All patients presented with severe abdominal pain. Pneumoperitoneum on erect chest radiographs was seen in five (41.7%) patients while emergency confirmatory computed tomographic (CT) scans were performed in the rest. Majority (n = 9, 75.0%) of patients underwent surgery within 24 hours of presentation. Table 2 highlights the various characteristics of the study group.

Intra-operatively, seven (59.3%) patients have severe peritoneal contamination. Ten (83.3%) had partial or subtotal gastrectomy performed with Bilroth II anastomosis, while the remaining two (16.7%) underwent total gastrectomy with a resulting Roux-en-Y anastomosis.

Two patients died from septic complications contributed by pneumonia and intra-abdominal sepsis, one of whom had a duodenal stump leak which necessitated a subsequent laparotomy, drainage of the intra-abdominal collections and repair of duodenal stump dehiscence. The remaining ten patients were discharged well after a median length of stay of 16 (range: 8~100) days. Table 3 illustrates the surgical observations, procedure and outcome.sification proposed by Clavien and group (8-10) (Table 1).

Though malignancy has been quoted as an independent factor predicting worse outcome in gastric perforation, other more commonly associated adverse factors would include pre-operative shock, poor pre-morbid condition, advanced age, delayed presentation and resection surgery (11-16). Over the years, several scoring systems have been advocated in the prognostication of patients with gastric perforation, with Boey score being commonly adopted and validated in several reports (15,16).

Boey score utilized three independent factors of concomitant severe medical illness, pre-operative shock and long-standing perforation with predicted mortality rate of over 80% if all three factors are present. However, one of its main criticisms has been its inability to consider other physiological and intraoperative parameters. This has resulted in the numerous other scoring systems such as the Mannheim peritonitis Index (MPI), ASA score and APACHE II being adopted, each with its advantages and limitations. Suffice to say, the outcome in these patients are dependent on a combination of patient, disease and surgeon factors.

To make matter worse, in the absence of a known pre-operative gastric malignancy, it may be difficult to accurately diagnose the presence of malignancy in any gastric perforation (1,2). Mistaking a benign ulcer perforation as malignant is not impossible given the significant surrounding induration and enlarged inf lammatory lymph nodes. This may subject the patient to an unnecessary extensive and resection surgery with its numerous associated complications (1-6,17). Some of the clues suggestive of a malignant perforation would include advanced age, size of ulcer > 6cm and size of perforation > 0.5cm, raised white cell counts and longer duration of symptoms (1). The importance of frozen section intraoperatively has been emphasised to clinch the diagnosis but it may not be always available and false negative is also possible. In our series, frozen section was not performed in any patients as it was either not available or deemed not necessary by the primary surgeon because of the size of the ulcer and perforation, or if the malignancy was clinically suspected or already diagnosed. These would have supported the decision for gastrectomy regardless of the outcome of frozen section.

Even when the malignant perforation could be accurately diagnosed, the surgical procedures of choice in these patients are often dependent on various factors. These would include the presence of metastatic disease, expertise of the surgeon in performing an oncologic resection, the degree of contamination and perhaps most importantly, the intra-operative haemodynamic status of the patient.

At one stage, malignant gastric perforation has been deemed as terminal disease due to the associated peritoneal dissemination and early recurrences (18-20). This had led to the practice of simple closure of the perforation (21,22). However, this technique has been associated with unacceptable peri-operative complications and hence abandoned. Perhaps this should only be considered when the patient is extremely haemodynamically unstable to withstand any resection.

Over the years, the morbidity following emergency gastrectomy has been improving due to improving surgical technique and advancement in critical care (23). This has become the preferred surgical option in patients with malignant gastric perforation. Not only is it able to tackle the perforation, it can also remove the underlying pathology. However, the extent of radical oncologic surgery is perhaps dependent on the aforementioned factors. While it may be dangerous to embark on a major radical oncologic resection, the implications of a limited procedure may seriously impact the long term survival in patients with potentially curable gastric malignancy. This had led to the adoption of a two-stage procedure in handling this perplexing situation (3,24). While the first stage aimed to tackle the peritoneal contamination and the gastrectomy, the second procedure would be performed at a later date to ensure adequate lymph node clearance. However, the problems of such a staged procedure would include the significant postoperative adhesions from the first surgery, and also the fitness of the patient to withstand another extensive surgery. In addition, this could delay the commencement of any chemo- and radio-therapy, especially if any complications were encountered.

Recent data have disproved the notion that gastric perforation often resulted in increased risks of recurrences and peritoneal disease. The long term survival of patients with per forated gastric adenocarcinoma is actually comparable to patients performed electively (3-6). The only factor determining long term survival is the stage of the malignancy. As seen in our series, the majority of our patients had very advanced disease on diagnosis and fared badly subsequently with almost all the patients developing disease recurrences. Though several of our patients developed peritoneal disease subsequently, it could be related to the advanced staging and progression of the primary malignancy rather than contributed by the perforation. Unfortunately, large series is not available in the literature to shed more light into this.