In this prospective, observational study, we describe 6 of 154 patients in whom Candida spp. appeared to play a contributory etiologic role in CAP. These patients shared clinical and laboratory features that distinguished them from usual CAP patients. (1) All had conditions that suggested a diagnosis of aspiration pneumonia to their admitting physicians. (2) The Charlson Comorbidity Index was substantially higher than is usual for CAP patients. (3) Gram stains of sputum showed large numbers of yeast forms, many within PMNs; (4) Quantitative cultures yielded > 106 CFU Candida per ml sputum. (5) Pseudohyphae were seen in 5 of 6 cases. (6) 1,3-beta-D-glucan was strongly positive in the 3 patients in whom it was tested.

Importantly, a Candida spp. was never detected as a sole infecting organism. Greater than 105 CFU/ml of recognized bacterial pathogens or > 106 CFU/ml of bacteria generally identified as ‘normal respiratory flora’ were also present in every case. Viral PCR was positive in 2 of 3 cases in which it was tested. In fact, 3 patients appeared to be infected with more than one Candida spp. In Case 6, two different Candida spp. were identified in sputum. In Cases 1 and 4, one Candida spp. was grown from blood and another was identified in the sputum. This seeming discrepancy results from the fact that only 1 or 2 colonies are selected for identification by MALDI-TOF. In young cultures, colonies of C. albicans and C. glabrata look alike, and the one growing in the blood might simply have been missed in the sputum. This explanation is supported by the finding of pseudohyphae in sputum from Case 1; the patient was infected with both, but only C. glabrata, which does not make pseudohyphae, was identified in sputum, whereas the blood culture yielded C. albicans. While these results do not prove that Candida spp. alone causes CAP, they suggest that Candida spp. may be a contributory cause of CAP, especially in patients who have a history of chronic aspiration.

Evidence opposing an etioloic role for Candida spp. in CAP has led to the teaching that the finding of Candida spp. in sputum culture simply reflects contamination by oropharyngeal colonization. Candida spp. regularly colonize the upper respiratory tract, with higher rates of colonization in sicker patients. An early study reported the presence of Candida in the sputum in increasing proportions of medical students, hospital employees, and patients, respectively [18]. Sputum cultures of 55% of medical inpatients yielded Candida; quantitative cultures were not done and the quality of the sputum sample was not addressed [18]. Rello et al reported that, in 28 adults with suspected pneumonia and positive sputum cultures for Candida spp., protected brush bronchoscopic specimens yielded > 103 CFU/ml Candida spp. in 24 cases, but most of these were regarded as contaminants [19]. Of 135 autopsies done on patients with evidence of pneumonia, respiratory samples from 77 had been positive for Candida spp., but none had histologic evidence of Candida pneumonia [20]. Conversely, most patients with histologic evidence of Candida pneumonia on lung biopsy did not grow Candida on premortem cultures [21]. In contrast to our work, only one of these studies quantitated Candida, and used a cutoff of 103 CFU/ml, lower than our cutoff of 106 CFU/ml.

In patients with CAP, sputum is the expectorated material that has collected in alveoli, including plasma, PMNs, and microbes. A sample that shows pure, or nearly pure PMNs and large numbers of microbial forms should reflect alveolar exudate. We have used these criteria in the past to show that non-typeable Haemophilus influenzae [13], Moraxella catarrhalis [10, 22], Corynebacterium [16] and other bacteria generally dismissed by microbiology laboratories as ‘normal respiratory flora’ [10] all may cause pneumonia. To our knowledge, no one has previously reported Gram stains and quantitative cultures of Candida in high-quality sputum samples, while requiring large numbers of organisms including intracellular forms and pseudohyphae to be seen microscopically and > 106 CFU/ml to be present. In the present study, the high quality of the sputum samples was shown by the presence of 1.3 × 107 WBC per ml (median value) and the absence of epithelial cells in microscopic fields at 1000 x magnification. We identified intracellular yeast forms within PMNs in all samples and pseudohyphae in 5 samples, further supporting a pathogenic role for Candida.

Some of these approaches have been utilized in intubated patients with suspected ventilator-associated pneumonia (VAP). In one study, the presence of intracellular organisms in at least 2% of cells had a sensitivity of 84% and a specificity of 80% for VAP [23]. Using a cutoff of 5% intracellular organisms, Torres et al, found a positive predictive value of 75% for diagnosing VAP using protected bronchoalveolar lavage [24]. In another study, a cutoff of 7% was 97% specific for diagnosing VAP [25]. We observed intracellular yeast in every sputum sample.

A limitation of the current study is the absence of a diagnosis of Candida pneumonia by lung biopsy. Without histologic evidence of invasion of yeast into lung parenchyma, Candida pneumonia cannot be diagnosed with certainty. However, most etiologic diagnoses of pneumonia are presumptive, unless organisms are also grown from a normally sterile site. Additionally, not all patients had the same evaluation including 1,3-beta-D-glucan, procalcitonin, and viral PCR. Treatment decisions were made by managing clinicians; all patients received both antibacterial and antifungal agents, and care was withdrawn in 2 cases, making it impossible to determine which treatments were beneficial.

The present study provides evidence suggesting that, in patients who have risk factors for chronic aspiration Candida spp. plays a contributory role in the etiology of CAP. We found that patients with CAP who had large numbers of Candida in a high-quality sputum sample all had a history consistent with aspiration, and in each case the infection was polymicrobial. These findings support the principle that a sufficient inoculum of organisms of low virulence into the lower respiratory tract may suffice to cause pneumonia.

The subject of yeast-bacteria interaction has been extensively studied. Roux et al have shown in rats that infection with Candida facilitates bacterial infection by interfering with the function of alveolar macrophages [6, 8], and Neely et al [26] showed that bacterial colonization of burn wounds rendered Candida more invasive [7]. In critically ill patients, the presence of Candida in sputum cultures has been associated with increased risk of VAP due to Pseudomonas aeruginosa [27].

In conclusion, by examining sputum samples submitted for Gram stain and culture, we found that, of 154 high-quality specimens (> 20 WBC per epithelial cell) from patients who were hospitalized with a diagnosis of CAP, 6 contained large numbers of intra- and extracellular yeast forms. Admitting physicians, unaware of microscopic findings, diagnosed chronic aspiration in all cases. Quantitative cultures revealed > 106 CFU Candida spp. per ml. Sputum from 5 samples showed pseudohyphae. One patient had candidemia, and in 3 whose serum was assayed for 1,3-beta-D-glucan, the level was markedly elevated. In every case bacterial coinfection was present and 2 patients had viral coinfection. These findings suggest that Candida spp. may play a contributory role in the etiology of CAP in patients who have a history of chronic aspiration. Future studies are needed to address whether patients with this constellation of findings benefit from antifungal therapy.