Skip Navigation

Penicillins for Treatment of Pneumococcal Pneumonia: Does In Vitro Resistance Really Matter?

  1. Lance R. Peterson
  1. Evanston Northwestern Healthcare Research Institute, Evanston Northwestern Healthcare, Department of Pathology and Laboratory Medicine, Division of Microbiology, and Department of Medicine, Division of Infectious Diseases, Northwestern University, Evanston, Illinois
  1. Reprints or correspondence: Dr. Lance R. Peterson, Clinical Microbiology Laboratory, Evanston Northwestern Healthcare, 2650 Ridge Ave., Evanston, IL 60201 (lancer{at}northwestern.edu).

Abstract

Background. The rate of in vitro bacterial resistance to antimicrobial agents is escalating among pathogens that cause the most serious respiratory tract infections. Many reports published during the past few years suggest that this has direct clinical implications. In particular, resistance of Streptococcus pneumoniae to β-lactam antibiotic therapy has assumed a prominent role in the evolution of guidelines for the initial treatment of respiratory tract infection.

Methods. I conducted a critical review of the published medical literature.

Results. There is only a single report of documented microbiologic failure of parenteral penicillin-class antibiotics in the treatment of pneumococcal pneumonia in patients with or without bacteremia, whereas there are numerous well-documented reports of treatment failure with quinolone-class (n ⩾ 21) and macrolide-class (n ⩾ 33) antibiotics in the treatment of pneumococcal pneumonia.

Conclusion. The recommended optimal in-hospital therapy for community-acquired pneumonia should continue to be a β-lactam antibiotic (penicillin, aminopenicillin, cefotaxime, or ceftriaxone) administered with a macrolide or a fluoroquinolone agent for adjunctive treatment of infection with potential atypical pathogens.

Respiratory tract infections caused nearly 125 million adults to visit physicians' offices in the United States during 1997 [1]. Among these infections, Streptococcus pneumoniae caused ∼7 million cases of otitis media, 500,000 cases of pneumonia, 3000 cases of meningitis, and ∼50,000 deaths each year [2]. It is the most frequently detected pathogen responsible for community-acquired pneumonia (CAP), with an overall mortality rate of ∼12% [3, 4]. A meta-analysis of 7000 cases of CAP found that S. pneumoniae was the causative agent in 66% of patients for which a pathogen could be identified and was responsible for nearly two-thirds of all deaths [3]. Other pathogens are associated with CAP, but they do not appear to result in the same degree of illness; thus, proper initial treatment of pneumococcal infection is paramount.

Alexander Fleming and Howard Florey recognized the tremendous ability of bacteria to adapt and survive [5] and wrote about the need for rapid, lethal therapy. To understand the microbial capacity for adaptive survival, one only needs to realize that microbial fossils have been found that date back 3.8 billion years, in earth's 4.5 billion—year history. In contrast, metazoan embryogenesis has existed for 500 million years [6], and we humans, as evidenced by the emergence of our Y chromosome, only have been residents here for ∼50,000–150,000 years [7]. Therefore, it is no surprise that bacteria can adapt in a myriad of exquisite ways, underscoring the need for us to carefully consider our therapeutic decisions regarding which antimicrobial agents to use for treatment of common, serious infections.

The purpose of this commentary is to help improve our understanding of the clinical implications regarding increasing resistance in pneumococci to the earliest and perhaps best of our antibiotics that are used to treat infection due to this key pathogen, and to suggest options for choosing antimicrobial agents in the therapy of hospitalized patients with CAP.

Emerging in Vitro Resistance

Microbial resistance to therapeutic agents can be either intrinsic or acquired. For pneumococcus, the latter pathway is primarily taken, with the key factor being the use of antimicrobial agents in both the hospital and the community [8]. Unfortunately, surveillance of resistance in pneumococcus has been sporadic. During the early 1980s, this was performed by the Centers for Disease Control and Prevention, and no in vitro resistance against penicillin was detected [9]. Thus, in the late 1980s, surveillance became a low priority and was discontinued. When surveillance resumed in 1992, a full 1% of invasive isolates appeared to be no longer susceptible to penicillin, and the rate of antibiotic resistance has steadily increased since then [9], leading to questions regarding antipneumococcal therapy.

Subsequently, several multicenter surveillance projects were conducted, according to pneumococcal in vitro susceptibility patterns, to determine the percentages of organisms exhibiting resistance [1012]. All findings reveal continually increasing levels of penicillin resistance in S. pneumoniae. By 2000, rates of high-level enicillin resistance ranged from 5.2% to 56.1%, with other antibiotics faring equally poorly [13]. Compounding the problem was the observation that in vitro macrolide resistance rates ranged from 6.1% to 53.7%, in vitro tetracycline resistance rates ranged from 7.3% to 50%, and in vitro trimethoprim-sulfamethoxazole resistance rates ranged from 13.5% to 50%, with 22.4% of strains being no longer susceptible to multiple classes of antibiotics [13].

Now, the occurrence of multidrug-resistant pneumococci has been documented worldwide, from Asia, where the rate of susceptibility to penicillin is as low as 20.3% [14, 15], to Canada, where the rate of susceptibility to penicillin decreased to 86.4% by 1998 [16]. Findings of longitudinal surveillance studies have demonstrated that the rate of macrolide resistance has increased rapidly, doubling from 16% to 32% between 1994 and 1999 [17], with 68% to 81% of strains tested in Hong Kong having become resistant to erythromycin by the year 2000 [18, 19]. A critical factor not to be overlooked is that in vitro resistance to penicillin and macrolides tends to occur in the same organism [13, 20]. A report from the Centers for Disease Control and Prevention that evaluated 4013 invasive strains of S. pneumoniae found that only 63% of isolates tested in 1998 were still fully susceptible to penicillin and that 15% of isolates were not susceptible to cefotaxime, 16% were not susceptible to meropenem, 16% were not susceptible to erythromycin, and 29% were not susceptible to trimethoprim-sulfamethoxazole [21]. During the period 1995–1998, the prevalence of isolates resistant to ⩾3 drug classes increased from 9% to 14%, suggesting that some cases of pneumonia may be very difficult to treat with standard therapy.

In response to this problem, Austin et al. commented that “Given the observed decline in the rate of development of novel antibiotics [see table 1], in the future more attention must be given to the development of prescribing policies designed to maximize the life expectancy of a given compound” [22, p. 1156]. A key element, which is the focus of this review, is discordant therapy with penicillins, for which there is little reported evidence of an increased rate of treatment failure in patients with penicillin-resistant pneumococcal pneumonia.

Table 1

New antibacterial agents approved in the United States since 1991.

Clinical Implications of Antibiotic Resistance

Lower respiratory tract infections remain the only communicable illness among the top 10 causes of death in the United States [23]. S. pneumoniae, the most important cause of CAP and meningitis, is resurgent in the form of bacteremic pneumonia [24]. Not only is this pathogen infecting diverse, compromised populations, but its increasing resistance to penicillin is also reported to be associated with an increase in mortality rates [4, 2527]. Importantly, during the 1930s, the case-fatality rate for pneumococcal pneumonia was 30%–35%, and after the introduction of penicillin, the rate had decreased to 5%–8% [23]. People with bacteremia fared worse, with a mortality rate that approached 85% at 3 weeks after onset, before a specific treatment was introduced, and that decreased to 50% among persons who received serum therapy and to 15% after penicillin became available [28, 29].

How we interpret and apply the information regarding emerging in vitro resistance will be crucial to our evolving use of antimicrobial agents. Virk and Steckelberg began their review of the clinical importance of resistance with the statement, “Antimicrobial resistance is a serious emerging global problem that physicians and patients have to contend with on an almost daily basis” [30, p. 200]. A key to the selection of optimal therapy is knowing how to translate findings of laboratory detection of resistance into an effect on clinical outcome. Although it is clear that S. pneumoniae is gaining resistance to the in vitro action of several antimicrobial agents, questions remain regarding the clinical impact of this “laboratory” phenomenon and the relevant level of resistance to specific β-lactams.

The publication with the most convincing evidence of an increased mortality rate associated with reduced susceptibility to penicillin is from the period after 1995, when the rate of resistance to penicillin has been observed to escalate most dramatically [4]. However, Feikin et al. [4] point out that if severity of illness factors such as age and underlying disease were controlled for, then the overall increased mortality rate would not be associated with penicillin resistance. They also comment that because no treatment data were available and because penicillin-resistant pneumococci are often multidrug-resistant, the small number of cases with poor outcome observed after 4 days were likely caused by more factors than only penicillin or cefotaxime treatment failure [4]. Table 2 summarizes published reports that correlate in vitro (β-lactam) susceptibility of S. pneumoniae with therapeutic outcome. In these reports, several striking observations are made. First, many of the studies that assess outcome of infection with penicillin-resistant pneumococci do not correlate specific therapies with success or failure. Second, prospective trials have been unable to associate discordant penicillin treatment with an adverse outcome. And third, there is only a single case reported in which clinical and microbiologic treatment failure of pneumococcal pneumonia is associated with the use of an (amino) penicillin in a patient infected with a drug-resistant strain [37]. This occurred in a patient who developed an empyema on day 5 of treatment with 2000 mg of amoxicillin and 200 mg of clavulanate administered every 8 h, and the recovered strain of S. pneumoniae had the same in vitro susceptibility (MIC, 8 µg/mL) as that of the initial infecting isolate [37].

Table 2

Summary of reports relating treatment failure of pneumonia with in vitro β-lactam—resistant Streptococcus pneumoniae.

The published reports that correlate in vitro resistance with bacteriologic treatment failure with the β-lactam class of compounds describe failures in the treatment of pneumococcal pneumonia with ceftazidime, cefazolin, cefuroxime, and cefamandole [25, 4549]. The single report implicating treatment failure with cefotaxime involved a patient with an (undrained) empyema [50], for which antibiotic treatment failure should have been the expected result. Another report links the failure of oral therapy and in vitro susceptibility with the development of bacteremic pneumococcal pneumonia (table 2). These authors concluded that clarithromycin resistance was clinically important and lead to therapeutic failures, whereas nonresponse to treatment with amoxicillin/clavulanate was most likely related to a suboptimal dosage of the antibiotic [31]. Gonzales et al. [34] came to a similar conclusion while studying azithromycin and β-lactam treatment failures (table 2). Thus, although there is anecdotal evidence suggesting that resistance to β-lactam causes failure in the treatment of respiratory tract infection due to S. pneumoniae, documentation of penicillin treatment failure, particularly with aminopenicillins administered at adequate dosages (e.g., parenterally), remains virtually nonexistent.

There are a limited number of published observations regarding the use of β-lactam (aminopenicillin)/β-lactamase inhibitor combinations for the treatment of severe CAP (i.e., that which requires hospitalization), and many do not specifically address the issue of discordant therapy. However, the information available indicates that drugs with an aminopenicillin component are highly useful in treating pneumonia. Although most practitioners do not consider β-lactamase inhibitor compounds to be particularly beneficial against non–β-lactamase-producing organisms, such as pneumococcus, experimental evidence has shown that these agents enhance intracellular killing of pneumococcus [51] and in vivo bactericidal action against β-lactamase—negative Staphylococcus aureus [52, 53]. Therefore, their usefulness in treating pneumonia is not surprising. One small study of 75 patients compared treatment with 1–2 mg of ampicillin plus 500–1000 mg of sulbactam every 6 h and treatment with 1–2 g of cefamandole every 6 h [49]. Seven patients presented with pneumococcal pneumonia, and although all patients achieved clinical and microbiological cure on receipt of ampicillin/sulbactam therapy, there was 1 episode of clinical treatment failure involving a patient with S. pneumoniae infection in the cefamandole arm of this prospective trial [49]. Four additional reports describe 110 treated patients infected with S. pneumoniae, 97 of whom were treated with piperacillin/tazobactam (dosage, 4 g of piperacillin plus 500 mg of tazobactam every 8 h or 3 g of piperacillin plus 375 mg of tazobactam every 6 h) and 13 of whom were treated with parenteral amoxicillin/clavulanate (dosage, 2 g of amoxicillin plus 200 mg of clavulanate every 8 h). No patients experienced clinical or bacteriologic treatment failure (1 patient was documented as being infected with a strain that was nonsusceptible to piperacillin), including at least 19 patients with bacteremia [5457]. Only 1 person died within 30 days after completion of therapy: an 84-year-old man who was dying of cardiorespiratory collapse that occurred 6 days after completion of therapy with piperacillin/tazobactam [57]. The success of this treatment was not unexpected, because piperacillin/tazobactam has antipneumococcal activity within a 2-fold dilution of amoxicillin and is administered at a much higher dose [58]. Additional trials of these agents, in which resistance was assessed, are included in table 2. Altogether, the available information indicates that at least the newer β-lactam/β-lactamase inhibitor combinations are highly effective in the treatment of hospitalized patients with pneumococcal pneumonia.

The mechanisms of macrolide resistance in S. pneumoniae include the following: target-site modification encoded by erythromycin ribosome methylation genes that provide resistance to all macrolides, lincosamides, and streptogramin B (MLSB); active efflux encoded by macrolide efflux genes that causes in vitro resistance to 14- and 15-member ring macrolides; and other, uncommon ribosomal mutation(s) in the 23S rRNA gene for ribosomal protein L4 or L22. MLSB—type-resistance leads to the highest level of resistance and is considered to be most clinically relevant. Importantly, demonstrating the clinical impact of in vitro macrolide resistance has been easier to document. Initially, treatment failure was only described in a single case report [59]. More recently, Lonks et al. [60] reviewed medical records of 76 patients with macrolide-resistant pneumococci in their bloodstream. They found that 18 patients had received treatment with a macrolide and had developed bacteremia after 3–5 days of therapy. Occurrence of treatment failure was evenly divided among patients who received older and newer macrolides and patients infected with pneumococcal strains with resistance based on either drug efflux or target mutation [60]. In 2004, Rzeszutek et al. [61] reviewed 33 published case reports of macrolide treatment failure, 89% of which involved CAP. In the 19 cases for which it was assessed, most (63%) of the reports associated treatment failure with a pneumococcal strain that demonstrated the MLSB phenotype [61]. Thus, the published data make a strong case that in vitro detection of macrolide resistance is relevant.

The problem of clinical treatment failure with fluoroquinolone antimicrobial agents was recently reviewed by Low [62]. Although the reports that document microbiologic treatment failure of this class of drugs are a series of retrospective case reports [6369], the documented evidence of clinical and microbiologic treatment failure with quinolone antimicrobial agents in the treatment of at least 21 patients infected with pneumococcus is also clear and substantial.

Current Options for Managing Emerging Resistance

New agent development. One way to manage the development of antimicrobial agent resistance is the development of new, potent drugs with novel antibacterial targets. As can be seen in table 1, this strategy has not been very successful in the past 15 years. Therefore, for the immediate future, we cannot assume that a new class (or classes) of compounds or even novel new drugs in an established class can alone solve the problem of emerging resistance.

Diagnosis as a critical element. Prescription of broad-spectrum therapy in excess of that which is needed is a prevalent and continuing problem for the management of respiratory tract infection in the community. Steinman et al. [70] found that 46% of patients with symptoms of cold (or other nonspecific complaints) received antimicrobial therapy on visiting their physician, and that 51% of patients who were prescribed an antibiotic received a broad-spectrum agent.

To minimize unnecessary prescribing of antimicrobial compounds, an accurate diagnosis of respiratory symptomatology is crucial [71]. Gram staining of properly collected sputum specimens was recently reassessed [72]. Although sensitivity for the detection of pneumococci and Haemophilusspecies was modest (57% and 82%, respectively), the specificity was high (97% and 99%, respectively), making specific therapy possible if the typical organism was observed on a Gram stain [72]. The Infectious Diseases Society of America's guidelines for the treatment of CAP [73] endorse the use of this simple, rapid test, combined with the use of culture, for patients who require hospitalization. Novel use of a new, commercially available urine test for the detection of pneumococcal antigen was reported by Guchev et al. [74] as a means to target treatment of CAP. However, they found that detection was only successful for 48 (63%) of 76 patients with pneumococcal pneumonia, once again indicating that this test should be used to augment standard diagnostic testing [73, 74]. Rapid testing for influenza virus is available as well, but it also has low sensitivity, as noted in the evaluation by Cazacu et al. [75], who found its sensitivity to be <45% during the 2002–2003 influenza season. Expanded use of even basic laboratory testing can be a step in the right direction while we await a future in which enhanced diagnostics may improve therapeutics [76].

Treatment options. With the increasing complexity of health care, there are no easy answers to the selection of treatment options. One must begin with an understanding of resistance in important pathogens in the community and of what the in vitro detection of resistance means in relation to clinical response. Many treatment guidelines appear to have evolved largely on the basis of trends of diminishing in vitro activity of penicillin. An example of this trend is shown in table 3, in which frequent updates of the Medical Letter Guide [7784] demonstrate how changes have occurred in response to in vitro development of resistance. Current recommendations for the prescription of a β-lactam combined with a macrolide or a newer respiratory fluoroquinolone (gatifloxacin or moxifloxacin) are reasonable for ambulatory patients with documented cases of pneumonia. However, on the basis of the clinical efficacy of selected β-lactam antibiotics, it appears that choosing 1 of these compounds (table 4) combined with another agent active against atypical pathogens is still considered to be optimal treatment for hospitalized patients with pneumonia. This recommendation is supported by the recent report by Baddour et al. [85], who observed a 3-fold reduction in mortality when critically ill patients with pneumococcal bacteremia received combination therapy, most often a β-lactam combined with a macrolide. Importantly, the impact on mortality was most pronounced when a second agent was added to β-lactam treatment [85]. On the basis of my review, the choice of the β-lactam should be limited to penicillin, an aminopenicillin, ceftriaxone, or cefotaxime. Nuermberger and Bishai [87] have recently reviewed the kinetics of orally administered agents for the treatment of pneumococcal pneumonia; table 4 contains a similar review of the parenteral β-lactam agents available for this infection, with correlations of dose that provide for the antimicrobial drug concentration in the blood to be above the MIC for ⩾50% of the dosing interval, where clinical success should be expected. The specific agent selected for initial therapy is dependent on which other microbial etiologies are suspected as the cause of pneumonia. It should be remembered that abscess collections, such as empyema, must be drained. Also, if endocarditis, meningitis, or other disseminated infections are suspected, use of ceftriaxone and/or inclusion of vancomycin as an additional agent should be considered whenever penicillin-resistant pneumococci are prevalent, until bacterial identification and susceptibility testing are complete.

Table 3

Summary of evolving recommendations for the therapy of pneumonia by The Medical Letter.

Table 4

Suitable β-lactam agents for the treatment of hospitalized patients with pneumonia when Streptococcus pneumonia infection is part of the differential diagnosis.

Conclusion

A recent review by Dagan et al. [88] emphasized the mounting clinical evidence that supports the need for bacterial eradication as part of treatment of respiratory tract infection. They conclude that “The aim of antimicrobial therapy in respiratory tract infections should be the eradication of the infecting organism” [88, p. 129]. In the case of pneumococcus, it appears that penicillins are still most capable of accomplishing this goal, clinically outperforming both macrolides and quinolones in the treatment of pneumococcal pneumonia.

The era of antibiotics has now matured for >60 years. However, we still have much to learn about how to use them effectively to preserve this precious resource that has served us so well. We need to fully comprehend our microbial adversaries' capacity for survival when they invade to cause infection. The emergence of significant resistance in many pathogenic bacterial species does have direct, serious, clinical, and economic implications that must be effectively dealt with as we plan for future management of infectious diseases.

Acknowledgments

I thank Kyle D. Gadbois and Glenn S. Tillotson for their critique of portions of this article.

Financial support. Evanston Northwestern Healthcare.

Potential conflicts of interest. L.R.P.: no conflicts.

  • Received August 12, 2004.
  • Revision received August 2, 2005.
  • Accepted December 13, 2005.

References

| Table of Contents