| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
Puget Sound Veterans Affairs Medical Center Seattle, Washington
Correspondence to: Jan V. Hirschmann, MD, Medical Service (III), Puget Sound Veterans Affairs Medical Center, 1660 S. Columbian Way, Seattle, WA 98108-1532; e-mail: pepsi{at}u.washington.edu
To the Editor:
To label me or my viewpoints as "pessimistic," "naysayer," "nihilistic," and "negative," as Murphy et al1 do in their response (July 2000) to my article in CHEST,2 is no substitute for thoughtful critical analysis, and it is the kind of personal attack that debases medical discourse. Instead, the discussion should focus on the nature, quality, and validity of the evidence about whether bacteria cause COPD exacerbations. Accordingly, the substantive assertions that Murphy et al1 make in their response and the references that they use to support them deserve careful scrutiny to determine the meticulousness of their scholarship and credibility of their conclusions. As with all scrupulous critical analysis, their statements should follow certain axioms: (1) the assertions should be relevant to the issues being discussed; (2) the authors should provide evidence for any substantive statement that is not self-evident; (3) the cited references should support the assertion being made; (4) the authors should be accurate and thorough in summarizing the relevant information that the references provide; (5) the cited articles should be scientifically rigorous; if not, the authors should describe the weaknesses and limitations of the studies; and (6) the authors should be accurate and fair in summarizing viewpoints that differ from their own. A careful reading of their response demonstrates that Murphy et al1 repeatedly fail to follow these principles.
Microbiologic Studies
Murphy et al1 state that "cultures of expectorated sputum samples ... do not reliably reflect conditions in the lower airways," but they provide no references for that assertion. In fact, this topic has received little exploration in chronic bronchitis. I know of only one relevant study, published in 1961.3 It examined the bacteriology of expectorated sputum, pharyngeal cultures, and swabs taken from the trachea and bronchi through a bronchoscope or through rubber catheters introduced fluoroscopically in 52 hospitalized patients, 8 of whom had uncomplicated chronic bronchitis. All eight patients had either Haemophilus influenzae or Streptococcus pneumoniae in the bronchi; in seven patients, these organisms were also present in the expectorated sputum, and in all eight patients they were cultured from the pharynx. This small study is hardly definitive, but it does not support the assertion that Murphy et al1 make about the unreliability of expectorated sputum samples in revealing the putative pathogens in chronic bronchitis.
In furthering their argument that sputum cultures are unreliable, Murphy et al1 state that the article on bacteremic pneumococcal pneumonia by Austrian and Gold4 showed that S pneumoniae is absent from the sputum cultures in one fourth of the patients. This statement is misleading in two ways. It is factually incorrect—nowhere in that report do the authors discuss the frequency of positive sputum cultures in their patients with bacteremic pneumococcal pneumonia. Second, the issue is irrelevant in any case. Using studies of sputum cultures in pneumonia to argue about their validity in exacerbations of chronic bronchitis is inappropriate, because these two respiratory disorders differ in the patient populations affected, pathogenesis, pathology, and locations of inflammation. In addition, one major problem in patients with pneumonia is that many cannot provide adequate sputum samples.5 Since an increase in the amount or a change in the color of sputum are defining features of exacerbations of chronic bronchitis, obtaining an adequate specimen during these episodes is rarely difficult.
Murphy et al1 state that studies of quantitative cultures of sputum "show that it is an oversimplification to assume that pathogens are more likely to be present, or to be present in higher numbers, in sputum cultures during exacerbations than in stable periods." They do not indicate which studies demonstrate that point or in what way the investigations allow such a conclusion. Indeed, I cannot conceive of how the available evidence justifies their statement.
They then cite as "innovative" the study showing that 25% of patients with stable COPD had > 1,000 cfu/mL of bacteria recovered from protected-specimen brush technique compared to another group of patients with exacerbations, in whom 52% had this number of bacteria.6 They assert that the data in this study are "relevant to the issue being considered," but do not elucidate how they help resolve the question of whether bacteria cause COPD exacerbations. Of course, this is another investigation using quantitative cultures, which they had previously dismissed as failing to provide "definitive answers."
In fact, no technique currently available can definitively determine whether bacteria cause exacerbations, which is why I suggested using several different criteria in my article. One of those was that the putative pathogens should be present more frequently in patients during exacerbations than remissions and that, among those chronically colonized, the density of organisms should increase. The available studies do not demonstrate such changes—an important, but not definitive, piece of evidence against the thesis that bacteria cause exacerbations. Certainly, if the investigations had been positive, proponents of that thesis would justifiably cite the information as data supporting their viewpoint.
Serologic Studies
Investigations examining serologic evidence for bacterial infections in exacerbations of chronic bronchitis should include three control groups to determine the specificity of any changes observed: (1) normal hosts; (2) patients with chronic bronchitis who are colonized with the organism being examined, but who are not currently experiencing an exacerbation; and (3) patients with chronic bronchitis who do not have the organism present in their sputum, but who are experiencing an exacerbation. Such controls are important because studies in COPD patients have demonstrated that significant changes in antibody titers to the putative pathogens can occur in the absence of symptoms, ie, during periods of disease quiescence,7 and during exacerbations even when the sputum cultures were negative for the bacterium being studied.8 Murphy et al1 cite three papers to support their statement that a specific immune response occurs in the bacteria present in the sputum of patients during exacerbations. In one, the investigators studied 14 patients with what they called "acute febrile tracheobronchitis due to H influenzae."9 Eleven patients had COPD, and 6 patients (the number of them with COPD is unspecified) had serologic studies of opsonizing antibodies against the H influenzae cultured from their sputum. Nine patients with stable COPD, but with sputum cultures positive for H influenzae, were control subjects. Of the six patients with tracheobronchitis, four patients had an increase in opsonizing activity in their convalescent, compared to their acute, sera. In the nine patients with stable COPD, the opsonizing activity was higher overall, but did not change in two samples taken 2 to 3 weeks apart. Unfortunately, the authors did not study patients with COPD exacerbations, but without H influenzae in their sputum, to determine whether the increased opsonizing activity observed was confined to those with H influenzae in their sputum or was a nonspecific increase in opsonization that occurs with exacerbations independent of what organisms are present. Moreover, the patients examined were not typical patients with COPD exacerbations, but they were distinctive enough for the authors to create a new diagnostic title for their disorder.
In another study of H influenzae, Yi et al10 examined the immune response in only two patients with COPD, both of whom grew the organism during a purulent exacerbation, but not from a sputum culture 1 month previously. Several serologic studies demonstrated that both patients developed new antibodies to a limited number of bacterial antigens of H influenzae. The investigation included no control groups to determine the specificity of this finding.
A study of the immune response to Moraxella (Branhamella) catarrhalis, isolated from the sputum in 13 patients with COPD exacerbations, examined the bactericidal activity of their serum against the organism.11 The investigation also included eight patients with other respiratory infections from whom M catarrhalis was cultured. The investigators used pooled serum from 20 normal, healthy adults as control subjects. All but one of the patients with COPD exacerbations either had bactericidal antibodies in their acute sera or developed them in their convalescent sera. The pooled serum from the healthy adults showed no bactericidal activity against 20 of the 21 isolates of M catarrhalis tested. Unfortunately, the study did not examine the sera from patients with COPD exacerbations, but without M catarrhalis in their sputum, to determine the specificity of their findings.
All three of these reports had inadequate controls. This deficiency, plus the extremely small number of patients studied, limits any confident conclusion from these studies, which hardly provide "another line of compelling evidence that bacteria cause exacerbations of COPD," as Murphy et al1 assert.
Vaccine Studies
Murphy et al,1 alluding to the trials of an oral vaccine against H influenzae discussed in my article, called them "promising," but did not address any of the problems or limitations of the studies that I delineated.
Pathologic Investigations
Murphy et al1 state that in all studies of sputum, BAL fluid, and bronchial biopsy specimens of COPD exacerbations, neutrophils outnumber eosinophils. The important issue, however, is what change in the numbers occurs in exacerbations compared with baseline. The studies cited show an increase in the proportion of eosinophils, but not in neutrophils.12 13 In a similar situation, when a patient with a previously normal white cell differential count develops blood eosinophilia of 30%, for example, clinicians do not dismiss this finding because neutrophils still outnumber the eosinophils.
Murphy et al1 cite five studies that they claim show an "association of eosinophilia with a variety of viral and bacterial infections." Three of those articles do not examine blood or tissue eosinophilia at all. One is an in vitro study of the antibacterial actions of eosinophilic major basic protein and eosinophilic cationic protein against Staphylococcus aureus and Escherichia coli14 ; another is an in vitro study examining the release of eosinophilic cationic protein from eosinophils, when cultured with cells infected with respiratory syncytial virus15 ; and the third measures eosinophil granule proteins, but not eosinophils themselves, in the blood of 46 patients with bacterial and virus infections.16 The two other studies cited include a report of 77 HIV-infected patients in Venezuela, in which 20 subjects had blood eosinophilia,17 and an article on the serum concentration of eosinophilic cationic protein in 25 children with Mycoplasma pneumoniae pneumonia, some of whom had blood eosinophilia.18 None of these references is germane to answering a critical question that any proponent of a bacterial cause of exacerbations must confront—why should the pyogenic organisms, H influenzae, S pneumoniae, and M catarrhalis, which Murphy et al1 believe cause COPD exacerbations, provoke eosinophilic inflammation when they involve bronchi, yet produce neutrophilic inflammation without eosinophilia when they infect any other tissue in the body?
Murphy et al1 cite an article in which lung explants from patients with COPD undergoing lung transplantation were studied for evidence of H influenzae by immunoperoxidase staining and polymerase chain reaction (PCR).19 They failed to give adequate details or to explore the meaning of the results. Tissue sections were positive by staining and PCR in 8 of 16 patients with COPD, but they were also positive in 4 of 12 patients with "noninfectious" pulmonary diseases, including pulmonary hypertension, pulmonary fibrosis, and Langerhans cell histiocytosis. A higher percentage of tissue sections (94%) stained positive for COPD than for the noninfectious diseases (88%), a statistically significant difference. The fact that the studies were positive in a substantial number of patients with disorders where H influenzae has no known or proposed pathogenic role makes it difficult to discern how this article can provide useful information about the role of this bacterium in causing exacerbations of chronic bronchitis. Certainly, Murphy et al1 provide no interpretation to elucidate this issue.
Antibiotic Studies
In discussing the Canadian trial,20 Murphy et al1 state that the difference in peak flow of 10 L/min "certainly could be clinically relevant." Anyone familiar with peak flow rates recognizes that not only is this difference attributable to expected variations in testing but it is also clearly inconsequential, both clinically and physiologically.
Murphy et al1 misrepresent my position when they say that I stated that "the use of corticosteroids confused the trial ... " The problem was not that the Canadian study included patients receiving corticosteroids, but that the trial design did not account for the possibility these medications could have an independent effect on the resolution of exacerbations. The authors should have ensured that patients received standard doses and duration of corticosteroids and that these patients were randomized to placebo vs antibiotics separately from those not receiving corticosteroids. This important defect in the study is baffling, for a prior study had indicated a benefit for corticosteroids in hospitalized patients with exacerbations,21 and many clinicians believed that they were effective in treating outpatients as well, a point demonstrated in this trial by the fact that the practitioners involved prescribed them for about 43% of the exacerbations.
Murphy et al1 contend that I incompletely described this trial when I stated that "antibiotics were not better than placebo in those patients receiving corticosteroids." The authors of the trial state, "In exacerbations treated with steroids, the success rate with antibiotic was higher (though not statistically so) than with placebo." Since they provide no data, it is uncertain whether the advantage was slight or marked. Patients received corticosteroids in about 43% of the 362 exacerbations, or approximately 155 episodes, in which about one half received antibiotic, and one half received placebo. If the therapeutic effects of antibiotics had been substantial, one would have expected a statistical difference to have emerged from such a large experience. Obviously, no matter how interpreted, this study did not provide a scientifically rigorous answer to the question of whether antibiotics provide any additional benefit in COPD exacerbations when patients receive systemic corticosteroids, which several studies have demonstrated are highly beneficial and should now be standard therapy.21 22 23 24 Resolving this question will require a large randomized trial, an effort that Murphy et al1 appear unwilling to endorse, when they state that "future studies should not be asking whether antibiotics are helpful ... " Given the available evidence, such an attitude is scientifically indefensible.
Murphy et al1 argue that certain antibiotics are superior to others in subsets of patients. They cite two studies. One is a randomized, controlled trial comparing ciprofloxacin vs other nonquinolone antibiotics, which purportedly shows that ciprofloxacin is better in a group with moderate or severe COPD and at least four exacerbations in the previous year.25 Murphy et al1 fail to delineate the serious problems in this investigation, which the manufacturers of ciprofloxacin funded. No microbiologic studies were done; the treatment of exacerbations was not standardized; and the use of systemic corticosteroids was neither regulated nor recorded. Both patients and clinicians were aware of the antibiotic prescribed, and the assessment of the duration of exacerbations rested exclusively on patient self-reporting. These last two features allow a tremendous potential for bias. For example, the practitioners could have implied, or the patients inferred, that the newer antibiotic, with a broader spectrum of activity, was inherently superior. Such a severely flawed investigation does not meet even minimal standards of scientific rigor.
The other study that Murphy et al1 cite is even worse. This retrospective investigation examined the medical records of 60 outpatients with 224 exacerbations and characterized their antibiotic treatment as belonging to one of three groups: (1) first-line agents (amoxicillin, tetracycline, erythromycin); (2) second-line agents (cephradine, cefuroxime, cefaclor, cefprozil); or (3) third-line agents (amoxicillin-clavulanate, azithromycin, ciprofloxacin).26 The study found a significantly higher failure rate within 2 weeks of initiating the antibiotic in those receiving the first-line agents than in those receiving third-line agents. The study has numerous problems. Bacteriologic results were not reported. Important baseline measurements, such as spirometry, were not available for all patients. Among the FEV1 values recorded, those of patients receiving first-line agents averaged 1.27, compared with 1.71 in those receiving third-line agents, suggesting that the latter had milder attacks, which alone might explain the difference in outcome. The authors do not indicate the nature of other treatments given for the exacerbations, including systemic corticosteroids. The authors provide no assurance that the patients actually took the medications prescribed. Furthermore, a chart review such as this does not ensure that follow-up of patients was adequate: some patients who did not improve may have gone for treatment elsewhere, others may have not returned and, instead, just waited for the attack to subside spontaneously. These patients would not have been detected as treatment failures. Obviously, this study has debilitating defects of design and execution.
That Murphy et al1
should base their contentions on two
such severely flawed studies reveals the weakness of their position and
their lack of critical analysis in assessing the scientific merit of
their sources. A better way to determine whether amoxicillin or other
first-line agents are clinically inferior to third-line antibiotics is
to examine prospective, randomized, double-blind studies conducted at
the same time as the retrospective study mentioned above. Trials that
test one antibiotic against another have several problems, which I
discussed in my article.2
Compared to the retrospective
study, however, at least they have the advantage of predefined
criteria, bacteriologic results, careful follow-up, blinded evaluation
of patients, and standardized methods of collecting and recording data.
I could find two such investigations.27
28
In the first,
218 patients received amoxicillin and 538 received grepafloxacin, a new
fluoroquinolone, in two different doses. Clinical response was about
85% in the three groups, even though the microbiologic success was
significantly higher (98%) in the grepafloxacin group than in the
amoxicillin group (89%). In the amoxicillin group, 27 isolates were
-lactamase positive M catarrhalis or H
influenzae. In the second trial, 132 patients received amoxicillin
and 279 patients received 2 different doses of trovafloxacin, another
fluoroquinolone. Clinical success was about 90% in each group, and all
three groups had equivalent bacteriologic efficacy. Even though some
authors have urged the use of broader spectrum agents for acute
exacerbations and have cited growing antibiotic resistance among the
putative pathogens, I know of no well-designed prospective studies that
show clinical advantages for the newer macrolides, the
fluoroquinolones, or amoxicillin-clavulanate when compared to older
agents, such as amoxicillin or tetracycline. For those who are
skeptical about the role of bacteria in acute exacerbations and the
efficacy of antibiotics in treating them, it is unsurprising that
studies fail to show superiority of one antibiotic over another.
In contrast to my analysis, Murphy et al1 contend that their assessment of the evidence is more "objective," "comprehensive," "critical," "open-minded," and "constructive." An impartial, but rigorous evaluation of my article, their response, this rejoinder, and most importantly, the relevant references will indicate to the reader that their self-adulation is undeserved—a large percentage of the substantive statements that Murphy et al1 make are irrelevant to answering the questions that my article addressed, or they violate one or more principles of sound scholarship that I delineated earlier. The most striking defect is the virtual absence of any searching critical analysis of the references that they cite.
References
SUNY Buffalo and VA Western New York Healthcare System and Winthrop University Hospital Mineola, NY
Correspondence to: Michael S. Niederman, MD, FCCP, Chief Division of Pulmonary and Critical Care Medicine, Winthrop University Hospital, 222 Station Plaza North #400, Mineola, NY 11501-3893; e-mail: mniederman{at}winthrop.org
To the Editor:
Clearly, Dr. Hirshmann disagrees with statements in our article and we disagree with his interpretation of the literature. However, our article contained no personal attacks and we apologize if our statements were perceived as such. A point-by-point rebuttal of Dr. Hirshmann’s current letter is not likely to contribute to the debate in a meaningful way.
Suffice it to say that we stand by all of the statements and conclusions in our article and refer the reader to the article for a more detailed and literature-based consideration of the issues. We have attempted to synthesize a broad range of information into coherent themes with particular attention to proposing potentially fruitful avenues of research. As investigators and clinicians who work in this area, we believe that the most constructive approach to dealing with the limitations of the current literature is to design and perform laboratory experiments and clinical trials to answer the critical unanswered questions. Perhaps the most important conclusion from this debate is that an urgent need exists for innovative original research to more precisely define the role of bacteria in exacerbations of COPD and in the course and pathogenesis of the disease.
This article has been cited by other articles:
|
|
 |
|
  M. C. Bruce and E. N. Bruce Analysis of factors that influence rates of carbon monoxide uptake, distribution, and washout from blood and extravascular tissues using a multicompartment model J Appl Physiol, April 1, 2006; 100(4): 1171 - 1180. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
