The place of a D2 gastrectomy for dealing with an antral stomach cancer inpatients in the West is a matter of continuing debate. This article describes a standardised operative procedure with the aim of achieving a satisfactory clearance of the gastric cancer in the antrum and a systematic removalof the regional draining lymphatics.

The first priority of surgery for gastric cancer must be to resect all of the primary tumour such that there is no macroscopic or microscopic tumour remaining (Ro resection). (1) There is some controversy regarding whether this is best achieved by a total or subtotal gastrectomy but, even more controversial, is the additional value of systematic lymphadenectomy (D2 resection). The Japanese literature and series from specialist units in the West indicate that a D2 gastrectomy can be performed safely without added morbidity compared with less radical lymphatic dissections. (2-4) Furthermore, it would appear that any survival benefit of a radical lymphadenectomy is likely to accrue only in patients with stage II and IIIa disease. (5) The prospective randomised controlled trials that have been performed show little benefit and, indeed, excess morbidity and mortality have been documented with D2 resections. (6,7) However, both of these trials have been criticised because of the lack of quality control between participating centres; it is not clear that all surgeons were doing the same operation when describing it as a D2 gastrectomy. Similarly, the morbidity and mortality reported from D2 gastrectomies within these trials (6,7) was far in excess of that reported in single institutions. (2-5) The case for D2 gastrectomy, therefore, must be regarded as unproven. This will remain the case until the procedure is standardised and evaluated in a trial. The purpose of this article, therefore, is not to advocate the performance or otherwise of this operation but rather to describe a standardised approach to a D2 gastrectomy. The authors have concentrated on the operation as performed for an antral tumour but will describe modifications, depending upon the position of the tumour. It should be noted at this stage that this operation is time consuming and that unless sufficient attention is paid to ensuring a good quality dissection the procedure will not achieve its goal of a systematic lymph node clearance.

Furthermore, there is great potential for morbidity and increased risk of mortality with such an extensive dissection and only by careful patient selection and meticulous technique can the results reported in the Japanese literature be reproduced.

The pattern of lymphatic involvement from gastric cancers has been best studied in Japan. From very detailed mapping of lymph nodes at the time of surgery Japanese surgeons have developed the concept of tiers of lymph nodes (see Table 1 and Figure 1). The following nomenclature is used:

Table 1: The incidence of lymph node metastases at each lymph node station
for tumours of the gastric antrum and fundus of the stomach (8)

Node station	Antrum	Fundus
1. Right cardiac	7	31
2. Left cardiac	0	13
3. Lesser curve	38	39
4. Greater curve	35	11
5. Suprapyloric	12	2
6. Subpyloric	49	3
7. Left gastric artery	23	19
8. Common hepatic artery	25	7
9. Coeliac artery	13	13
10. Splenic hilum	0	10
11. Splenic artery	4	12
12. Porta hepatis	8	1

It should be noted that there are differences between this staging system and the TNM staging system; readers are referred to discussion of this topic in larger texts.(9)

The extent of the lymphadenectomy can then be defined in terms of the level of en-bloc dissection of lymph nodes:

A D2 resection, therefore, will vary according to the position of the primary tumour. For example, for an antral carcinoma a D2 resection should include supra- and sub-pyloric lymph nodes but need not include lymph nodes at the cardia, whereas for a tumour of the gastric fundus nodes at the cardia are dissected as part of a D2 resection and the pyloric nodes are optional (Figure 2).

As with all surgery, patient selection is of paramount importance if good results are to be obtained from a D2 gastrectomy. Leaving aside the controversy as to the value of the procedure, even the most stringent advocates of radical lymph node resection would only feel it is warranted in patients in whom there was a reasonable chance of long-term survival. It is clear that this procedure should only be considered in patients with resectable tumours and without significant co-morbid disease. Accordingly, pre-operative assessment should concentrate on: (1) staging of the tumour and (2) evaluation of the patient’s fitness.

Many series have attested to the difficulty in pre-operative assessment of resectability with conventional staging modalities. (7,10) Percutaneous ultrasonography is usually good at detecting liver metastases. Assessment of the primary tumour and lymph nodes, however, is perhaps best carried out using a combination of water contrast computerised tomography (CT), and laparoscopic and endoscopic ultrasonography, although the latter is still not widely available. (11,12)

Special attention should be paid to assessment of pulmonary and cardiovascular function in patients being considered for major upper gastrointestinal surgery. As a base line, patients are screened with a chest radiograph, pulmonary function tests, arterial blood gases and an electrocardiograph. If there is any doubt regarding cardiac function it is usually prudent to obtain a functional assessment with an echocardiogram. Clearly, any correctable abnormalities are dealt with preoperatively and the patient’s general condition optimised.

It is important to assess each patient’s nutritional status and to determine a strategy for nutritional support. Many patients are malnourished at presentation and, even when recovery is straightforward, it is likely to be several days before an adequate diet can be ingested. As with all major abdominal operations thromboembolic prophylaxis is required; as a standard we would fit patients with graduated compression stockings and start subcutaneaous heparin with the induction of anaesthesia - the timing is important, particularly if an epidural anaesthetic is being used. We also use a single dose of prophylactic Augmentin (1.2g) administered with the induction of the anaesthetic.

For patients who have received pre-operative chemotherapy (as part of the MRC STO-2 trial), particular attention must be focussed on the pre-operative full blood count to ensure there is no evidence of continued myelosuppression.

Epidural analgesia is employed routinely in our unit to provide intra-operative and post-operative analgesia. The epidural catheter is inserted at the mid-thoracic level prior to the induction of general anaesthesia. A moderate volume (10mls/hr) of a low concentration of local anaesthetic (0.1% bupivicaine) will produce an appropriate sensory block in the majority of cases. Post-operatively, the epidural anaesthetic is supplemented with rectal paracetamol and diclofenac, together with patient controlled analgesia with morphine. This is commenced in the recovery unit.

General analgesia is induced with cricoid pressure. A double lumen endobronchial tube is only employed if a thoraco-abdominal approach is planned from the outset. Opioid analgesia is administered at induction and anaesthesia is maintained with a vapour in an air and oxygen mixture. Desfurane has proved useful and allows cricothyrotomy to be performed at the end of a long procedure immediately prior to the termination of the anaesthetic, if required. Extubation is undertaken when the patient is awake and in the sitting position after adequate ventilation and the return of airway reflexes are confirmed.

Throughout the procedure and in the post-operative period, invasive cardiovascular monitoring is employed. Urinary catheterisation (with hourly urine outputs), arterial and central venous lines are utilised as a routine; additional monitoring may be used in patients with high risk factors for cardiac disease.

In most cases a “roof top” or bilateral subcostal incision, to give maximal exposure to the hiatus and upper abdomen, is employed. After insertion of “Turkish” wound towels the costal margin is retracted upwards and outwards either with a multi-bladed Omnitract retractor (CLS Medical, Newcastle upon Tyne, UK) (Figure 3) -

or using two middle blades from a Balfour retractor secured with cord to the vertical limbs of a sternal bar (Figure 4).

An alternative incision which may be useful in patients with tumours at or around the gastro-oesophageal junction is a left thoraco-abdominal incision. Although this approach does give excellent exposure there is increased morbidity associated with it and it is probably best reserved for obese patients with a very deep chest.

As alluded to above, pre-operative tumour staging can be unpredictable. It is necessary, therefore, to perform an accurate intra-operative assessment of the tumour before deciding on the operative strategy. After opening the abdomen, distant metastases are sought in the liver or as peritoneal deposits which may be indicated by the pressure of peritoneal free fluid - particular attention is paid to the pouch of Douglas and the ovaries in a woman. Clearly, the detection of distant metastases will preclude further attempts at radical surgery. The tumour itself is assessed to determine: (1) the position in the stomach (antral, body or cardia); (2) the extent of the tumour (circumferential and involvement of more than one sector of the stomach) and (3) the depth of invasion (involvement of the serosa or otherwise, direct invasion of adjacent structures).

After inspection and palpation of the regional lymph nodes a decision is made as to the extent of the gastric resection (including the need to resect any adjacent organs - transverse colon, tail of the pancreas or right lobe of the liver). This will also allow a decision to be made regarding the advisability of a D2 lymphadenectomy. The strategy for this will be determined by the position of the tumour, as described above. In a patient with an antral tumour it is not usually necessary to perform a total gastrectomy to obtain adequate clearance and a sub-total gastrectomy can be performed, without compromising the extent of the lymphadenectomy. The presence of involved lymph nodes at the root of the mesentery or along the aorta would usually preclude further consideration of systemic lymph node dissection. Care must be exercised, however, as enlarged nodes may be reactive. If in doubt, samples should be sent for histological assessment of frozen sections. (13) In patients with tumours of the cardia or body of the stomach it has been our practise to perform total gastrectomies.

After a decision has been made to proceed with the resection, the first step is to mobilise the hepatic flexure of the colon from the underlying duodenum (Figure 5a),

The omentum is now separated from the colon and dissection is continued on to the transverse mesocolon - taking the anterior layer of the meso-colon with the omentum and taking care not to damage the colonic vasculature. Although this is said to be an avascular plain small vessels are frequently encountered which require to be coagulated with diathermy; in this regard bipolar dissecting scissors (PowerStar, Ethicon, Edinburgh, UK) can hasten an often tiresome dissection (Figure 6).

It is important to ensure that the correct plane is entered at the beginning of this dissection. To this end, it is often useful to start separating the omentum at the mid-colonic level and be guided by the middle colic vessels. It is helpful to lift the transverse colon and the omentum out of the abdomen and to have an assistant apply gentle traction on the colon as the omentum is lifted upwards. This dissection is carried to the right and the right gastroepiploic vessels are identified - the surrounding lymphatics (including the sub-pyloric lymph nodes) are swept up with the vessels; these are ligated with 2/0 Nurolon (Ethicon, Edinburgh, UK) at their origin (Figure 7).

The dissection is carried onto the pancreas taking the pancreatic capsule with the specimen, it is all too easy to get into the wrong plain at this point and find that the pancreas was been mobilised with the omentum!

If it is proving difficult to separate the pancreatic capsule at this point, attention is directed to the lesser omentum. The omentum is placed back in the abdomen and an assistant retracts the liver in a cephalid direction; it may be helpful to have a reverse Trendelenberg tilt on the table at this time. The reflection of the lesser omentum on the liver is divided (Figure 8),

starting at the hiatus and working to the right side (an accessory left gastric artery may have to be ligated in the process). A decision has to be made at this stage as to whether the porta requires dissection (the group 12 lymph nodes). This is certainly not required for tumours of the body or cardia and is not part of the standard D2 resection, even for tumours of the gastric antrum. However, approximately 10% of antral carcinomas will have nodes in the free edge of the lesser omentum (8) and dissection is carried out in the porta if the tumour is antral. In dissecting from above downwards the right gastric artery is ligated at its origin and divided. An elongated lymph node is encountered behind the common hepatic artery and the portal vein, gentle traction and rotation on this may be used to expose the posterior aspect of the hepatoduodenal ligament - care should be taken to ensure haemostasis.

If a total gastrectomy is being performed for a more proximal tumour then this sequence of events can be changed.

Dissection of the porta is unlikely to be required; after division of the lesser omentum it is often helpful to dissect the hiatus mobilising the lower oesophagus. For tumours at the oesophago-gastric junction a cuff of diaphragm should be resected in continuity with the tumour.

After ligation of the right gastric artery the duodenum is mobilised from the pancreas. At this stage, a number of small vessels are often encountered and should be ligated and not diathermised. The duodenum (>2 cm from the pylorus) can now be divided between two crushing clamps (Schoemaker or Payr’s clamps). The duodenal stump is closed with a continuous 3/0 PDS suture which is then inverted with a second 3/0 PDS layer of closure. Alternatively, a linear cutting/stapling device can be used to divide the duodenum (Figure 9); however, it is our practise to bury the staple line with a continuous 3/0 PDS suture.

Separation of the pancreatic capsule and dissection along the common hepatic artery is now carried out. The transected duodenum and stomach are lifted out of the wound and to the left, the common hepatic artery identified and cleared, removing the tissue inferior to this with the pancreatic capsule. It is important at this phase in the operation to be particularly fastidious about haemostasis. It may be helpful to display the plain with a right-angled clamp before division of the tissue. As the coeliac axis is approached, the left gastric vein is identified and ligated (Figure 10). Superiorly, the oesophagus and proximal stomach are retracted towards the left and the right crus of the diaphragm identified; dissection along the crus from above downwards allows the coelic trunk to be exposed from above (if a total gastrectomy has been planned then this will already have been performed).

The remainder of the omentum is freed from the transverse colon and mesocolon and the splenic hilum is approached from below. It is often useful at the beginning of this stage to divide adhesions to the lower aspect of the spleen so minimising the risk of traction trauma. The left branch of the gastroepiploic arcade is divided and the lowest short gastric vessels ligated and divided. The stomach and omentum can then be lifted forwards exposing the ceoliac axis and left gastric artery from below. The tissue from around the coeliac axis and splenic artery are swept forward and included with the left gastric artery which is double ligated with 2/0 Nurolon and divided. At the end of this dissection the pancreas is skeletalised as are the hepatic and splenic arteries (Figure 11).

The stomach, at this point, is prepared for the transection by ensuring that it is cleaned to its wall - this will involve dividing tissue adherent to the high lesser curve which will include the right cardiac nodes. On the greater curve side of the stomach the remaining short gastric vessels supply the gastric remnant. If a total gastrectomy is planned the remaining short gastric vessels are ligated as close to the spleen as possible and the remainder of the pancreatic capsule stripped from the pancreatic tail. With the stomach lifted upwards stay sutures are applied at the greater and lesser curves. The posterior serosa and muscular layers of the stomach are divided leaving the mucosa intact so that it is possible to identify and coagulate submucosal vessels with diathermy before finally dividing the mucosa. The posterior wall of the stomach can be anastamosed to the jejunum with the anterior wall being used to lift the stomach remnant upwards.

Our preferred method of reconstruction is with a retro-colic Roux-en-Y loop. This is fashioned from the first loop of jejunum - dividing the first arcade of vessels and dividing the loop with a GIA-60 linear cutting stapling device (Autosuture, Ascot, UK). The distal end is brought through a hole in the transverse mesocolon and anastamosed end to side with the gastric remnant. A continuous sero-mubmucosal suture using 3/0 PDS is carried out. As this is not haemostatic, it is essential that haemostasis is secured prior to the anastamosis being commenced.

At this stage the jejunal loop to be anastamosed is laid along the length of the stomach where the posterior gastrotomy has been made. A longitudinal jejunotomy, of equal length to the stomach remnant to be anastamosed, is fashioned. With the anterior wall of the stomach still attached, the posterior layer of the anastamosis is commenced. When complete, the anterior stomach wall is divided as above and a continuous sero-submucosal suture used to complete the join.

After a total gastrectomy a Roux-en-Y reconstruction is fashioned, but more mobilisation will be required than simply division of the first arcade of vessels, to ensure that the anastamosis is not under any tension. If a long loop is required it is useful to use soft bulldog clamps (such as Vascu-Statt arterial bulldog clamps, Scalan Int. Minnesota, USA) to occlude vessels, to determine the effect on the viability of the jejunum before permanently dividing them. An end-to-side oesophagojejunostomy is fashioned using 3/0 Nurolon vertical mattress sutures. Placement of the sutures is greatly helped by use of a suture rack (Gabbay-Frater suture guide, Genzyme surgical products, Lubeck, Germany).

A side-to-side jejunojejunostomy is fashioned 50 cm distal to the gastrojejunal (or oesophagojejunal) anastamosis, again using a continuous sero-submucosal technique with 3/0 PDS. Before closing the abdomen a double lumen nasojejunal (NJ) tube (Entrafeed NJ feeding tube, Medicina Ce, Berkshire, UK) is passed and positioned so that the end of the jejunal tube lies distal to the jejunojejunostomy and the gastric component lies across the gastrojejunostomy. The abdomen is closed in layers with 1 PDS and the skin closed with sub-cuticular 2/0 Vicryl or skin staples. Drains are not used routinely.

Initial post-operative care focuses on adequate analgesia, maintenance of fluid and electrolyte balance and chest physiotherapy. The patients are routinely nursed in a high dependency unit and monitored with hourly blood pressure, pulse, urinary volumes and central venous pressure recordings for the first 24 hours. Humidified oxygen is administered at an appropriate concentration to ensure an oxygen saturation of >95mmHg. (measured continuously with a pulse oximeter). The gastric lumen of the NJ tube is left on free drainage and the jejunal lumen spiggoted until there are signs that bowel sounds are returning; thereafter, feeding is introduced (Clinifeed, starting at 30ml/hr and increasing to full nutritional requirements).

In patients undergoing a sub-total gastrectomy, contrast radiology is not carried out, but oral intake is recommenced when bowel sounds are detected. The NJ feeding can be left in place until patients are taking an adequate oral intake. Prior to discharge from hospital, it is essential that the patient and their relatives understand the potential nutritional sequelae of the operation and the need to adjust their eating pattern to small frequent meals. The input of a dietician with community liaison can be particularly helpful in this regard.

Patients in whom a total gastrectomy has been performed will require three-monthly vitamin B12 and folate supplements. Those patients who had a subtotal gastrectomy carried out should have their full blood count and serum B12 and folate levels checked at regular intervals.

As with all major intra-abdominal operations, the non-specific early complications will include: deep venous thrombosis, wound infections and problems related to fluid and electrolyte balance.

Although not specific to gastric surgery, respiratory problems are particularly prevalent due to the upper abdominal incision, upward traction on the costal margins and pain. In the case of a thoraco-abdominal incision there is the added risk of one lung anaesthesia. Peri-operative physiotherapy, combined with adequate analgesia, are essential in minimising the risks involved. In high-risk patients, the perioperative insertion of a mini-tracheostomy can be extremely valuable.

This may be reactive or secondary to infective complications. Reactive haemorrhage may result from the gastric resection line, the extensive dissection or unrecognised damage to the spleen at the time of the operation. Prevention is obviously better than cure!

There are four possible sites for this; the duodenal stump, the gastrojejunostomy (or oesophagojejunostomy), the blind end of the proximal or distal jejunum (unlikely) or the jejunjejunostomy (unlikely). Leakage from the duodenal stump is usually due to afferent loop obstruction - this risk is reduced by the formation of a Roux-en-Y reconstruction. If there is a controlled leak without sepsis or generalised peritonitis conservative treatment is indicated. Otherwise, exploration is required. The aims of treatment should be: (1) to drain any sepsis; (2) to establish drainage of the afferent loop - a useful technique is to insert a T tube into the duodenum and, therefore, establish a controlled fistula, and to decompress the afferent loop obstruction. It is unlikely that direct suture of a leak will be feasible.

Treatment of a gastrojejunal or oesophagojejunal leak is problematic. If the leak has occurred early in the postoperative phase, then re-operation, peritoneal lavage and possible patching/resuturing may be possible. More commonly, the leak is delayed, occurring 7-14 days postoperatively. If there are clinical features to indicate that this is limited and localised, then conservative treatment with nutritional support, gastrointestinal decompression and antibiotics is indicated. If re-operation is required then it is unlikely that the tissues will support sutures and adequate drainage should be established together with peritoneal toilet. It should be noted at this point that leakage from an oesophagojejunal anastamosis is more likely (although hopefully a rare event) than from a gastrojejunal anastamosis following a sub total gastrectomy.

It is almost inevitable that patients having undergone a gastric resection will suffer nutritional consequences due to a reduced (or absent) stomach capacity. Unfortunately, the use of gastric pouches or jejunal interpositions have not significantly improved on the results of more standard reconstructions. In order to avoid megaloblastic anaemia patients having undergone a total gastrectomy will require vitamin B12 and folate replacement, in addition iron supplements may be required. Anastamotic strictures may lead to dysphagia in patients following a total gastrectomy. This is more likely to be significant in patients with a small oesophageal lumen. The aetiology of anastomotic strictures is likely to relate to either anastomotic leakage (which may be subclinical) or biliary reflux. The use of a Roux-en-Y reconstruction virtually abolishes biliary reflux as a problem.

Patients having undergone a D2 gastrectomy will inevitably undergo a vagotomy. This in turn may lead to post-operative problems with diarrhoea. Other causes of diarrhoea will include bacterial overgrowth and early dumping syndrome. Apart from diarrhoea, early dumping is associated with light-headedness, flushing and fainting. If this does occur most patients can be managed with dietary modifications and advice. Similar symptoms, which affect patients 1-2 hours after eating, may be due to a reactive hypoglycaemia (late dumping). Treatment is usually successful with dietary modifications.

Adequate preoperative assessment and preparation are essential
Trans-abdominal approach should be used, if possible; this may be facilitated by good retraction
Initial intra-operative assessment should be undertaken to determine the likelihood of a complete resection of the tumour and the optimal operative strategy. A D2 resection should only be considered in a patient in whom a Ro resection is possible (even then it remains controversial).
The operative strategy to achieve an en-bloc resection of N1/N2 tier of lymph nodes is dependent upon the position of the tumour
For an antral tumour the spleen and tail of the pancreas should not be resected routinely as part of a D2 gastrectomy
Attention to detail both intra-operatively and post-operatively is essential to reduce morbidity and detect any complications at an early stage
Careful follow-up is required to detect long-term complications and ensure adequate nutrition

Return to specialty index

Return to detailed index

REFERENCES

D2 gastrectomy for an antral stomach tumour A Qadir, C Trotter and KGM Park Aberdeen Royal Infirmary, Foresterhill, Aberdeen, U.K. J.R.Coll.Surg.Edinb., 45, June 2000, 242-251

Return to specialty index

Return to specialty index

D₂ gastrectomy for an antral stomach tumour
A Qadir, C Trotter and KGM Park
Aberdeen Royal Infirmary, Foresterhill, Aberdeen, U.K.
J.R.Coll.Surg.Edinb., 45, June 2000, 242-251