Streptococcus agalactiae (group B Streptococcus [GBS]) is considered a leading cause of morbidity and mortality in neonates and pregnant women. Recommendations for diagnosing maternal GBS colonization and administering intrapartum antimicrobial prophylaxis have led to a significant decrease in these infections (Schrag et al., 2002). Nonetheless, the rate of invasive GBS disease in non-pregnant adults continues to climb (Phares et al., 2008). Elderly persons and those with underlying diseases – two expanding segments of the population – are at increased risk (Skoff et al., 2009). This epidemiological shift is associated with uncertainty in clinical management. Because treatment concepts in adults are not established, those used for neonates – i.e., the combination of β-lactams plus an aminoglycoside (Polin, 2012) – are transferred to adults. For example, some experts advocate this combination therapy for at least the first 2 weeks of treatment for infective endocarditis (IE) (Baddour, 1998; Westling et al., 2007) and periprosthetic joint infection (PJI) (Zimmerli et al., 2004). These recommendations are based on a postulated synergistic effect with penicillin (PEN) and gentamicin (GEN) observed in in vitro studies (Cooper et al., 1979; Baker et al., 1981; Swingle et al., 1985). However, adults, especially elderly persons, are more prone to develop side effects caused by aminoglycosides (e.g., nephrotoxicity and ototoxicity) than are neonates (Gonzalez and Spencer, 1998). Here, we evaluated the synergistic effect of PEN and GEN, using contemporary clinical isolates obtained from adults with IE and PJI, and the same antimicrobial products that are administered in clinical practice.

Rates of invasive GBS disease in non-pregnant adults continue to climb. Hence, clinicians are faced with the challenge of transferring therapeutic knowledge from neonatal sepsis to infections in adults. Other groups previously reported synergism with PEN plus GEN toward GBS (Overturf et al., 1977; Cooper et al., 1979; Baker et al., 1981; Swingle et al., 1985). We used two different methods with isolates obtained from adults and could not confirm these results. The following arguments may explain the differences between the results of previous studies and those provided in the present study.

The definitions and methods used for synergism vary and all have limitations. This again points toward the difficulty to transfer synergy assay results from the laboratory to patient treatment concepts. The two commonly used laboratory methods to evaluate synergism are checkerboard assays and TKAs over a predetermined time interval, as used in the present study. Although the checkerboard method has been reported to have reproducibility problems (Rand et al., 1993), the range of results in our multiple repeated assays was small (Table 1). Because the effect of the antibiotic combination in the checkerboard method can be observed at only a single time point, in the TKAs, we used the advantage of measuring the colony counts at various time points up to 24 h after antibiotic exposure. With the last time point, we were able to show a parallel in the results of both methods. In contrast to previous analyses showing a lack of correlation between FICIs and killing curves (Hallander et al., 1982; Bayer and Morrison, 1984; Swingle et al., 1985), our results for both methods were congruent in that they showed no synergistic response.

Some investigators use the first 4–8 h after exposure to an antimicrobial agent or a combination of agents to compare differences in colony counts (Cooper et al., 1979). In our view, this method reflects rapid killing, as shown in our experiments with PEN 0.048 mg/L plus GEN 4 mg/L or 12.5 mg/L (Figure 1), or PEN 0.2 m/L plus GEN 12.5 mg/L (Figure 2). These results may be helpful to describe potentially beneficial clinical interactions in the very early stage of treatment. Though, they are difficult to standardize (e.g., dynamic of bacterial replication can be variable at different time points). Therefore, differences in colony count measurements at 4 h to 8 h are commonly not used for the definition of synergism (Pillai et al., 2005; Moody and Knapp, 2010). Other means to determine synergy may be more accurate from today’s perspective (e.g., molecular synergy) (McCafferty et al., 1999).

As a consequence of lack of synergism, these results theoretically argue against prolonged combination therapy. The clinical relevance of the laboratory phenomenon “synergism” is difficult to estimate. The bacterial inoculum in time-kill experiments is up to 100,000 times higher than that found in human sepsis (Yagupsky and Nolte, 1990; Puttaswamy et al., 2011). Also, the antibiotic concentrations found in humans treated for bacteremia are significantly higher than the ones used in TKAs. Ten minutes after completion of i.v., administration of 5 million U PEN, the mean serum concentration is 273 mg/L (Plaut et al., 1969; Geddes and Gould, 2010). This corresponds to more than 5,500 times the PEN concentration used in this study. In addition, in severe GBS disease (e.g., IE, PJI), PEN is commonly administered i.v., every 4–6 h (e.g., 18–24 million U/day i.v., in six doses) (Zimmerli et al., 2004; Habib et al., 2009). In vivo PEN concentrations decrease in human serum over time. Nonetheless, prior to administration of the next dose, they rarely fall to those levels used in experimental settings. For example, 4 hours after completion of i.v., administration of 5 million IU PEN, the mean serum concentration is 3 mg/L (i.e., >60 times higher than the concentration used in this study) (Plaut et al., 1969; Geddes and Gould, 2010).

In our study, we saw rapid killing of GBS with 0.2 mg/L PEN monotherapy. The killing rate was fastened by 2–4 h when GEN 12.5 mg/L was added. There are no standard concentrations at which antibiotics are tested for synergism. The results with PEN concentrations in the range of 1 × MIC have to be interpreted with caution because of the following reasons. Experiments depend on the precise inoculum (Brook, 1989), the exact determination of MIC, and the antibiotic concentration. In our isolates, PEN MICs were determined by using two different methods and multiple measurements. Moreover, the antibiotic concentrations used were based on calculated dilutions, but confirmed with measurements via fluorescence polarization immunoassay and HPLC.

Various GEN concentrations and products have been used in previous in vitro experiments, ranging from 0.5 mg/L to 13 mg/L (Overturf et al., 1977; Cooper et al., 1979; Baker et al., 1981; Kim and Anthony, 1981; Kim, 1987). Since GEN alone in non-toxic systemic concentrations has little or no effect on GBS, the optimal setting for testing synergism in TKAs is unknown. The antibiotic concentrations of combinations used in in vitro settings are often not physiological, because PEN is far below achievable serum concentrations, although GEN is within the expected range. On the other hand, higher in vitro PEN concentrations diminish the visible GEN effect, and lower GEN concentrations have no effect. The benefit of rapid killing within the first few hours when adding GEN is difficult to interpret for clinical practice, because of the low PEN concentrations used in this study (0.048 mg/L and 0.2 mg/L). Considering the high PEN concentrations and the low inoculum during bacteremia found in humans [e.g., ≤10² cfu/mL; (Yagupsky and Nolte, 1990; Puttaswamy et al., 2011)], we hypothesize no beneficial clinical effect for planktonic bacteria when GEN is added. A clinical trial is required to confirm or reject this hypothesis.

The stability of our results is supported by the small variation seen in multiple assays, comparing TKA results under various conditions (Ruppen and Sendi, 2015) and the use of a HLGR GBS isolate (Sendi et al., 2016).

Another reason for the differences in results found in our study in comparison to previous investigations might be the PEN product. The efficacy of a drug depends largely on the purity of the active ingredient. It is conceivable that the manufacturing process has improved over the decades (van der Beek and Roels, 1984; Penalva et al., 1998). Here, we used PEN products that are administered to patients. However, because we cannot compare older products with the ones used in this study, this statement remains speculative.

We cannot uncritically extrapolate our findings for five isolates to other GBS isolates. Nonetheless, the isolates were obtained from two patients with invasive diseases in which the addition of gentamicin is recommended (IE and PJI). In addition, we used 3 control isolates [two of them previously investigated (Glaser et al., 2002; Sendi et al., 2016)]. The number of isolates used in our study is small. Though, analyzes were performed at multiple time points within the first 8 h for every single isolates. In our view, this information is important in the light that – in clinical practice – PEN is administered every 4–6 h.

Our study reinvestigated the synergism of PEN plus GEN with two common laboratory methods, clinical isolates in mid-log growth phase and antimicrobial products administered in clinical practice. Synergism according to definition was not observed with either of the methods. In view of the potential nephrotoxicity of aminoglycosides and the increasing elderly population at risk for invasive GBS disease, our findings may have implications for the decision to administer or withhold aminoglycosides.

All authors significantly contributed to the generation of this manuscript und fulfilled the criteria for authorship. CR and AL performed the experiments, interpreted the results and co-wrote the manuscript. LD contributed to the acquisition and analyses of the results, co-drafted the work. PS contributed to the conception and design of and interpretation of data for the work, and co-wrote the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.