There are few randomized clinical trials about the use of antibiotics in AP. Pederzoli et al. in 1993 [¹⁶] conducted the first of them, which analyzed 74 patients with SAP in six medical centers in Italy. The patients were randomized in two groups: one control with 33 patients and the treatment group with 41 patients that received Imipenem 3 ? 500 mg/day intravenously (i.v.) during 14 days. They observed reduction in pancreatic sepsis in patients that received antibiotics (30.3% vs. 12.2%, p < 0.01). However, they did not find reduction in mortality between the groups (12% vs. 7%, respectively).

Sainio et al. carried out another study in 1995 [¹⁷], analyzing data from 60 patients with SAP. One group control and the other treated with Cefuroxime 3 ? 1.5 g/day i.v. They observed reduction in the number of pancreatic operations (36 vs. 8, p = 0.012) and in mortality (7 vs. 1, p = 0,028) in the treatment group. However, 66.7% of patients in this group had their antibiotics changed during the treatment. Furthermore, 23 out of 30 patients in the control group received antibiotics in an average period of six days after the beginning of the treatment. The use of this cephalosporin is also questionable due to low pancreatic penetration.

Delcenserie et al. in 1996 [¹⁸] analyzed prospectively 23 patients with alcoholic SAP that were divided into two groups: one control and one that was treated with Ceftazidime 3 ? 2 g/day i.v., Amicacin 2 ? 7.5 mg/kg i.v., and Metronidazole 3 ? 500 mg i.v. for 10 days. They observed a decrease in septic complications in the treatment group (7 vs. 0, p = 0.03). Nevertheless, there was no difference in mortality rate between the groups. In another clinical trial, Schwarz et al. described a decrease in pancreatic infection in a group of 29 patients, utilizing Ofloxacin and Metronidazole (7% ? 46%, respectively) [¹⁹].

In summary, the studies of Pederzoli, Delcenserie e Schwarz demonstrated reduction in pancreatic infection, although without decrease in mortality. Only Sainio observed reduction in mortality with the use of antibiotics. In spite of these results, a solid answer to the question about the use of antibiotics in AP was still missing [¹,²⁰] (Table 1).

In 2004, the Ulm group in Germany published the best study regarding the use of antibiotics in AP [²¹]. It was a prospective randomized double-blind trial that analyzed 114 patients, 58 that received antibiotics (Ciprofloxacin 2 ? 400 mg/day i.v. + Metronidazole 2 ? 500 mg/day i.v.) and 56 that received placebo. It was established that if any patient developed systemic inflammatory response, organic failure, any kind of infection or clinical deterioration, this patient would be discontinued from the protocol with open antibiotic treatment. Their results showed that the use of antibiotics did not reduce pancreatic infection (12% antibiotics vs. 9% placebo, p = n.s.) and mortality (5% antibiotics vs. 7% placebo, p = n.s.). However, 28% of patients in the group that received antibiotics had their protocol opened, versus 46% in the placebo group (p = 0.037). Furthermore, the mean time to open the protocol was 11.5 days in the treatment group and 5 days in the placebo group (Table 2). Based on these data, it is reasonable to suppose that a group of patients had benefits receiving early antibiotics. This is the same opinion of other authors [²²,²³].

Although the risk for the development of pancreatic infection is higher in the third week, microorganisms can be found in the pancreatic tissue in the first week [²⁴].

De Souza et al. demonstrated in an experimental model that bacteria could be detected in the pancreatic tissue six hours after the induction of AP [²⁵]. Similarly, Schwarz et al. observed bacteria in the pancreatic necrosis between 8 and 16 hours after the pancreatitis induction [²⁶].

These data suggest that the best time to introduce antibiotics is immediately after the diagnosis of AP and the evaluation of its severity. However, the confirmation of necrotizing pancreatitis by CT scan can take at least 72 hours since the onset of symptoms. C-reactive protein is a sensitive marker of pancreatic necrosis and it starts to increase significantly 48 hours after the onset of symptoms. Thus, C-reactive protein can be useful in the identification of patients with high possibility to develop necrosis, in particular when the value is over 150 mg/dl, and subsequently define which patients are candidates to receive early antibiotics [²⁷].

Nordback et al, in a clinical trial observed advantages in the early use of Imipenem in patients with pancreatic necrosis when compared with the late use [²⁸].

The main arguments against the use of antibiotics are the increase of fungal infection, and the increase of bacteria resistance, with more Gram-positive infections [¹].

There are evidences in the literature that fungal infection has to be considered as an additional factor which influences the outcome of the patients [²⁹].

One paper analyzing data of 46 patients with infected pancreatic necrosis receiving antibiotics showed that 17 (37%) of them developed fungal infection. Nevertheless, the fungal infection did not contribute to increase mortality in this group of patients [³⁰].

Gloor et al. analyzed 103 patients with necrotizing AP that received antibiotics (Imipenem/cilastatin) and concluded that fungal infections when treated properly do not contribute to a worst prognosis. This same paper showed that the presence of multi-resistant bacteria was rare (2.9%) [³].

Several guidelines about the treatment of AP can be found in the literature, and the majority of them recommend the early use of antibiotics in patients with pancreatic necrosis (Table 3) [^31-39]. The International Association of Pancreatology guidelines for the management of AP recommends the use of prophylactic broadspectrum antibiotics to reduce infection rates in CT-proven necrotizing pancreatitis as a recommendation grade A [³¹]. The Cochrane review about the use of antibiotics in AP concluded that despite variations in drug agent, duration of treatment and methodological quality of clinical trials, there is strong evidence that intravenous antibiotic prophylactic therapy for 10 to 14 days decrease the risk of superinfection of necrotic tissue and mortality in patients with SAP with proven pancreatic necrosis at CT [³⁷].

Moreover, the UK guidelines for the management of AP advocate that the risk of infected necrosis is very small when there is less than 30% necrosis [³⁴]. Although there is no support in the literature, it is reasonable to assume that prophylactic antibiotic therapy should be considered only for patients with more than 30% pancreatic necrosis. In contrast, there is no evidence in the literature that the use of antibiotics in the absence of necrosis is beneficial.

Based on these studies we believe that the use of antibiotics is benefic to patients with SAP with more than 30% pancreatic necrosis, and starting as soon as possible.