1298 THE JOURNAL OF BONE AND JOINT SURGERY

The surgical anatomy of the blood supply to the femoral headDESCRIPTION OF THE ANASTOMOSIS BETWEEN THE MEDIAL FEMORAL CIRCUMFLEX AND INFERIOR GLUTEAL ARTERIES AT THE HIP

A. W. Grose, M. J. Gardner, P. S. Sussmann, D. L. Helfet, D. G. Lorich

From the Hospital for Special Surgery, New York, USA

A. W. Grose, MD, Assistant Professor of Orthopaedic SurgeryNew York Medical College, Westchester, 95 Grasslands Road, Valhalla, New York 10595, USA.

M. J. Gardner, MD, Orthopaedic Trauma FellowHarborview Medical Center, 325 9th Avenue, Seattle, Washington 98105, USA.

P. S. Sussmann, MD, Orthopaedic Trauma SurgeonOrthopaedic DepartmentUniversitätsklinik Balgrist, University of Zurich, Forchstrasse 340, 8008 Zurich, Switzerland.

D. L. Helfet, MD, Director of the Orthopaedic Trauma Service, Professor of Orthopaedic Surgery

D. G. Lorich, MD, Associate Director of the Orthopaedic Trauma Service, Assistant Professor of Orthopaedic SurgeryThe Hospital for Special Surgery and New York Presbyterian Hospital, 520 East 70th Street, New York 10021, USA.

Correspondence should be sent to Professor D. L. Helfet; e-mail: helfetd@hss.edu

©2008 British Editorial Society of Bone and Joint Surgerydoi:10.1302/0301-620X.90B10. 20983 $2.00

J Bone Joint Surg [Br] 2008;90-B:1298-303.Received 28 February 2008; Accepted 23 April 2008

The inferior gluteal artery is described in standard anatomy textbooks as contributing to the blood supply of the hip through an anastomosis with the medial femoral circumflex artery. The site(s) of the anastomosis has not been described previously. We undertook an injection study to define the anastomotic connections between these two arteries and to determine whether the inferior gluteal artery could supply the lateral epiphyseal arteries alone. From eight fresh-frozen cadaver pelvic specimens we were able to inject the vessels in 14 hips with latex moulding compound through either the medial femoral circumflex artery or the inferior gluteal artery. Injected vessels around the hip were then carefully exposed and documented photographically.

In seven of the eight specimens a clear anastomosis was shown between the two arteries adjacent to the tendon of obturator externus. The terminal vessel arising from this anastomosis was noted to pass directly beneath the posterior capsule of the hip before ascending the superior aspect of the femoral neck and terminating in the lateral epiphyseal vessels. At no point was the terminal vessel found between the capsule and the conjoined tendon. The medial femoral circumflex artery receives a direct supply from the inferior gluteal artery immediately before passing beneath the capsule of the hip.

Detailed knowledge of this anatomy may help to explain the development of avascular necrosis after hip trauma, as well as to allow additional safe surgical exposure of the femoral neck and head.

It is accepted that the primary blood supply tothe femoral head in the adult is the terminalextent of the deep branch (also known as theramus profundus) of the medial femoral cir-cumflex artery.1-3 Standard textbooks describethe additional contribution of the cruciateanastomosis to the blood supply to the hip ingeneral.4,5 The vessels contributing to thisanastomosis have been variously reported asincluding the obturator, the lateral femoralcircumflex, the inferior gluteal, the superiorgluteal and possibly the first perforating arter-ies in addition to the medial femoral circum-flex artery.1,2,4,6,7 Recently, Gautier et al2

reported that the inferior gluteal portion ofthis anastomosis was “constant” and couldpotentially provide adequate blood to the fem-oral head if the medial femoral circumflexartery was damaged. Despite the potentialimportance of this portion of the cruciateanastomosis, there remains no detaileddescription of either the course of the inferiorgluteal artery in the trochanteric region or ofconnections between it and the medial femoralcircumflex artery.

We performed a series of injections into themedial femoral circumflex artery and inferiorgluteal artery in an attempt to delineate pre-cisely the site or sites of connection betweenthe deep branch of the medial femoralcircumflex artery and the inferior glutealartery.

Materials and MethodsWe obtained eight fresh-frozen pelves withtheir hips (Anatomic Gifts Registry, Hanover,Maryland), none of which were known to havehip disease. Two specimens were shared withother researchers, in which only one single sidewas injected and dissected in each. Bilateralstudies were carried out in the six remainingspecimens. Thus a total of 14 investigationswere performed.

Vascular injections were through either themedial femoral circumflex artery in threespecimens and the inferior gluteal artery in 11specimens. Access to the former was obtainedby dissection through the femoral triangle.Angiotomy was performed in either the super-ficial or deep femoral artery at the level of

THE SURGICAL ANATOMY OF THE BLOOD SUPPLY TO THE FEMORAL HEAD 1299

VOL. 90-B, No. 10, OCTOBER 2008

branching of the lateral femoral circumflex artery at whichpoint the medial femoral circumflex artery was easily iden-tified as exiting posteriorly. Access to the inferior glutealartery was obtained through either intrapelvic or extrapel-vic dissection as it crossed behind piriformis. All the speci-mens were injected in a standardised fashion aftercannulation with a 14-gauge angiocatheter. Industrial liq-uid rubber compound PMC-780 (Smooth-On, Easton,Pennsylvania) was prepared and coloured with the dye So-Strong (Smooth-On) following the recommendations of themanufacturer and injected using a pressurising cement gun.Once injected the latex was allowed to cure for at least 16hours in accordance with the manufacturer’s instructions.

All the dissections were done through an extended poste-rior approach to the hip. In each case the vessels weredissected to their terminal branches in the trochanteric regionto ascertain if the ramus profundus of the medial femoral cir-cumflex artery was perfused. Additionally, the hip capsulewas opened to evaluate perfusion of the lateral epiphysealarteries along the superior aspect of the femoral neck.

ResultsOf the 14 possible specimens, 11 were successfully injectedthrough either the medial femoral circumflex artery or theinferior gluteal artery (Table I). There were three technicalfailures, one due to poor pressurisation and two to failureto cannulate the inferior gluteal artery. Of the successfulinjections, ten specimens showed a clear anastomosisbetween the medial femoral circumflex artery and the infe-rior gluteal artery as documented by retrograde flow in thealternate vessel (Table I).Retrograde injections into the vasculature of the inferior glu-teal artery. In all three medial femoral circumflex arteryinjections, retrograde flow was observed into the inferiorgluteal artery in a characteristic pattern through one tothree vessels located either superior or inferior to theconjoined tendon.Antegrade injections through the inferior gluteal artery. Ineight of nine inferior gluteal artery injections a directanastomosis was observed with the ramus profundus of

the medial femoral circumflex artery through a medium-sized vessel of 1 mm to 3 mm in diameter at or about thepoint where it crossed over the tendon of obturator exter-nus. The pattern of these anastomoses was identical tothat seen in those specimens in which filling of the inferiorgluteal artery occurred in a retrograde fashion afterinjection of the medial femoral circumflex artery.Course of the inferior gluteal artery and description of anas-tomotic sites. Most of the course of the inferior glutealartery has been described well in standard texts.4,5 In thegluteal region, it descends in the company of the sciaticnerve, sending two to four branches posteriorly and laterallyto gluteus maximus, a branch to the sciatic nerve, andbranches into the sacrotuberous ligament. In addition, fromour latex injections we identified three branches which con-tributed to the cruciate anastomosis. These ran from poster-omedial to anterolateral on either side of the conjoinedtendon. Two of these branches have been reported as the‘anastomotic’ and ‘articular branches’, respectively in anearly edition of Gray’s Anatomy.4 We have attempted toretain that nomenclature for consistency, although the ter-minology seems to be inappropriate given the distributionwhich we found. We consider the anastomotic branch to bethe piriformis branch described by Gautier et al2 and thearticular branch to be one of two contributions.

The piriformis branch (‘anastomotic branch’) divergedfrom the main trunk of the inferior gluteal artery at aboutthe level of the superior gemellus. It crossed posteriorly(superficially) to the sciatic nerve in all but one specimen,and continued laterally, running between piriformis and thesuperior gemellus (Fig. 1). Along the course of this vessel,many small branches were given off passing to the adjacentmuscles. Early in its course, a branch was given off whichproceeded deeply between piriformis and the superiorgemellus to the posterosuperior capsule and labrum. In thearea of the greater trochanter the piriformis branch splitinto three or four smaller branches. Some of these arborisedto supply the posterior aspect of the tendon of gluteusmedius just proximal to the greater trochanter and otherscoursed posteriorly to the tendon of piriformis. A direct

Table I. For each specimen, the vessel injected is listed for each side. The presence and site of the observed anastomosisbetween the medial femoral circumflex artery (MFCA) and the inferior gluteal artery (IGA) are listed as described in the text

Specimen Right side Left side Anastomosis Site of anastomosis

1 MFCA LFCA* MFCA only Piriformis and capsular branches2 IGA IGA Yes bilaterally Piriformis and inferior branches

gemellar branches3 SGA† IGA No4 IGA IGA Yes bilaterally Inferior gemellar branch5 IGA MFCA Yes bilaterally Inferior gemellus, capsular

and piriformis branches6 Not injected IGA Yes Capsular branch7 IGA (poor injection) IGA Yes bilaterally Piriformis & capsular branches8 MFCA Not injected Yes Piriformis branch

* LFCA, lateral femoral circumflex artery† SGA, superior gluteal artery

1300 A. W. GROSE, M. J. GARDNER, P. S. SUSSMANN, D. L. HELFET, D. G. LORICH

THE JOURNAL OF BONE AND JOINT SURGERY

anastomosis between the ramus profundus and the piri-formis branch (Figs 1 and 2) occurred in five of sevenspecimens in which an anastomosis was found. Each timesuch an anastomosis occurred, the piriformis branchtravelled posteriorly across the piriformis and conjoinedtendons and met the medial femoral circumflex artery inthe interval between the inferior gemellus and the tendon ofobturator externus.

The inferior anastomotic supply consisted of one or twovessels, one superficial and one deep, both running in theinterval between the inferior gemellus and quadratus femo-ris. We consider that the deeper branch represented the‘capsular branch’ described previously,4 and the moresuperficial branch the ‘quadratus branch’ which we havenamed based on the structures with which the vessels wasmost intimately associated.

Fig. 1a

a) Photograph and b) diagram showing antegrade injection through the inferior gluteal artery. The abductors are to the left and the sciatic nerve canbe seen centrally in the top of the picture (A). The injected vessel descending from the top of the picture and crossing over the conjoined tendon isthe piriformis branch of the inferior gluteal artery (B). The quadratus branch (C) can be seen running across quadratus femoris. Other labelledstructures are: D, obturator externus; E, greater trochanter; F, gluteus medius; G, quadratus femoris; H, inferior gemellus; I, superior gemellus; J,obturator internus; K, ramus profundus of the medial femoral circumflex artery; L, piriformis.

Fig. 1b

Fig. 2a

a) Photograph and b) diagram showing antegrade injection through the medial femoral circumflex artery. The trochanter is inferior (A). Quadratusfemoris has been reflected to the right (B), exposing the capsule (C) and the tendon of obturator externus (D). The blue locator paper in the top of thepicture is placed beneath the piriformis branch of the inferior gluteal artery (E, superior cruciate supply), seen here running laterally between thepiriformis (F) and conjoined tendon (G). This vessel has been filled retrograde after injection into the medial femoral circumflex artery. The ramusprofundus (H) can be seen crossing over the tendon of obturator externus and passing deep beneath the posterior capsule. It disappears deep to thecapsule before passing deep to the conjoined and piriformis tendons (seen here still attached to the trochanter). The capsular branch of the inferiorgluteal artery (I) can be seen running transversely along the capsule, having been filled retrograde from this medial femoral circumflex artery (J,gluteus medus; K, inferior gemellus).

Fig. 2b

THE SURGICAL ANATOMY OF THE BLOOD SUPPLY TO THE FEMORAL HEAD 1301

VOL. 90-B, No. 10, OCTOBER 2008

The capsular branch was present in all specimens. Itarose from the inferior gluteal artery at about the level ofthe inferior border of the inferior gemellus, and coursedanteriorly (deep) to the sciatic nerve. It always ramified tosupply the proximal capsule and labrum, and continuedacross the capsule laterally, just cranial to the tendon ofobturator externus (Fig. 2). In four specimens this vesselterminated in the lateral portion of the capsule. Equally fre-quently, it continued transversely and made a direct con-nection with the ramus profundus at the posteroinferiormargin of the capsule of the hip, just cranial to the tendonof obturator externus. In other specimens there was asecond, more superficial inferior supply which we havereferred to as the quadratus branch. This was quite largebilaterally and crossed quadratus femoris in a retrogradefashion. When present, this branch always travelleddirectly to the region of the tendon of obturator externusanastomosing in that area with the ramus profundus(Fig. 3). In two specimens, it travelled a noticeably moredirect route to the hip capsule than the medial femoral cir-cumflex artery. In these specimens the inferior gemellar

branches terminated as the lateral epiphyseal arteries, andthe deep branch of the medial femoral circumflex arteryappeared to enter the inferior gemellar branch perpendicu-larly, in a ‘T’ fashion.

In all cases, regardless of which branch of the inferiorgluteal artery provided the anastomotic vessel, the anasto-mosis occurred as the ramus profundus crossed behind andover the tendon of obturator of externus (Figs 2 and 3).This vessel then passed directly under the posterior orbicu-lar fibres of the hip capsule while still on the postero-superior aspect of the femoral neck. Once under thecapsule, it coursed subsynovially and obliquely upwardsalong the femoral neck, and finally split into two to fourlateral epiphyseal arteries. We found that the ramus profun-dus always proceeded beneath the capsule before arrivingat the superior aspect of the neck. As the vessel passed deepto the conjoined tendon, it was always protected by a por-tion of the hip capsule.

In the three examples in which the medial femoral circum-flex artery was injected, we also noted latex filling of themedial epiphyseal arteries. This was occasionally noted even

Fig. 3a

a) Photograph and b) diagram showing antegrade injection through the inferior gluteal artery. This is a deeper dissection of the same specimen shownin Figure 1, in the same orientation. The short external rotators have been removed, except for the tendon of obturator externus. The capsule has beenopened showing lateral epiphyseal vessels running from right to left up the neck. Labelled structures are as follows: A, obturator externus; B, greatertrochanter; C, gluteus medius; D, quadratus femoris; E, inferior gemellus; F, femoral neck; G, labrum; H, hip capsule; I, lateral epiphyseal arteries; J,ramus profundus; K, synovium; L, sciatic nerve; M, piriformis.

Fig. 3b

1302 A. W. GROSE, M. J. GARDNER, P. S. SUSSMANN, D. L. HELFET, D. G. LORICH

THE JOURNAL OF BONE AND JOINT SURGERY

in specimens injected through the inferior gluteal artery.These vessels were found to course in a vincular reflection ofthe synovial fold at the posteroinferior aspect of the neck.They have been well described previously by Harty8 andSevitt and Thompson.9 We noted one or two vincular vesselswhich ramified within 3 mm to 5 mm from the physeal scarinto four or more terminal epiphyseal vessels. These vesselswere distributed along the posteroinferior aspect of the head-neck junction at the five or seven o’clock position, respec-tively, according to the side of the specimen.

DiscussionVarious aspects of the cruciate anastomosis between theinferior gluteal and medial femoral circumflex arteries havebeen described for at least 100 years.4 However, the surgicalimplications have only recently been considered. In the 1918edition of Gray’s Anatomy,4 anastomotic connections weredescribed between the internal and external iliac systemsthrough the medial femoral circumflex and the inferior arter-ies with the anastomotic branches of the inferior glutealartery being described as “directed downward across theexternal rotators, and assist(ing) in forming the so-called cru-cial anastomosis”. This description has persisted largelyunchanged into the 38th edition of this classic text.10 This‘crucial’ or ‘cruciate’ anastomosis has many differentdescriptions.5 Constant constituents are variously said to bethe medial femoral circumflex artery and the inferior glutealartery,4,7,11 the lateral femoral circumflex artery,4,7,11,12 theobturator7,9 the superior gluteal artery11 and the first perfo-rator.11 However, in the many descriptions of this anastomo-sis, neither the precise course of the respective arteries, northe specific sites of anastomosis are described.

Gautier et al2 most recently mentioned the anastomosisbetween the inferior gluteal and medial femoral circumflexarteries in a report detailing the course of the medial fem-oral circumflex artery. They noted that there was a consis-tent blood supply to the femoral head through the medialfemoral circumflex artery without any apparent contribu-tion from the lateral femoral circumflex artery. In addition,they described a significant constant anastomosis with theinferior gluteal artery occurring through a branch whichthey indicated as travelling along piriformis. Without morespecific comment on that anastomosis, they stated theirbelief that it “may be capable of compensating after injuryto the deep branch of the MFCA”.2 Nonetheless, no spe-cific description was offered of the precise location of theanastomotic connections between these two vessels.

We were able to produce clear evidence of retrogradeflow from the external iliac system, through the medialfemoral circumflex artery, into the internal iliac systemthrough the inferior gluteal artery, in all three specimens inwhich the medial femoral circumflex artery had beeninjected. The pattern was similar in all three injections, andresulted in filling of the inferior gluteal artery with latexthrough branches extending from the trochanteric fossa.One branch ran between the superior gemellus and piri-

formis and the second ran either along the capsule adjacentto the tendon of obturator externus or more superficiallybetween the inferior gemellus and quadratus femoris. Wewere able to elaborate on those anastomoses through ante-grade injections directly into the inferior gluteal artery. Thisallowed precise determination of the source and sites of anyconnection between the deep branch of the medial femoralcircumflex artery and the inferior gluteal artery. Our dissec-tions established the extent of the ramus profundus andlateral epiphyseal arteries to confirm the hypothesis ofGautier et al2 that the inferior gluteal artery is able to per-fuse the femoral head in the absence of medial femoral cir-cumflex artery inflow. In all successfully-injected specimenswith an anastomosis we were able to fill the lateralepiphyseal and also the medial epiphyseal arteries.

Our dissections showed that the ramus profundus stayedclose to the level of the tendon of obturator externus in thetransverse plane until it ran under the posterior aspect ofthe hip capsule, the zona orbicularis. The vessel coursed rel-atively anteromedially along the neck until it reached theposterior capsule, but it never travelled in a significantly-cranial direction, and was never found in the intervalbetween the capsule and the gemelli and obturator internus.

In summary, the ramus profundus regularly receives flowfrom at least one vessel from the inferior gluteal artery. Wefound both the course of these tributary vessels as well asthe site of the anastomoses to be consistent and easily iden-tifiable. All anastomoses occurred in an extracapsular posi-tion immediately adjacent to the tendon of obturatorexternus in the trochanteric fossa. The ramus profunduspassed deep to the posterior, orbicular fibres of the hip cap-sule in every case and was never found in the intervalbetween the short external rotators and the hip capsule andthe inferior epiphyseal arteries were noted in a vincular foldof synovial tissue.

We have hypothesised on the surgical and pathologicalimplications of these findings. Jedral et al13 described the fetalpattern of the blood supply to the hip as being dominated bythe inferior gluteal artery in more than 50% of the cases theyexamined. In some of our specimens the inferior gluteal arteryappeared to be the more dominant vessel in terms of both thedirect course of specific inferior gluteal artery branches to thelateral epiphyseal arteries and the many sources of inflow tothe ramus profundus from the inferior gluteal artery relativeto the single medial femoral circumflex artery source. It maybe that some aspect of this variability in fetal developmentleaves some hips more susceptible to vascular insult in thisregion. In addition, we believe that our dissections provide abetter understanding of the final blood supply to the femoralhead. In all seven specimens with an anastomosis, the deepbranch of the medial femoral circumflex artery received signif-icant additional inflow from vessels derived from inferior glu-teal artery immediately after it crossed the tendon of obturatorexternus. It may be more accurate to consider the ramus pro-fundus, which is destined to become the lateral epiphysealarteries, to be a common vessel with input from many sources.

THE SURGICAL ANATOMY OF THE BLOOD SUPPLY TO THE FEMORAL HEAD 1303

VOL. 90-B, No. 10, OCTOBER 2008

The sequelae of traumatic dislocation of the hip may bebetter understood based on this concept of the anatomy. Itmay be that the presence of a proximally-based blood flowat the level of the posterior hip capsule is protective againstvascular insult. In the event of a posterior dislocation, suchvessels based on the inferior gluteal artery should have moreability to accommodate the extreme displacement of thefemoral head. It is apparent from clinical reports that not allposterior dislocations, with or without accompanying frac-tures, result in avascular necrosis of the femoral head.14-16

In these circumstances perhaps it is this added supply to theramus profundus of the medial femoral circumflex arterywhich prevents the occurrence of avascular necrosis.

In conclusion, we reliably cannulated the inferior glutealartery and were thus able to document the precise courseand site of anastomoses of branches contributing to the cru-ciate anastomosis. We found both the occurrence and thesite of these anastomoses to be remarkably consistent andlocated near easily identifiable local landmarks. This infor-mation may be relevant in surgical practice to decrease therisk of iatrogenic avascular necrosis during posteriorapproaches to the hip.

No benefits in any form have been received or will be received from a commer-cial party related directly or indirectly to the subject of this article.

References1. Trueta J, Harrison M. The normal vascular anatomy of the femoral head in adult

man. J Bone Joint Surg [Br] 1953;35-B:442-61.

2. Gautier E, Ganz K, Krugel N, Gill T, Ganz R. Anatomy of the medial femoral cir-cumflex artery and its surgical implications. J Bone Joint Surg [Br] 2000;82-B:679-83.

3. Dora C, Leunig LM, Beck M, Rothenfluh D, Ganz R. Entry point soft tissue dam-age in antegrade femoral nailing: a cadaver study. J Orthop Trauma 2001;15:488-93.

4. Gray H. Anatomy of the human body. Twentieth ed. Philadelphia: Lea and Febiger,1918.

5. Hollinshead WH. Textbook of anatomy. Fourth ed. Philadelphia: Harper and Row,1985.

6. Chung S. The arterial supply of the developing proximal end of the human femur. JBone Joint Surg [Am] 1976;58-A:961-70.

7. Judet J, Judet R, LaGrange J, Dunoyer J. A study of the arterial vascularization ofthe femoral neck in the adult. J Bone Joint Surg [Am] 1955;37-A:663-80.

8. Harty M. Some aspects of the surgical anatomy of the hip joint. J Bone Joint Surg[Am] 1966;48-A:197-202.

9. Sevitt S, Thompson RG. The distribution and anastomoses of arteries supplying thehead and neck of the femur. J Bone Joint Surg [Br] 1965;47-B:560-73.

10. Williams PL, Bannister R. Grays anatomy. Thirty-eighth ed. London: Churchill Liv-ingstone, 1995.

11. Howe WWJ, Lacey T, Schwartz P. A study of the gross anatomy of the arteriessupplying the proximal portion of the femur and the acetabulum. J Bone Joint Surg[Am] 1950;32-A:856-66.

12. Crock HV. An atlas of the arterial supply of the head and neck of the femur in man.Clin Orthop 1980;152:17-27.

13. Jedral T, Anyzewski AP, Cisek B, Benke G. Vascularization of the hip joint in thehuman fetuses. Folia Morphol (Warsz) 1996;55:293-4.

14. Letournel E, Judet R. Fractures of the acetabulum. Second ed. Berlin: Springer-Ver-lag, 1993.

15. Epstein HC. Posterior fracture-dislocation of the hip: long-term follow-up. J BoneJoint Surg [Am] 1974;56-A:1103-27.

16. Epstein H. Traumatic anterior and simple posterior dislocations of the hip in adultsand children. Instr Course Lect. Vol 22. St Louis: CV Mosby, 1973.