Candida Pneumonia— Does it Exist?


Synopsis: The recovery of Candida from the respiratory secretions of non-neutropenic mechanically ventilated patients is seldom indicative of invasive pulmonary infection with this organism.

Source: el-Ebiary M, et al. Significance of the isolation of Candida species from respiratory samples in critically ill, non-neutropenic patients. Am J Respir Crit Care Med 1997; 156:583-590.

El-ebiary and colleagues in barcelona intensively studied the role of Candida species in producing parenchymal lung disease in 25 patients who had been mechanically ventilated for at least 72 hours prior to death. Patients with immunosuppression, hematologic malignancy, or neutropenia were excluded. Twenty-one (84%) of the 25 had chest radiographic abnormalities present within 24 hours prior to death—seven with diffuse and 14 with localized infiltrates. Only eight of the patients had not received antibiotic therapy in the 24 hours before death.

Mechanical ventilation was continued for up to 90 minutes after death, during which time respiratory samples were obtained by: 1) endotracheal aspiration (EA); 2) protected brush specimens (PBS) from area of maximal radiographic abnormalities via bilateral bronchoscopies; 3) bronchoalveolar lavage (BAL); 4) bilateral open lung biopsies guided to the area of maximal pulmonary infiltrates by simultaneous bronchoscopy; and 5) bilateral, unguided open lung biopsies.

Candida was recovered from a post mortem pulmonary biopsy specimen from 10 (40%) of the patients, but only one (4%) had histologic evidence of Candida pneumonia. Another had had a positive antemortem pleural fluid culture but no post-mortem histologic evidence of pneumonia. The patient with pneumonia had been treated empirically with fluconazole, but his infecting organism proved to be C. krusei. Multivariate analysis examining seven variables such as antibiotic therapy, duration of mechanical ventilation, and receipt of parenteral nutrition failed to indicate that any was an independent risk factor for having a Candida present in respiratory samples. All 10 also had bacteria recovered from at least one respiratory sample.

Using the presence of histologically confirmed pulmonary or pleural candidiasis as a gold standard, the sensitivities of EA, PBS, BAL, guided lung biopsy, and blind lung biopsy were, respectively, 100%, 50%, 50%, 75%, and 100%, while their respective specificities were only 67%, 55%, 70%, 33%, and 20%. Quantitation of culture results was not helpful.


Infectious disease specialists are often asked to interpret the significance of the recovery of Candida from respiratory secretions in critically ill patients. My usual response is that, while detection of this organism may be a clue to the presence of invasive infection elsewhere, that pulmonary parenchymal infection, at least in non-neutropenic patients, is seldom present.

This study, performed in mechanically ventilated patients, generally confirms this assessment. Thus, of 25 non-neutropenic patients who died after at least three days of mechanical ventilation, only one had postmortem biopsy evidence of pulmonary parenchymal candidiasis, while another had antemortem evidence of pleural space infection at postmortem. Since the latter was known prior to the study to have pleural infection as well as a positive blood culture, his inclusion in the analysis is somewhat questionable. Thus, in a situation in which the appropriate interpretation of, for example, an EA culture yielding Candida is not obvious, the likelihood of true pneumonia due to this organism is quite low.

It is possible, however, that repeated isolation of Candida from respiratory secretions may be an indication of disseminated candidiasis. Unfortunately, other organs were not examined in this study. Only one patient, however, had had a positive blood culture.

The only means of definitively determining the presence of Candida pneumonia, whether caused by aspiration or hematogenous dissemination, is by the histological demonstration of the organism in pulmonary parenchymal tissue obtained by biopsy. Biopsy is seldom, if ever, indicated for this purpose alone, however.

Since the recovery of Candida from respiratory secretions may be indicative of the possible presence of disseminated candidiasis, this finding must be examined in conjunction with additional clinical and laboratory factors. These include whether the patient is septic, the spectrum and duration of antibacterial therapy, the degree of immunocompromise, and, importantly, the recovery of Candida from additional sites, such as urine. Combining this information with a careful mucocutaneous and fundoscopic examination should allow the clinician a reasonable chance of making a correct decision concerning the empirical institution of antifungal therapy.