Fig. 1 Chronically infected median sternotomy wound following cardiac surgery

On referral the patients had chronically infected median ster notomy wounds. They were up to 23 cm long, 9 cm deep and were from 4 to 8cm wide (Fig. 1) The base of the wound was formed by thick fibrinous exudate, underneath which lay the heart, great vessels and coronary artery bypass grafts.

'Acute' group - The Problem

Five patients formed the 'Acute' group (table 1) . Four out of five had undergone aorta-coronary bypass grafting. they presented with mediastinitis and discharging wounds; all showed signs of systemic toxicity with fever and leucocytosis. In 3 the whole sternum was necrotic; one of these patients had been undergoing polyvidone-iodine irrigation until massive haemorrhage started from an infected aorta-saphenous anastomosis. The fifth patient's wound completely dehisced at 5 days; he had undergone emergency replacement of a degenerated aortic xenograft, implanted 4 years previously. The wound had been chronically discharging since that time.

Total sternectomy and omentopexy is inappropriate for most early deep sternal infections; the treatment of choice is debridement and primary closure. However, if necrotic bone or old suture material is left behind, or a space which cannot be obliterated by drainage, healing is precluded. The judgement was made at these acute group operations that, unless omentopexy was carried out, the problems of the 'Chronic' group would be encountered or worse.

Fig. 2 The defect after excision of the wound margins and curettage of the base

'Chronic' group - The Reconstruction

The wound margins, necrotic bone and any exposed costal cartilage were excised. The base was only gently curetted to avoid damage to underlying structures (Fig. 2). The abdomen was opened by an upper right paramedian or midline incision and the omentum inspected. In one case prior abdominal surgery had left adhesions; these did not prevent use of the omentum, which was always cleared off the transverse colon after division of adhesions.

Fig. 3 Mobilisation of the greater omentum out of the abdomen (based on the right gastro-epiploic vessels)

Further mobilisation was carried out by dissecting the omentum from the greater curvature of the stomach and dividing the gastro-epiploic arcade to base the graft on the usually larger right vessel in 10 of 12 cases. In the 'Chronic' group it was then brought out through the upper part of the abdominal wound, great care being taken to avoid torsion (Fig. 3).

Fig. 4 The postoperative result showing good skin graft take ('Chronic' group)

The omentum was then draped into the sternal defect by folding it in on itself in concertina fashion. The abdominal wound was closed around the omental pedicle, and drains placed in the depths of the chest wound. One and a half times expanded meshed split skin grafts were then applied (Fig. 4). Postoperatively there was a profuse discharge from the exteriorised omentum which required frequent changes of the outer dressings. This usually settled by the fifth day.

'Acute' group - The Reconstruction

In four cases in the 'Acute' group a total sternectomy was performed and left a large defect. The other had a limited debridement but a large post-sternal defect with relatively fixed walls remained. These were filled by slightly smaller omental flaps prepared as follows. Routine anti-stapylococcal cover is started with induction of anaesthesia. The wound is opened carefully and debrided with removal of all suture material. The sternum is next re-opened paying attention to possible adherence of grafts or cardiac chambers. After mobilisation of the sternal halves, necrotic bone is removed with rongeurs until tissues which are bleeding freshly are encountered. Assessment is then made of the defect produced; frequently the whole sternum has been removed with costal cartilages, but even if no sternum needs excision, a large post-sternal space may persist. This is likely to be relatively rigid in reoperations. The incision is extended inferiorly in the midline about one third of the way to the umbilicus and the peritoneum opened. The greater omentum is mobilised, if necessary with division of adhesions, and delivered through the wound. Next the avascular plane between the omentum and transverse colon and mesocolon is opened until the right side of the omentum remains attached to the stomach by the gastro-epiploic arcade. Depending upon size, the last two or three vessels of the omental arcades are separated of and the resulting small flap is swung up to the chest wall defect. It usually fits without further mobilisation, but, if necessary, the gastro-epiploic may be divided and the flap based on the further mobilised right gastro-epiploic artery. The flap is tacked to the top of the defect after incising the diaphragm posteriorly to avoid strangulation of the pedicle. By removing the sternum and debriding the edges. musculo-cutaneous flaps have already been raised but may require a further mobilisation. These are re-approximated using mass vertical mattress sutures of No. 1 nylon; the recti are closed with figure-of-eight monofilament sutures. Two silicone drains are positioned anterior and posterior to the flap; suction is not used.

Results

Omentopexy has now been used successfully to close 12 infected median sternotomy wounds in post cardiac surgical patients (Table 1). In the 'Acute' group, four of the five wounds were colonised by staphylococcal organisms, whereas in the 'Chronic' group most (72%) grew copious gram-negative organisms (Table 2). One of the 'Acute' group had been operated upon through a chronically coliform infected wound following valve replacement four years previously.

Fig. 5 Six months after acute omentopexy

Discussion

The management of a patient with a large irregular defect and mediastinitis is a difficult problem. Debridement and prolonged open dressing may take up to 16 months to heal (11). Further debridements may expose and infect costal cartilages, leading to the formation of tracks and sinuses branching off the main wound. Infection of the great vessels or graft thrombosis threatens (18) with inevitable further morbidity and mortality (11, 36).

What are the options?

The full-thickness chest wall defect resulting from the unhealed, infected median sternotomy wound can be reconstructed in a variety of ways.After thorough surgical toilet, muscle flaps such as the pectoralis major either turned over (21) or transposed (26) are useful for upper sternal defects but may need the addition of the rectus abdominis to close the lower region. The latissimus dorsi and the islanded rectus abdominis are other potential fillers of the mediastinal dead space (22, 42).

The pectoralis turnover operation (15, 21) bases the flap on the ipsilateral internal mammary artery; this artery by its continuation as the superior epigastric artery supplies the rectus abdominis on that side. This muscle may also be turned up to fill lower sternal defects (15). Clear-up rates using these techniques have been good (21, 27). However the mammary may have been used for grafting, or may easily have been damaged by repeated closure attempts and debridements. This makes the pectoralis turnover operation a less attractive option than using a pectoralis flap based on the acromiothoracic trunk. Unfortunately upturned rectus is usually required. Mobilisation of pectoralis major may require separate skin incisions. Extensive exploration of the tissue planes is necessary to free the muscle from its attachments including the humerus.

There is thus a risk of more widely spread infection occurring. The raising of flaps needs postoperative vacuum drainage and pressure bandaging. The cosmetic defect caused by loss of the anterior axillary skin fold needs to be considered, but maybe preserved by some techniques (21, 33). These tend, however, to reduce the amount of muscle available to fill the space.

In our 'Chronic' group the pectoralis major had been widely undermined and advanced by repeated debridements and closure attempts. The viability of the rectus abdominis (either ipsi- or contralateral) was in doubt. In six cases the internal mammary artery had been used in the myocardial revascularisation (2 bilateral and 4 unilateral left grafts). In the remainder its patency was in doubt due to possible thrombosis from infection or damage during surgical debridements. In view of these circumstances, it was decided to fill dead space with a vascularised graft.

The greater omentum is a unique organ with multipotential properties (30) and its use in abdominal surgery is well known (38). It is easily mobilised as a first stage by separation from the transverse colon. In a number of subjects we have observed this to give enough length to reach the neck. Das (6) observed that when it is further mobilised by basing it on the right gastro-epiploic artery, in 88% of subjects it can be made to reach the back of the neck. The artery, which is dominant in 90% of cases, should be taken back to its origin from the gastro-duodenal artery after establishing dominance by palpation, occlusion and observation (16). Previous abdominal operations usually result in easily separable adhesions. Useful amounts of omentum are reportedly available even after partial gastrectomy (38). Exteriorised omental transposition to provide cover to the chest, neck, and axilla (39, 41) is well established. Retrosternal transposition into the chest following cephalad mobilisation of the diaphragm is described (4) for repair of radionecrosis of the anterior chest wall; they also found that previous abdominal surgery was not a contra-indication to its use. The use of omentum in cardiac surgery was first described by O'Shaunghnessy in 1937 (25) as a method of myocardial revascularisation following successful work in dogs (24).

A disadvantage of omentum is that, to mobilise it, a second body cavity must be opened. The known complica tions of laparatomy need to be borne in mind but the risk of peritoneal contamination does not appear to be great. More specifically the possibility of herniation and gastric traction need consideration. Herniation of hollow viscera can be avoided by directing the omental flap to lie parallel to the falciform ligament, and secondly by ensuring that the route to the chest wound is a snug fit. Omentum readily attaches to raw surfaces (24) and there has been no incidence of intra abdominal obstruction.

Despite the removal of the sternum and costal cartilages and their replacement by soft omentum, chest wall instability causing respiratory embarrassment is not often a problem. This is probably due to rigidity engendered by inflammation and fixity of the mediastinum from adhesions. It was therefore surprising that one of the 'Chronic' group patients developed this complication following omentopexy when there had been no preoperative difficulty.

The formation of ventral hernia through upper abdominal incisions is reported to be low (8) when closed by mass monofilament sutures. However when the anterior chest wall is unstable, the normally fixed upper end of the wound is unsupported. This is the likeliest cause of the rather high incidence of incisional hernia reported here. It is proposed in future to mobilise the omentum through a right upper transverse abdominal (4) or left paramedian incision as attempts to stabilise the chest wall by wire sutures have been ineffective or, in one case, dangerous. Due to the lack of bone they may also penetrate otherwise clean costal cartilage.

Traction on the stomach by the omental graft was reported by Vineberg (40); symptoms were vague but persistent. He used no mobilising procedures to lengthen the graft as advocated by Goldsmith (10) whereas in our series all were mobilised off the transverse colon as minimum. The only patient with nausea responded to withdrawal of dipyridamole.

The defect following chronic sternotomy wound break down poses a challenging reconstructive problem, not only because of the peculiarity of its shape, but also because it is also difficult to perform proper surgical toilet in the base of the wound which overlies the heart, great vessels, and bypass grafts. The plasticity of omentum appeared to be its great advantage. Large and irregular defects were filled merely by laying in the graft which at one conformed to the wound's shape. In the acute group, the procedure always took less than two and a quarter hours; silicone drains were used but drainage suction could not be used as the omentum obstructed the drains. There is now corroberative evidence that drainage is not required (13) and we have stopped using it routinely to promote early ambulation. Being well vascularised the omentum would be expected to deliver white cells and fibroblasts; being also the major site of intraperitoneal lymphatic absorbtion (39), it was surprising that by contrast the 'Chronic' group had copious discharge through the meshed split skin grafts for 5 days. By virtue of the meshing, graft take was not compromised, but in one case there was a small area of graft loss due to underlying omental venous congestion; it is recommended that such areas are excised before skin grafting. The reason for less drainage in the 'Acute' group may be the lesser amount of chronic inflammatory tissue present, or be due to the smaller omental graft used.

What are the reasons for early radical treatment? Experience of wound breakdown had taught us that proper debridement is mandatory. It has been suggested that very early intervention may decrease the amount of tissue to be debrided (7); despite this, when sutures have cut through the bone, some fragments are non-viable. This may be especially true after internal mammary artery grafting. When proper necrotomy has been performed an irregular space which is not amenable to closure may be left. Irrigation regimens (32, 33, 37), extensive rewiring (29) or prolonged vacuum drainage (7) are unlikely to be effective due to the size of the defect. It has always been our policy to perform the most appropriate of these procedures whenever feasible. The 'Acute' group cases in this paper represent those who would have been left open and one case of polyvidone iodine irrigation failure. This last required vascularised cover of an infected great vessel in addition to space filling and skin cover to remove the risk of further haemorrhage (2, 20). It was not thought, at the time, that muscle grafting would be as effective as recently reported (35) in one case, and there were precedents for omentopexy (19, 34). Once it was recognised that the debridement had made the simpler closure procedures impracticable, two options were apparent. The first was open dressing with its risk of graft thrombosis (18) and increased morbidity and mortality; the second was immediate reconstruction as recommended by Grossi (12). It was not necessary to rotate skin flaps to achieve full thickness skin cover, as also found by Lee. (17) When applied to 'Acute' group wounds this resulted in complete healing at 14 days when Sutures were removed; one had a persistent discharge at the level of the diaphragm for 2 months. This was the result of exploring a sterile pericardial effusion in a non-dependent manner; it is no longer practised.

The results are presented by groups and are compared as such. While these groups are not from a controlled or even randornised trial, they overlap temporally. Further more the patients are from the same hospital and are very similar by comparison of population parameters (see Table 1). The use of statistical analysis is to demonstrate that for the same problem a new and more aggressive course of surgical treatment was advantageous in that the wounds healed quickly, there were fewer complications, and hospital stay was reduced. In addition we consider that the acute group had a better cosmetic result (Fig. 4 and 5). In the case with an infected aorto-saphenous anastomosis we consider the procedure to have been life-saving (19, 34).

References