Mechanical plus oral bowel preparation with paromomycin and metronidazole reduces infectious complications in elective colorectal surgery: a matched case-control study

Colorectal surgery is an operation prone to complications, in which up to 30% of all cases involve infectious complications such as anastomotic leakage, organ space abscesses, and superficial surgical site infections being present in up to 30% of all cases [1, 2].

In recent years, there has been an ongoing discussion over the past years about the benefits of oral antibiotics (OA) with or without mechanical bowel preparation (MBP) on infectious postoperative complications following colorectal resections. Over 90% of US colorectal surgeons apply mechanical as well as decontaminating bowel preparation [3]. Based on literature review, there have been recommendations to omit MBP [4] as an analysis of existing data in 2010 could not provide benefits on surgical site infections or anastomotic leakage in open colorectal surgery. Large-scale data analysis from the USA, taken from the ACS-NSQIP, has shown that MBP plus oral antibiotics prior to elective colorectal surgery was able to reduce overall surgical site infections by about 50% (14.7% vs 6.2%) [5]. Several other studies have based their analyses on this particular registry with different aspects highlighted, but all judged antibiotic preparation beneficial [5‐12]. In general, a protective effect could be found in open and laparoscopic procedures [13]. A similar protective effect of oral antibiotic bowel preparation (OABP) without MBP could be shown in a meta-analysis that tried to point out the importance of OABP only for bowel preparation [14]. Recently, a large randomized controlled trial from Spain promoted the use of OABP without MBP to reduce SSIs in colorectal surgery [15].

In general, different antibiotic regimens are used in accordance with local availability of drugs and possible side effects. Most of the studies report the use of either an aminoglycoside (i.e., neomycine [16‐18], kanamycine [19], streptomycin [20], gentamicine [21]) or a fluoroquinolone (levofloxacine [1] or ciprofloxacine [15]) plus metronidazole. Additionally, in all present studies, single-shot antibiotics were used preoperatively, mostly consisting of an IV application of a second-generation cephalosporine plus metronidazole intravenously. The ideal decontaminating agent remains in the bowel after oral ingestion. In Germany, most of the aminoglycosides used in other studies are only available as topic agents but not in an oral formulation designed for intestinal effects. An old but still common aminoglycoside is paromomycin which is used as a soluble powder approved for the prevention of hepatic encephalopathy and which is available in Germany. The only study that evaluated paromomycin for bowel decontamination in colorectal surgery is about 40 years old but already provided evidence for the possible use in colorectal surgery to reduce surgical site infections [22].

Since no data is available on the combined use of paromomycin and metronidazole for preoperative intestinal decontamination in colorectal surgery to prevent infectious complications, we conducted this prospective case-control study with matched-pair analysis.

The aim was to show a significant reduction of postoperative infectious complications by using paromomycin and metronidazole preoperatively after mechanical bowel preparation in colorectal surgery.

Between 1 January 2019 and 31 January 2020, a total of 120 patients were found eligible for inclusion. Of those, 19 had to be excluded for various reasons (i.e., no OA received, refusal of informed consent, inability to consent, postponement of surgery). Thus, 101 patients (71 males, 30 females) were included in the study and completed the study protocol. After the completion of the trial period, we performed the matching process in accordance with the before-mentioned parameters. After the primary analysis of the main outcome parameter, i.e., reduction in SSI, we concluded further inclusion of patients as power levels were reached. A total of 92 of those 101 patients were matchable, excluding 5 male and 4 female patients because of the set matching variables’ limits (Fig. 1). The patient characteristics of our intervention group (IG) and the matched control group (CG) are listed in Table 1. The matching process resulted in equally distributed comorbidities.

Most common indications for surgery were colon and rectal cancer as well as sigmoid diverticulitis. This is reflected in the resections performed, with rectal resection (35.1%), sigmoid resection/left colectomy (35.1%), and right colectomy (25%) being the most common interventions (Table 2).

In the IG, the overall rate of total infections (15.2% vs 30.4%, p = 0.013) was significantly reduced with an odds ratio (OR) of 0.333 (95% CI 0.142–0.784; power 0.86) (Table 3). The incidence of superficial SSI (8.7 vs 19.6%, p = 0.041) was significantly lower in the IG with an OR of 0.333 (95% CI 0.121–0.917, power 0.74). The CG had predominantly superficial SSI involving the abdominal laparotomy wound (p = 0.049). Nonetheless, wound measures did not differ between both groups. With regard to the surgical approach, preoperative bowel decontamination showed a significant decrease in SSI in open surgery (p = 0.018), but a less remarkable effect in laparoscopic surgery (p = 0.263).

All other postoperative complications were similar in both groups except for cardioembolic events of which 3 occurred in the IG.

In this prospective matched case-control study, we sought to evaluate the effects of preoperatively administered oral paromomycin and metronidazole together with mechanical bowel preparation and IV ertapenem to reduce infectious complications in elective colorectal surgery.

Recent literature on preoperative oral antibiotic bowel preparation (OABP) heavily relies on results taken from the ACS-NSQIP registry for different colorectal procedures [5‐12]. These studies prefer OABP + MBP as protective regimen with regard to infectious postoperative complications. The recent meta-analysis carried out by Mulder et al. [14] excluded most of the ACS-NSQIP studies to prevent bias by multiple inclusions of the same patients. Nonetheless, they found a benefit of preoperative OABP. Those registry data only take into account if patients received any kind of antibiotic bowel preparation without paying giving any further detail in the specific regimen given. Unfortunately, those data only promote OABP but do not show any superior regimen so that colorectal surgeons cannot use those data as a proper guideline. Furthermore, OABP without MBP has to be taken into consideration, as recent evidence by the ORALEV trial exists, that the mechanical preparation maybe does not exert any additional benefits in reducing the postoperative infectious complications [14, 15].

Internationally, a wide variety of antibiotic regimen can be found. The common combination of an orally available aminoglycoside along with metronidazole, beginning 1 or 2 days preoperatively, is influenced by local drug legislation. For example, several Asian studies report the use of kanamycine [19, 26], whereas the Angloamerican studies mostly have used neomycine [27, 28]. In Germany, most of the studied aminoglycosides do not exist as orally available drug formulations. We therefore chose paromomycin as an orally available aminoglycoside, which has been proven to reduce infectious complications in colorectal surgery used in bowel irrigation preoperatively [22]. Nonetheless, data is scarce on its potential in combination with metronidazole, which is a common adjunct antibiotic in bowel preparation. Metronidazole is an antibiotic drug that especially is effective against anaerobic bacteria such as Bacteroides spp. and Clostridoides spp., common intestinal bacteria that are isolated from SSI [29]. In our intervention group (IG), we found an OR of 0.333 for total and superficial surgical site infections when taking MBP + OA, which is a similar effect compared to the study by Abis et al. [30], but superior to the results demonstrated by Mulder et al. [25, 31]. Abis et al. performed a proper selective intestinal decontamination by prolonged perioperative administration of the drugs for at least 6 days. Both groups used tobramycin and colistin for 1 and 3 days respectively, but Abis et al. also added amphotericin B as an anti-fungal agent. This raises the question of the need for additional anti-fungal bowel preparation. We found some yeast species in the swaps of the CG but none in the IG. This is most likely due to the small number of swaps taken in the IG. At least all fungi (Candida albicans and Candida krusei) would have been sensitive to amphotericin B. Unfortunately, the studies using amphotericin B as an oral decontamination agent do not describe wound cultures in order to justify the necessity of addition of anti-fungal agents.

We used ertapenem as a single-shot perioperative IV antibiotic. Other studies routinely used cephalosporines combined with metronidazole. We purposefully chose this first-generation carbapenem to reach a better coverage of the intestinal microbiota including extended-spectrum beta-lactamases (ESBL)–producing enterobacteria combined with the longer plasma half-life of ertapenem of about 4 h [32]. Additionally, ertapenem is effective against most of the gram-negative bacteria that often cause SSI in colorectal surgery (i.e., Escherichia coli, Klebsiella pneumonia), gram-positive anaerobes, and gram-positive cocci such as Streptococcus pneumoniae [29]. The regime of cefuroxime and metronidazole is less effective against those gram-positive and gram-negative anaerobes as well as enterobacterales according to the most recent EUCAST report [33]. Even some Enterococcus faecalis strains are susceptible to ertapenem. The only available study comparing different single-shot antibiotic regimen in colorectal surgery found a slightly higher SSI rate using ertapenem compared to cefazoline plus metronidazole (OR 1.48, not significant) [34]. Overall, they attributed the greatest effect in SSI prevention to the oral decontamination and not to IV single-shots. By using ertapenem instead of cefotaxime preoperatively, we could cover all resistant gram-negative bacteria, which we had identified preoperatively by routine screenings. Unfortunately, ertapenem does not cover Pseudomonas aeruginosa which may be responsible for AL development under certain circumstances [35]. Still, it seems impossible to cover any possible bacteria with a single-shot antibiotic with the intention to prevent SSI, especially since some of the data on Pseudomonas aeruginosa and Enterococcus faecalis and their role on anastomotic leakage are in vitro experiments [35, 36]. After all, you have to keep in mind that a responsible administration of reserve antibiotics to prevent bacterial resistance generation is mandatory, and therefore, clinical evaluation of bacterial prevalences in SSI is necessary.

The most pronounced effect of preoperative bowel preparation could be seen in the open surgical group, which was by far the minority of surgical procedures (14%). The respective infection rates were 70% compared to 23% in the laparoscopic surgery group. This is in line with common knowledge on the advantages of laparoscopic surgery. Nevertheless, our results reflect those of Özdemir et al. [21] who evaluated OABP (gentamycine and metronidazole) in open colorectal surgery. They also demonstrated a decrease in SSI from 71 to 36% by using OABP and calculated subsequent savings of about roughly 1700 USD per patient per hospital stay. Similarly, Suzuki et al. [19] found the majority of infections in the open surgical subgroup of their cohort of both open and laparoscopic colorectal cancer surgeries, but they attributed the reduction in infection rates to the addition of mechanical bowel preparation. The SSI rates we found in the CG are relatively high but still in the range of the aforementioned studies. But this elevated rate of SSI stresses the need for measures to improve postoperative patient outcome with regard to infectious complications.

Reduction of anastomotic leakage (AL) has been considered a primary end point in several studies that evaluated preoperative bowel decontamination. Schardey et al. [37] terminated their study after an interim analysis because of a significant reduction in AL in rectal resections. They used a very broad combination of topical antibiotic agents comprising tobramycin, vancomycin, polymyxin B, and amphotericin B versus placebo plus amphotericin B. Their intention was to double cover gram-positive microorganisms after elimination of gram-negative flora. Another randomized study to assess anastomotic leakage in colorectal surgery was recently carried out in the Netherlands [31]. The study was terminated prematurely, as an OR of 0.8 for deep SSI and 0.5 for anastomotic leakage was found. At the point of study cessation, they had included 80 of 966 estimated patients and found 1 versus 2 anastomotic leakages. Hence, it seems doubtful if the authors’ calculated ORs really represent a statistical effect that justifies the termination of the study. Other data hints at reduced AL rates after MBP + OA. Even less aggressive bowel preparation regime (neomycine and metronidazole) could show a reduction in AL in a very recent randomized trial [38]. But almost simultaneously, a different study group published contradictory data with no reduction in AL but slightly lower rates of SSI [39]. We could show a reduced AL rate with less severe AL grades in the IG which did not reach significance levels. The necessary treatment differed remarkably with 5 relaparotomies in the CG versus 1 in the IG, but AL treatment overall did not reach significant levels (p = 0.287). This difference is due to the different localization of anastomosis, i.e., rectal versus colonic anastomosis. The detected trend towards a shorter LOS in the IG might also be attributed to the higher rate of reoperation. Although, conservative treatment of an AL (grades A and B) with endoscopic sponge therapy might be even more time-consuming until a sufficient granulation of anastomosis is assured. In total, 10 procedures per group were performed without an anastomosis reducing the subgroup for the analysis of AL. Therefore, a greater cohort is needed to demonstrate a statistically significant effect of our bowel preparation regime with regard to AL. Still, our data taken together with existing data provides diverse recommendations with regard to AL reduction which further stresses the necessity of studies on the best antibiotic regime for perioperative bowel preparation.

Regardless of the statistically significant changes due to the introduction of OA with MBP + OA discovered in our study, the external validation using the national surveillance data of our institution confirms our observations and stresses the importance of MBP + OA.

Our study has several limitations. First and foremost, we conducted a prospective study with matching of a historical control group. The matching process was performed by the authors manually in a blinded fashion from a patient list maintaining the matching variables only. Due to the manual selection pattern, a possible matching bias cannot be excluded. The inclusion of several rare procedures such as proctocolectomy or Hartmann’s reversal might influence the results as well. However, since those patients were properly matched, we do not assume great effects, especially since those procedures consist of the main surgical steps as the more common procedures (i.e., surgical approach, extracorporeal bowel resection, transanal anastomosis, placement of ostomy).

As several studies hint at the benefits of OABP in the prevention of infectious complications in colorectal surgery, future studies should focus on the best antibiotic regimen for preoperative bowel decontamination depending on the microbiological background epidemiology. To date, the inhomogeneity of the regimens used with respect to antimicrobials and drug dosages is still a major problem in the field of OABP.

Based on the results of our matched-pair analysis, we propose oral paromomycin and metronidazole in single doses for oral antibiotic bowel preparation, paired with a single shot of IV ertapenem preoperatively in colorectal surgery as a promising option to effectively reduce postoperative infectious complications.