Introduction

As the musculoskeletal system is clearly designed for weight bearing, studies of intraosseous pressure (IOP) should be undertaken, not only under static conditions, but also under conditions of loading or during activity in vivo.^1,2 Denham, as well as others,^3-5 have calculated that a considerable force, perhaps five or more times body weight, is transmitted through weight bearing joints in use. Previous authors^6-8 have explored the possibility that bone might be hydraulically strengthened, but those tests used non-physiological techniques such as grease-saturated dry bones, or bone plugs in mechanical testing jigs.

A raised IOP has been associated with bone pain in osteoarthritis and osteonecrosis, and in vascular diseases such as osteonecrosis and Perthes’ disease of the hip in children. Embolic bone diseases, for example sickle cell and caisson, and storage diseases, such as Gaucher’s, have been reported to have a raised IOP.⁹ Alcohol, corticosteroid use and diabetes have also been associated with a raised IOP.¹⁰ Osteonecrosis incidence varies across cultures, but is said to affect 10 000 to 20 000 new patients a year in the United States. Worldwide, osteoarthritis affects the majority of the older population.¹¹

Since Clopton Havers (1657 to 1702)¹² first described canals in cortical bone, the anatomy of the blood supply in bone has been widely investigated. Trueta and Caladias¹³ and Brookes and Revell¹⁴ used a barium injection with decalcification. Others used microsphere¹⁵ and isotope¹⁶ studies. Experimental models of osteonecrosis or Perthes’ disease have been developed by obliterating the blood supply, by cryotherapy and by using steroids.¹⁷ Ficat¹⁸ developed a means of classifying and treating osteonecrosis, but did not consider the possible effect of weight bearing on IOP. The physiological relationship between subchondral IOP, loading and blood supply in normal healthy bone has received little attention.

The factors that usually control IOP physiology in normal bone, and whether IOP might change during ordinary activity, remain largely unknown. In 1964, Azuma¹⁹ reported fluctuating IOP in vivo at rest. More recently, a falling IOP has been noted in exsanguination studies.²⁰ It is possible that loading itself might change IOP in the subchondral bone below the cartilage.

In concurrent work using the same animal model, we found that at rest, in addition to a wave form synchronous with the arterial pulse, there is often an underlying wave synchronous with respiration. Intraosseous pressure also reflected other circulatory changes, for example those due to anaesthetic drugs. We found that in our model, IOP was not significantly affected by gender, weight, needle size or site of measurement. Technique affects IOP measurement with a prolonged recovery after saline injection but rapid recovery after aspiration. We confirmed that IOP is proportional to systemic blood pressure, and that whatever the initial or basal IOP value, the associated pulse pressure correlates well with IOP.²¹

Our hypothesis is that loading forces might be transmitted by hydraulic pressure through bone fat and marrow from the subchondral region to the trabeculae, and on to the cortical shaft during weight bearing. We therefore studied the effect of physiological loads on subchondral IOP. We then studied the effects combined with vascular occlusion. We also assessed possible compartmentalization within bones by saline injections.

Materials and Methods

Intraosseous pressure was measured experimentally in the subchondral bone of the femoral condyle and proximal tibia of five anaesthetized adult New Zealand White rabbits (two male, three female; Royal Postgraduate Medical School, Home Office licence ELA 24/4994). Induction of anaesthesia was by IV fentanyl (Sublimaze) 2 ml of 0.05 mg/ml solution, depending on the size of animal, with a top-up IV infusion of diazepam (Valium) 0.5 ml of 5 mg/ml solution, alternating with fentanyl (0.5 ml to 1.0 ml), given slowly on an approximately half-hourly basis.

The femoral vessels were exposed at the inguinal ligament. Sheathed ‘bulldog’ vascular clips could then be applied as required to the proximal femoral artery or vein. A 23G saline-filled venesection needle was pushed into the subchondral bone at the femoral condyle, or proximal tibia, percutaneously by rocking the needle through a 5° to 10° arc along the line of the bevel of the needle. Femoral head needle insertion was by an anterior approach dissection and direct puncture through the capsule.

A heparinized saline-filled line was connected from the intraosseous needle to a pressure transducer (Bell and Howell, Wheeling, Illinois or Druck PDCR75, Druck & Temperatur, Leitenberger, Germany) and to a four-channel chart recorder (Lectromed MX4P- 31, Jersey, United Kingdom). The transducers were calibrated on a 0 mmHg to 100 mmHg scale. Intraosseous pressure recordings were made at each site. For each set of recordings, the pressure transducers were zeroed to air, calibrated and kept level with the site measured.

Loading was carried out using a reversed spring pusher. A weighing spring and its calibration scale were mounted in concentric tubes, one end attached to the inside of each tube such that longitudinal compression of a set force could be applied. The inner tube contained and controlled the foot, preventing collapse into a flexed posture. The tube prevented movement of the hind limb and avoided direct squeezing of limb muscle. The subject was supine, with the leg extended. A steady load of one body weight (between 4520 g and 5400 g) was applied down the limb as in Figure 1. The loading recordings were carried out with the leg in an extended position and the pusher in place, without movement of the limb.

Fig. 1

Diagram showing experimental set up. The spring-loaded pusher covers the foot and prevents flexion and extension while applying a longitudinal one body weight load (4520 g to 5400 g; IOP, intraosseous pressure).

A series of experiments was undertaken to investigate the following:

Results

1a. The effect on IOP of loading by one body weight

We compared the basal IOP at the femoral condyle and proximal tibia. There was a wide range of values but no significant difference (p = 0.159, t-test) between them. The tibial and femoral results were therefore pooled.

When a load of one body weight was applied longitudinally without moving the limb, the IOPb (mean 11.7 mmHg, sd 7.1 mmHg) rose to IOPb+Ld (17.9 mmHg, sd 8.1 mmHg; n = 12, p < 0.0002, t-test), as in Figures 2 and 3 and Table I.

Fig. 2

Graph showing the basal intraosseous pressure (IOP) during perfusion at rest (IOPb), an increase in IOP with loading by one body weight (IOPb+Ld) p < 0.0002; IOP during proximal arterial occlusion (IOPa) and with loading during arterial occlusion (IOPa+Ld) p < 0.002 and IOP during venous occlusion (IOPv) and loading during venous occlusion (IOPv+Ld) p < 0.003; (all p-values t-test). Error bars are standard error of the mean.

Fig. 3

Graph showing the effect of loading alone with one body weight. Upper trace - femoral head, middle trace - femoral condyle and lower trace - proximal tibia. All traces shows basal intraosseous pressure, loading by one body weight for two minutes, rest for six minutes and load for three minutes. Vertical scale 0 mmHg to 100 mmHg on all traces, trace speed 12.5 mm/min.

Table I.

Values for intraosseous pressure (IOP) (mmHg) at 12 sites among five subjects. Columns are for basal IOP at rest (IOPb), with only loading (IOPb)+Ld (difference p < 0.0002), proximal arterial occlusion (IOPa), arterial occlusion with load (IOPa)+Ld (difference p < 0.002), venous (IOPv) occlusion and venous occlusion and load (IOPv)+Ld (difference p < 0.003; all t-test)

n = 12 for all	IOPb mmHg	IOPb+Ld mmHg	IOPa mmHg	IOPa+Ld mmHg	IOPv mmHg	IOPv+Ld mmHg
Mean	11.7	17.9	5.3	7.5	20.2	26.3
sd	7.1	8.1	4.1	4.5	5.8	6.3
Range	5 to 27	7 to 30	2 to 17	4 to 18	14 to 35	16 to 34

1b. The effect of arterial occlusion then loading

When a clip was applied to the proximal femoral artery, there was a fall in IOP (IOPb mean 11.7 mmHg, sd 7.1) to IOPa (mean 5.3 mmHg, sd 4.1; n = 12, p < 0.005, t-test). The fall in IOP was proportional to the initial or basal IOPb (Pearson correlation 0.81). Loading the limb under these conditions caused a small rise in IOPa to IOPa+Ld (mean 7.6 mmHg, sd 4.5; n = 12, p < 0.002, t-test) as in Figures 2 and 4 and Table I.

Fig. 4

Graphs showing the upper trace from the femoral head, middle trace femoral condyle and lower trace proximal tibia. From left to right, the traces show the effect of arterial occlusion, arterial occlusion with loading, load removal and arterial clamp removal. The second part shows changes at the same sites after venous occlusion, venous occlusion with loading, removal of load and removal of the occluding clamp. The intraosseous pressure changes are seen to be similar at all three sites and simultaneous. Vertical scale 0 mmHg to 100 mmHg on all traces, trace speed 12.5 mm/min.

3. Saline injection at different sites

Saline 0.5 ml was injected into the femoral head, femoral condyle and proximal tibia in turn. Pressure was not recorded at the injection site itself, but pressure was recorded at the other two IOP needles. There was a pressure rise in the femoral condyle when the femoral head was injected, and vice versa. Neither caused a rise in proximal tibial pressure. Proximal tibial injection caused no rise across the knee joint in the femur, as shown in Figure 5.

Fig. 5

Graph showing the upper trace from the femoral head, middle trace femoral condyle and lower trace proximal tibia. Left: injection (between the arrows) at the femoral head caused a rise in intraosseous pressure (IOP) at the femoral condyle, but not the proximal tibia. Centre: injection at the femoral condyle caused a rise in IOP at the femoral head, but not the proximal tibia (some artefactual shake is seen on the femoral condyle trace during injection). Right: injection at the proximal tibia had no effect on femoral head or femoral condyle IOP. Vertical scale 0 mmHg to 100 mmHg on all traces, trace speed 12.5 mm/min.

Discussion

We can find no previous references to direct IOP measurement during physiological loading, still less with and without vascular occlusion.²² Although individual measurements of IOP vary considerably, as other authors have noted,²³ we have shown that subchondral IOP is increased by physical load, providing the bone is perfused. The effect of loading during venous occlusion is more marked, and IOP may then even exceed the perfusion pressure. The pressure change is less marked during arterial occlusion. Subchondral bone appears to be acting as a perfused tissue, in an enclosed space, with additional IOP rises during weight bearing.

The finding that loading results in a marked increase in IOP suggests that there is transfer of force by hydraulic pressure from a slightly flexible subchondral plate through the perfused soft or semiliquid marrow fat contained between the trabeculae and onwards, to the rigid cortical shaft. It therefore suggests that not all of the load is transmitted through the trabeculae, but that a substantial proportion is transmitted by hydraulic pressure. This hydraulic load transfer would also be effective at dissipating the energy associated with impact loading, as proposed by Simkin,²⁴ and would reduce the risk of trabecular fracture.

Other authors who experimented with grease-saturated dry bones, or cancellous bone discs in compression jigs, concluded that load in bone was not transferred by hydraulic pressure.^7,8 We believe the reason why their conclusion was different from ours is that their bone was not perfused, and there was no pressure within the bone before load was applied. In our experiments, when the bone was perfused and there was normal IOP or elevated IOP resulting from venous inclusion, mechanical loading increased the IOP, substantially demonstrating transmission of load by hydraulic pressure. With arterial occlusion and a very low IOP, loading caused only a small increase in pressure, suggesting that only a small amount of load was transmitted hydraulically. Therefore, when there was no resting IOP, as in the grease experiments, we would expect no hydraulic load transmission. As a result, these non-physiological experiments support our conclusion that, in the normal physiological situation when bone is perfused, load is transmitted by hydraulic pressure.

With loading, the IOP rises, sometimes to above-perfusion pressure. It is therefore likely that there are circulatory adaptations to withstand these high pressures. Burkhardt²⁵ described histological features which we believe could protect the delicate subchondral capillaries and fat cells while they operate at rest with ordinary perfusion pressures, while at the same time allowing higher pressures to be transmitted through the tissues during weight bearing. These include ‘choke’ cells at 90° capillary branching points, and apparent valves or muscular cuffs at exit veins. In addition, there may be adaptations in vessels in the subchondral plane adjacent to the cortex where they exit bone to limit venous outflow under load to maintain IOP for load transmission.

The effects are seen simultaneously at different points in the limb as shown in Figures 3 and 4. The question arises as to the compartmentalization of bones or separation of one from another by joints. Our injection study demonstrated that while the load bearing pressure change is found throughout the bony weight bearing chain as shown in Figure 5, each bone forms a relatively enclosed and separate compartment. We found that pressure could be transmitted from one end of the bone to the other. As pressure did not cross the knee joint in either an antegrade or retrograde manner, as shown in Figure 5, we conclude that joints form a hydraulic barrier to pressure transmission. We can find no previous reference to suggest that load bearing increases IOP, which is transmitted through the bone, or that joints act as pressure transmission barriers. We note that intraosseous saline injection continues to be used clinically for resuscitation.^26,27 We injected relatively small volumes (0.5 ml) but consider that virtually any saline injection into the delicate bone vascular tree is potentially damaging. There is now some evidence that larger volumes of intraosseous saline injection may be harmful.²⁸

The clinical relevance of our work is that it shows that IOP is not a constant, as previously supposed, but fluctuates in vivo with local perfusion conditions at the needle tip. Load bearing causes a rise in IOP, with hydraulic pressure transmitted through subchondral bone, rather than the trabeculae alone taking the full load. Subchondral bone is not an inert region with a fixed IOP, but a slightly flexible structure which transfers load partly by hydraulic pressure through the soft or semifluid bone marrow to the cortical shaft or diaphysis. This potentially opens a new field of subchondral perfusion physiology. There are likely to be circulatory modifications to allow the relatively delicate subchondral capillaries and fat cells to cope with fluctuating high pressures during load bearing. Once the normal physiology of subchondral perfusion and load transmission is more clearly understood, a better appreciation of the vascular contribution to the pathology of bone diseases, such as osteonecrosis and arthritis, may follow. Future work should explore this vasculo-mechanical model of joint physiology.

There are limitations in this work. The animals were not identical in terms of age or weight. Needle placements could never be identical. Anaesthesia without specialized equipment in these subjects is recognized to be brittle. As the duration of each experiment varied, an average value for IOPb from early, middle and late in each subject was used. The vascular occlusion and loading values were recorded once only at the steady state achieved within 30 seconds of applying the clamp or load. For the main study of loading, we had five subjects with a total of 12 IOP recordings, from the femoral condyle and the proximal tibial. Although there was a wide range of values for IOP at both sites, we found no statistical difference between the femoral condyle and the proximal tibial sites, thus we used data from both. This reflects the clinical situation where IOP may be measured at virtually any site.

In conclusion, our work demonstrates that IOP is not a constant, but is reduced during proximal arterial occlusion, and increased with proximal venous occlusion. Loading increases IOP, whatever the perfusion state. With loading, force is transferred through fluid marrow and fat constrained within the bone. Load is thereby transmitted partly by means of hydraulic pressure from the subchondral plate to the diaphyseal shaft. In addition, we suggest that each bone represents a relatively enclosed space, through which pressure is transmitted, and that joints act as hydraulic barriers to pressure transmission.