Introduction

External fistula formation is a serious complication of gastrointestinal and pancreatic surgery since mortality ranges between 5 and 30% in the case of enterocutaneous and between 5 and 10% in the case of pancreatic fistulae₁,₂,₃. Spontaneous fistula closure occurs in most cases when essential treatment principles are followed, such as elimination of sepsis, adequate drainage, proper nutritional support, correction of fluid and electrolyte disturbances, and skin care. However, significant morbidity and mortality rates, increased length of hospitalisation, the cost and the psychological consequences to the patient render the conservative medical treatment insufficient₁,₄.

Somatostatin (SMS) and its synthetic analogue, octreotide (SMS 201-995, sandostatin), are known vasoactive drugs that reduce gastrointestinal, biliary and pancreatic secretions₅,₆,₇. These properties render them good candidates for the treatment of gastrointestinal and pancreatic fistulae. Their effectiveness has been examined in several clinical trials₇,₈,₉,₁₀,₁₁,₁₂,₁₃,₁₄. The results, though, have been rather inconclusive. This study was conducted in order to evaluate the potential benefit from the administration of these two agents to patients with gastrointestinal and pancreatic fistulae, as well as to compare them to conservative treatment with total parenteral nutrition (TPN) alone.

Patients And Methods

In this prospective, randomised controlled study, 51 consecutive patients with gastrointestinal (38 patients, 74.5%) or pancreatic fistulae (13 patients, 25.5%) were included. All patients were treated in the surgical clinic at hippocration hospital, athens medical school, greece, between december 1995 and april 1999. They received standard medical treatment (TPN, skin care, infection control) plus SMS (16 patients), or octreotide (17 patients), or standard medical treatment only (control group, 15 patients). SMS was given by continuous intravenous infusion at a dose of 6000 IU/day, while octreotide was given at a dose of 100 µg three times a day (tid), subcutaneously. The local ethics committee approved the design of this study.

Demographic and clinical characteristics were recorded for all our patients (table 1). The diagnosis of fistula origin was done, when necessary by ct-scan or fistulography (Table 1). The classification of the fistulae was done according to the type, pancreatic or enteric. Enterocutaneous fistulae were further subdivided by origin (Table 1). Fistulae were also classified as high or low output on presentation (Table 1). Enterocutaneous fistulae were classified as high-output fistulae when the daily output on presentation exceeded 500 ml, while pancreaticocutaneous fistulae were considered as high-output fistulae when the daily output exceeded 200 ml₁₂,₁₅.

The parameters recorded in order to evaluate the results from the administration of these two agents were: length of hospital stay, days until restoration of regular per os nutrition, complications, fistula outcome on discharge (spontaneous closure or surgical/other treatment), days until fistula closure and outcome on discharge. Length of hospital stay, time until restoration of per os nutrition and time to fistula closure, were recorded from the day of fistula onset for patients with a fistula of recent onset (41 patients, 80.4%) and from the day of admission for patients with a fistula of non-recent onset, respectively (more than 8 days prior to admission, 10 patients, 19.6%). The total cost for treatment was calculated based on data provided by the administrative department of our hospital. The figures provided included the total cost of the admission to the health insurance agency.

The statistical tests used were the chi-square test for categorical data and the Kruskal Wallis test for numerical data, while a p of less than 0.05 was considered to indicate statistical significance. All tests were double-sided.

Results

The three patients' groups that were recruited in this study did not actually have the same demographic characteristics. Age varied significantly among the three groups (Kruskal Wallis test, p = 0.049, Table 1), the patients in the standard medical treatment group being older (median age 72 years), followed by the patients in the octreotide group (median age 70 years) and in the SMS group (median age 61.5 years). Due to the relatively small number of patients in each subgroup and the p--value of 0.049 being close to the border of 0.05 this statistically significant result was not considered to reflect non-homogeneity between our groups. Accordingly, the subgroups were well matched for gender (chi-square = 0.671, p = 0.715, Table 1).

The clinical characteristics of the three subgroups in our study were similar. Enterocutaneous fistulae outnumbered pancreatic ones in all groups and no differences could be established between the three groups (chi-square = 2.8, p = 0.253, Table 1). The low- to high- output ratio was approximately 1:1 in all groups, and again no statistically significant differences could be established (chi-square = 1.245, p = 0.537, Table 1).

Figure 1

Table 1: Demographic and clinical characteristics of 51 patients with enterocutaneous or pancreaticocutaneous fistulae, treated with somatostatin, octreotide or standard medical treatment only

Median length of hospital stay was 20 days (Table 2) and was 15, 24 and 28 for the somatostatin, the octreotide and the standard medical treatment group, respectively. The difference was of statistical significance (Kruskal Wallis test = 7.42, p = 0.024, Table 2.

Figure 2

Table 2: The comparative evaluation of somatostatin, octreotide and standard medical treatment groups, through several characteristics

A difference between the two therapy arms was not established (Mann Whitney u-test, z = -1.745, exact p = 0.08). The time interval until restoration of per os nutrition also did not vary significantly between the three subgroups (overall median 13.0 days, Kruskal Wallis test = 1.6, p = 0.448, Table 2). Similar results were obtained when the outcome on hospital discharge was evaluated. Overall death rate was 3.9% (2 out of 51 patients) with both deaths having occurred in the standard medical treatment group. However, this difference was not statistically significant.

Overall complication rate in our series was 49.0% (25 out of 51 patients), while in the standard medical treatment group it was 66.7% (10 out of 15 patients). In the somatostatin and octreotide group, the complication rate was 42.1% (8 out of 19 patients) and 41.2% (7 out of 17 patients), respectively. Even though variation in the complication rate between the three groups was seen, these differences were not statistically significant (chi-square test = 2.116, p = 0.347). Overall closure rate was 60.1% (31 out of 51 patients). In the standard medical treatment group it was 26.7% (4 out of 15 patients), while in the SMS and octreotide group it was 84.2% (16 out of 19 patients) and 64.7% (11 out of 17 patients), respectively, with these differences being statistically significant (Kruskal Wallis test, p = 0.003). Conversely, the difference in closure rate between the SMS and the octreotide group was not statistically significant (fisher's exact test, p = 0.255). The median time for fistula closure was 16 days (Table 1i), while in the standard medical treatment group it was 18 days, in the SMS group 10.5 days and 16.5 days in the octreotide group, respectively. These differences did not reach statistical significance (Kruskal Wallis test, p = 0.140).

The median cost of hospital stay was €6944. A small variation was observed between the three subgroups and median cost was €6636, €7111 and €7812 for the somatostatin, the octreotide and the standard medical treatment group, respectively. These differences were not statistically significant (Kruskal Wallis test, p = 0.459).

Discussion

Gastrointestinal and pancreatic fistula formation is a major complication of abdominal surgery associated with significant morbidity and mortality. The use of total parenteral nutrition was a major advancement in the field, since it improved spontaneous fistula closure rates from 10-20% to 60-70% and reduce mortality from 16-62% to 10-20%₂,₁₆. Nevertheless, the prolonged hospital stay along with the improved, but still significant mortality of fistulae, renders them a challenge in modern therapeutics. The reduction of gastrointestinal and pancreatic secretions has been a major goal in the pursuit of the optimal treatment and SMS, as well as octreotide, have been used on this basis with variable results. In several centres, the administration of octreotide or somatostatin has been proposed as the treatment of choice for gastrointestinal and pancreatic fistulae, since they have been reported to reduce time of closure, despite the non-significant influence on mortality and complication rates₉,₁₂. In the present clinical trial, 51 patients with gastrointestinal and pancreatic fistula were recruited in order to evaluate and compare the potential clinical benefit from the administration of SMS and octreotide.

The overall death rate of less than 4% (2 out of 51 patients), the fistula closure rate of approximately 60% (31 out of 51 patients) and the median time to fistula closure of 16 days observed in our study are within international standards₁₇. Even though we were unable to detect statistically significant differences in terms of mortality and complication rate, the differences in closure rate were statistically significant (p = 0.003). In fact, spontaneous fistula closure rate was 27% in the group receiving standard medical therapy only, much less than the SMS and the octreotide groups (84% and 65%, respectively). This difference in closure rate has not been reported in certain comparative studies₉,₁₂,₁₈. This could be attributed to the standard medical group's relatively low closure rate in our study (27%). Nevertheless, the aforementioned studies reported that SMS led to strongest inhibition of fistulae output and to significantly lower healing periods. On the other hand, in accordance with our results, one prospective study comparing standard medical treatment with SMS for postoperative enterocutaneous fistulae, demonstrated the significant superiority of SMS in rates and time to closure ₁₁ . In addition, a multicenter, prospective, double-blinded, placebo-controlled trial with octreotide showed statistical significance in the difference of spontaneous closure rates of 57% and 35% in the octreotide and in the standard medical treatment group, respectively₈. The latter, though, could not be confirmed by another european study₁₉.

The differences in complication rate between the control group and the two groups receiving SMS and octreotide were noticeable, despite not being statistically significant. While in the control group the complication rate was 67%, in the somatostatin and the octreotide group it was 42% and 41%, respectively. These results, along with the fact that no adverse effects were observed for either drug, are consistent with the idea that there is a clinical benefit from the administration of these two agents. However, the number of patients included in this study is not sufficient to enable satisfactory statistical power and thus this clinical benefit could not be shown in terms of mortality and complication rate.

An important parameter in the treatment of fistulae is the cost--effectiveness of the therapeutic modality ₂₀,₂₁. For SMS and octreotide, the issue of their significant cost has always been a major drawback for their clinical use. We could not find differences in the cost of therapy based on economic data provided by the administrative authority of our hospital. It appears that therapy with either somatostatin or octreotide at a similar cost to standard medical treatment achieve higher closure rates at a shorter length of hospital stay.

International literature lacks data on direct comparison between SMS and octreotide as far as it concerns fistulae management. In the current study the comparison between SMS and octreotide yielded insignificant results for all the recorded parameters, which could also be attributed to the fact that only a small cohort was included. SMS achieved a better spontaneous closure rate (84% vs 65%) than octreotide and provided equivalent results in terms of complication rate (42% vs 41%). Let us point out that SMS is a useful and effective therapy for severe acute pancreatitis and in preventing complications following ercp, whereas octreotide has no beneficial effect and may be deleterious in both these indications ₁₇ . Altough SMS and octreotide are both effective in promoting the closure of pancreatic fistulae, the time to closure after commencement of therapy is much more variable and longer in patients treated with subcutaneous octreotide than those receiving intravenous SMS, possibly as a result of fluctuations in pancreatic enzyme secretion between consecutive administration of the hormone ₁₇ . It has also been shown that subcutaneous administration of octreotide produces a sustained decrease in the volume of pancreatic juice secreted, but enzyme secretion rises progressively between consecutive administration of the analogue, which may delay the healing of fistula tract ₂₀ .

Conclusion

In conclusion, 51 patients with gastrointestinal and pancreatic fistulae were enrolled and received SMS, octreotide or nothing in combination with standard medical treatment. The spontaneous fistula closure rate was significantly higher in the subgroup of patients receiving either SMS or octreotide compared to the controls at a similar cost. SMS might also have a superior effect over octreotide and could be regarded as the pharmaceutical agent of choice in the conservative management of gastrointestinal fistulae.

Correspondence to

Panoussopoulos D. 21 LYDIAS STR CHALANDRI, ATHENS 15232 GREECE Email: dpanous@med.uoa.gr Tel/fax: +2106833755