Collateral Damage

For most of the infectious diseases on the wards of the Boston City Hospital in 1937, there was nothing to be done beyond bed rest and good nursing care. Then came the explosive news of sulfanilamide, and the start of the real revolution in medicine. I remember the astonishment when the first cases of pneumococcal and streptococcal septicemia were treated in Boston in 1937. The phenomenon was almost beyond belief. Here were moribund patients, who would surely have died without treatment, improving in their appearance within a matter of hours of being given the medicine and feeling entirely well within the next day or so … we became convinced, overnight, that nothing lay beyond reach for the future.

Lewis Thomas, The Youngest Science: Notes of a Medicine Watcher(1)

In his splendid memoir, Lewis Thomas movingly describes what must have been the prevailing optimism at the time. Bacterial infections had been a tremendous scourge for young and old alike, and physicians had become accustomed to watching helplessly as patients succumbed to community-acquired infections caused by Streptococcus pyogenes, Staphylococcus aureus, Streptococcus pneumoniae, and Mycobacterium tuberculosis. In relatively short order, however, therapeutic failures due to the emergence of antibiotic resistance were recognized. Patients whose tuberculosis had responded to streptomycin therapy returned months later with progressive disease due to streptomycin-resistant M. tuberculosis. Resistance to penicillin in Staphylococcus aureus spread rapidly, and within a few years more than 50% of nosocomial Staphylococcus aureus isolates were penicillin resistant.

The emergence and spread of antimicrobial resistance over the ensuing decades, first in the hospital and more recently in the community, led to a concerted effort on the part of academia and industry to understand the mechanisms underlying resistance and to develop new antimicrobial agents effective for treating resistant strains. By any measure, efforts in both areas have been an enormous success. We now know in detail the mechanisms by which many bacteria develop resistance to antibiotics and the means by which they distribute resistance determinants among and between species. We also have a large number of highly potent and effective antimicrobial agents for treating most infections encountered in the community or in the hospital. If bacteria are perceived as the enemy, we could posit that the major fighting in the battle is over, although it is clear that pockets of resistance (some strong and growing) remain.

Pogo stated, “We have met the enemy, and he is us!” Our anthropocentric view of the universe generally specifies as human those cells that contain human DNA. A broader view of our existence would acknowledge that we also consist of a multitude of cells that do not contain human DNA—our abundant “colonizing” microflora. In a sense, this microflora is yet another organ whose proper functioning is essential to our well-being. Among other functions, bacteria participate in our enteropathic circulation, as well as helping to break down oligosaccharides in the colon as fatty acids. Unfortunately, just as our properly functioning organs are sometimes the source of deadly cancer, our microflora is occasionally the source of the bacteria that cause serious infection. Antimicrobial agents do not distinguish colonizing from infecting flora; they simply inhibit or kill any susceptible microbe with which they come in contact. In short, it is rarely possible to successfully treat an infecting bacterium without also doing significant “collateral damage” to our friendly microflora.

Two examples of collateral damage associated with the use of antimicrobial agents are presented in this issue. McMahon and colleagues (2) described a relationship between previous exposure to antimicrobial agents and resistance in gastric Helicobacter pylori. This relationship was particularly strong for previous exposure to macrolides and resistance to clarithromycin; the risk increased with repeated exposure to this class of antibiotics. A similar association was noted for previous exposure to metronidazole and resistance to this agent. The practical consequences of these resistance phenotypes were more obvious for clarithromycin than for metronidazole. Seventy-seven percent of patients infected with clarithromycin-resistant H. pylori (10 of 13) experienced treatment failure with clarithromycin-containing regimens, compared with only 13% (5 of 40) of those infected with susceptible isolates. In contrast, only 11% of patients infected with metronidazole-resistant isolates (2 of 18) experienced treatment failure with metronidazole-containing regimens.

Macrolide resistance in H. pylori results from point mutations within the 23S ribosomal RNA genes. The frequency with which McMahon and colleagues found macrolide resistance in isolated H. pylori demonstrates the ease with which resistance due to point mutations may arise. More disturbing than the appearance of the resistant mutants is their apparent persistence over time, presumably in the absence of further antimicrobial selective pressure. The persistence of the resistant phenotype may be due to the accumulation of additional “compensatory” mutations that reestablish fitness in the presence of resistance (3), or it may reflect the fact that repeated exposures to H. pylori among adults in industrialized countries appear to be rare. Therefore, without susceptible competitors to displace them, the resistant variants persist.

In a separate report from Sweden, Sjölund and colleagues (4) reported on fecal analysis in 5 patients after treatment with a 1-week regimen of clarithromycin, metronidazole, and omeprazole for H. pylori-related ulcer disease. Few if any clarithromycin-resistant enterococci were isolated from the patients' feces before treatment. After treatment, however, clarithromycin-resistant enterococci were isolated from all patients. In 1 patient, the presence of genetically similar clarithromycin-resistant Enterococcus faecalis was noted 3 years later. Clarithromycin-resistant enterococci were not isolated from a group of 5 untreated controls examined over a similar period.

What is the likely source of the macrolide-resistant enterococci in these patients' gastrointestinal flora? The antimicrobial agent tylosin (a macrolide) was used in many European countries as a growth promoter in food animals in the 1990s. In a study from Switzerland, nearly 100% of enterococci isolated from pigs before tylosin was banned were resistant to erythromycin. After tylosin use was banned in 1999, rates of resistance were reduced to roughly 35% (5). Since enterococci can be acquired through the food chain, it seems likely that most if not all of the patients in Sjölund and colleagues' study may have been colonized at low levels by macrolide-resistant enterococci before H. pylori treatment was administered. The use of a combination of metronidazole, which promotes overgrowth of enterococci in the gastrointestinal tract through its strong antianaerobic activity (6), and erythromycin would predictably lead to overgrowth of enterococcal strains that are resistant to macrolides. Increasing the quantity of resistant enterococci in the stool has clinical consequences. Donskey and colleagues (7) demonstrated a correlation between the numbers of vancomycin-resistant enterococci in patients' feces and the ability to isolate the same bacteria from environmental surfaces in hospital rooms. Surface contamination, in turn, promotes nosocomial spread.

So there is no free lunch. Antimicrobial therapy always incurs a significant risk for adverse effects, either because of the drug's interaction with the patient (generally rare) or the drug's interaction with the normal flora (virtually guaranteed). Against these risks must be weighed the benefits of therapy. Macrolide prescriptions to treat viral upper respiratory tract infections offer little hope of benefit and may doom the patient to later difficulties if H. pylori-related gastrointestinal disease develops. Conversely, widespread treatment of H. pylori disease exposes the patient to risks not only of resistant H. pylori but also of potentially long-term colonization with other resistant bacteria. These results should give pause to those who would promote prolonged antimicrobial therapy for treatment of diseases, such as atherosclerosis, for which a specific infectious cause is far from established.

Consensus panels have generally recognized the inherent risk–benefit ratios of antimicrobial therapy for H. pylori disease by recommending treatment only for patients in whom there is a demonstrable benefit, such as those with history of active peptic ulcer disease, history of documented peptic ulcer, or gastric mucosa-associated lymphoid tissue lymphoma (8). Complicating such limited recommendations is the possibility, as suggested in 1 study (9), that patients with H. pylori-associated nonulcer dyspepsia may be at greater risk for gastric cancer. It is clear that continued research into bacterial and host characteristics that predispose patients to serious sequelae of H. pylori infection will be critical to defining the patient population that will benefit most from treatment. More effective and tolerable therapeutic regimens must also be developed.

As Lewis Thomas surmised, antibiotics are exceedingly powerful tools that can alter the natural relationships between humans and bacteria. But they are indiscriminate, and kill friend and foe alike. As such, they are best used sparingly and, most important, wisely.

• Copyright ©2004 by the American College of Physicians