U.S. Army Medical Department, Office of Medical History
Skip Navigation, go to content







AMEDD MEDAL OF HONOR RECIPIENTS External Link, Opens in New Window






Chapter VII



Arterial Aneurysms and Arteriovenous Fistulas

Technical Consideration in Their Surgical Management, With Special Reference to Exposure of the Blood Vessels

Daniel C. Elkin, M. D. and Harris B. Shumacker, Jr., M. D.

    The safe and successful accomplishment of operations on blood vessels requires careful anatomic exposure. Young surgeons assigned to the vascular centers established in the Zone of Interior were frequently woefully lacking in anatomic knowledge. In years past, textbooks have dealt with anatomic considerations as matters of first importance and devoted considerable space to proper approaches to the blood vessels and methods of ligation. When World War II began, however, textbooks of surgery were concerned for the most part with anatomy as applied to diagnosis and treatment, and the young military surgeon could get no adequate guidance from them. Though many of these men had served internships and even surgical residencies they lacked anatomic knowledge as it relates to the peripheral nerves, blood vessels, and the musculature of the extremities.

   To overcome this lack it was necessary for skilled vascular surgeons assigned to the centers to train these officers not only in the specialized techniques of vascular surgery, but often in the elementary details of such techniques.

  In the following pages are set forth the important principles and techniques taught and practiced in the vascular centers.


    One of the cardinal principles of the surgery of aneurysms is that incisions must provide easy and complete access to the vessels involved. The planning of an approach which permits isolation and control of the involved vessels is often more important than the surgical treatment of the lesion itself. Furthermore, the incision must be so placed that by lengthening it the vessels distal and proximal to the site of the suspected lesion are made accessible without difficulty. Unless the plan of exposure is an integral part of every procedure, hemorrhage may so interfere that the operation cannot be carried out successfully; even death of the patient may result. Whenever the exposure and control of major vascular trunks are regarded as preliminary steps to the repair of aneurysms and arteriovenous fistulas, the hazards of operation are greatly reduced.


    These considerations do not obviate in any sense, however, the advisability of planning skin incisions which yield a good plastic result, which do not cross normal skin folds and creases, and which minimize the hazards of keloid formation, scarring, and contractures-provided such incisions sacrifice nothing essential in regard to the exposure of the vessels. Adequate exposure is not necessarily achieved by an incision which lies directly over the vessels involved; only rarely does a properly placed plastic incision offer any real limitation upon the achievement of easy and complete access to the vessels.


Femoral Vessels

The guide to the superficial femoral vessels is the sartorius muscle which in the upper portion of the thigh lies lateral to the vessels, and in the middle and lower portion overlaps the vessels in their course in Hunter's canal. The upper popliteal and lower femoral vessels are best reached through an incision just anterior to the sartorius muscle. When the knee is flexed and the thigh rotated externally, the sartorius is easily displaced posteriorly upon opening the deep fascia (Fig.17).  For lesions of the common femoral artery, and occasionally for those of the proximal portion of the femoral or profunda femoris artery, an oblique incision just below Poupart's ligament is preferred. If the lesion is too high to permit adequate exposure without proximal extension of a longitudinal incision, it is preferable to curve the incision up along the inguinal fold rather than to cross the crease.

Popliteal Vessels

Although it is common practice to explore the popliteal vessels through a longitudinal incision  crossing the popliteal crease, a high percentage of patients thus treated subsequently develop     heavy scars or keloids, contractures, and even ulcerative lesions.  These sequelae are disabling,      their remedy by plastic procedures is time-consuming and not always  successful, and exposure         of the popliteal vessels can quite as well be secured by more desirable methods.

    Lesions of the first portion of the popliteal and of the femoropopliteal vessels can be exposed by a longitudinal anteromedial incision extending down to the knee. When the thigh is rotated externally and the knee flexed, the sartorius muscle is relaxed and easily retracted. The femoral vessels are then accessible proximal to the point at which they pass under the adductor tendon and the popliteal vessels are accessible to the point at which they emerge beneath this tendon.

    When the lesion lies more distally, in the midpopliteal space, some type of transverse incision crossing the popliteal space in one of the natural creases is used (Fig. 18). A simple transverse incision is sometimes adequate, particularly when a large saccular aneurysm is approached by the intrasaccular route and


Figure 17. Exposure of lower femoral and popliteal vessels by medial approach anterior to sartorius muscle.

the blood supply is controlled by a tourniquet. More often the incision must be extended upward along the medial aspect of the thigh and downward along the lateral surface of the leg (Fig.18A). The skin flaps are reflected and the fascia incised longitudinally in the midline. The biceps is retracted laterally and the semimembranosus and semitendinosus are retracted medially. After the tibial and peroneal nerves are identified, which is not difficult, the vein and artery are readily accessible.

    When the lesion is in the distal portion of the popliteal space, exposure by retraction or division of the heads of the gastrocnemius muscle is difficult. Such a procedure entails a definite hazard since the innervation of the gastrocnemius, plantaris, and popliteus muscles may be injured.

    The difficulty is obviated and exploration made easy by the use of an L-shaped incision crossing the lateral half of the popliteal space and extending downward along the posterolateral aspect of the calf (Fig.18C and D). The fascia is divided beneath the skin incision. The lateral head of the gastrocnemius along with the plantaris is retracted, as is the medial head. The popliteal


Figure 18.   Skin incisions for exploration of popliteal vessels. A. Incision for lesion in popliteal vessels in midpopliteal space. B. Incision for lesion higher in popliteal fossa. C,  D. Incisions for lesions of distal popliteal vessels. E. Modified incision useful for exposure of associated nerve lesion.

artery and vein and the tibial nerve are thus brought into view and can be followed as they cross over the popliteus muscle and pass through the tendinous arch of the soleus muscle (Fig.19).

  When the lesion is in that portion of the vessels covered by the soleus muscle, the skin incision must be extended down the leg. Exposure of the vessels generally necessitates division of the lateral fibers of the soleus attached to the fibula, with or without division of that portion arising from the head of the fibula. Sometimes the vessels can be exposed satisfactorily by incising the muscle longitudinally directly over the vessels.

Vessels of the Leg

    In general, exposure of vascular lesions in the lower extremity can be accomplished without difficulty by the use of properly placed longitudinal incisions which do not cross skin creases and which therefore heal without sequelae. This is particularly true of lesions of the lower third of the leg which are easily reached by direct approach along the posteromedial aspect. The incision is begun one fingerbreadth posterior to the border of the tibia and can be carried upward by detaching a portion of the soleus muscle from its origin on the tibia.

    Lesions of the posterior tibial vessels and of the anterior tibial and peroneal vessels in the upper part of their course present a different problem. To distinguish clinically which of these vessels is involved is frequently impossible because of their close proximity to each other, but difficulties of exposure are the same for all. The tibia and fibula are closely apposed in this area and there is a dense interosseous membrane between them. Hemorrhage is a serious


Figure 19.  Exposure of the popliteal vessels. The lateral and medial heads of the gastrocnemius are retracted and nerve and vessels are seen lying upon popliteus muscle and as they enter the soleus.

possibility. The rich collateral anastomosis which develops as the result of an arteriovenous communication, together with dilatation of the vessels including those which perforate the interosseous membrane, demands direct visualization of these vessels and their careful ligation and division. Otherwise the retraction of vascular channels through the interosseous membrane may


result in serious or even uncontrollable hemorrhage. This would necessitate a second incision along the front of the leg, or the removal of the fibula in the presence of hemorrhage and at an inopportune time during the course of the operation. Deliberate removal of the upper portion of the fibula to secure adequate exposure is often, therefore, the part of wisdom.

At the vascular centers in World War II the incision used for exposure of the posterior tibial and upper portion of the anterior tibial and peroneal vessels was a modification of the technique described by Henry 1 in 1927. In this procedure the upper portion of the fibula is removed subperiosteally which insures continued stability of the knee joint. An additional advantage of this method is that the peroneal nerve, which is frequently injured concomitantly with the blood vessels, can be exposed and treated as necessary through the same incision.

    Continuous spinal is the anesthetic of choice for this operation. The patient lies on the unaffected side with the knee slightly flexed. A pheumatic tourniquet is applied to the thigh but is not inflated unless severe hemorrhage occurs.

    The incision is begun about 2 inches above the head of the fibula and carried distally as far as may be indicated by the position of the aneurysm. After the skin and superficial fascia are divided at the upper end of the incision, the deep fascia is opened at the medial edge of the biceps tendon. The common peroneal nerve is exposed, and a rubber strip passed around it for aid in immobilization (Fig.20). Division of the deep fascia is continued downward along the course of the nerve at the posterior margin of the biceps tendon. The fascial origin of the peroneus longus muscle, which lies directly over the groove in which the nerve passes forward across the neck of the fibula, is divided. A definite plane, the lateral intermuscular septum, which lies between the soleus muscle posteriorly and the peroneus longus muscle anteriorly, is easily developed. When the muscles are separated the lateral border of the fibula immediately comes into view. The periosteum is readily stripped from the fibula by sharp dissection, aided by a periosteal elevator, and the fibula divided by means of a Gigli saw. The removal of the head subperiosteally is more difficult and is best carried out by dissection with a sharp knife. The blade of the knife is kept directly against the bone and the peroneal nerve is retracted completely out of the field of incision. The removal of the head and of the upper portion of the fibula exposes the lower end of the popliteal artery with its terminal branches, namely, the posterior tibial, anterior tibial (Fig.21), and peroneal vessels. These are easily seen when the soleus muscle is retracted posteriorly and the peroneus longus anteriorly.

    As the first step in excision of the arteriovenous fistula, the artery proximal to the fistula is secured. A ligature is passed around it to control any possible bleeding, but is left untied. The vessels distal to the lesion are isolated,

Henry, Arnold K.:  Exposure of Long Bones and Other Surgical Methods.  Bristol, John Wright and Sons, Ltd, 1927.


Figure 20. Resection of head of fibula for exposure of tibial and peroneal vessels. Note sharp dissection of periosteum covering head of fibula and complete mobilization of common peroneal nerve.

ligated, and divided. Following this the proximal artery is ligated and divided. The fistula is then removed from below upward, all communicating vessels being ligated and divided. The proximal vein is divided as the last step in the operation. It is of utmost importance that the region of the fistula be avoided until its principal blood supply is completely controlled.

  The resected portion of fibula is not replaced. The question of the stability of the knee joint naturally arises following removal of the insertion of the fibular collateral ligament, but none of the patients in whom the operation was performed in the vascular centers had any complaints referable to the knee joint on the operated side. Furthermore, examination failed to reveal any loss of stability. Comparison of the fibular collateral ligaments by palpation, with the ligament under stress, usually disclosed as tense a ligament on one side as on the other.

    Whenever possible, incisions on the sole of the foot are to be avoided since the scar is usually painful when the foot is bearing weight. Approach to the plantar vessels is through an incision on the medial aspect of the foot. The


Figure 21.  Exposure of arteriovenous fistula of posterior tibial vessels by removal of upper third of fibula. Insert shows line of incision.

abductor hallucis muscle is detached from its origin after which the vessels, with the tibial nerve, are easily traced into the deeper structures of the foot (Fig. 22).


Axillary Vessels

    For lesions involving the axillary vessels an incision is made over, and parallel to, the upper fibers of the pectoralis major muscle beginning about the midportion of the clavicle. If the first portion of the axillary vessels is to be exposed, usually all that is necessary is to separate the upper fibers of the pectoralis major through this type of incision and to retract or transect the muscle through its insertion, the pectoralis minor. If the distal portion of the axillary vessels is involved, the incision described is continued out over the insertion of the pectoralis major muscle and this muscle is divided through its tendinous insertion (Fig.23).   Often the incision used for exposure of the first portion of the brachial vessels is adequate for the exposure of the distal end of the axillary vessels. The incision begins over the insertion of the pectoralis major muscle, curves down the inner aspect of the arm distal to the axilla to the region of the neurovascular bundle, and then downward for a short distance over the course of


Figure 22.  (After Henry.)  Exposure of plantar vessels through medial approach, which is facilitated by detachment of abductor hallucis muscle from its origin.

these vessels (Fig. 24). When the skin flaps are reflected the distal portion of the axillary vessels and the first portion of the brachial vessels are readily accessible. Incision through the axilla is thus avoided. This is most desirable since keloids, heavy scars, and disabling contractures are particularly troublesome in this region. An additional advantage of the incision described is that it permits exposure of any concomitant nerve lesions.

Vessels of the Antecubital Fossa

    The brachial vessels are satisfactorily approached for most of their course on the inner aspect of the arm through longitudinal incisions overlying their course. The biceps muscle is used as a guide. Lesions in the first portion of the brachial vessels are best handled by an incision which does not cross the axilla. This incision is also used for exposure of the subclavian and axillary arteries (q. v.).

    Lesions in and near the antecubital fossa are exposed through incisions similar to those employed in the popliteal fossa. A transverse incision is made across this space in the line of a natural skin crease. Depending upon the requirements in the special case under treatment, it is extended upward


Figure 23.  Incisions for exposure of A innominate, B subclavian, and C axillary vessels.

Figure 24. Incision for exploration of distal portion of axillary, and first portion of brachial, vessels.

for a short distance along the anteromedial aspect of the arm or downward along the anterolateral aspect of the forearm (Fig. 25). The skin flaps are reflected. The communications between the cephalic and basilic veins are divided and the lacertus fibrosus severed. The cephalic vein and the lateral antebrachial cutaneous nerve are retracted laterally, and the basilic vein and medial antebrachial cutaneous nerve, medially. The brachial artery and vein, with the median nerve, then come into full view (Fig. 26) and can easily be exposed down to a point below their bifurcation into radial and ulnar branches.


Figure 25.   Incisions for exposure of vessels in antecubual fossa. A. Usual incision.  B. Incision for lesion presumed to be in proximal portion of ulnar vessels. C. Incision for lesion of brachial vessels just proximal to antecubual crease.  D. Incision for exploration of distal end of brachial, or proximal end of radial, vessels.

    When the lesion is in the proximal portion of the radial vessels it is sometimes helpful to supplement an anterolateral longitudinal incision by curving it across a portion of the antecubital space.

    One of us used these incisions in 11 aneurysms of the radial and ulnar vessels. They proved entirely satisfactory. Prompt healing occurred in all, and in none was there any scarring or contractures. The only untoward occurrence was a superficial slough about 1 cm. in diameter in one case in which the corner of an L-shaped incision was perhaps traumatized during operation. The slough, however, did not involve the entire thickness of the skin and epithelization was rapid.

   Lesions of the lower portion of the brachial artery and of the upper portion of the radial and ulnar arteries can be approached through an S-shaped incision in the cubital space. Separation of the brachioradialis and pronator teres muscles will expose the radial artery in the upper portion of its course,

Shumacker, H. B., Jr.:   Incisions in surgery of aneurysms, with special reference to exploration in the antecubital and popliteal fossae.  Ann. Surg. 124: 586-598, Sep 1946.


Figure 26.  Exposure of fistula between brachial artery and vein in antecubital space. The communications between the basilic and cephalic veins have been divided, and the lacertus fibrosus has been severed.  Note exposure of cutaneous and median nerves.

and separation of the flexor carpi ulnaris and the flexor digitorum sublimis will reveal the ulnar artery throughout its course in the forearm.


  Cirsoid aneurysms of the scalp may arise independently or may be preceded by a congenital angiomatous vascular abnormality. Since trauma, frequently inconsequential, may be the starting point, one might expect many of them to follow injury by military missiles. In all, however, only nine aneurysms of this kind were observed at the vascular centers during World War II.  Cirsoid aneurysm, which is also known as serpentine or racemose aneurysm, may originate in congenital telangiectases, nevi, or angiomas, particularly if intermediary trauma or long continued irritation has occurred. Multiple communications between vessels may, however, follow trauma in the absence of preexisting vascular lesion. The term cirsoid is generally limited to a lesion which is in effect a diffuse arteriovenous fistula formed by multiple anastomoses between contributing 'arteries and outgoing veins.


    The possibility of an intracranial extension of such a lesion or of the extracranial manifestation of a primary intracranial varix must be borne in mind whenever a cirsoid aneurysm is encountered.

    A variety of methods has been employed in the management of cirsoid aneurysms of the scalp, including carotid ligation, ligation of the multiple vessels connected with the lesion, obliteration through galvanocauterization or by the injection of thrombosing solutions, and extirpation. The method last named is undoubtedly more certain of cure and therefore the procedure of choice. It is performed as follows:

    As a preliminary step, the principal artery (arteries) leading to the lesion is independently ligated. A horseshoe flap of scalp is turned down to expose the lesion from the underside; care is taken to include the galea in the flap. This step amounts to carrying the incision down to the pericranium and reflecting with the lesion all tissues superficial to it. Hemostasis is obtained by digital pressure and by individual ligation and division of each vessel as it is encountered in the incision. The main vessels, including the central portion of the lesion, are completely excised from the underside of the flap. Particular care is taken not to buttonhole the skin. After excision of the lesion the flap is replaced with interrupted stitches of silk in the galea and in the skin.

    Case 1.  This 25-year-old soldier sustained an injury to the right parietal region in 1937, when he was struck on the head by a fist during an altercation. Following this a small pulsating hematoma, 0.5 cm. in diameter, developed which gave him no trouble. On 12 March 1945 while in combat training he was accidentally struck on the right side of his scalp by the elbow of his opponent. Three hours later he noticed that the mass was considerably increased in size, although it was neither painful nor tender. Within the next month the swelling increased rapidly and he became conscious of a discomforting pulsation in his scalp, and a throbbing, buzzing sensation in his left ear. He was admitted to Ashford General Hospital on 23 April 1945.

    On examination, a large tortuous system of convoluted blood vessels covered the whole right side of his scalp (Fig. 27A). The main tumor mass was centered in the right parietal region and the vessels which radiated to and from it were apparently derived mainly from both superficial temporal, frontal, and the left occipital veins and arteries (Fig. 27B).  From the center tumor in the right parietal region vessels radiated over the entire scalp. These appeared to be venous and were readily compressible, although all of them pulsated. Over the right parietal region a harsh, continuous bruit and the thrill could be heard. The bruit was transmitted throughout the scalp but was readily obliterated by pressure of the right carotid vessels in the neck.

    Operation was performed on 8 May 1945.  Preliminary to excision of the aneurysm the right external carotid artery was ligated and the left occipital and left superficial temporal arteries were ligated and divided.  A horseshoe-shaped incision, with its base above the right ear, was then turned downward.  The incision was carried through the galea and numerous large tortuous vessels were ligated and divided as they were reached.  The galea was then opened and a mass of dilated vessels removed from the underside of the flap.  In spite of previous ligations there was considerable bleeding during the operation which was controlled by sutures and coagulation.  The flap was replaced by suturing the galea and skin with interrupted sutures of silk.

    Recovery was uneventful and 3 months after operation there was no evidence of recurrence. (Fig. 27C and D.)


Figure 27.  (Case 1.)   A and B show cirsoid aneurysm of the scalp condition before operation, 8 May 1945.  C. and D.  Appearance of head 3 months after operation.


    Although the great vessels of the neck are frequently exposed by longitudinal incisions along their courses, the resulting scars are often heavy and wide and associated with keloid formation and actual contractures. On the other hand, general experience with oblique skin incisions following natural skin creases of the neck has been that scars are inconspicuous and never cause


contractures or deformity. When such incisions are properly placed, exposure of the carotid artery and jugular vein is entirely adequate. Localization of the level of the lesion is usually possible prior to exploration. If this is done, and if the skin incision is properly placed, the whole procedure is simple. The upper and lower skin flaps are widely mobilized, the fascia is incised along the anterior border of the sternocleidomastoid muscle which is retracted posteriorly, and the carotid sheath is opened longitudinally. The vessels above, below, and at the site of the lesion can thus be exposed adequately. The vagus nerve can also be visualized and protected against injury.

    Resection of the Clavicle.  In the approach to the great vessels at the base of the neck (that is, the subclavian, axillary, carotid, innominates, and their branches), the clavicle frequently prevents adequate exposure and control of the vessels proximal and distal to the vascular lesion. The safety of vascular surgery in this region, however, depends largely upon the accuracy with which isolation of these vessels is accomplished. Removal of a portion of the clavicle is often the solution of the problem. This procedure greatly facilitates all vascular operations at the base of the neck, does not interfere with local motion, produces no noticeable deformity, and leaves no residual pain. During World War II this technique was employed in almost 50 cases at the vascular centers in the Zone of Interior with uniformly good results. Resection of the medial half of the clavicle with its sternal articulation improves the exposure of lesions of the base of the neck and the superior mediastinum. Exposure of the subclavian and axillary vessels requires resection of only the central third of the bone.

    Incision is made over the most prominent portion of the clavicle to be reflected (Fig. 28) and is extended as necessary along the course of the vessels. When skin flaps are reflected the clavicle, the manubrium, the sternum, and the muscles attached to these structures are readily accessible. Incision of the periosteum on the anterior surface of the clavicle, away from the sites of muscular attachment, greatly reduces the amount of bleeding. The periosteum strips freely in this area, but the transverse scapular vessels run close to the posterior surface of the clavicle and are easily injured if the periosteum is torn. If the clavicle itself forms part of the false sac of an arterial aneurysm, as it not infrequently does, its removal may be accompanied by profuse bleeding which must be controlled at the site of the opening by digital pressure. The mobilized portion of the bone is divided with a Gigli saw.  If the sternal end is removed the cartilage should be excised with it.

    A vertical incision beneath the sternal end of the clavicle will expose the innominate and carotid vessels and the first portion of the subclavian artery and its branches. The sternum is divided in the midline down to the level of the second interspace and is transected at this level. It is then spread apart with traction upon bone hooks, or is separated with a rib spreader. Either method provides good visualization of the vessels. A vertical incision through the central portion of the clavicular bed will expose the subclavian vessels


Figure 28.   Resection of clavicle subperiostally to expose aneurysm of subclavian artery.

and the brachial plexus. Division of the anterior scalene muscle facilitates exposure of the first and second portions of the subclavian artery.  It also facilitates exposure of the vertebral artery and the thyrocervical trunk. Extension of the incision laterally, with the cephalic vein as a landmark, and division of the costocoracoid membrane, reveals the route to the axillary vessels.

   Closure of the wound is effected by layers. For the repair of the periosteum, interrupted sutures of fine silk are used.

    The resected segment is not replaced. Experience has shown that results are poor if the resected bone is reinserted and fixed in place by approximating it to the remaining bone with wire or other sutures. When this is attempted, proper immobilization is often not secured and painful nonunion or malunion occurs so commonly as to make the procedure eminently unsatisfactory. Other complications such as osteomyelitis and subluxation of the sternoclavicular joint may result. In fact it is sometimes necessary to remove the replaced portion of the clavicle to give the patient relief.

    At the conclusion of the operation the dressing is reinforced with adhesive tape, the arm and forearm being left free. The patient is kept in bed for 2


weeks. He is asked to lie flat on his back and to hold his shoulders well back. (No external apparatus to insure immobilization is used.) During the third week he is allowed to move about, but is encouraged to rest a considerable part of the time on his back.

    There is little discomfort after this operation, and activity is not associated with pain. The area of the clavicular defect rapidly becomes firm and on clinical examination the absence of the bone cannot be detected, even on careful palpation. There is little external deformity and asymmetry is less than when the resected segment of clavicle is replaced. Resection of the articular portion of the clavicle makes no difference in the postoperative results. The patient returns to normal activity within 6 to 8 weeks.

    Although new bone can be palpated in the clavicle within a few weeks of operation, several months must elapse before it can be demonstrated by roentgenogram. During this time, as already mentioned, patients are active and have no discomfort on motion. These results suggest that the concept of the function of the clavicle as a fulcrum to provide better lateral motion of the arm should be revised. The suggestion is supported by the fact that the occasional congenital absence of this bone is known to be of no great consequence, and that in many carnivore it is normally rudimentary.

>     In a case observed at Mayo General Hospital excessive bony regeneration had occurred at the site where the periosteum of the clavicle had been divided. While the patient bad no complaints at time of disposition, it is conceivable, on the analogy with the excessive callus formation which sometimes follows fractures of the clavicle, that pressure upon the brachial plexus might occur later.  In view of this possibility the periosteum should be so transacted that the incision does not lie over this plexus.

    Replacement of the Clavicle by Bone Chips. Although patients who have had the clavicle removed have normal configuration of shoulders, normal motion, and are entirely comfortable, regeneration of the bolae is slow and it would be desirable to hasten it if this could be accomplished without undue risk. For this reason, one of us devised the plan of filling the periosteal bed with bone chips secured from the excised bone.3   In the 4 patients in which the method was employed at the vascular center of Mayo General Hospital, the results were excellent.

In each instance the clavicle was excised subperiosteally and care taken to preserve the periosteum intact. In one patient the periosteum was also transacted to facilitate exposure. The bone was divided with a Gigli saw. The excised segment was kept in sterile physiologic salt solution until the vascular phase of the operation had been completed, and was then converted into small bone chips, 2 to 3 mm. in diameter, with a rongeur. These chips were packed into the periosteal bed, a sufficient quantity for this purpose being obtained    

3   Shumacker, H. B., Jr.:  Resection of clavicle with particular reference to use of bone chips in periosteal bed.   Surg., Cynec. & Obst. 84: 245-248, Feb 1947.


from one-half to two-thirds of the resected bone. The periosteum was closed with interrupted sutures of fine silk.

   Postoperative treatment followed the usual regimen after resection of the clavicle. Definite firmness was noted in the region of the excised bone earlier than in patients who had been subjected to similar resections without replacement of bone chips. Osseous regeneration was rapid and complete. Beginning callus formation was evident in 2 patients 3 weeks after operation; complete regeneration was observed in all 4 patients by the 18th postoperative week. Roentgenologic studies in 10 patients in whom the excised bone was not replaced by bone chips showed in most instances no callus formation at periods ranging from 5 to 34 weeks after operation. In these patients, furthermore, regeneration sometimes proceeded from the ends of the remaining bone rather than throughout the periosteal bed, as happened when bone chips were used. In 1 patient eburnation of the bone-ends 34 weeks after operation made it doubtful that union would be achieved at all.


    Reports of the successful treatment of vascular lesions of the vessels which lie within the mediastinum are not common, though almost every conceivable type of operative approach has been employed since Valentine Mott 4 first ligated the innominate artery in 1818 in an attempt to cure a traumatic subclavian aneurysm. In Greenough's review 5 of the literature on the subject, published in 1929, the fact is emphasized that safety can be achieved only by adequate resection of the overlying bony framework.  In only 37 of the 91 cases of innominate ligation or attempted ligation which Greenough collected was resection, or ostectomy, of the overlying bones performed. Inadequate exposure of the vessels was the primary factor in a number of the 18 fatalities in this group.

Technical Considerations in Exposure of Vessels in the Mediastinum

    In Greenough's review of the operations performed upon the innominate artery it is interesting to note the wide variety of procedures employed to expose the vessel. In 11 patients a portion of the clavicle and the manubrium was resected. In 11 others only a part of the clavicle was resected; in 1 patient this procedure was combined with partial resection of the first costal cartilage. Other maneuvers included sternoclavicular disarticulation; splitting of the sternum in the midline; partial excision of the manubrium, clavicle, and first rib; and resection of the manubrium with or without excision of the first two costal cartilages.


Mott, V.:   Reflections on securing in a ligature the arteria innominata, to which is added a case in which this artery was tied by a surgical operation. Med. and Surg. Register, New York 1: 9-54, 1818.

5 Greenough, J.:  Operations on innominate artery; report of successful ligation. Arch. Surg. 19: 1484-1544, pt. 2 Dec 1929.


    The operation described by Bardenheuer illustrates exposure by resection of bone. The inner part of the clavicle and first rib are removed on one side and the manubrium is sectioned transversely about an inch below the superior border. The opposite clavicle and the first and second ribs on that side are then divided and the freed manubrium is excised. Various modifications of the procedure have been employed.

Exposure of about the same extent is obtained by the osteoplastic operation of Kocher.7  The manubrium is reflected as a flap attached to the costal cartilages on one side after division or disarticulation of the clavicles, the first and second cartilages on one side, and transection of the sternum at the level of the second interspace. The exposure devised by Sauerbruch and Schumacher s for mediastinal explorations has also been used for vascular operations in the mediastinum. It requires splitting of the sternum longitudinally down to the level of the third interspace and cutting it across into this interspace. In spite of the excellent exposure thus secured of the arch of the aorta and the proximal and midportions of the innominate vessels, it is impossible even with wide retraction of the divided sternum to obtain adequate visualization of the subclavian and carotid arteries or of lesions hidden under the upper part of the manubrium and the sternoclavicular joint.

    Most textbooks of surgery in discussions of aneurysms and other lesions of the innominate artery describe methods of exposure directed toward ligating that vessel by resection of the clavicle and disarticulation of its medial end. This exposure, however, is inadequate in the presence of an aneurysm or arteriovenous fistula near the aorta in which constant risk of hemorrhage makes adequate exposure particularly imperative. Manipulations must be conducted under full vision because the vessels are dilated and thin-walled. Each case must be considered an individual problem and effort made to choose the ideal operative approach.  To illustrate: Within a short period, three cases involving innominate aneurysms and arteriovenous fistulas were reported and in each a different approach was used. One of us reflected the inner half of the clavicle, the second costal cartilage and the inner portion of the accompanying rib, divided the manubrium, and excised a portion of its right half. Trent, 10 after resection of the second rib and cartilage and division of the third, used a transpleural approach. Lindskog 11 excised the proximal portion of the second costal cartilage, partially divided the sternum by rongeur at this level, split the manubrium vertically with a Gigli saw introduced from above, and divided the clavicle.

An exposure employed at Mayo General Hospital which proved extremely satisfactory, is carried out as follows: An incision is made from the midportion of the clavicle down over the sternoclavicular joint to the midline. From

6   Bardenheuer:   Die Resection des Manubrium Sterni.  Deutsche med. Wchnschr. 11: 688-690, 1885.

7   Kocher, Theodor: Textbook of Operative Surgery.New York, The Macmillan Co., 1911.

8   Sauerbruch, F., and Schumacher, E. D.: Technik der Thoraxchirurgie. Berlin, J. Springer, 1911.

9   Elkin, D. C.: Arteriovenous aneurysm, the approach to the innominate vessels.  J.A.M.A. 129: 26-27, Sep 1945.

10  Trent, J. C.: Ligation of the innominate artery for innominate aneurysm using rubber bands; report of a case.  Surg., Gynec. & Obst. 82: 463-470, Apr 1946.

11  Lindskog, G. E.:   Surgery of the innominate artery.  New England J. Med. 235: 71-76, Jul 1946.


this point it is continued down over the sternum to the level of the third or fourth interspace. The platysma is divided. The inner third of the clavicle is resected subperiosteally by disarticulation of the sternal end and division of the other end with a Gigli saw. The sternal portion of the sternocleidomastoid muscle is divided, as is a portion of its clavicular origin. The sternohyoid and sternothyroid muscles are similarly severed near their points of origin. It is necessary ordinarily to divide these only on the side on which the clavicle is removed. A finger is gently passed behind the manubrium from above; through this maneuver it is usually possible to separate the manubrium from underlying structures. If a blunt dissector is used for the purpose, dissection must be cautious and under the constant guidance of the palpating finger to avoid trauma to the underlying vessels. The periosteum of the sternum is incised anteriorly in the midline and the sternum split down to the level of the second or third interspace. The Shoemaker sternal shears are ideal for this purpose. The Lebsche knife is less satisfactory, since the bone tends to split in advance of the cut and not always in the midline.

Once the sternum has been divided for some distance its edges can be retracted with bone hooks. Underlying structures can then be palpated satisfactorily and decision made as to whether division of the sternum should be carried out at the second or third interspace to provide the required exposure. The sternum can also be freed more completely from underlying tissues. After complete incision of the periosteum and freeing of the outer borders, the sternum is divided across into both interspaces at the appropriate level. The halves of the divided sternum are now widely retracted. The loose areolar tissue and the remains of the thymus gland are pushed to the side or retracted, and the great vessels thus brought into view (Fig. 29). A small rib spreader is substituted for the bone hooks at this point.

    It must be borne in mind in all mediastinal surgery that there is wide variation in the position of the mediastinal structures in relation to the overlying bony framework. The arch of the aorta is occasionally as high as the superior border of the manubrium. The pericardium is sometimes barely in view. The superior vena cava may be of considerable length, while in other cases the innominate veins appear to join more caudally into a short vena cava.

    After the necessary vascular surgery is accomplished, the wound is closed in layers. The sternum is reapproximated with wire sutures placed through drill holes or with silk sutures placed in the periosteum. The clavicular periosteal bed is then carefully closed with interrupted silk sutures, or prior to suturing, it may be filled with bone chips from the excised portion of the clavicle to hasten bony repair. The muscles, fascia, and skin are brought together with silk sutures. The skin incision is entirely satisfactory and there is normal range of shoulder motion (Fig. 30) when this technique is used.


Figure 29. Operative exposure of vessels of superior anterior mediastinum by division of sternum and resection of inner third of left clavicle.

Technical Considerations in Management of Vessels in Mediastinum

    In 1945, one of us analyzed 38 reported cases of attempted surgical treatment for innominate aneurysms (36 cases) and arteriovenous fistulas (2 cases).12 In these, 18 deaths occurred. Thirteen took place among the 20 patients treated prior to 1924, and 4 in the 15 treated after that date (the date was not mentioned in 3 patients reported by Matas, 1 of whom died). Undoubtedly inadequate exposure played an important role in these fatalities. Causes of

12   Shumacker, H. B., Jr.:   Surgical cure of innominate aneurysm.  Surgery 22: 729-739, Nov 1947.


Figure 30.  Normal range of shoulder motion in patient submitted to exposure of blood vessels in superior anterior mediastinum by technique shown in Figure 29.  In this patient the clavicular periosteal bed was filled with bone chips.


death included tracheal obstruction, uremia, infection, cerebral ischemic difficulties, and hemorrhage and shock. Hemorrhage and shock were reported as causing death in 8 patients and would also seem responsible in 2 or 3 others in which the cause was not given.

    With present methods of therapy some of these causes are preventable. Adequate operative exposure, gentle dissection of vessels, and transfusions of whole blood have reduced the hazard of bleeding. Aseptic technique, chemotherapy, and antibiotics have reduced the danger of infection. While cerebral complications are still a danger, this hazard can be reduced to a minimum by repeated preoperative compression of the carotid artery carried out until the patient can tolerate occlusion for a long period. In certain instances it may be advisable, before operation for innominate aneurysms, to perform a preliminary ligation of the carotid artery with a removable metal band, or with a fascial or tape ligature. No instance of gangrene of the upper extremity has been reported after such operations, but other evidences of ischemic have not infrequently been observed. Before permanent ligation the efficacy of the collateral circulation should always be tested by observation of the hand during temporary occlusion of the artery.

    There are too few cases in each category of treatment in Shumacker's collected series to permit conclusions about the safety and effectiveness of the various methods. If the artery could be safely and securely repaired, repair would naturally be the method of choice, but it is doubtful that this method can be employed except in occasional cases. It was used in only 1 of the 38 cases collected from the literature and was not practical in any of the cases observed at the vascular centers in World War II. The best procedure from the standpoint of safety to the patient and likelihood of cure of the lesion seems to be proximal and distal ligation (Table 28), combined, when feasible, with excision of the sac, or with evacuation of the thrombus within the sac. Distal ligation will usually involve ligation of the carotid and subclavian arteries. In such cases, as in all instances of arterial surgery, it is wise to divide and transfix the arteries rather than to ligate them in continuity. Opinions differ whether the concomitant vein should be ligated when occlusion of a major artery is necessary.

    Of these 38 cases reported up to 1945 only 2 had aneurysms involving the innominate vessel at       its point of origin. One of these patients died before ligation could be accomplished; the other, reported by Trent,13  survived.  Several unreported fatalities have also followed attempts at ligation. Unless constriction with a rubber band, as suggested by Trent, proves the answer to the problem-experience with this method in aneurysms of the aorta suggests that it may end in disaster-some other more indirect method of controlling the aneurysm must be sought. Babcock's surgical production of a carotidjugular arteriovenous fistula 14 is, according to Matas,       a hazardous operation   

13   See footnote 10, p. 243.

14  (1) Babcock, W. W.: A new treatment for thoracic aneurysm.   Ann. Clin. Med. 4: 933-942, 1926.

      (2) McCarthy, P. A.:   Treatment of aneurysms of thoracic aorta and innominate artery by distal arteriovenous anastomosis. Ann. Surg. 91: 161-196,1930.



likely to be followed by functional difficulties.15   Distal ligation of the carotid and subclavian arteries by the method of Brasdor-Guinard 16 has produced some excellent results 17 but is not generally applicable and the moderate chance of relief offered is quite out of proportion to the gravity of the risk involved.

Wiring combined with coagulation, as developed by Blakemore and King,18 seems the method of choice when direct surgical intervention on the aneurysm is not feasible. It cannot be considered a curative measure, but is likely to arrest the expansion of the aneurysm and to bring about relief of symptoms. Possibly a combination of wiring and coagulation, with distal ligation of the carotid and subclavian arteries, may yield better results than wiring and coagulation alone. Once the sac wall has been reinforced by an increase in the intramural thrombus following this procedure there should be little hazard associated with distal ligation.After distal ligation there should be a good possibility of obtaining complete saccular thrombosis with further wiring, though this outcome would be impossible with wiring alone.

15  Matas, R.:   On the treatment of aortic aneurysm by method of jugulo-carotid anastomosis; discussion.  New Orleans M. & S. J. 84: 445-454, Dec 1931.

16  Matas, R.  In Keen, W. W.:   Surgery, Its Principles and Practice.  Philadelphia, W. B. Saunders Co., 1920.

17 (1) Baldwin, J. F.:   Aneurysm of the innominate; synchronous ligation of subclavian and common carotid with long survival; unusual incision.  Am. J. Surg. 5: 517, Nov 1928.

     (2) Schwyzer, A.:  Aneurysm of innominate artery.  Ann. Surg. 96: 666-669, Oct 1932.

18   Blakemore, A. H., and King, B. G.:  Electrothermic coagulation of aortic aneurysms. J.A.M.A. 111: 1821-1827, Nov 1938.


    In view of the limited number of reports in the literature of surgical repair of lesions of the innominate artery, and particularly in view of the small number of successful cases reported, histories of certain of the patients with these lesions observed in the vascular centers in the Zone of Interior in World War II are herewith presented.

Case 2.   A 25-year-old soldier was stabbed in the right upper chest with sharp-pointed barber scissors 11 November 1943.  He became unconscious from loss of blood and was treated for shock by transfusions of plasma and whole blood. Recovery was prompt and he was returned to full duty in 4 days.  When he was reexamined 1 April 1944, a murmur was discovered in the anterior part of the chest.  His only complaint was dyspnea on exertion.  He was hospitalized at once and subsequently transferred to the vascular center at Ashford General Hospital.

    Examination at this center on 6 November 1944 revealed a small healed wound on the anterior wall of the chest just below the right clavicle and 1 inch lateral to the sternal border. In this region could be felt a well-defined continuous thrill, which was less distinct in the neck and along the clavicle.  A loud, continuous murmur, intensified in systole, was audible. The bruit was transmitted over the entire anterior part of the chest, both shoulders, and both sides of the neck. It was also heard over the posterior part of the thoracic wall. Neither the thrill nor the bruit could be obliterated by pressure.  The blood pressure in the left brachial artery was 135 mm. of mercury systolic and 50 diastolic and in the right 140 mm. of mercury systolic and 40 diastolic. The venous pressure in the left and right cubital veins was 210 and 220 mm. of water respectively. A teleoroentgenogram of the heart which had been made 2 October showed a transverse diameter of 16.3 centimeters. After a month of rest a second teleoroentgenogram, made a few days after admittance to Ashford, showed that this measurement had decreased to 15.2 centimeters. (Figs. 31A and B.)

    Operation was performed 18 January 1945. An incision beginning at the midportion of the clavicle on the right was carried medially to the sternum and downward to the fourth rib. The skin, with the pectoral muscles, was reflected laterally. The inner half of the

Figure 31. (Case 2.) Roentgenograms showing successive changes in size of heart in traumatic arteriovenous fistula involving innominate vessels.  A. Eleven months after injury. B. Twelve months after injury and after 1 month of rest.  C. One month after correction of fistula.


clavicle was resected subperiosteally and disarticulated at the sternoclavicular joint (Fig. 32). The second costal cartilage was resected together with about 2 inches of the second rib. The space beneath the manubrium sterni was entered by blunt dissection and the right half of the manubrium divided, its connection with the first rib being left intact. A portion of the right half of the manubrium was removed with a rongeur. The pleura was dissected back by blunt dissection from the midline under the sternum, and the first rib with the attached portion of manubrium was retracted upward and outward. The internal mammary artery and vein, with their branches, were ligated and divided. The regional fat and remains of the thymus gland were removed. The innominate vessels were thus brought into view. After considerable dissection a fistula was found between the innominate artery and the two innominate veins at their point of junction just above the arch of the aorta. Both veins were greatly dilated, as was the artery proximal to the fistula. A false sac had been produced by local dilatation of the veins at the point of the fistula.

Figure 32.  (Case 2.)   Exposure of innominate vessels for correction of arteriovenous fistula.   A. Removal of portion of clavicle, first and second ribs, and sternum.  B. Fistula with false venous sac. C. Transverse section showing false sac and communication between artery and vein.  D. Ligation and division of affected vessels.

    The right innominate vein distal to the fistula was ligated and divided. The false sac was separated from the aorta to which it was firmly bound, and the innominate artery doubly ligated at its point of origin from the aorta. The thrill and bruit ceased completely when this step of the procedure had been concluded. After ligation of the artery the pulse rate dropped from 84 to 72 beats per minute and the blood pressure rose from 136 mm. of mercury systolic and 60 diastolic to 156 min. of mercury systolic and 100 diastolic. The innominate artery distal to the fistula was then doubly ligated. It was not thought feasible, however, to ligate the left innominate vein or the superior vena cava. Complete excision


of  the fistula was considered, but it would have required ligation and division of both innominate veins, the superior vena cava and the innominate artery, and would have introduced the risk of cerebral edema. The operation was therefore discontinued at this point. Since no branches of the innominate artery communicated with the fistula it was hoped that the communication would be eliminated by what had already been done.

   The sternum with the attached first rib was replaced and sutured with tantalum wire. The musculocutaneous flap was replaced with interrupted silk sutures.

    At the end of the operation, which required 7 hours, the pulse rate was 68 beats per minute and the blood pressure 140 mm. of mercury systolic and 82 diastolic. The hand and fingers were warm.

    Recovery was uneventful except that it was necessary to aspirate collections of blood beneath the pectoral muscles on three occasions. The temperature did not rise above normal at any time during the patient's convalescence and the highest pulse rate was 100 beats per minute. The hand and fingers on the affected side were always warm. Teleoroentgenograms of the heart made on 13 February 1945, 3 months after last preoperative determination, showed the transverse diameter to be 13.1 cm., a reduction of 2.1 cm. (Fig. 31C).  The venous pressure in the cubital veins on the same date was 154 mm. of water on the left and 234 mm. on the right. Hand, arm, and shoulder movements were all within normal range. Examination of the chest 17 May 1945 revealed neither bruit nor thrill and the patient was considered cured.

    Case 3.   A 35-year-old officer was wounded 19 August 1944 by shell fragments which caused a large avulsed wound of the left thigh and a small wound in the right infraclavicular area near the sternum.  When he arrived at a general hospital in the Zone of Interior he complained of a buzzing, throbbing sensation in the right sternoclavicular area, but had noted no other symptoms.  Examination revealed an arteriovenous fistula in this area and he was transferred to the vascular center at Mayo General Hospital.

    Examination at this center revealed a continuous thrill and murmur, most prominent just below the sternal end of the clavicle. The bruit was transmitted downward toward the heart, outward over the course of the subclavian and axillary vessels, and upward into the neck, though in the latter region it was heard somewhat less clearly.   Neither thrill nor bruit could be eliminated by digital compression. There were no evident circulatory changes in the right upper extremity.  The electrocardiogram revealed no abnormality.  Roentgenograms revealed the presence of a small fragment of metal below the sternoclavicular joint, otherwise they showed nothing informative.
    Exploration 16 February 1945 through the incision described in Case 2 revealed a fistula between the internal mammary artery and the overlying innominate vein. The shell frag ment was imbedded in the posterior wall of the vein. The vessel was indurated and scarred in this area. The fistula was excised with quadruple ligation of the affected vessels.

    Convalescence was uneventful and upon examination 10 months after operation, the patient appeared cured.  

     Case 4.   A 24-year-old soldier was injured by shell fragments 20 October 1944. He received penetrating wounds of the left malar region, the left side of the neck and the left supraclavicular fossa. The shell blast caused temporary loss of consciousness. Shortly after injury he lost his hearing on the left side and became hoarse.  He was found to have a fracture of the maxilla and a massive hemothorax. The fracture was reduced through a Caldwell-Luc approach and the hemothorax was treated by thoracentesis.  As a result, his hearing began to improve, but a distressing tinnitus and hoarseness persisted.  He also developed a throbbing, aching pain in the left arm, and subsequently a persistent left ulnar hypesthesia.  Decortication of the left lung was performed and after recovery he was transferred to the vascular center of Mayo General Hospital.

    The patient was admitted to this vascular center 9 March 1945 with a diagnosis of left carotid aneurysm.  At this time he complained of abnormal pulsation in the left side


of the neck, throbbing pain in the left arm, numbness in the left u1nar area, tinnitus in the left ear, watering in the left eye, and hoarseness.

    Examination revealed a pulsating mass at the base of the left side of the neck.  No thrill could be demonstrated, but there was a loud systolic bruit. The bruit could be obliterated by digital compression.  Pressure over the inner part of the mass interrupted carotid pulsation and reduced the radial pulse.  No circulatory disturbance of the left upper extremity was evident, but left ulnar hypesthesia was demonstrable.  The patient also presented left recurrent laryngeal paralysis and bilateral deafness of the nerve type. The blood pressure was 116 mm. of mercury systolic and 70 diastolic.  Electrocardiograms gave no evidence of abnormality and roentgenograms showed nothing informative. The reactive hyperemia test, with the left subclavian artery occluded, gave evidence of good collateral circulation.   

     The condition was diagnosed as traumatic aneurysm of the subclavian or the common carotid artery.After repeated digital occlusion of the carotid artery, which was tolerated without symptoms, operation was performed 4 May 1945.  A large subclavian aneurysm was found over which the left vagus nerve was stretched.  Except that it was thinner than normal the nerve seemed uninjured. The carotid artery was not involved.

    The operative approach described in Case 2 was used.  The first portion of the sub clavian artery was isolated medial and inferior to the sac and a tape placed about it.  The subclavian artery was isolated distal to the sac and dissected back toward the involved area beneath the aneurysm.  When the artery was occluded proximally and distally the aneurysmal sac, though it ceased to pulsate, continued to fill rapidly with bright arterial blood.  This was demonstrated by needle aspiration.  Since it was evident that the vertebral artery also opened into the sac, it was feared that simple proximal and distal ligation of the subclavian artery would not effect a cure. The vertebral artery, however, was inaccessible because of the overlying aneurysmal sac and the adjacent scarring.  The sac was therefore carefully freed back to its mouth during temporary occlusion of the subclavian artery on either side of the aneurysm, after which it was opened widely.  Back bleeding was prevented by digital pressure while the subclavian artery was ligated just proximally and distally. The vertebral artery could then be dissected free and ligated. The sac was totally resected after the overlying innominate, subclavian, and internal jugular veins had been ligated and divided.

    The patient was in good condition throughout the procedure, but remained drowsy afterward.  There was no  paralysis of the extremities. About 8 hours after operation he complained of severe headache.  He became first stuporous, then comatose, and death occurred 14 hours after operation. At necropsy extensive encephalomalacia of the left cerebellum was found.  No embolus was demonstrated, there was no thrombosis of the left vertebral artery, and the         circle of Willis was normal.    

    Case 5.   A 25-year-old soldier was struck in the right suprasternal region by a small shell fragment 14 July 1944.There was no external bleeding, but thoracentesis was required twice because of hemothorax.  For a while he progressed satisfactorily, but 9 days after injury when told that he was being returned to duty he became very hoarse and for several days could not speak at all.  Four months later, after he had been passed through several replacement centers and hospitals, including a psychiatric installation, a right recurrent laryngeal paralysis and a mass in the anterior mediastinum were discovered.  He arrived in the Zone of Interior 10 December 1944.

    The patient reached the vascular center of Mayo General Hospital 22 December 1944. By this time he had regained his normal voice, but complained of a constant sense of oppression accentuated by bending forward and localized in the upper anterior part of the chest, also of substernal pain on deep inspiration, and some weakness of the right upper extremity. He stated that this extremity was often darker in color than the left and that the veins of the hand often appeared distended. He attributed a weight loss of some thirty pounds to poor appetite.


    Examination revealed a loud systolic bruit in the right supraclavicular fossa. A similar sound was present to a lesser extent in the suprasternal notch and to the right of, and over, the upper sternum. Pulses in the upper extremities were about equal in force.  Blood pressure was 134 mm. of mercury systolic and 80 diastolic in the right arm and 146 systolic and 90 diastolic in the left.  No color changes were noted in the extremities and sweating was normal, but the veins of the right hand were somewhat distended. In a room at 22 o C. the fingers of this hand were from 2 to 5 degrees cooler than those of the left hand. Oscillometric readings revealed that the oscillations were equal in both arms, but slightly less at the right wrist than at the left.  Tracings of the ergograph revealed some weakness of the right hand.  A traumatic arterial aneurysm with right recurrent laryngeal palsy was suspected.  He withstood prolonged right carotid compression repeatedly without difficulty.  Laryngoscopic examination showed paralysis of the right vocal cord. Electrocardiograms revealed no abnormality.  Roentgenograms demonstrated a mass in the upper right section of the anterior mediastinum (Fig. 33A).

Figure 33.  (Case 5.)  Roentgenograms showing innominate aneurysm.  A. Approximately 51/2  months after injury.  B. Fifteen days after partial ligation of innominate artery.  Note that the aneurysm is unchanged in size. C. Eleven weeks after second operation, at which complete correction of the vascular lesion was accomplished.

    Exploration was carried out 15 February 1945 through a sternum-splitting incision  The sternum was divided down to the level of the third interspace and across into the interspace on both sides.  This approach permitted the exposure of a large saccular aneurysm of the innominate artery.  The lesion originated about 2 cm. distal to the origin of the vessel and involved a considerable portion of its length.  When the artery was occluded proximally with a rubber-shod clamp, the right hand immediately became extremely pale and cold, and neither color nor warmth returned during the 15-minute period of occlusion. Because of this observation it was thought unwise to carry out complete ligation.  The artery was therefore constricted to about one-half its diameter with a strip of fascia secured with interrupted mattress sutures of silk; over this a band of cellophane was placed. At the conclusion of the procedure the radial pulse was smaller on the right than on the left, and oscillometric studies showed oscillations in the right arm reduced by about 40 percent as compared with the left.

    By the 15th postoperative day, oscillometric studies showed that the discrepancy between the two arms had completely disappeared.  The bruit in the right side of the chest had, however, increased in intensity and was now the same as at the time of admittance.  Roentgenograms revealed no reduction in the size of the aneurysm (Fig. 33B).  It was thought possible that the band placed about the artery had given way, but, whatever the


reason, there was no evidence at this time that the artery was partially constricted. In view of the ischemia which had occurred during complete temporary arterial occlusion, it was decided to perform sympathectomy in the hope that it might render the collateral circulation adequate.  The operation was performed 9 April, and thereafter the right hand was warmer and had a better color than the left.

    The mediastinum was again explored 2 June.  At the first operation the proximal artery had been readily accessible, but not the subclavian which lay beneath the retracted right side of the manubrium and the sternoclavicular joint.  At the second operation, in order to provide better exposure, division of the sternum was supplemented by resection of the inner third of the right clavicle subperiosteally and by section of the anterior scalene muscle.  Exposure by this method revealed that the entire aneurysm, which lay in the right side of the mediastinum, extended l cm. above the upper border of the divided manubrium and was about 10 cm. long and 6 cm. in diameter.  It lay lateral to the artery and was partially covered by the innominate vein.

    The subclavian, internal jugular, and innominate veins were divided between transfixing ligatures and the intervening segment removed. Isolation of the proximal artery was difficult because of the scarring in the vicinity of the previous partial arterial ligation. In the process of isolating the artery the sac was torn just at its origin.  Bleeding was effectively controlled by digital pressure while a clamp was placed across the innominate artery, and the carotid and subclavian arteries were divided between transfixing ligatures of silk.  The proximal innominate artery was then ligated with umbilical tape and transfixed and ligated with silk just distal to the tape.  The aneurysm was now completely stilled and there was no bleeding.  The sac was opened widely and a large amount of laminated thrombus removed.  Fibrin foam was placed in the wound before closure.  The sternum was brought together with three wire sutures, and the clavicular periosteal bed, the fascia, and the skin closed with silk sutures.  Two pints of whole blood were administered during the operation and the blood pressure did not fall below 130 mm. of mercury systolic.

    Two hours after operation a nonpulsatile hematoma appeared in the wound. The wound was opened, the clot evacuated, and the skin sutures replaced after thorough irrigation. Otherwise, the postoperative course was uneventful and the end result was satisfactory in every way except for persistence of slight general weakness and slight fatigability of the right hand. When the patient was separated from service in September 1945 he still had a vocal cord paralysis, though he spoke well.  Roentgenograms showed no mass in the area formerly occupied by the aneurysm (Fig. 33C).

    Case 6.   A 27-year-old soldier received shell-fragment wounds in the right thigh, left buttock, back, and posterior aspect of the neck 28 January 1945.  He was evacuated to the Zone of Interior and was admitted to the vascular center of Mayo General Hospital with a diagnosis of right subclavian arteriovenous fistula.

    When admitted to the hospital he complained of a buzzing sensation in the right sternoclavicular area, numbness in the right ulna distribution, and hyperhidrosis of the right hand.

    Examination revealed a continuous thrill centered over the sternoclavicular area, and a continuous bruit so loud that it masked all respiratory sounds in the chest.  Neither bruit nor thrill could be obliterated by digital compression.  In comparison with the left hand, the right hand was cooler and oscillometric studies showed a reduction of oscillations.  Blood pressure in the right upper extremity was slightly reduced: 142 mm. of mercury systolic and 76 diastolic. The electrocardiogram revealed no abnormality, and a roentgenogram revealed nothing remarkable except for a metallic foreign body under the sternal end of the right clavicle.

    Exploration on 13 June disclosed an arteriovenous fistula between the subclavian artery at its point of origin and the innominate vein at the point at which it received the internal jugular and the two subclavian veins. A large anomalous vein entered the vena cava proximally. The five veins involved could all be ligated without special difficulty.


    The subclavian artery was ligated just distal to the fistula as it emerged from the innominate artery; the vertebral and internal mammary arteries, which originated in proximity to the fistulous opening, were also ligated. The distal subclavian artery and veins were divided between transfixing ligatures and part of the venous sac excised.

    Convalescence was uneventful.  Excellent circulation in the right upper extremity was maintained.

    Case 7.   A 24-year-old soldier sustained shell-fragment wounds of the abdomen, chest, hands, right forearm and arm, and right leg 24 January 1945.   These were debrided, an exploratory laparotomy performed, and numerous thoracenteses done for bilateral hemothorax.  Upon the diagnosis of an arteriovenous fistula he was evacuated to the vascular center of Mayo General Hospital, arriving there 21 May 1945.

    Examination revealed a continuous murmur audible over the upper anterior part of the chest and maximal over the right border of the sternum at the level of the second interspace.  No thrill was palpable.  There were no circulatory disturbances of the extremities.  Blood pressure was 124 mm. of mercury systolic and 60 diastolic.  Electrocardiograms showed nothing abnormal.  Roentgenograms of the chest revealed a mass in the upper anterior mediastinum.  The impression was that the condition was an arteriovenous and saccular aneurysm, probably involving the innominate vessels.

    It was impossible to close the fistula by compression and thus test the collateral circulation to the right upper extremity.  The patient, however, tolerated carotid compression without symptoms.  Dorsal sympathectomy was performed 5 July and followed by mediastinal exploration 17 September through an approach as described in Case 2.

    The upper portion of the mediastinum was found to be filled with a firm, pulsating aneurysmal mass about 7 by 6 cm. in size. Associated with this was a continuous thrill.  The carotid and subclavian arteries and the internal jugular veins could be identified at the tipper pole of the mass on each side; they appeared to emerge from the aneurysm itself. 

    The superior vena cava was free proximally but disappeared into the mass about 2 cm. from its cardiac end.  The superior border of the arch of the aorta was fused with, and partly buried beneath, the mass.  Apparently the arteriovenous and saccular aneurysm involved the innominate vessels, the origins of both carotid and subclavian arteries, the terminal portions of the internal jugular veins, the superior vena cava, and possibly the arch of the aorta.  Since surgical extirpation was obviously impossible the wound was closed at once.  Convalescence was uncomplicated and the patient after his recovery was able to do light work about a farm.

    Case 8.   A 24-year-old soldier was injured in the left thigh, leg, arm, and scapular area by shell fragments 10 December 1944.  A femoral fracture and a compound fracture of the tibia and fibula were incurred.   Following debridement of the wounds, the left upper and lower extremities were placed in plaster casts.  Soon after injury an arteriovenous fistula in the left sternoclavicular region was found and the patient evacuated to the vascular center of Mayo General Hospital.  He arrived at this center in May of 1945.

    Examination disclosed that the femoral fracture had progressed satisfactorily, but that osteomyelitis had developed following the compound fracture of the tibia and fibula.  The patient had no complaints referable to the fistula, but left ulnar paralysis was present and he complained of weakness of the left upper extremity.  A continuous thrill was palpable and a loud continuous bruit was heard over the left anterior chest wall and the left side of the neck, most prominently in the sternoclavicular region.  Neither bruit nor thrill could be obliterated by digital compression. The fingers on the left hand were cooler than on the right, and pulses and oscillometric readings presented evidence of reduced circulation in the left upper extremity as compared with the right.  After treatment on the orthopedic service and left ulnar neurorrhaphy, dorsal sympathectomy was performed 30 August because of the evidence of reduced circulation in the left upper extremity and the impossibility of testing the collateral circulation.


    A preoperative diagnosis of arteriovenous fistula of the proximal portion of the subclavian vessels was made, and on 16 September mediastinal exploration was undertaken.  A fistula was found between the subclavian artery just distal to the vertebral branch, and the innominate vein near its distal end.  After the vessels had been isolated proximally and distally and the fistula dissected free and transfixed, it was discovered that there was also present a saccular aneurysm arising from the posterior surface of the artery opposite the fistulous opening. The artery was badly damaged over a considerable distance. It was therefore ligated on either side of the aneurysm.  The innominate, internal jugular, subclavian, and internal mammary veins were divided and the aneurysm and fistula excised.

    After an uneventful convalescence the patient was returned to the orthopedic service for further treatment for osteomyelitis.


    During World War II, 13 arteriovenous fistulas of the vertebral vessels were observed at the vascular centers in the Zone of Interior. This was an unusual experience. The collected reviews of Matas in 1893,19 of Perrig in 1932,20 and of Heifetz in 1945,21 indicate that up to the date last named only 67 instances of this type of aneurysm had been recorded in the literature. The differentiation between arteriovenous fistulas and false arterial aneurysms is not brought out in all of these cases, but it seems clear that the great majority were aneurysms. Arteriovenous fistulas involving the vertebral artery therefore seem to be among the most uncommon of arterial injuries, probably because of the protection afforded by the peculiar anatomic location of that vessel. The outcome is also not clear in all reported cases, but apparently more than half of the patients died as the direct or indirect result of the vascular condition. This is a much higher death rate than is associated with aneurysms or arteriovenous fistulas in almost any other location, and the explanation, again, seems to be the anatomic position of the vessels, with the resulting difficulties at operation.

Diagnosis of aneurysms and fistulas of the vertebral vessels is difficult to make. It was made with certainty in only 5 of the 13 patients observed at the vascular centers in World War II, though suspected in others. The most common erroneous diagnosis is a fistula between the internal jugular vein and the branches of the external carotid artery. Vertebral arterial lesions may also be confused with fistulas arising from other branches of the subclavian artery, particularly the inferior thyroid, transverse scapular, and transverse cervical vessels. Differentiation can usually be made on the basis of whether or not the bruit disappears on compression of the common carotid artery. If it does not, the vertebral vessels should be suspected as the site of the lesion They lie deep in the neck, and obliteration of the bruit by pressure is difficult although not impossible.

19  Matas, R.:  Traumatisms and traumatic aneurisms of the vertebral artery and their surgical treatment; report of a cured case. Ann. Surg. 18: 477-521, 1893.

20  Perrig, H.: Zur Anatomie, Klinik and Therapie der Verletzungen and Aneurysmen der Arteria Vertebralis.  Beitr. z. Klin. Chir. 154: 272-307, 1931.

21 Heifetz, C. J.: Traumatic aneurysm of first portion of left vertebral artery; case report. Ann. Surg. 122: 102-110, Jul 1945.


    At operation, confirmation of the diagnosis of vertebral arteriovenous fistula can usually be made by a process of exclusion. When the other vessels in the neck are isolated and temporarily occluded without effect on the thrill and bruit, the chances are that the vertebral artery is involved. Occasionally, however, collateral circulation is so ample that the bruit persists even during proximal occlusion of the vertebral artery and vein.

Technical Considerations in Exposure of Vertebral Vessels

    To comprehend the difficulties in exposing the vertebral artery, a discussion of its anatomic location is necessary. The extracranial portion of the artery is divided, for descriptive purposes, into three parts

    The first (extraforaminal) portion is about 3 cm. long. It arises from the subclavian artery and ascends, on each side of the neck, to the lowermost foramen in the transverse process usually of the sixth cervical vertebra. This part of the vessel runs upward and backward between the longus colli and the anterior scalene muscles. Between it are the transverse process of the seventh cervical vertebra and the sympathetic chain. In front are the vagus nerve, the internal jugular vein, and the sternomastoid muscle which lies in front of the internal jugular vein. The upper portion of the first part of the vertebral artery is crossed by the inferior thyroid artery as it passes medially. It does not enter the foramen in the transverse process of the seventh cervical vertebra, but lies in front of it.

    The second portion of the vertebral artery passes upward through the transverse processes of the upper six cervical vertebrae. In this part of its course it lies anterior to the cervical nerves and inferior to the anterior scalene muscle. It is surrounded by a venous complex which terminates below in the vertebral vein. Overlying this part of the artery and slightly to its medial side is the internal jugular vein and, more superficially, the sternomastoid muscle. In the upper part of the second portion of the vertebral artery the spinal accessory nerve is found about 2 fingerbreadths below the mastoid process. Underneath the sternomastoid muscle the dense prevertebral fascia covers the levator scapulae muscle which arises from the transverse process of the upper four cervical vertebrae and therefore directly overlies the vertebral vessels in their course through the upper four foramina.

    The third portion of the vertebral artery begins at the point at which i t emerges from the foramen of the atlas. It turns backward and lies in a groove on the upper surface of the atlas where it is covered by the semispinalis capitis in the suboccipital triangle. The guide to the artery in this portion of its course is the transverse process of the atlas which can easily be palpated 1 fingerbreadth below, and 1 fingerbreadth in front of, the tip of the mastoid bone. The vessel is more easily identified when the sternomastoid muscle is detached from its origin.

    Approach to any portion of the vertebral vessels is difficult because of the deep, position and the vital nature of the surrounding structures. Approach to


each portion, moreover, must be considered separately because the anatomy of each portion varies.

In the first portion, exposure is best achieved through an incision parallel with the fibers of the sternomastoid muscle directly over the interval between the sternal and clavicular heads of that muscle. After incision of the deep fascia, the heads of the muscle are separated and retracted (Fig. 34). Some of the fibers may be divided transversely if better exposure is necessary. The internal jugular vein, when it is encountered in the depth of the incision, is retracted medially and the triangular interval between the longus colli and the anterior scalene muscles is developed. The inferior thyroid artery, which will be found running upward and medially and crossing in front of the vertebral artery in this region, must be carefully isolated. The vertebral artery arises medial to the inferior thyroid artery and can be identified in this location by its pulsation. The vertebral vein lies anterior to the artery, and on the left the thoracic duct passes in front of it. This portion of the vertebral artery must always be identified as a preliminary to operations on the vessel anywhere in its course. A ligature passed around it, but left untied, is a protection against hemorrhage later in the operation, though it will only diminish bleeding, not control it, because of the rich anastomotic blood supply in the vertebral system. Ligation of the artery should not be done at this point in the operation since the thrill and bruit would thus be obliterated and identification of the fistula thus become more difficult.

    The second portion of the vertebral artery is reached by an incision along the anterior border of the sternomastoid muscle. After the deep fascia is opened the muscle is retracted laterally and the carotid sheath identified and retracted medially. It is usually necessary to divide the omohyoid muscle as it passes across the carotid sheath. The anterior scalene muscle, which arises from the transverse processes of the third, fourth, fifth, and sixth cervical vertebrae, is retracted laterally and may be detached from its origin in order to expose fully the transverse processes through which the vertebral vessels pass. The interval between the transverse processes is less than a fingerbreadth, which makes ligation in this region particularly difficult unless one (or more) of the transverse processes is removed with a rongeur.  If this is done a ligature may be passed about the vessel preferably on an aneurysm needle, or the vessel may be occluded by the use of heavy metal clips.

    Exposure of the upper portion of the second part and of the third part of the vertebral vessels is much more difficult than exposure of the other portions. The technique suggested by Henry 22 is most satisfactory. The patient lies on his back with his neck extended and his chin turned to the side opposite the involved vessels. An incision is made along the anterior border of the sternomastoid muscle from the middle of the neck to the mastoid process. The deep fascia is opened. The insertion of the sternomastoid muscle is cut away


22  See footnote 1, p. 230.


Figure 34.  Surgical approach to a fistula between lower portions of the right vertebral vessels.

from the bone and the muscle reflected laterally. The spinal accessory nerve is isolated and mobilized as it passes laterally and downward beneath the sternomastoid muscle (Fig. 35). Once the sternomastoid muscle is reflected, the tip of the transverse process of the atlas is easily identified. The prevertebral fascia is divided from above downward, the origin of the levator scapulae muscle cut away from the atlas and axis, and turned downward. Further dissection exposes the vertebral artery and surrounding venous plexus in the interval between the atlas and axis, as well as in its third portion.

     Whenever necessary, the transverse processes of the cervical vertebrae should be removed in order to expose completely the lesion and its contributing vessels. If the lesion is close to the origin of the vessels from the subclavian artery, resection of the inner third of the clavicle facilitates control of the blood supply. This is a useful step when the lesion is in the intraforaminal portion of


Figure 35. Surgical approach to a fistula between upper portions of the left vertebral vessels.

the artery. Before it is carried out, however, the slips of the scalene, the longus colli, and longus capitis muscles which are attached to the transverse processes must be stripped away by sharp dissection and the use of a periosteal elevator. In three cases observed at the vascular centers the overlying transverse processes could be removed without injury to the underlying vessels by gently depressing the vessels during the excision of the bony processes with a small rongeur.

    If there is difficulty in controlling the spinal branches, the branches can be isolated and either ligated with silk or clipped with silver clips. When this is not possible, bleeding can be controlled by packing with muscle or fibrin foam.

Technical Considerations in Management of Vessels

    There was no evidence of cerebral circulatory difficulties in any of the cases observed at the vascular centers. All of the patients had practiced prolonged carotid compression before operation and could withstand it for practically half an hour without symptoms. The abundant collateral circulation about the vertebral vessels in the neck and the communication in the circle of Willis would lead one to assume that cerebral ischemia would not usually follow ligation. One of us, however, observed a case in which aneurysmorrhaphy of a


large subclavian aneurysm required ligation of a single vertebral artery, but of no other vessel supplying the brain.23 The patient died within 24 hours, and necropsy revealed homolateral cerebellar necrosis without evidence of thrombosis or embolism.

   In the 13 patients with vertebral arteriovenous fistula treated in the vascular centers of the Zone of Interior in World War II, the lesion was known to be located in the first portion of the vessel in 2 patients, in the second portion in 6, and in the third portion in 3.  No attempt was made in any instance to repair the fistula. The position of the vessels as well as their small size precluded the effort. Anatomic reasons prevented actual excision of the fistula except in 7 instances, and obliteration of the fistula was therefore usually accomplished by proximal and distal ligation of the vessels and by the placing of mass ligatures.

    Since arteriovenous fistulas involving the vertebral artery are so rare, three case reports from the experience at Ashford General Hospital are herewith presented.

    Case 9. On 28 June 1944, a soldier was struck in the right arm and neck by multiple fragments of an artillery shell. There was profuse bleeding from all wounds. The neck wound was debrided and healed promptly. Approximately 3 weeks later he noted a buzzing sensation in the left side of the neck, but did not hear any noise. He reported this to his medical officer who made a diagnosis of arteriovenous fistula. On admittance to the vascular center of Ashford General Hospital 5 months later, examination revealed that there was a small scar overlying the upper portion of the left sternomastoid muscle about 4 cm. below the angle of the mandible. There was a continuous bruit and thrill heard and felt over this area which were transmitted up the scalp to the parietal region and into the neck to the clavicle.  Deep pressure failed to obliterate the bruit and thrill. It was thought that this was an arteriovenous fistula of the second portion of the left vertebral vessels.

    On 30 December 1944, the first portion of the vertebral artery was ligated and divided. This caused the thrill and bruit to be almost completely obliterated. On 3 February 1945, an incision was made from the level of the thyroid cartilage upward along the anterior border of the sternomastoid muscle to the mastoid process and then turned backward severing the sternomastoid from the skull at its insertion. The muscle was retracted laterally and the third cervical nerve and the spinal accessory nerve isolated. The levator scapulae muscle was detached from its origin from the atlas, axis, and third and fourth cervical vertebrae. The vertebral vessels were then isolated between the transverse processes of the atlas and axis. The sensory branch of the second cervical nerve overlying the vessels was dissected free and the vertebral artery divided between ligatures.  Several smaller vessels in this region were also ligated. The vein was not ligated. This stopped the thrill, but a bruit was still present.  As time passed the bruit increased in intensity.

    On 25 April 1945, the scar on the left side of the neck was excised and the sternomastoid muscle retracted laterally. The transverse processes of the first, second, third, and fourth vertebrae were exposed. The levator scapulae muscle was detached from the transverse process of the atlas and the spinalis group of muscles freed from the transverse processes of the exposed vertebrae by sharp and blunt dissection. The lateral roofs of the foramina transversarium of the third and fourth cervical vertebrae were removed by rongeur. This exposed the vertebral vessels for a distance of 4 or 5 centimeters. The fistula was apparently

23 Shurnacker, H. B., Jr.:  Arteriovenous fistulas of the cervical portion of the vertebral vessels. Surg., Gynec. & Obst. 83: 625-630, Nov 1946.


at the level of the inferior border of the canal of the third vertebra. Ligatures were passed about the vessels proximal and distal to the fistula, and the vessels ligated. Silver clips were applied above and below the fistula. This caused the bruit and thrill to disappear completely. At the time of discharge 3 months later there had been no recurrence of the bruit or thrill.

    Case 10.  A soldier was wounded in July 1944 by fragments of a high explosive shell. There was a wound below and behind the right mastoid process and one on the back of the right hand. He was unconscious for 15 minutes and bled profusely from the neck wound. The wounds were debrided and he was returned to England where the fragment was removed from the right hand. Forty-three days after injury he was returned to combat duty.  Four months later he was evacuated because of combat exhaustion. Upon routine physical examination the presence of an aneruysm of the right side of the neck was discovered and he was returned to the United States.

    The patient was admitted to the vascular center of Ashford General Hospital 21 November 1945. His only complaint at, that time was of a buzzing sensation in his right ear when lying on that side.  Examination revealed a healed wound just posterior to the upper third of the right sternomastoid muscle. In this region a thrill was felt and a continuous bruit heard. The bruit was transmitted throughout the neck, upward behind the ear. In was easily obliterated by pressure over the right common carotid artery. Upon obliteration of the fistula the pulse fell from 80 to 72 beats per minute and the blood pressure rose from 120 mm. of mercury systolic and 68 diastolic to 120 systolic and 80 diastolic. A diagnosis of arteriovenous fistula of the internal carotid artery and internal jugular vein was made.

    On 8 December 1945, an incision was made parallel to the anterior border of the right sternomastoid muscle. Since it was thought that the fistula involved the internal carotid artery, the carotid vessels were identified at their bifurcation and ligatures passed about the common, internal, and external carotid arteries. Occlusion of these vessels singly or in combination did not diminish the bruit. Compression at a slightly lower level in the neck obliterated the thrill. Dissection was carried out inferiorly and the vertebral artery identified. It was found that occlusion of this vessel stopped the bruit and thrill. Ligatures were passed about the vertebral vessels. The incision was then extended upward with detachment of the sternomastoid, longus colli, splenius capitis, and levator scapulae muscles from their origins. In this manner the transverse processes of the upper two cervical vertebrae were exposed. The transverse processes of the second cervical vertebra were removed and a ligature passed about the vertebral vessels at this point. Occlusion of the vessels did not result in cessation of the thrill. The third transverse process was similarly removed with isolation of the vessels at the level of the third cervical vertebra. Compression at this point resulted in cessation of the thrill. The vertebral vessels were then ligated at the site of the second and third transverse processes. A mass ligature was tied in the interval between these sutures with further obliteration of a segment of vessel.  Slight bleeding was encountered at this time due to the tearing of small branches of the vertebral artery.  Further to insure occlusion, a tantalum clip was applied at the site of the fistula. The vertebral artery was doubly ligated in its first portion and a tantalum clip applied at the point at which it entered the sixth vertebra.

    Recovery was uneventful. There was no recurrence of the bruit. At the time of the patient's discharge 2 months later, there was no evidence of recurrence of the fistula.

    Case 11.   A soldier was wounded in action 18 December 1944, when he was struck in the left side of the neck by a shell fragment. The wound of entrance was just inferior and posterior to the angle of the mandible. There was no wound of exit. There was no excessive bleeding and no loss of consciousness. The wound was allowed to heal by second intention. There was marked diminution in the hearing of the left ear after the injury, and he was conscious of a continuous buzzing in that ear.  He was evacuated to the United


States and admitted to the vascular center of Ashford General Hospital 8 November 1945.  Examination revealed a scar about 1 cm. in diameter just inferior and posterior to the angle of the left mandible.  There was a small sinus in the region of this scar from which a limited amount of serous fluid could be expressed.  A continuous bruit transmitted to the base of the neck, was heard over this region.  It could not be obliterated by occlusion of the common carotid artery or by pressure anywhere in the neck.   A diagnosis of vertebral arteriovenous fistula was made.

   On 31 January 1946, a 6-cm. transverse incision was made above the medial end of the left clavicle. The two heads of the sternomastoid muscle were separated and the vertebral artery isolated near its origin from the subclavian.  Occlusion of this artery did not obliterate the bruit, however it did diminish it.  This wound was covered, and a 10-cm. longitudinal incision was made along the medial edge of the sternomastoid muscle extending from the level of the thyroid cartilage upward to the mastoid process, then curved laterally over the mastoid process so that the sternomastoid muscle was detached at its insertion. The spinal accessory nerve was isolated and the sternomastoid muscle reflected outward and downward.  The splenius capitis muscle was reflected downward from the transverse process of the atlas. This exposed the vertebral vessels between the axis and atlas. Compression of the vessels in this region did not obliterate the bruit, although it did diminish it. It was then felt that the fistula involved the third portion of the vertebral vessel. The rectus capitis muscle was detached from its origin to the transverse processes of the atlas and axis and reflected downward. This exposed only approximately 2 cm. of the third portion of the vessels. It was therefore necessary to remove the transverse process of the atlas in order to expose adequately the third portion of the vessels. The vessels were then mass ligated distally and proximally to the fistula with two ligatures which were passed about the vessels. This caused the thrill to disappear completely, but the bruit was still slightly audible. It was assumed that there were branches between the proximal and distal ligatures. Three other sutures were passed about the vessels between the proximal and distal ligatures and this caused the bruit to disappear completely. The vessels were not divided. The divided muscles were approximated with interrupted silk sutures.

    The postoperative course was uneventful. There was no evidence of recurrence of the bruit or thrill 2 months after operation.


    Approach to the iliac vessels is through a retroperitoneal incision. The peritoneum and abdominal contents are displaced medially to obtain exposure of the ureter in its lower portion. Lesions of the distal external iliac vessels are approached through an oblique incision just above Poupart's ligament. The fascia of the external oblique muscle is divided and the external ring opened. The internal oblique muscle and the transversalis fascia are divided in the direction of their fibers. The iliac vessels are then readily exposed above the inguinal ligament, and the femoral vessels can be exposed below the ligament, if necessary. If exposure is inadequate, the inguinal ligament can be transected to obtain more complete exposure and resutured at the close of the procedure. Closure of the wound is carried out as in herniorrhaphy.