Bacterial ventriculoperitoneal shunt infections: changing trends in antimicrobial susceptibility, a 7-year retrospective study from Pakistan

Abstract

Background

Ventriculoperitoneal (VP) shunt infections in adults represent a severe complication and make treatment more challenging. Therefore, drug susceptibility patterns are crucial for therapeutic decisions and infection control in neurosurgical centers. This 7-year retrospective study aimed to identify the bacteria responsible for adult VP shunt infections and determine their drug susceptibility patterns.

Methods

This single-center study was performed from 2015 to 2021 in Lahore, Pakistan, and included CSF cultures from VP shunt infections. Demographic data, causative organisms, and antimicrobial susceptibility testing results were collected. Multivariate analysis of variance (MANOVA) and two-sample t-tests were used to analyze and compare the antibiotic sensitivity trends over the study period.

Results

14,473 isolates recovered from 13,937 CSF samples of VP shunt infections were identified and analyzed for their susceptibility patterns to antimicrobials. The proportion of Gram-negative and Gram-positive bacteria were 11,030 (76%) and 3443 (24)%, respectively. The predominant bacteria were Acinetobacter species (n = 5898, 41%), followed by Pseudomonas species (n = 2368, 16%) and coagulase-negative Staphylococcus (CoNS) (n = 1880, 13%). 100% of Staphylococcus aureus (S.aureus) and CoNS were sensitive to vancomycin and linezolid (n = 2580). However, 52% of S. aureus (719/1,343) were methicillin-resistant Staphylococcus aureus (MRSA). Acinetobacter showed maximum sensitivity to meropenem at 69% (2759/4768). Pseudomonas was 80% (1385/1863 sensitive to piperacillin-tazobactam, Escherichia coli (E. coli) showed 72% to amikacin (748/1055), while Klebsiella spp. was 57% (574/1170) sensitive to piperacillin-tazobactam. The sensitivity of piperacillin-tazobactam and meropenem for Gram-negative bacteria decreased significantly (p < 0.05) over 7 years, with 92.2% and 88.91% sensitive in 2015 and 66.7% and 62.8% sensitive in 2021, respectively.

Conclusion

The significant decrease in the effectiveness of carbapenem and beta-lactam/beta-lactamase inhibitor combination drugs for the common Gram-negative causative agents of VP shunt infections suggests that alternative antibiotics such as colistin, fosfomycin, ceftazidime/avibactam, ceftolozane/tazobactam, and tigecycline should be considered and in consequence included in testing panels. Additionally, it is recommended to adopt care bundles for the prevention of VP shunt infection.

Introduction

Ventriculoperitoneal (VP) shunt insertion is one of the most frequently performed neurosurgical interventions worldwide [1, 2], where shunt insertion can restore the elevated intracranial pressure associated with hydrocephalus and is chiefly used for its management [3, 4]. Unfortunately, complications related to the VP shunt placement are common. One severe consequence is the development of infection after shunting, despite the availability of new antibiotics and advanced neurosurgical techniques. Shunt infection rates range from 5 to 15% of the patients undergoing the procedure and are often associated with adverse outcomes [5]. Independent risk factors for VP shunt infections include the initial indication of shunt placement, revision or replacement for dysfunction, previous shunt-associated infection, postoperative CSF leakage, extreme age groups, procedure duration, the neurosurgeon's experience, and use of a neuro endoscope [6,7,8] More than 60% of the shunt infections occur within the first four to five weeks after the shunt placement. However, late shunt infections after some years are also observed [1, 2, 7]. Early shunt infections are often initiated during shunt insertion whereas, late infections are associated with unconnected pathologies, e.g. peritonitis and bowel perforation [7, 9]. The VP shunt infection frequently leads to ventriculitis and meningitis [10]. Any delay in effective treatment can have a poor prognosis with a mortality rate of 20–50% [10, 11]. Therefore, when the infection is clinically apparent, antimicrobial therapy should be started immediately along with the removal of shunt where applicable [12, 13]. and empirical antimicrobial treatment based on regional epidemiology, the prevalence of potential bacteria, and antimicrobial susceptibility patterns is essential [14, 15]. During the last decade, the infectious bacterial spectrum in VP shunt infection has started shifting from previously common causative agents such as Staphylococcus aureus, coagulase-negative Staphylococcus, and Enterococcus Gram-positive bacteria, to Gram-negative bacilli, especially Acinetobacter species, Pseudomonas species, and Enterobacterales [10].

Prescribing empirical antibiotics for this acute illness remains challenging. Antimicrobial resistance (AMR) surveillance data of microorganisms and the antibiotic susceptibility profile should be made available as a limited number of drugs can penetrate the central nervous system (blood–brain barrier). The emergence of multi-drug resistance (MDR), extensively drug-resistant (XDR), and even pan-drug-resistant microorganisms is catastrophic [11]. The mortality rate can extend to 60% to 70% in neurosurgical infection with carbapenem-resistant Gram-negative bacteria [10, 11]. Once the causative pathogen and antimicrobial susceptibility pattern have been determined by microbiology, every effort should be made to tailor the empiric treatment as per the sensitivity spectrum for the particular bacterium [16].

Our literature review revealed that limited international studies had been conducted on antibiotic susceptibility of VP shunt isolates in adults [5, 17], without any previous investigation in Pakistan, the fifth most populous country in the world. Herein, to fill the knowledge gap, we report epidemiological surveillance data at the leading neurosurgical institute of the country and assess the causative pathogens and their antimicrobial susceptibility patterns of antibiotics for VP shunt infections.

Materials and methods

Setting

The Punjab Institute of Neurosciences (PINS), Lahore General Hospital,located in the Lahore city of Punjab Province, is the largest and premier specialized neurosurgical center in Pakistan for more than fifty years. It has a capacity of five hundred beds and equipped with eight theatres for elective and two theatres for emergency surgeries. Each day, 700–800 outpatients and emergency patients are taken care of with output of nearly 7000 elective brain and spine operations in a year. Out of these, 10 to 20% of patients have shunt-related illnesses and they come in to PINS either directly or are referred to as complicated cases from other healthcare centers, specifically from Punjab province (population around 110 million) and generally from all around Pakistan (population around 207.8 million). [18]

Study design and data collection

In this retrospective study, we reviewed the records of clinically diagnosed cases of VP shunt infection and their respective reports of CSF culture and sensitivity from January 2015 to December 2021. The VP shunts inserted were plain. Some samples were excluded from the study based on incomplete information, for example, duplicate isolates within 7 days, and mismatched medical record numbers (Fig. 1). CSF samples from patients with diagnosed bacterial VP shunt infection and complete demographic and medical information were included in the study for final evaluation. Extracted data showed: age, gender, organism identified, and antimicrobial susceptibility patterns.

Fig. 1

Flow chart for cerebrospinal fluid (CSF) sample selection to be included in the study

Antimicrobial susceptibility results were further analyzed only for the isolates recovered from the CSF of patients with VP shunt infection, each year from 2015 to 2021 (bacteria 30 or more) [19]. So data included for Gram-negative bacteria, including Acinetobacter species, Pseudomonas species, Escherichia coli, Klebsiella species, and Gram-positive bacteria (Staphylococcus aureus, and Coagulase-negative Staphylococcus).

Identification of bacterial isolates and antimicrobial susceptibility testing

All CSF samples were processed in the microbiology laboratory according to the standard operating procedure [20]. Briefly, CSF samples were inoculated on sheep blood agar, chocolate agar, and MacConkey agar. Bacterial identification was performed by analytical profile index (API) (Biomerieux) [20]. Antibiotic susceptibility was determined by the Kirby-Bauer disc diffusion method and minimum inhibitory concentration (MIC) determination according to the Clinical and Laboratory Standards Institute (CLSI) guidelines [13] The laboratory deployed antibacterial testing of the drugs per the CLSI criteria for each bacterium and the laboratory's availability of antibiotic discs for the given years. Agents administered by oral routes only, first and second-generation cephalosporins and cephamycins, doripenem, ertapenem, imipenem and lefamulin, clindamycin, macrolides, tetracyclines, fluoroquinolones were excluded for the CSF isolates as per CLSI recommendation [13]. Based on a review of clinical practice in PINS throughout the study, the following antibiotics were tested, amikacin, gentamicin, cotrimoxazole (trimethoprim-sulphamethoxazole), ceftriaxone, ceftazidime, cefoperazone, cefotaxime, cefepime, piperacillin, ampicillin, amoxicillin-clavulanic acid, piperacillin-tazobactam, ampicillin-sulbactam, meropenem, oxacillin, penicillin, vancomycin, and linezolid.

Statistical analysis

The statistical results for continuous variables were presented as mean ± SD, range, or median (IQR) according to the statistical distribution. Categorical variables were presented as frequencies and percentages. Antimicrobial Susceptibility patterns of the bacteria were presented over time (years). The difference in sensitivity trends between 2015 and 2021 was examined using the multivariate analysis of variance (MANOVA), and two-sided p-values < 0.05 were considered statistically significant. The percentage of sensitive isolates was calculated as the sum of all sensitive bacteria (excluding both intermediately susceptible and resistant isolates) relative to the total number of bacteria tested against a particular drug. The sensitivity percentage was compared between 2015 and 2021 by a two-sample t-test and p-values < 0.05 were considered statistically significant. SPSS (IBM SPSS Statistics 23.0), Minitab version 17, and Microsoft Excel 2019 were used for statistical analyses and graphical presentation.

Results

During 7 years (2015–2021), 14,473 aerobic bacterial isolates were recovered from 13,937 CSF samples from patients with clinically diagnosed VP shunt infection; 536 (3.7%) of the CSF specimens showed the growth of more than one organism. The CSF samples came from 8,514 (59.9%) males and 5959 (40.1%) females with a mean age of 36.7 ± 19.3 years (range 15–92 years). Of 14,473 bacterial isolates analyzed, 11,030 (76%) were Gram-negative bacteria, and 3,443 (24%) were Gram-positive. The proportion of Gram-negative bacteria relative to the total number of bacteria increased over the course of the study: 57.2%, 77%, and 85.3 in 2015, 2017, and 2021 respectively (Fig. 2). Acinetobacter species were found to be predominant (41%) from 2015 to 2021, followed by Pseudomonas species (16%), Coagulase-negative Staphylococcus (13%), Staphylococcus aureus (10%), Klebsiella species (10%), Escherichia coli (8%) and others. An increasing trend was observed in the Acinetobacter species (Fig. 3).

Fig. 2

Gram-negative bacteria rate progression as compared to Gram-positive bacteria in VP shunt infections over 7 years (2015–2021) (N = 14,473)

Fig. 3

Frequency of isolated bacteria causing VP shunt infections in adults over 7 years (2015–2021) (N = 14,473)

Trends of antimicrobial susceptibility among bacteria

A total of 14,473 bacteria were tested against 14 clinically significant antimicrobials. Bacteria showed an overall susceptibility of ≥ 48.1%, with Gram-positive being 57.1% sensitive and Gram-negative bacteria being 39.0% sensitive. Antimicrobial susceptibility patterns for each bacterial species are presented in (Table 1). In 7 years, the highest frequency of sensitivity of Gram-negative pathogens to antibiotics was seen towards meropenem, piperacillin-tazobactam, and ampicillin-sulbactam by Acinetobacter, 69%, 64%, and 53%, respectively; piperacillin-tazobactam, meropenem, and amikacin by Pseudomonas, 80%, 71%, and 67% respectively; piperacillin-tazobactam, amikacin, and meropenem by Klebsiella, 57%,56%, and 50% respectively; amikacin, meropenem, and piperacillin-tazobactam by E. coli, 72%, 68%, and 67% respectively. The Gram-positive bacteria, including S. aureus and CoNS, were seen to be completely sensitive (100%) toward vancomycin and linezolid. 52% of S. aureus were MRSA, while methicillin resistance was found in 69.5% of CoNS.

Table 1 Antimicrobial susceptibility rates among the bacteria causing ventriculoperitoneal shunt infections over 7 years (2015–2021)

Conversely, the lowest frequency of sensitivity of Gram-negative bacteria to antimicrobials was seen towards amoxicillin-clavulanic acid by Klebsiella species and E. coli, being 11.8% and 12.3% sensitive, respectively: ceftriaxone by Acinetobacter species (13.4%) and Ceftazidime by Pseudomonas species (30.1%). Cumulatively, the frequency of sensitivity of all Gram-negative bacteria was less than 30% towards the third-generation cephalosporins (ceftazidime, cefotaxime, ceftriaxone, cefoperazone). However, in Gram-positive bacteria, both S. aureus and CoNS were least sensitive towards penicillin, with 25.2% and 16.7% sensitivity, respectively.

Trends of antibiotics

Year-wise frequency of sensitivity of the drugs commonly prescribed during the study period for Gram-negative bacteria against which the drugs have been reported during the study period (Fig. 4) showed a falling trend of sensitivity over 7 years (2015–2021). A significant decrease in the frequency of sensitivity for piperacillin-tazobactam (p = 0.0003) and meropenem (p = 0.0007) by all the Gram-negative bacteria collectively occurred in 2021 compared to 2015, piperacillin-tazobactam losing its sensitivity by 32.92% and meropenem by 26.11%. A prominent insignificant decrease in sensitivity frequency was shown by amikacin, 15.97%, followed by third-generation and fourth-generation cephalosporins, losing 14.90%, 14.82%, and 14.66% sensitivity by ceftriaxone, ceftazidime, and cefepime respectively. Cotrimoxazole showed 11.33% less sensitivity in 2021 compared to 2015, while gentamicin lost its efficacy by 7.06%. Amoxicillin-clavulanic acid showed low sensitivity throughout the 7 years without prominent variation, being 16.77% sensitive in 2015 and 11.50% in 2021, with a 5.25% loss in susceptibility.

Fig. 4

Boxplots showing yearly antimicrobial effectiveness of antibiotics in terms of sensitivity for Gram-negative bacteria cumulatively (for which the drug has been reported) each year. For each antibiotic, boxes represent the sensitivity rate at the 25–75th percentiles (interquartile range), and the ends of vertical lines represent values at the 10^–90th percentiles for the respective year. Horizontal lines represent median values for each year. The comparison of the efficacy of the drug between 2015 and 2021 was done by a two-sample t-test. P =  < 0.05 was considered significant

Discussion

Infection is a severe complication after VP shunting which may lead to prolonged hospital stay, increased medical costs, or even death. [2]. However, the data regarding the etiology of this infection is scarce, especially in the adult population. In this study, CSF culture results and antibiotic susceptibility were analyzed over 7 years. Here, 3.7% of the CSF samples revealed more than one organism, similar to some previous studies [21, 22], although, a single organism has been reported in the literature more often [23,24,25]. This discrepancy may be due to reporting in clinical practice where more than one organism is often considered a contaminated sample and reported as such. The changing spectrum of VP shunt infection-causing bacteria from Gram-positive to Gram-negative, as seen in our study as well (Fig. 2), might be because of the complex neurosurgery, neurocritical care, extended hospital stays, healthcare-associated infections, and antibiotic prophylaxis targeting Gram-positive bacteria [26,27,28,29].

In this study, most of the Gram-negative pathogens, including Acinetobacter spp.., Pseudomonas spp, Klebsiella spp, and Escherichia coli, showed an overall trend of increased resistance towards all the drugs used for the empirical treatment of VP shunt infection included in this study. Meropenem is the primary empirical and targeted treatment, consistent with the recommended guidelines [5]. Other recommended antimicrobial agents [5] include cefepime and ceftazidime. However, PINS rarely uses them as empirical treatments because of their high resistance rates. (Table 1). Unfortunately, the most significant decrease in sensitivity was seen for Gram-negative bacilli collectively (p < 0.05) against meropenem (26.11%) and piperacillin-tazobactam (32.92%). When individual isolates were tested for meropenem susceptibility, Acinetobacter susceptibility was reduced by 50% over the course of the study. Sensitivity to meropenem declined for Klebsiella spp and E. coli by 25.4% and 24.27%, respectively. High-level carbapenem resistance is on the rise and has been reported in the literature [10, 30, 31]. Of all the antibiotics compared for the difference in susceptibility over the study period, gentamicin showed the least change, being 50% sensitive in 2015 and 42% sensitive in 2021. Although such a phenomenon in treating VP shunt infections has not been reported before, further studies should be done to assess its significance.

We had some limitations while concluding the results. As it is a retrospective study and our center receives referral infected and complicated cases from other healthcare facilities as well, we donot have exact data about how many VP shunt infections were relapses or reinfections.

Based on our results, the management of patients with VP shunt infections should be guided by some fundamental principles for improving empirical therapy. The currently prescribed drug (meropenem) gives Gram-negative coverage, but it has lost its efficacy considerably. Therefore, antibiotics including colistin, fosfomycin, ceftazidime/avibactam, ceftolozane/tazobactam, and tigecycline should be evaluated to have more effective treatment of infections caused by multidrug-resistant Gram-negative bacilli. However, the intravenous (IV) administration of antibiotics like colistin and tigecycline is associated with a very low CNS transfer. Consequently, a concomitant intrathecal or intraventricular administration route is required for the treatment of severe ventriculitis in patients with VP shunt infection [16]. It should be noted that although tigecycline and colistin have been used clinically for the last two years in our center for highly drug-resistant Gram-negative bacteria in VP shunt infections, data about their susceptibility patterns are unavailable due to inadequate guidelines on reporting these drugs. The synergistic action of antibiotics like meropenem–amikacin and meropenem–colistin combinations, ampicillin-sulbactam, and aminoglycosides combination therapy, should be explored. Furthermore, the clinical literature is emerging on using extended-infusion β-lactams to treat Gram-negative bacteria, especially with cefepime, piperacillin-tazobactam, and carbapenems (meropenem, imipenem, and doripenem). One of the key advantages of extended-infusion β-lactams is the ability to achieve drug concentrations above the MIC for a longer time for less susceptible organisms, especially those with a MIC between 4 and 16 µg/mL [32]. In addition, according to Infectious Diseases Society of America (IDSA) practice guidelines [33], intrathecal administration of anti-infectives should be considered for patients with healthcare-associated ventriculitis and meningitis in which the infection responds poorly to systemic antimicrobial therapy alone despite shunt removal in the setting of highly resistant organisms susceptible only to antibiotics with poor CSF penetration or in situations where devices cannot be removed.

In addition to addressing infections, we suggest the implementation of care bundles to decrease the frequency of VP shunt infections. Interventions that combine different prevention strategies appeared to be effective in certain settings. These bundles should include the enforcement of strict infection control protocols, emphasizing proper hand washing techniques while scrubbing and the use of strict sterile techniques during surgery, among other measures. We advocate for the use of antibiotic-impregnated shunt devices as they have the potential to reduce the incidence of CSF shunt infections [5, 34]. Furthermore, we support hair clipping instead of shaving, minimal trafficking during surgery, double gloving by all team members, the use of antibiotic-impregnated sutures and considering injecting vancomycin/gentamicin into the shunt reservoir as these measures have been shown to be effective in reducing the incidence of CSF infections [35].