et al. 1997 157 667
intraoperative cadaver femur Reading et al. 2000 131 cadaver femur
Dozier et al.2000 486 cadaver femur
Churchill et al. 2001 1447 cadaver femur
15
preparation jet-lavage cup- (n=14) cup-trial cup-
implantation positioning of the leg stem-
preparation plug-
insertion stem-trial stem-
implantation relocation R1
11,6 7,1
1,1
29,8
9,6 13,8
1,1 3,2
72,3 70,2
0 10 20 30 40 50 60 70 80
percentage
embolic events during cemented total hip replacement
⊡ Fig. 15.2. Time and operative step dependent embolic events during cemented THA (n = 96)
⊡ Fig. 15.3. »Snow flurry«, the echocardiographic correlate of air embolism
⊡ Fig. 15.4. Filamentous embolus in the right atrium
⊡ Fig. 15.5. Massive embolus circulating in the right heart for more
than 40 seconds after trial reduction
»Snow flurries« were observed in 85% of stem imp- lantations and emboli similarly, in 72%. Some very small emboli occurred during pulsatile lavage of the acetabulum in 1/14 patients [22]. At the moment of relocation of the implanted components, 93% of patients who were subject to echocardiographic monitoring displayed »snow flurries«, while embolic events could be found in 70%. It is, however, of note that no pulsatile lavage had been used in 82/96 pati- ents in this multicentre study, hence representing a worse case scenario.
Cementing Techniques Influencing Intraoperative Embolism
There are several factors influencing intramedullary pressu- re and therefore intraoperative embolism. In a new animal
model we were able to compare certain methods and ope- ration/cementing techniques and it was possible to quan- tify the amount of fat and bone marrow intravasation.
Animal Model
A new sheep model was developed ( ⊡ Fig. 15.6 ), which allowed for standardised bilateral, simultaneous cement pressurisation. The operative procedure involved bilat- eral placement of intravenous catheters into the external iliac veins via retroperitoneal approach ( ⊡ Fig. 15.7 ). A specially designed cementing apparatus was used to allow for bilateral simultaneous cement pressurisation. Venous blood from both iliac catheters was then collected during cementing ( ⊡ Fig. 15.8 ), anticoagulated and a quantitative and qualitative fat analysis was performed.
Cement pressurization apparatus
Reducing valves and manometers Blood
collection in aliquots with sodiumcitrate Catheters in the
external iliac veins
⊡ Fig. 15.6. Schematic drawing of the sheep model which allows for bilateral simultaneous cement pressurization and collection of blood via the external iliac veins
⊡ Fig. 15.7. Intraoperative view of the retroperitoneal situs. On the left the ballooned iliac vein after proximal ligation is shown, on the right the
catheter is placed and secured
cement penetration under in vivo bleeding conditions despite adequate and sustained pressurisation. Cement application at a higher viscosity state carried a higher risk of fat embolism but resulted in improved cement inter- digitation – a finding in contrast to many in-vitro cement studies! Cement applied at low viscosity state seems to take the path of least resistance into the venous system ( ⊡ Fig. 15.10 ) before deeper cement penetration into bone can occur [8]. Accordingly, three cases have been reported in literature of intraoperative death, where bone cement was found in the lung at autopsy [77].
Influence of Pressure and Pressurisation
Intramedullary, pressure is dependent upon a number of factors such as the magnitude of pressure, the duration of pressurisation, the cement viscosity as well as the stem de- sign and the relationship between the prosthesis and the medullary cavity [21]. Tapered stem designs resulted in significantly higher pressures and higher intrusion rates.
The larger the prosthesis in relation to the medullary cav-
15
⊡ Fig. 15.8. Animal model setup during operation. On top cement apparatus with bilateral simultaneous pressurisation is shown. Via the retroperitoneal iliac vein catheter the draining blood could be collected during cementing to allow determination of the fatty contents
Influence of Pulsatile Lavage
Using the sheep model described above, we studied the effectiveness of both pulsatile and syringe lavage of equal volume with regard to their cleansing capabilities as mea- sured by fat and bone-marrow intravasation. After ran- domisation, one side was lavaged with 250 ml irrigation using a bladder syringe, the contralateral femur with the identical volume but using a pulsatile lavage. De- spite equal volume manual lavage produced significantly higher fat and bone marrow intravasation (p <0.001) than pulsatile lavage ( ⊡ Fig. 15.9 ) [12].
It is well known that lavage is necessary to clean the cancellous bone in order to obtain a strong bone-cement interface [11, 13]. Further, our model was able to dem- onstrate, that not only the volume but also the quality of bone lavage is an essential factor influencing the risk of fat embolism and adverse cardiorespiratory effects.
Manual lavage carries a significantly increased risk of fat and bone-marrow intravasation [12].
Influence of Cement Viscosity
Pressurisation is necessary in order to force the cement into the bone. Bone cements displace blood and bone marrow when pushed into the medullary cavity. Little is known about the in vivo impact of cement viscosity on cement interdigitation and fat embolism.
Using this model, we also studied the influence of dif- ferent cement viscosities on fat intravasation and cement penetration in vivo [8]. As high viscosity cement we used Palacos in comparison to low viscosity Osteopal cement.
In this model, low viscosity cement yielded signif- icantly lower rates of microradiographically measured
⊡ Fig. 15.9a,b. Macroscopic appearance of blood samples collected.
a More supernatant fat on the blood surface in the syringe lavage group
(right) than in pulsatile lavage group (left). b Higher magnification and comparison of pulsatile (left) versus manual lavage (right) groups
a
b
ity, the higher is the measured pressure and the amount of embolised material. Large stem sizes should be inserted slowly and particular emphasis should lie on prior me- ticulous canal cleansing with pulsatile lavage.
Influence of a Cement Restrictor
Contemporary cementing techniques include the use of a cement restrictor to occlude the intramedullary canal to allow for cement containment, cement pressurisation and improvement of cement interdigitation [73]. These intramedullary plugs seal the distal medullary cavity and therefore lead to higher intramedullary pressures during pressurisation and implantation of the femoral prosthesis.
A good cementation result requires the ability of the re- strictor to withstand displacement forces during cementa- tion [47]. In order to get a good fit in the femur, it is usually necessary to advance the plug distally with force, resulting in high intramedullary pressure peaks leading to pulmo- nary embolisation [22]. Obviously, there are differences between the designs of the cement restrictors. Our own investigations in cadaver femora, showed that expandable restrictors have favourable characteristics and the potential to reduce the risk of fat embolism [10]. In a further in-vivo animal study [46], utilising the same principle of venous blood collection and fat sampling, we found in eight of thirteen evaluated animals a peak in the fat intravasation caused by the application of the cement restrictor.
Our results emphasise the importance of a thor- ough preparation of the intramedullary canal, particularly when cemented fixation is performed. As a consequence of these studies we concluded that pulsatile lavage, which
should be considered a mandatory standard in cemented THA, should be implemented prior to the insertion of the cement restrictor (and even templating for it!) in order to further reduce the risk of fat embolism.
Influence of Femoral Cement Filling Technique In the early years of hip replacement, cement was intro- duced in an antegrade fashion from the proximal end into the femoral cavity before insertion of the femoral component. Charnley’s »thumbing« method induced in- tramedullary pressure peaks with cement alone [73, 74].
Use of a venting tube during cement introduction was able to reduce the pressure peaks [26, 74], it did, however, cause cement mantle defects [30].
Glen proposed as early as 1970, that holes could be drilled into the distal femur in order to relieve pres- sure build-ups [38]. This enabled a fourfold reduction in intramedullary pressure [7, 56, 74, 111] and furthermore provided a route via which material from within the mar- row cavity could escape. Numerous animal models [7, 56]
and clinical studies [44, 108] have established the efficacy of the distal venting hole.
The logical continuation of this concept was the vacu- um technique according to Draenert [31]. The distal vent- ing hole, as stipulated by many authors, is connected to a vacuum and thus sucks the cement into the cancellous bone and, additionally, should serve to reduce pressure during cement and stem implantation. A further cannula located proximally in the fossa piriformis was intended to improve the cement filling of cancellous honeycomb structures and at the same time reduce intramedullary
⊡ Fig. 15.10. a Postoperative radiograph taken after implantation and pressurisati- on of low viscosity bone cement showing cement intravasation into the venous sys- tem. b After dissection of the specimen the full extent of cement escape is revealed
a b