Pediatric Radiology

, Volume 40, Issue 9, pp 1517–1525

Biphasic threat to femoral head perfusion in abduction: arterial hypoperfusion and venous congestion

Authors

    • Department of RadiologyComer Children’s Hospital
    • Department of RadiologyThe University of Chicago
  • Diego Jaramillo
    • Department of RadiologyChildren’s Hospital of Philadelphia
  • Neil Johnson
    • Department of RadiologyCincinnati Children’s Hospital
  • Kirk Doerger
    • Radiology Associates of Northern Kentucky
  • Christopher Sullivan
    • Department of SurgeryThe University of Chicago
Original Article

DOI: 10.1007/s00247-010-1602-1

Cite this article as:
Yousefzadeh, D.K., Jaramillo, D., Johnson, N. et al. Pediatr Radiol (2010) 40: 1517. doi:10.1007/s00247-010-1602-1

Abstract

Background

Hip abduction can cause avascular necrosis (AVN) of the femoral head in infants.

Objective

To compare the US perfusion pattern of femoral head cartilage in neutral position with that in different degrees and duration of abduction, testing the venous congestion theory of post-abduction ischemia.

Materials and methods

In 20 neonates, the Doppler flow characteristics of the posterosuperior (PS) branch of the femoral head cartilage feeding vessels were evaluated in neutral and at 30°, 45°, and 60° abduction. In three neonates the leg was held in 45-degree abduction and flow was assessed at 5, 10, and 15 min.

Results

Male/female ratio was 11/9 with a mean age of 1.86 ± 0.7 weeks. The peak systolic velocities (PSV) declined in all three degrees of abduction. After 15 min of 45-degree abduction, the mean PSV declined and showed an absent or reversed diastolic component and undetectable venous return. No perfusion was detected at 60-degree abduction.

Conclusion

Abduction-induced femoral head ischemia is biphasic and degree- and duration-dependent. In phase I there is arterial hypoperfusion and in phase II there is venous congestion. A new pathogeneses for femoral head ischemia is offered.

Keywords

Femoral head ischemiaAbductionAVNDoppler USVenous hypertensionChildren

Introduction

Abduction treatment in infants with developmental hip dysplasia (DDH) can cause AVN of the femoral head [1, 2]. Femoral head ischemia can also occur in a variety of childhood disorders such as Legg-Perthes, sickle cell disease, steroid therapy, synovitis, and storage diseases.

Currently, childhood and adult AVN are thought to be caused by venous hypertension of the femoral head and secondary arterial hypoperfusion [318]. The cause of AVN in abduction treatment for DDH, however, is presumed to be the stretching of the medial circumflex artery (MCA) [17] or complete obstruction of the profunda femoris artery [18].

In previous studies, gadolinium-enhanced MR imaging after extreme abduction has shown femoral head ischemia and enhancement defects, presumably caused by arterial compression and obstruction [1]. To the best of our knowledge, abduction-induced femoral head ischemia has not been linked to femoral head venous congestion and intracartilaginous venous hypertension. Because the draining veins always accompany the feeding arteries with a given magnitude of external pressure during abduction, the veins should occlude much sooner than the arteries because of their lesser vessel wall resistance and intraluminal pressure.

The purpose of this report is to compare the blood flow characteristics of the posterosuperior (PS) branch of the medial circumflex artery (MCA) in neutral position with the data obtained immediately after different degrees of hip abduction, and at different times after abduction.

Material and methods

Institutional review board approval and maternal consent were obtained and Health Insurance Portability and Accountability Act (HIPAA) regulations were observed. Volunteers were recruited from well-baby clinics and the neonatal nursery. The babies were bottle fed or nursed during the procedure; no sedation was used. No US contrast agents were used to enhance the Doppler signal or achieve greater visibility of the feeding vessels.

The chondroepiphyses of the proximal left femur in 20 African-American neonates, newborn to 11 weeks old, were prospectively studied and the flow characteristics of the posterosuperior (PS) branch of the MCA were evaluated. Only the spectral Doppler flow characteristics of the PS branch of MCA were studied because: (1) the PS branch is the most instrumental feeding vessel of the femoral head [1921], (2) the PS branch and its accompanying draining veins are anatomically situated so that they are most adversely affected during abduction, and (3) the PS branch is the most sonographically accessible vessel that can be studied both in neutral and abduction positions.

In neutral position, linear transducers of 8–13 MHz of an Acuson Sequoia 512 unit (Mountain View, CA, USA) were used by a single radiologist. During abduction, occasionally a sector transducer was used because of better fit. The PSV, the width of the systolic base, the characteristics of the diastolic flow, and the venous return were evaluated in neutral position and at three abduction angles. Power Doppler study of the femoral head, neck, and the greater trochanter was evaluated to assess the effect of abduction on overall perfusion of the entire chondroepiphysis.

After obtaining baseline data at 0-degree angle abduction using a goniometer, the leg was flexed 90° and abducted 30°, 45°, and 60° and the flow characteristics of the same artery were recorded. In three babies, legs were held at 45-degree angle of abduction and flow pattern was evaluated at 5, 10, and 15 min.

In a single case the color Doppler scale settings were lowered to illustrate the draining veins that accompany the PS branch of the MCA. Last, the angles of the arcs at the periphery of the left femoral head cartilage, between the lower edge of the limbus and the entry point of the PS branch, were measured in two babies.

Results

There were 11 boys and 9 girls with a mean age of 1.86 ± 0.7 weeks (range 0.1–11 weeks). Color and power Doppler imaging could illustrate the PS branch of the MCA in all 20 infants. For waveform analysis, the correction angle was <60-degrees in all cases.

In neutral position, the PS branch of the MCA was the most dominant feeding artery perfusing the femoral head epiphysis (Fig. 1), entering at the lateral aspect of the femoral chondroepiphysis shortly distal to the lower end of the triangular labrum (Fig. 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig1_HTML.gif
Fig. 1

Neutral position, epiphyseal perfusion. a Coronal view of the left femoral head demonstrates the dominance of the PS branch of the MCA in perfusion of the femoral head cartilage of a newborn (arrows) (reproduced with permission [21]). b Transverse image shows the dominant contribution of the PS branch of MCA (arrowheads) versus the smaller and less contributing branches of the PI branch with lesser dimensions and velocities (smaller arrows). The ossific nucleus can also be seen (large arrow) (reproduced with permission [21])

https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig2_HTML.gif
Fig. 2

Neutral position, distance between the lowest point of the labrum (L) and the entry point of the PS branch of the MCA (E), shown by the dotted line. The femoral head cartilage (H), the greater trochanter (T), and the PS branches to the head and greater trochanter are shown (arrows) (reproduced with permission [21])

The PSVs were 4–22 cm/s, with the lowest values in the first week of life, increasing with age up to 11 weeks, with a positive correlation value of 0.879 (Fig. 3). The systolic component had a broad base followed by sustained diastolic flow. The venous return was always present on spectral Doppler above the baseline with a slow velocity of 2–3 cm/s (Fig. 4).
https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig3_HTML.gif
Fig. 3

PSV in neutral position had the lowest values in the first week, increasing with age up to the 11th week

https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig4_HTML.gif
Fig. 4

Neutral position image shows the systolic flow with a broad base, positive diastolic flow, and slow and wavy venous flow velocity above the baseline (reproduced with permission [21])

Immediately after abduction, the peak systolic velocities declined at all three degrees of abduction (Fig. 5). The P values were 0.15 for 30-degree of abduction and 0.003 for 45-degree of abduction. Only three babies tolerated the 60-degree abduction, at which the PSV declined to 0–4 cm/s (Fig. 6).
https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig5_HTML.gif
Fig. 5

PSV is progressively declining but there is sustained diastolic flow immediately after 30°, 45° and 60° of abduction

https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig6_HTML.gif
Fig. 6

Comparison of PSV in neutral and three abducted positions shows an insignificant decline in PSV at 30-degree angle (P = 0.15) but a significant decline in PSV at 45-degree angle of abduction (P = 0.003)

In three babies, the leg could be held in abduction up to 15 min. Comparing to the baseline values in neutral position, the mean PSV value declined by nearly 20%, 40% and 100% by 5, 10, and 15 min, respectively, with a P value of 0.0015 (Fig. 7).
https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig7_HTML.gif
Fig. 7

Decline of femoral head PSV at 45-degree abduction with time. Steady and statistically significant decline of femoral head PSV at 45-degree abduction with time (P = 0.0015)

The systolic flow became narrow-based with needle-point systolic peak. The diastolic flow was either completely absent (Fig. 8) or it reversed and the venous return became undetectable (Fig. 9).
https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig8_HTML.gif
Fig. 8

Systolic and diastolic flow changes in 45-degree abduction. a Broad-base systolic and sustained diastolic flow (below the baseline) in neutral position. b Dampened PSV with a narrower base than in neutral position (A) and loss of diastolic flow 15 min after 45-degree abduction

https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig9_HTML.gif
Fig. 9

Systolic, diastolic and venous flow changes at 45-degree abduction. a Broad-base systolic and sustained diastolic flow in neutral position, below the baseline, and sluggish and wavy venous return above the baseline. b Narrowed systolic base, needle-point PSV below the baseline, and reversed diastolic flow above the baseline. The venous return was no longer detectable

Power Doppler of the entire femoral chondroepiphysis at 45-degree abduction showed perfusion void in the domain of the PS branch of the MCA. However, the domain of the posteroinferior (PI) branch of the MCA, the femoral neck and the greater trochanter remained perfused (Fig. 10).
https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig10_HTML.gif
Fig. 10

Loss of blood flow in the PS branch of the MCA at 45-degree abduction. Power Doppler image of the entire chondroepiphysis in 45-degree abduction shows cessation of the femoral head perfusion in the domain of the PS branch of the MCA. However, the domain of the posteroinferior (PI) branch of the MCA, the flow within the femoral neck and the greater trochanter remain intact

In a recent single case in which lower velocity settings for color Doppler were chosen, the draining veins were illustrated, either intimately next to the arterial branches or intertwined with them (Fig. 11). The angles of the arcs at the periphery of the left femoral head cartilage, between the lower edge of the limbus and the entry point of the PS branch, were 35° and 45°, respectively (Fig. 12).
https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig11_HTML.gif
Fig. 11

Relationship of draining veins to perfusing arteries. a Color Doppler image in a full-term baby in neutral position shows the PS branches of the left MCA (blue) accompanied by intimately close draining veins (red) with the scale set at 5.5 cm/s. b Color Doppler image in a full-term baby in neutral position shows the PS branches of the left MCA (blue) intertwined with intimately close draining veins (orange) with the scale set at 2.8 cm/s

https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig12_HTML.gif
Fig. 12

Relationship of perfusing arteries to labrum. a The angle of the arc (dotted line) between the lower edge of the labrum and the entry point of the PS is 45°. b The angle of the arc (dotted line) between the lower edge of the labrum and the entry point of the PS is 35°

Discussion

Femoral head ischemia after abduction is presumed to have an arterial pathogenesis, either by stretching of the MCA during non-extreme abduction [17] or by complete occlusion of the profunda femoris with extreme abduction [18].

Similar to our US experience, MRI has shown that the severity of cartilaginous ischemia increases with both degree and duration of abduction [1]. Using a piglet model with 70-degree abduction, segmental ischemia developed in 100% of the cartilaginous epiphyses and 85% of the physes. Marrow abnormalities were seen in 69% and the ischemia of the secondary ossification center was encountered in 56%. A smaller area of ischemia occurred in the posterior part of the femoral head and the opposing acetabular rim. Interestingly, no enhancement defects were seen if the abduction angle was less than 50º [1].

In the same piglet model, [18] angiographic studies showed that with extreme abduction, the profunda femoris artery becomes occluded completely, but not at a site where the MCA was presumed to be stretched [17]. The site of complete obstruction appeared to be too proximal to explain a localized segmental ischemia of the femoral head. Nonetheless, the experiment supported the arterial obstruction pathogenesis for femoral head ischemia after abduction.

To the best of our knowledge, the role of venous congestion in post-abduction ischemia of the femoral head has not been explored, despite the fact that intraepiphyseal venous hypertension is believed to be the cause of femoral head AVN in disorders such as Legg-Perthes, sickle cell anemia, steroid therapy, septic arthritis, synovitis, and storage diseases [318].

The rationale behind the present study was that if it takes an extreme abduction of 70° or more to obstruct the main feeding artery and cause ischemia, then less extreme but prolonged abduction must have ill effects on the drainage of the veins that, as illustrated, accompany the PS branch of the MCA. After all, given an equal external compression force, the veins should collapse and obstruct much earlier than the arteries.

Although gadolinium-enhanced MR illustrates enhancement defects far better than US, our US study had unique advantages in elucidating the mechanism of ischemia.

The zone of the perfusion alterations on US was limited to the domain of the PS branch of the MCA, while the rest of the chondroepiphysis remained perfused. The segmental nature of enhancement defects on MRI and perfusion defects on US indicates that neither the theory of the obstruction of the main artery [1, 18] nor Ogden’s [17] theory of the stretching or compression of the MCA branch can fully explain the cause of the ischemia. If Ogden’s theory were correct, the domain of the posteroinferior (PI) branch of the MCA should not have remained normally perfused in our experiments.

In this study, Doppler US offers a biphasic pathogenesis for explaining femoral head ischemia after abduction. In phase I, immediately after abduction, decreased PSV and maintained diastolic flow indicates arterial compression. However, in phase II, 15 min after abduction, the systolic base narrows with needle-pointed PSV and abolished or reversed diastolic flow, and undetectable venous return. This indicates increased intracartilaginous impedance secondary to venous hypertension.

Delayed conversion of the phase I arterial hypoperfusion to phase II of ensuing venous hypertension should not be surprising because the signs of venous congestion are always duration-dependent. In femoral head cartilage, histology shows venous lakes within the vascular canals that course through the femoral head cartilage that are disproportionally larger than the diameter of the arterial lumen (Fig. 13) [1, 17]. This excess venous capacity provides a reserve that protects the cartilage from sudden venous hypertension and tissue edema. When this protective venous capacitance is overcome over time, cartilage impedance caused by venous congestion is critically increased and the flow pattern converts from the phase I to the phase II pattern, seen 15 min after abduction in our study.
https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig13_HTML.gif
Fig. 13

Histological sections of the femoral head cartilage. a Photomicrograph of a cartilaginous vascular canal. Multiple vessels (v) are surrounded by fibrous stroma. Normal-appearing matrix and chondrocytes surround the canal. Toluidine blue stain (reproduced with permission [1]). b Cartilage near an ossification center. Note that the lumen of the vein (v) in cartilaginous vascular canal is disproportionately larger than the artery (a) (reproduced with permission [17])

We believe that the greater MRI enhancement defects of the femoral head in the delayed phase, the abnormal marrow signal, and late recovery after cessation of abduction previously reported [1] might all be secondary to duration-dependant intracartilaginous venous hypertension, which could not have been recognized as a second phase of femoral head ischemia by MRI.

Therefore, the cartilage ischemia after abduction is caused first by arterial obstruction, followed by further reduction of arterial perfusion, this time because of intracanalicular epiphyseal venous hypertension. Doppler flow pattern in the immediate post-abduction period was identical to that described following compression of the feeding artery, i.e. dampening of the PSV with sustained diastolic flow [22]. By 15 min and beyond, however, the Doppler pattern changed and signified increased tissue impedance and vascular bed resistance seen in venous hypertension.

On MRI experiments, enhancement defects were not seen in abductions <50° [1]. However, the PSV declined from baseline values in our Doppler US experiments with abduction angles of 30° (P = 0.15) and 45° (P = 0.003). Furthermore, at 45-degree abduction, power Doppler demonstrated a perfusion defect in the domain of the PS branch of the MCA and at 60-degee abduction spectral Doppler did not detect any measurable flow.

Last, unlike our US observation, the MRI literature does not show the dominant perfusion role of the PS branch of the MCA and does not single out this artery as the main target of abduction’s ill effects.

The dysplastic femoral head develops AVN only when abducted in cast, at a much higher rate than the opposite normal femoral head [2, 17] because the abduction-induced venous hypertension threatens the inherently underperfused dysplastic cartilage far more than a well-perfused normal femoral head (Fig. 14).
https://static-content.springer.com/image/art%3A10.1007%2Fs00247-010-1602-1/MediaObjects/247_2010_1602_Fig14_HTML.gif
Fig. 14

Epiphyseal cartilaginous perfusion, normal vs. dysplasia. a Plethora of various feeding vessels of the femoral chondroepiphysis in a normal newborn including the PS branch of the MCA (long arrow), the transphyseal vessels (short arrows) and feeding vessels of the greater trochanter (GT) (reproduced with permission [1921]). b Marked paucity of intracartilaginous feeding vessels in an 18-month-old dysplastic hip. The PS and PI branches of the MCA don’t advance into the femoral head cartilage (reproduced with permission [17])

Exactly what compresses the PS branch of the MCA and the accompanying draining veins has not been satisfactorily explained. Neither the theory of stretching or compression of the MCA in abduction [17] nor the complete obstruction of the profunda femoris [1, 18] explains the selective perfusion defects seen only at the domain of the PS branch of the MCA.

We believe that the compressing structure is the lower edge of the labrum ring. As demonstrated in neutral position, there is a safe but short distance between the lower edge of the labrum ring and the entry point of the PS branch. As the abduction rotates the femoral head counterclockwise, the entry point of the PS branch and the exit point of the accompanying veins will come in contact with and be compressed by the labrum edge. This fully explains both phases of arterial hypoperfusion and venous hypertension of the femoral head, selectively in the domain of the PS branch of the MCA in both our clinical Doppler US observations and prior MRI experiments.

In infants, avascular necrosis only develops with abduction greater than 55° [23]. This is more compatible with prior MRI experiments than our current US observations. This might be because decreased perfusion by Doppler US precedes actual clinical tissue ischemia and enhancement defects by MRI. The perfusion void on power Doppler might represent current equipment limitations in detecting very low-velocity flow. It is true that by spectral Doppler, the blood flow markedly decreased at 45-degree abduction but never ceased to exist. Therefore, perfusion defects by power Doppler forecast impending tissue ischemia before it is well established.

In treating hip dysplasia, the practical challenge is to anticipate and choose an ideal angle of abduction that minimizes the risk of redislocation yet carries the lowest risk of developing avascular necrosis of the dysplastic and the normal hip [24]. US-assisted measurement of the distance between the edge of the labrum and the point of entry of the PS branch might finally enable us to predict the safe angle of abduction before these two reference points come into contact with each other and compromise femoral head perfusion. According to our study, the least-threatening angle of abduction is 30°, followed by 45°. Sixty degrees of abduction inflicted the worst ischemia and should be abandoned.

One of the limitations of our study is that all US studies were performed by a single radiologist. This study constitutes preliminary research in a small number of newborns and a smaller number of those who could tolerate 15 min of abduction. The limiting factors were the IRB-imposed conditions, lack of the use of sedation, and no use of a contrast agent, as well as the inherent limitations of clinical investigations. Further studies in infants with hip dysplasia before and after abduction, perhaps assisted by the use of US contrast agents, are needed.

Conclusion

Doppler studies can successfully evaluate the blood supply to the femoral head cartilage in newborns and young infants without the use of US contrast agents or sedation. The perfusion defects of the femoral head cartilage can be demonstrated convincingly in varying degrees of abduction. The abduction-induced ischemia is in the domain of the PS branch of the MCA, with the rest of the femoral chondroepiphysis least affected. Decreased femoral head perfusion after abduction is biphasic, not monophasic as it was presumed to be. The structure that compresses the feeding artery and the draining veins might be the rim of the labrum. Doppler illustration of the cessation of blood flow during hip dysplasia treatment justifies cast adjustment with lesser degree of abduction; 30-degree abduction carries the lowest and 60-degree the highest regarding the threat of ischemia. Using US might enable us to predict how much abduction is safe before the two reference points come in contact with each other. Femoral head cartilage venous hypertension in abduction should be added to the list of childhood disorders in which AVN is believed to be caused by venous hypertension of the femoral head epiphysis.

Copyright information

© Springer-Verlag 2010