The usual course of community-acquired pneumonia (CAP) after antimicrobial treatment shows progressive improvement in the clinical signs and symptoms of infection with a reduction in abnormal test results and total resolution of lung infiltrates on chest radiographs. In cases with an inadequate therapeutic response, the clinical symptoms and abnormal test results may persist, or the patient’s condition may even deteriorate with dissemination of the infection and/or multiple organ failure or even death.
Treatment failure has been recognized as one of the main reasons for poor prognosis in patients with most forms of severe CAP. Moreover, the mortality in pneumonia with higher initial severity as evaluated with prognostic scales is increased with treatment failure and is lower if the response to treatment is adequate. In all patients with CAP, including those with mild-to-moderate CAP, treatment failure is related to a longer time to clinical stability, longer hospital stay, and more complications . For the whole CAP population, treatment failure leads to higher morbidity, mortality, and costs.
TREATMENT FAILURE: CONCEPT AND DEFINITION
A crucial aspect of CAP that has not totally been resolved as yet is treatment failure. Since the previous empiric and arbitrary definitions described by Fein and Feinsilver  and several other authors, we have moved on to the latest complete and complex classification released by the Infectious Diseases Society of America (IDSA) in cooperation with the American Thoracic Society (ATS) consensus . Clinical response is difficult to define because of many factors related to the resolution of CAP. Thus, the time required to eliminate the microorganisms depends on the microorganisms, adequate treatment, and the condition of the host. Another decisive question is the most appropriate time to consider that an antibiotic has failed. The studies by Montravers et al  showed that after 72 hours of adequate treatment, respiratory samples obtained by protected brush specimen demonstrated control of infection in 88% of the patients with ventilator-associated pneumonia. Most of the studies addressing the issue of treatment failure have used 72 hours after initiation of treatment as the time to evaluate the clinical response in CAP patients [5-9].
New insights concerning clinical stability in CAP support this assumption. Halm et al  found that the median time to clinical stability in CAP was 3 days, corroborating the microbiological study of Montravers et al  that demonstrated a correlation between microbiological findings and clinical signs and symptoms. The criteria employed to define clinical stability included several clinical parameters such as the patient’s temperature, heart rate, systolic blood pressure, respiratory rate, and oxygen saturation, with various thresholds. These authors highlighted that clinical stability was related to initial severity, that is, the higher the initial severity, the higher the number of days required for achieving clinical stability. Nevertheless, other independent factors related to clinical stability were the presence of comorbidities such as chronic bronchitis, treatment not compliant with guidelines, and complications .
The new systematic classification of causes of nonresponding CAP from the IDSA/ATS consensus , took the time of onset of infection and type of failure into account (Table 1). For hospitalized CAP patients, 72 hours is the period most often used to evaluate response to therapy, with deterioration after this period often being related to complications, comorbid diseases, or nosocomial superinfection. Two different patterns of nonresponding pneumonia have been described in patients hospitalized in other than intensive care units (non-ICU). Progressive pneumonia is considered in cases with clinical deterioration with acute respiratory failure requiring ventilatory support and/or the appearance of septic shock, usually occurring within the first 72 hours of hospital admission. Persistent or nonresponding pneumonia is defined as the absence of or delay in achieving clinical stability. The term early failure (<72 hours) employed in two recent studies was used for cases of treatment failure before 72 hours of treatment and was similar to that of progressive pneumonia published in the consensus documents [5,6].
| ||TABLE 1. Classification of nonresponding pneumonia |
(Modified with permission from Mandell LA, Wunderink RG, Anzueto A, et al. Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of community-acquired pneumonia in adults. Clin Infect Dis 2007;44(suppl 2):S27-S72. Copyright © 2007, The University of Chicago Press)
INCIDENCE AND CAUSES
Due to the difficulties in defining treatment failure, its incidence has not been completely determined. In outpatients, the incidence of treatment failure in CAP estimated from patients who need subsequent hospital admission is not clearly known. In a retrospective database of 7526 outpatients, a frequency of about 21% was reported . In hospitalized CAP patients, the percentage ranges from 6 to 15% [5,6,9]. In about half of treatment failure cases, early failure develops within the first 72 hours [6,12]. In the ICU setting, the incidence of treatment failure is higher, about 40% .
The causes were classified as infectious in 40% of treatment failure cases, noninfectious in 15%, and not determined in 45% [5-7] (Table 2). Streptococcus pneumoniae, Legionella pneumophila, Staphylococcus aureus, and Pseudomonas species are the most common etiologic microorganisms in treatment failure in CAP patients [6,13]. In a study of nursing home patients hospitalized for CAP, El-Solh et al  reported that the most common microorganisms related to this disease were methicillin-resistant S. aureus (MRSA; 33%), enteric gram-negative bacilli (24%), and Pseudomonas aeruginosa (14%). Although resistance of S. pneumoniae against penicillin and other antibiotics was not demonstrated to be independently related to mortality [15,16], some isolated cases of treatment failure have been reported with levofloxacin and macrolides [17,18]. Rosón et al  reported that L. pneumophila is a possible cause of early failure when initial antimicrobial treatment does not include that specific etiology. Moreover, Legionella infection may develop into a progressive CAP with increasing infiltrates, more complications, and a higher mortality rate , mainly if several species of Legionella are involved. P. aeruginosa can cause treatment failure if the infection is persistent and has not previously been covered by antibiotics (10%) or if the infection appears later as a nosocomial complication . Unusual microorganisms that can cause treatment failure in CAP are mycobacteria, Nocardia species, Pneumocystis jirovecii, anaerobes, leptospires, endemic fungi, and others .
| ||TABLE 2. Infectious and noninfectious etiologies of treatment failure in community-acquired pneumonia |
Pleural effusion is frequently associated with treatment failure .
Noninfectious diseases cause almost 15% of treatment failures in CAP . The etiologies described in several reports include cancer, lung hemorrhage, bronchiolitis obliterans with organizing pneumonia, eosinophilic lung, hypersensitivity pneumonitis, and others [8,21].
RISK FACTORS OF TREATMENT FAILURE
In the last several years, various risk factors have been related to a poorer clinical response and a poor prognosis. In more recent studies we have identified the independent risk factors for treatment failure as well as independent protective factors (Table 3), which are related to the initial severity, patient characteristics and concomitant diseases, the causal microorganisms, and antimicrobial treatment.
|TABLE 3. Factors related to treatment failure |
Rosón et al  and Menéndez et al  reported that, as expected, the higher the initial severity measured by Pneumonia Severity Index (PSI), the higher the risk for treatment failure, with an odds ratio (OR) of between 1.3 and 2.7.
Comorbid conditions may also play a role in treatment failure. Liver diseases are associated with a higher risk of failure, whereas chronic obstructive pulmonary disease (COPD)  and heart failure have been associated with lower risks of treatment failure .
Inadequate antibiotic treatment and regimens not compliant with guidelines have been related to a worse outcome in CAP [6,23,24]. On the other hand, influenza vaccination has been found to be associated with a lower risk of treatment failure in CAP .
INFLAMMATORY RESPONSE OF HOST AND BIOLOGICAL MARKERS
In the complex response of host against microorganisms, a local inflammatory response develops in order to contain the infection. This response may be sufficient to control infection and to avoid dissemination, but the infection may also remain local and not extend to the systemic circulation. An excessive proinflammatory response has been related to a higher rate of complications, multiple organ failure, and death. In a multicenter study with about 1900 patients with CAP, Kellum et al  aimed to investigate the systemic inflammatory response and its impact on prognosis. The study involved determinations of both proinflammatory cytokines (tumor necrosis factor-alpha and interleukin [IL]-6) and anti-inflammatory cytokines (IL-10), tested daily for the first 7 days and then once a week. This study confirmed that systemic levels of cytokines were higher initially, followed by a rapid reduction the first 3 days then a slightly slower reduction thereafter. Interestingly, patients with sepsis and/or those who died exhibited the highest levels of cytokines, both IL-6 and IL-10. These results corroborate previous studies of CAP, sepsis, and ventilator-associated pneumonia, highlighting the influence of systemic inflammation on outcome.
In a prospective study, Menéndez et al  aimed to evaluate the systemic inflammatory profile of cytokines and biological markers initially and after 72 hours of antimicrobial treatment. Their results showed that both initial and persistently higher levels of IL-6 and IL-8 and the markers C-reactive protein (CRP) and procalcitonin (PCT) were associated with treatment failure. After adjustments were made for initial severity, multivariate analyses pointed out the independent effect of biological markers. For early failure (<72 hours) high levels of both CRP and PCT increased the risk of failure twofold to threefold (relative risk of 2.6 and 2.7, respectively). For failure after 72 hours, the CRP level retained its predictive value, whereas PCT was not independently associated, possibly related to the different kinetics between CRP and PCT, with PCT mainly being released more rapidly than CRP. The best threshold for predicting treatment failure has not been sufficiently addressed. In our study, we proposed a threshold value of 21.9 mg/dL for CRP on day 1, the 75th percentile of the group of patients with good response to treatment. This cut-off provided moderate sensitivity (57%) and specificity (72%) and a low positive predictive value (16%) but a very high negative predictive value (94%).
CRP and PCT levels have been monitored in several studies to determine their use in predicting a good or poor prognosis. The pattern of decreasing levels has been correlated with a better outcome. On the contrary, the lack of a reduction and persistent or increasing levels of these markers have been reported in CAP and in ventilator-associated pneumonia as independent risk factors of mortality. In a multicenter study of 1671 patients with CAP, Kruger et al  evaluated the value of PCT together with the CRB-65 scale to improve the prediction of death. The authors selected the value of 0.228 ng/mL as the cut-off for predicting mortality. The predictive value of this cut-off was similar to that of the CRB-65 scale, after the area under the receiver operating characteristic (AUC) curve (0.80 vs 0.79) was calculated. When PCT values were added to CRB-65 scores, the AUC curve increased to 0.83. Interestingly, in that study the low levels of PCT were associated with a significantly reduced risk factor for mortality even with the higher severity scores (CRB-65 >3).
New markers such as pro-atrial natriuretic peptide (pro-ANP) and pro-vasopressin [28,29] have shown promising results in predicting complications and poor outcome. Prat et al  reported that high levels of pro-ANP were related to a greater number of complications and a higher mortality rate but were not related to the causal microorganism. Similarly, Kruger et al  compared the predictive value of markers against that of the CRB-65 scale in 589 patients with CAP. The predictive value was calculated with the AUC, which was 0.86 for pro-vasopressin and 0.76 for pro-ANP; both were higher than the AUC for CRB-65 (0.73).
At present, the newest field of investigation regarding the inflammatory response in CAP is related to the impact of genetic polymorphisms of inflammatory response and host defense against microorganisms on prognosis .
In cases of treatment failure, a complete re-evaluation of the patient and the clinical history must be done (Table 3). The first clinical step is to re-evaluate whether the diagnosis of CAP is adequate and rule out other noninfectious etiologies mimicking CAP. The second step is to re-evaluate the possible microorganisms involved, because unusual, persistent, or resistant pathogens might be related to treatment failure. The presence of some microorganisms and host factors may explain the slower resolution of infectious parameters. Thus, CAP due to Legionella, bacteremic pneumonia, and other etiologies may be responsible for a protracted clinical course and delayed resolution. Moreover, elderly patients with comorbid conditions or immunosuppression may experience slower resolution of symptoms. In the presence of these circumstances and the absence of clinical deterioration, a conservative approach with clinical observation and serial radiographs may suffice. On the contrary, a new microbiological examination with noninvasive and invasive samples is recommended. A complete protocol for investigating the reasons for treatment failure, including examination of respiratory samples obtained with noninvasive and invasive techniques and radiographic studies provided a diagnosis in almost 70% of cases of treatment failure . Nevertheless, the beneficial effect of microbiological investigation on the outcome has not been completely demonstrated [14,31].
Noninvasive microbiological studies may rule out the persistence of infection, the appearance of resistance during treatment, or a new nosocomial infection. Respiratory samples may be obtained with invasive techniques including fiberbronchoscopy or fine-needle puncture aspiration. Bronchoscopy allows direct observation of the airways and recovery of samples directly in the area of CAP. In fact, it has been reported that examination of protected brush specimens and bronchoalveolar lavage (BAL) specimens had a diagnostic yield of 41% in cases of treatment failure [6-8]. Microbiological studies of protected brush specimens and BAL fluid may include stains and cultures for the usual bacteria, fungi, viruses, and opportunistic microorganisms. Conventional and modified Ziehl staining may be used when Nocardia is suspected, and direct immunofluorescence and culture should be used for the investigation of Legionella. For bacteria cultures, the results should be expressed as colony-forming units (CFU) per milliliter in order to differentiate between colonization and infection. However, the colony counts of conventional bacteria must be interpreted with caution and together with other tests, because previous antibiotic treatment can reduce the counts to below the established cut-off of 103 CFU/mL for protected brush specimens and 104 CFU/mL for BAL fluid. In patients with mechanical ventilation, examination of a tracheal aspirate using a cut-off of 105 CFU/mL had a good diagnostic yield in cases of treatment failure (sensitivity of 93%, specificity of 80%) .
BAL fluid study may also provide very useful diagnostic information concerning noninfectious etiologies mimicking CAP. The study and count of cells in BAL fluid allows for orientation of the differential diagnosis of noninfectious causes . Pulmonary hemorrhage is suggested by the presence of blood or >20% of hemosiderin-loaded macrophages . Eosinophil counts higher than 20% make it mandatory to rule out pulmonary eosinophilia, fungal infection, and drug-induced pneumonitis. An increase in the number of lymphocytes in BAL fluid indicates hypersensitivity pneumonitis, sarcoidosis, and pulmonary fibrosis as possible causes. When the leukocyte count is increased, the differential diagnosis should include bronchiolitis obliterans with organizing pneumonia.
Bronchial biopsy should be performed if abnormalities are found in the airways. The role and indication of transbronchial biopsy is not clear. It is indicated if airway examination rules out other findings and if there is persistence of infiltrates and no diagnosis has be achieved. Arancibia et al  obtained up to 57% of diagnoses with transbronchial biopsy in nonresponding pneumonia, although this procedure was done only in 25% of their cases.
The computed tomography (CT) scan allows complete examination of the lung parenchyma, pleural space, mediastinum, and heart. If thromboembolism is suspected, a helicoidal CT may be indicated. CT can disclose reasons for treatment failure such as loculated pleural effusions, lung abscess, adenopathies, or even an obstructive tumor in the airway. The distribution and morphology of infiltrates in the parenchyma may indicate some specific microorganisms  and may be useful in selecting the lobes from which to obtain microbiological or histological samples.
In regard to the persistence of infiltrates in radiographs, the approach may be more conservative if the infiltrates are reducing, symptoms are ameliorating, and the time of infection is less than 6 weeks . The period for choosing a more invasive diagnostic process when residual radiographic infiltrates persist remains to be elucidated. Nevertheless, after this period a more complete evaluation with bronchoscopy and CT may be indicated. Although not very common, lung carcinoma may be the reason for unresolved infiltrates .
THERAPEUTIC APPROACH AND PROGNOSIS
To our knowledge, there are no evidence-based recommendations for antimicrobial therapies in treatment failure. Nonetheless, while microbiological study results are awaited, an empirical treatment change or escalation of antibiotic therapy is advisable in order to include resistant microorganisms, Pseudomonas, or anaerobes. An alternative may include use of beta-lactamase inhibitors (cefepime, imipenem, meropenem, piperacillin/tazobactam) plus intravenous fluoroquinolones and assessment and association of endovenous macrolides (azithromycin or clarithromycin). In institutionalized older patients with severe pneumonia and prior antibiotic exposure, El-Solh et al  recommend selecting two antipseudomonal agents in addition to vancomycin, until MRSA can be ruled out and microbiological study results are available.
If Aspergillus species is highly suspected (risk factors include severe COPD and previous use of corticosteroids or other immunosuppressants), antifungal therapy should be initiated before Aspergillus infection is ruled out.
Treatment failure increases the mortality in CAP from threefold to elevenfold. Arancibia et al  found a global mortality rate of 43% in patients hospitalized in conventional wards and the ICU, and the rate (42%) was similar in patients in nursing homes . The probability of death also depends on the cause of treatment failure, with a probability of 88% in nosocomial infection, 40% in persistent infection, and 27% when no diagnosis is reached.
1. Menéndez R, Torres A, Rodriguez de Castro F, et al. Reaching stability in community-acquired pneumonia: the effects of the severity of disease, treatment, and the characteristics of patients. Clin Infect Dis 2004;39(12):1783-1790. [Medline]
2. Fein AM, Feinsilver SH, Niederman, MS. Nonresolving and slowly resolving pneumonia: diagnosis and management in the elderly patient. Clin Chest Med 1993;14(3):555-569. [Medline]
3. Mandell LA, Wunderink RG, Anzueto A, et al. IDSA/ATS consensus guidelines on the management of community-acquired pneumonia in adults. Clin Infect Dis 2007;44(suppl 2):S27-S72. [Medline]
4. Montravers P, Fagon JY, Chastre J, et al. Follow-up protected specimen brushes to assess treatment in nosocomial pneumonia. Am Rev Respir Dis 1993;147(1):38-44. [Medline]
5. Menéndez R, Torres A. Treatment failure in community-acquired pneumonia. Chest 2007;132(4):1348-1355. [Medline]
6. Rosón B, Carratalà J, Fernández-Sabé N, et al. Causes and factors associated with early failure in hospitalized patients with community-acquired pneumonia. Arch Intern Med 2004;164(5):502-508. [Medline]
7. Ortqvist A, Kalin M, Lejdeborn L, et al. Diagnostic fiberoptic bronchoscopy and protected brush culture in patients with community-acquired pneumonia. Chest 1990;97(3):576-582. [Medline]
8. Arancibia F, Ewig S, Martinez JA, et al. Antimicrobial treatment failures in patients with community-acquired pneumonia: causes and prognostic implications. Am J Respir Crit Care Med 2000;162(1):154-160. [Medline]
9. Aliberti S, Amir A, Peyrani P, et al. Incidence, etiology, timing and risk factors for clinical failure in hospitalized patients with community-acquired pneumonia. Chest 2008;134:955-962. [Medline]
10. Halm EA, Fine MJ, Marrie TJ, et al. Time to clinical stability in patients hospitalized with community-acquired pneumonia: implications for practice guidelines. JAMA 1998;279(18):1452-1457. [Medline]
11. Ye X, Sikirica V, Schein JR, et al. Treatment failure rates and health care utilization and costs among patients with community-acquired pneumonia treated with levofloxacin or macrolides in an outpatient setting: a retrospective claims database analysis. Clin Ther 2008;30(2):358-371. [Medline]
12. Menéndez R, Torres A. Risk factors for early and late treatment failure in community-acquired pneumonia. Am J Respir Crit Care Med 2003;167(7):A560.
13. Menendez R, Torres A, Zalacain R, et al. Risk factors of treatment failure in community acquired pneumonia: implications for disease outcome. Thorax 2004;59(11):960-965. [Medline]
14. El-Solh AA, Aquilina AT, Dhillon RS, et al. Impact of invasive strategy on management of antimicrobial treatment failure in institutionalized older people with severe pneumonia. Am J Respir Crit Care Med 2002;166(8):1038-1043. [Medline]
15. Ewig S, Ruiz M, Torres A, et al. Pneumonia acquired in the community through drug-resistant Streptococcus pneumoniae. Am J Respir Crit Care Med 1999;159(6):1835-1842. [Medline]
16. Aspa J, Rajas O, de Castro FR. Pneumococcal antimicrobial resistance: therapeutic strategy and management in community-acquired pneumonia. Expert Opin Pharmacother 2008;9(2):229-241. [Medline]
17. Chen DK, McGeer A, de Azavedo JC, et al. Decreased susceptibility of Streptococcus pneumoniae to fluoroquinolones in Canada. Canadian Bacterial Surveillance Network. N Engl J Med 1999;341(4):233-239. [Medline]
18. Kelley MA, Weber DJ, Gilligan P, et al. Breakthrough pneumococcal bacteremia in patients being treated with azithromycin and clarithromycin. Clin Infect Dis 2000;31(4):1008-1011. [Medline]
19. Muder RR, Yu VL, Parry MF. The radiologic manifestations of Legionella pneumonia. Semin Respir Infect 1987;2(4):242-254. [Medline]
20. Menéndez R, Cordero PJ, Santos M, et al. Pulmonary infection with Nocardia species: a report of 10 cases and review. Eur Respir J 1997;10(7):1542-1546. [Medline]
21. Jacobs JA, De Brauwer EI, Ramsay G, et al. Detection of non-infectious conditions mimicking pneumonia in the intensive care setting: usefulness of bronchoalveolar fluid cytology. Respir Med 1999;93(8):571-578. [Medline]
22. Hoogewerf M, Oosterheert JJ, Hak E, et al. Prognostic factors for early clinical failure in patients with severe community-acquired pneumonia. Clin Microbiol Infect 2006;12(11):1097-1104. [Medline]
23. Mortensen EM, Restrepo M, Anzueto A, et al. Effects of guideline-concordant antimicrobial therapy on mortality among patients with community-acquired pneumonia. Am J Med 2004;117(10):726-731. [Medline]
24. Menéndez R, Torres A, Zalacain R, et al. Guidelines for the treatment of community-acquired pneumonia: predictors of adherence and outcome. Am J Respir Crit Care Med 2005;172(6):757-762. [Medline]
25. Kellum JA, Kong L, Fink MP, et al. Understanding the inflammatory cytokine response in pneumonia and sepsis: results of the Genetic and Inflammatory Markers of Sepsis (GenIMS) Study. Arch Intern Med 2007;167(15):1655-1663. [Medline]
26. Menéndez R, Cavalcanti M, Reyes S, et al. Markers of treatment failure in hospitalised community acquired pneumonia. Thorax 2008;63(5):447-452. [Medline]
27. Kruger S, Ewig S, Marre R, et al. Procalcitonin predicts patients at low risk of death from community-acquired pneumonia across all CRB-65 classes. Eur Respir J 2008;31(2):349-355. [Medline]
28. Prat C, Lacoma A, Dominguez J, et al. Midregional pro-atrial natriuretic peptide as a prognostic marker in pneumonia. J Infect 2007;55(5):400-407. [Medline]
29. Kruger S, Papassotiriou J, Marre R, et al. Pro-atrial natriuretic peptide and pro-vasopressin to predict severity and prognosis in community-acquired pneumonia: results from the German competence network CAPNETZ. Intensive Care Med 2007;33(12):2069-2078. [Medline]
30. Waterer GW, Wunderink RG. Genetic susceptibility to pneumonia. Clin Chest Med 2005;26(1):29-38. [Medline]
31. Sanyal S, Smith PR, Saha AC, et al. Initial microbiologic studies did not affect outcome in adults hospitalized with community-acquired pneumonia. Am J Respir Crit Care Med 1999;160(1):346-348. [Medline]
32. Wu CL, Yang D, Wang NY, et al. Quantitative culture of endotracheal aspirates in the diagnosis of ventilator-associated pneumonia in patients with treatment failure. Chest 2002;122(2):662-668. [Medline]
33. Pereira Gomes JC, Pedreira WL Jr, Araujo EM, et al. Impact of BAL in the management of pneumonia with treatment failure: positivity of BAL culture under antibiotic therapy. Chest 2000;118(6):1739-1746. [Medline]
34. De Lassence A, Fleury-Feith J, Escudier E, et al. Alveolar hemorrhage: diagnostic criteria and results in 194 immunocompromised hosts. Am J Respir Crit Care Med 1995;151(1):157-163. [Medline]
35. Franquet T. Imaging of pneumonia: trends and algorithms. Eur Respir J 2001;18(1):196-208. [Medline]
36. Kuru T, Lynch JP III. Nonresolving or slowly resolving pneumonia. Clin Chest Med 1999;20(3):623-651. [Medline]