Evaluating antimicrobial treatment for community-acquired pneumonia: clinical and microbiological responses

1. Evaluating antimicrobial treatment for community-acquired pneumonia: clinical and microbiological responses Daniel M. Musher, MD Head of Infectious Diseases, VA Medical Center, Houston Professor of Medicine Professor of Molecular Virology and Microbiology Baylor College of Medicine Disclosures: Research funding from Merck for followup of Pneumovax study; Romark for nitazoxanide in C. difficile

2. Evaluating Rx for pneumonia: philosophical problems 1. The natural history of infectious diseases: varying proportion resolve spontaneously 2. Generally a very high success rate of existing therapies for common pathogens (this could change with emergence of a new pathogenic organisms causing disease OR newly resistant organisms) 3. “Empiricism” = in many cases, we don’t know what infection we are treating. We unfortunately livewith empiricism, but we must continue to recognize that this increasingly pervasive approach is antithetical to scientific study of medicine

3. Evaluating Rx for pneumonia Without correct diagnoses we have no idea whether, if a patient gets better on treatment, our drug is responsible True cases of the disease are diluted by those that might not respond to, or get better without regard to, treatment Even if we know what we are treating and develop criteria to recognize therapeutic success/failure can we design studies that are large enough to provide meaningful results but still practicable

4. US Army pneumonia vaccine trials, 1942-4 MacLeod, Hodges, Heidelberger, Bernhard, J Exp Med 82:445, 1945 Pneumonia cases Controls Vaccinated Type Included n=8546 n=8449 1 yes 2 2 2 yes 14 1* 4 no 6 8 5 yes 4 1 7 yes 6 0* 12 no 24 21 other - 28 27 all pneumonia 84 60** *p<.05 **p>.05

5. Kayser Permanente study of 7-valent conjugate vaccine (38,000 infants): invasive pneumococcal disease in recipients * Vax Nonvax Infected with vax strain 4 ** 49 Infected, nonvax strain 3 6 * Ped Infect Dis J 19;187-195, 2000 **Only one of these had received the full set of three doses of vaccine

6. Kayser Permanente study of 7-valent conjugate vaccine: otitis media* Reduction by vaccine All visits for otitis media 8.9% OM 4 times per yr 9.3% OM 5 times per yr 11.9% OM 6 times per yr 22.8% Tube placement 20.1% Vaccine type pneumo in MEF 64.7% * Ped Infect Dis J 19;187-195, 2000

7. Evaluating Rx for pneumonia Thus, the goal for studying any new drug should be to eradicate disease for which the etiology is established Some clinicians object: this is not a ‘real life’ scenario If we were prescribing antibiotics only for patients who really needed them, the proposed approach would be much closer to a real life scenario

8. Clinical criteria to evaluate therapeutic success 1. Time to defervescence or mean rate of fall in temperature using Kaplan-Meier analysis of highest recorded daily temp 2. Time to clinical stability Halm et al JAMA 279:1452, 1998 3. Symptom questionnaire Lamping Chest 122:920, 2002

9. Median time to defervescence Welte et al CID 41:1697

10. Median time to defervescence Even when measuring time to defervescence: (a) in patients who are on their way to a cure, does a day or two of lower body temperature really matter? Yes. a. More rapid = more rapid b. Fewer days in hospital c. Probably fewer complications * (b) is the defervescence due to some other property of the antimicrobial agent? * Obviously, failure to defervesce is consistent with clinical failure, although other causes possible

11. Time to clinical stabilityHalm et al JAMA 279:1452, 1998 # abnl at Criterion baseline Median days Temperature <100 63 3 <99 80 3 Pulse <100 56 2 Systolic BP >90 7 2 Respiratory rate <24 49 3 <22 71 3 <20 78 4 O2 saturation >90 23 3 >92 31 3 >94 39 4 Able to eat 11 2 Mental status 8 3

12. Symptom questionnaireLamping Chest 122:920, 2002 Included: chills/sweats; cough; sputum production; chest pain; shortness of breath; vomiting/diarrhea; fatigue; trouble thinking; trouble sleeping In a comparative study of three antibiotic regimens questionnaire was easily administered and well-accepted Shown to be reproducible, reliable and to give valid results

13. Open label study, moxifloxacin vs ceftriaxone + erythromycin: patient diaries

14. Open label study, moxifloxacin vs ceftriax + erythromycin: patient diaries

15. Important to note: Duration of hospitalization was shorter in moxy group (p<.001), but there is no oral form of ceftriaxone, so the comparison is misleading [editors of respectable journals shouldn’t accept such stuff] Overall cure rate was identical in the two treatment groups (85.7% and 86.5%)

16. The problem of open-label studies Essentially not valid for comparative purposes, even if only include “objective observations” FDA simply should not endorse comparative studies that are not blinded; the results (if favorable) will be used for marketing purposes Examples: Moxifloxacin vs ceftriaxone + erythromycin, data obtained from patients diaries If doctors know which drug, so do patients, and all subjective data are invalid

17. What constitutes a clinical failure of treatment for pneumonia? 1. Death – 3-day, 7-10 day, 30-day? 2. Persistent or recurrent bacteremia by causative organism on Rx 3. Complication: necrotic lung, empyema, remote infection (joint, bone, heart valve) 4. Rate of resolution/progression of pneumonia 5. Delayed defervescence 6. Duration of hospitalization

18. What constitutes a clinical failure of treatment for pneumonia? 1. Death – 72 hours, 7-10 day, 30-day? Death within 72 hours due to overwhelming sepsis (cytokine storm) probably unaffected by Rx (Austrian and Gold, Ann Intern Med 60:759, 1964; Finland, Am Rev Resp Dis 120:481, 1979) Death between 72 hr and 10 days influenced by above, but probably pretty good indicator Death by 30 days probably determined by other comorbid conditions; questionable whether antibiotics will affect this, but should be covered by randomization

19. Survival in bacteremic pneumococcal pneumonia: no Rx, Rx serum, Rx penicillin Austrian and Gold (1964)

20. Caveats in studying death as an endpoint in pneumonia 1. Patients must be sick enough to for Rx to have an observable effect 2. The more broadly we cast our net in order to increase our numbers, the greater dilutional effect of death due to other causes. 3. Thus, a study designed to detect all deaths within 3 months may show no difference between treatments A and B, although one might be superior in treating the infection

21. What constitutes a clinical failure of treatment for pneumonia? 2. New, or persistent or recurrent bacteremia by causative organism, while patient is on Rx In CABP, a rare occurrence: e.g., Gram neg rods severely immunocompromised patients, repeated bouts of COPD/pneumonia on many courses of antibiotics and steroids Obviously if bacteremia recurs, it is a failure, but the percentage in which it will be seen is way too small to be useful

22. What constitutes a clinical failure of treatment for pneumonia? 3. Complication: necrotic lung, empyema, remote infection (joint, bone, heart valve) These are usually seen at the time of admission or they appear so soon afterwards that it is difficult to imagine they reflect poor Rx If they do appear on treatment, especially after 3-4 days, very reasonable to consider them as treatment failure Appearance on Rx is so uncommon that, if drug is reasonably effective, it would be difficult to measure without huge sample Finland, The J. Burns Amberson Lecture, Am Rev Resp Dis 20:481, 1979

23. What constitutes a clinical failure of treatment for pneumonia? 4. Rate of resolution/progression of pneumonia: Infiltrates may progress in first few days because inflammatory process continues despite effective antimicrobial agent Study variables in the PORT score (pulse, respiratory rate, temperature, BP, BUN, Na, oxygenation etc.) and apply Kaplan-Meier analysis VERY complicated; dependent upon intensity of Rx and skill of MDs, but in a blinded study, these should average out

24. What constitutes a clinical failure of treatment for pneumonia? Other possible considerations: Days in ICU (for those requiring ICU care) Days of intubation (ICU with intubation) Days of IV therapy (for protocols where switch to oral therapy is an option) CAN ONLY USE THESE IN BLINDED STUDIES Total days in hospital (too dependent on comorbidities)

25. What constitutes a bacteriological cure? First consider bacteriological diagnosis Extensive literature on the unreliability of sputum gram stain and culture in diagnosing bacterial pneumonia Problem is with the patients included in the study- the denominator

26. Results of sputum gram stain [clear bars] or culture [solid bars] showing pneumococci in proven pneumococcal pneumonia Musher CID 2005 All patients Any sputum Valid sample (70%) (55%)

27. Results of sputum gram stain [clear bars] or culture [solid bars] in proven pneumococcal pneumonia: relation to antibiotics Musher CID 2005

28. Bacteriological cure If it is difficult to establish the diagnosis in pneumonia, even more difficult to evaluate efficacy of antibiotic therapy in eradicating Most who could provide a sample before Rx can not do so afterwards Most who “can” → poor/useless sample FDA requirement encourages bad data Culture detects colonizing organisms a. original organism may persist as airway colonizerCalder Lancet 1:1156, 1971 b. new organism may colonize (Tillotson and Finland, J Infect Dis 119:597, 1969) and may not be able to exclude without molecular fingerprinting

29. Bacteriological failure is easier Failure to eradicate in absence of clinical failure: ? significance, but common sense dictates: a. Persistence of large numbers of the original infecting organism in purulent sputum (i.e. gram stain proof) suggests poor antimicrobial effect. This would most likely be associated with poor clinical response, but requires good micro b. Emergence of resistance in the original infecting organism (only if you know the original infecting organism)

30. Microbiological cure Note that these comments address bacterial pneumonia only; for nonbacterial causes, no one has even proposed studying this in pneumonia due to viruses, mycoplasma, chlamydia, or even Legionella

31. What about placebo studies? Ethical considerations: My opinion is simple. Unacceptable. Anyone who signs consent hasn’t been fully informed or isn’t competent to sign Scientific: Can design study of people who don’t have serious disease; spontaneous cures will dilute response. Some may progress to serious disease Must exclude pneumococcal pneumonia

32. Mortality in bacteremic pneumococcal pneumonia (Musher, Mandell ID Text, 2006)

33. Summary and Conclusions: evaluating clinical and microbiological responses during Rx of “CAP” Symptom questionnaire * Time to defervescence * Time to clinical stability * Mortality between 72 hr and 10 day Stay in ICU, days of intubation * Development of a complication on Rx Emergence of resistant bacterium (must prove that it is same organism) Persistent bacteremia *= only in double-blind studies

35. BACKUP SLIDES

36. Causes of “pneumonia” syndrome • Common Less common Streptococcus pneumoniae Moraxella catarrhalis Haemophilus influenzae Staphylococcus aureus Lung cancer Klebsiella pneumoniae Pneumocystis carinii Influenza virus Mycobacterium tuberculosis Legionella CHF, ARDS Pseudomonas aeruginosa Respiratory syncytial virus Microaerophilic/anaerobic Histoplasma, Coccidioides NonTB mycobacteria Chlamydia pneumoniae Nocardia Pulmonary infarction Hammann-Rich, UIP, DIP BOOP, etc.

37. Is microbiologic evaluation of sputum (Gram stain and culture) useful? A good quality specimen is obtained in only slightly >50% cases of pneumonia When obtained before antibiotics are given or within 6 hours of the first dose, and analyzed in an ordinary lab but with motivated laboratory technicians Has an 85% yield by gram stain and/or culture That’s not bad as diagnostic tests go

38. Causes of pneumonia, 1930’s (Heffron)

39. Bacteriological cure b. appearance of new potential pathogen May be S. pneumo (must plan to serotype to detect new type) (Finland) May be S. aureus, GNR, etc. (Tillotson Finland, J Infect Dis 119:597, 1969), either colonizing or causing disease; clinical response remains determining factor Nosocomial acquisition, likely to be resistant to antibiotics 3. A strong incentive to have sample → bad data on bacterial eradication

40. Appearance of new organisms in sputum during Rx Very common, especially in more debilitated and older patients. In the absence of clinical If patient has clinical failure AND now has pathogenic organisms in sputum, Is this failure of original Rx? (did organism develop resistance?) Is this ‘superinfection’?

41. Open label study, linezolid vs. vanco: length of hospital stay Itani, Int J Antimicrob Ther, 2005

42. Bacteriological cure Finland, The J. Burns Amberson Lecture, Am Rev Resp Dis 20:481, 1979 1. Bacteremia rapidly cleared, usually before second dose of penicillin 2. Also rapid eradication of organisms from sputum with modern doses 3. Clinical relapses in pneumococcal pneumonia also related to low doses of penicillin 4. Pneumonia due to different type S. pneumo soon after Rx → ? need to serotype 5. Extrapulmonary complications do not develop after initiation of antibiotics

43. What about placebo studies? Even seemingly simple ones: Retapamulin vs placebo for Rx impetigo, defined as a “superficial, usually self-limited infection” Treated 210 patients (2:1 drug vs placebo) Clinical success rate 85.6% vs 52.1% Two other trials of same drug vs cephalexin; each had about 90% cure rate [might raise objection in MRSA era that there was a placebo effect for some of these cases, which would justify a placebo study] But how self-limited is impetigo if 48% failure rate? And how honest was informed consent?

44. Considerations in the Design of CAP Studies Steve Gitterman DSPTP, FDA

45. Intertwined Considerations • Study design • Study population • Analysis populations • Clinical endpoints • Microbiology outcome • Non-inferiority margin • Inclusion Criteria • Exclusion criteria • Failure • Route of Administration • ‘Diagnostics’ • Blinding • Spectrum of approval for CAP

46. Outpatient (Oral) vs. Inpatient (IV) Studies

47. CAP Considerations Although challenges exist for both inpatient and outpatient studies….. The more difficult issue may be identifying an appropriate non-inferiority margin for drugs that have only oral formulations

48. Inpatient (Parenteral) Studies • Study design • Non-inferiority • Superiority • Study Population • PORT score as criterion (?) • PORT II or III as minimum • Analysis populations • Bacteriologically confirmed (exclude mycoplasma?) • Without bacteriological confirmation • Non-bacterial infections (safety only)

49. Inpatient (Parenteral) Studies • Clinical endpoints; • Failure/success • Mortality • Non-inferiority margin • IDSA recommendations presented

50. Outpatient (Oral) vs. Inpatient (IV) Studies