Candida in the Respiratory Tract Potentially Triggers Galactomannan Positivity in Nonhematological Patients

BAL fluid samples from critically ill patients shared a rate of 29% false-positive galactomannan results. We aimed to determine whether Candida species abundance in BAL fluid causes galactomannan (GM) positivity.

ABSTRACT

BAL fluid samples from critically ill patients shared a rate of 29% false-positive galactomannan results. We aimed to determine whether Candida species abundance in BAL fluid causes galactomannan (GM) positivity. A total of 89 Candida culture-positive BAL fluid samples from patients without suspicion of invasive aspergillosis (IA) were analyzed. GM results were correlated with Candida species abundance, Candida species quantity, and patient data. Candida species quantities of ≥10⁴/ml and Candida glabrata abundance were significantly associated with positive GM results. The added diagnostic value of GM in BAL fluid for diagnosing IA in critically ill patients is limited.

INTRODUCTION

Diagnosis of invasive aspergillosis (IA) remains challenging in the critical care setting. Established diagnostic criteria of IA, based on the presence of host factors and clinical and mycological criteria, are primarily proposed for severely immunocompromised patients (1). In critically ill patients, characteristic radiological signs suggestive for IA are usually absent, and heterogeneous underlying diseases lacking apparent immunodeficiency hinder a profound diagnosis of IA according to the established and recommended diagnostic criteria (2, 3). The galactomannan (GM) assay was proposed to be included in the intensive care unit (ICU) setting as an additional diagnostic criterion with the aim to increase accuracy for diagnosing IA (4). However, GM testing in BAL fluid is prone to falsely positive (38%) among nonneutropenic patients (5).

In daily clinical routine diagnostics, we observed that BAL fluid samples from critically ill patients had high GM values in the absence of a clinical suspicion of IA, but high quantities of Candida species were cultured from the BAL fluid samples of affected patients (unpublished data). This practical observation encouraged us to investigate the correlation between Candida culture-positive (CCP) BAL fluid and GM positivity. The study was restricted to nonhematological patients did not have signs, symptoms, or other hints suggesting IA.

A set of 268 randomly chosen BAL fluid samples from nonhematological patients from Tyrolean hospitals (University Hospital of Innsbruck, the associated secondary hospital of Natters, and the secondary Tyrolean hospitals Kufstein, Reutte, Schwaz, and Zams) without clinical suspicion of IA was investigated during a study period of 8 months. In all, 98 CCP and 170 Candida culture-negative (CCN) BAL fluid samples were gained from 69 nonhematological patients and 143 unselected controls, respectively. The exclusion criterion was cultural evidence of molds within 6 months before or after sample collection (Aspergillus spp. in 9 BAL fluid samples from 6 patients), resulting in 89 CCP BAL fluid samples from 63 patients and 170 CCN controls. Fungal load was determined by CFU counting; BAL fluid samples were concentrated (2,000 rpm for 10 min), and 100 μl was spread on Sabouraud agar and incubated for 24 h at 37°C. Species identification was performed by subcultivation on Candida CHROMagar (chromID Ref 43631; bioMérieux) and matrix-assisted laser desorption ionization–time of flight mass spectrometry (Microflex, Bruker Daltonics, MBT IVD library DB-6763). The GM assay was performed in duplicate according to the manufacturer’s instructions (Platelia Aspergillus EIA, Bio-Rad, France). A GM value of ≥0.5 was defined as positive. Patient- and sample-specific data (age, sex, date of sampling, CFU count, Candida species, and GM values) were entered into an Excel database. Descriptive and univariate analyses were performed using EpiInfo software, version 7.2.1.0 (Centers for Disease Control and Prevention). Odds ratios were calculated, and a 95% confidence interval was applied to test for statistical significance. A cutoff value of </≥10⁴/ml was used to categorize CFU.

GM results referring to demographic data, Candida species, and CFU count are summarized in Table 1 and Fig. S1 in the supplemental material. Of BAL fluid samples, 78% (49/63) contained one Candida species, 21% (13/63) contained two species, and one showed three species. Candida albicans was recovered most frequently (65%), followed by Candida glabrata (27%), Candida tropicalis (13%), and Issatchenkia orientalis (formerly, Candida krusei; 6%). Other, rare species identified comprised Saccharomyces cerevisiae (n = 2), Candida inconspicua (n = 2), Kluyveromyces marxianus (formerly, Candida kefyr; n = 2), Wickerhamomyces anomalus (formerly, Candida pelliculosa; n = 1), and Candida dubliniensis (n = 1). The median CFU count encountered was 0.2 · 10⁴/ml. With the GM positivity cutoff defined as 0.5, 29% (26/89) of the BAL fluid samples tested were valued as positive, with a median GM index of 2.4 (0.65 to 11.4). Only 1 GM value was observed within a higher cutoff between >0.5 and ≤1.0 (0.65). A CFU count of ≥10⁴/ml was significantly associated with positive GM results (odds ratio [OR], 4.3; 95% confidence interval [CI], 1.57 to 11.47). Among all Candida species, C. glabrata was the only one significantly associated with positive GM values (OR, 3.6; 95% CI, 1.35 to 9.85). Twelve CCN BAL fluid samples (7%) revealed positive GM results, with a median GM index of 1.89 (0.8 to 6.5). GM positivity was significantly associated with CCP BAL fluids (OR, 5.43; 95% CI, 2.58 to 11.43). GM positivity, shown in 29% of tested BAL fluid samples, was associated with a high CFU count (≥10⁴/ml) of Candida spp. and in particular with an abundance of C. glabrata. Whereas yeast suspension series were reported to remain GM negative (6), positive GM values of >0.5 of BAL fluid samples were linked to microscopic evidence of pseudomycelia-forming C. albicans (7). Inconsistencies in positive GM results among other pathogens, including Acremonium spp., Alternaria spp., Fusarium spp., Wangiella spp., Geotrichum spp., and Cryptococcus spp., were described (6, 8, 9). Variable cross-reactivity of non-aspergilli isolates with the GM assay was attributed to variable presence of exoantigens among species or even strains (6). The cell wall of C. glabrata was shown to contain 50% more protein than the cell wall of S. cerevisiae and C. albicans, resulting in a significantly larger amount of mannan (10–12). Subsequently, the higher mannose-to-glucose ratio in C. glabrata might enhance cross-reactivity with the GM assay. However, exoantigens causing GM positivity do not seem to appear commonly, if at all, in vitro. Numerous factors of complex interactions between host, pathogens (i.e., fungal species), and resident microbiota within the lung are not easy to mimic in vitro, which raises a wide array of possibilities leading to Candida-associated GM positivity in vivo.

TABLE 1.

Overview of demographic data and GM results referring to CFU count and Candida species

Variable	Candida positivea (n [%])		Odds ratio (95% CI)b	Candida negative (n [%]) (control group)a
	GM+	GM−		GM+	GM−
Mean age (yr) (range)	65 (18–96)			62 (16–90)
Male	18/64 (28)	46/64 (72)	0.53 (0.31–0.92)	8/98 (8)	90/98 (92)
Female	8/25 (32)	17/25 (68)	Ref	4/72 (6)	68/72 (94)
ICU	22/69 (32)	47/69 (68)	6.33 (3.51–11.4)	4/60 (10)	56/60 (93)
Non-ICU	4/20 (20)	16/20 (80)	Ref	8/110 (7)	102/110 (93)
University Hospital of Innsbruck	18/72 (25)	54/72 (75)		8/142 (6)	134/142 (94)
Surgery	5/23 (22)	18/23 (78)		1/12 (8)	11/12 (92)
Pulmonology	4/14 (29)	10/14 (71)		5/99 (5)	94/99 (95)
Traumatology	4/17 (24)	13/17 (76)		0/10 (0)	10/10 (100)
Internal medicine	4/8 (50)	4/8 (50)		2/8 (25)	6/8 (75)
Gastroenterology	0/4 (0)	4/4 (100)
Neurosurgery	1/3 (33)	2/3 (67)
Oncology	0/2 (0)	2/2 (100)
Pediatrics	0/1 (0)	1/1 (100)		0/2 (0)	2/2 (100)
Transplant				0/8 (0)	8/8 (100)
Secondary hospitals	8/17 (47)	9/17 (53)		4/28 (14)	24/28 (86)
Candida sp.	26/89 (29)	63/89 (71)	5.43 (2.58–11.43)
Candida negative	12/170 (10)	158/170 (90)	Ref
C. albicans	19/58 (33)	39/58 (67)	1.67 (0.61–4.56)
C. non-albicans			Ref
C. glabrata	12/24 (50)	12/24 (50)	3.64 (1.35–9.85)
C. non-glabrata			Ref
C. tropicalis	2/12 (17)	10/12 (83)	0.44 (0.09–2.17)
C. non-tropicalis			Ref
C. krusei	1/5 (20)	4/5 (80)	0.59 (0.06–5.55)
C. non-krusei			Ref
Other species	2/8 (25)	6/8 (75)	0.79 (0.15–4.21)
C. albicans, C. glabrata, C. tropicalis, C. krusei			Ref
CFU ≥104/ml	13/26 (50)		4.25 (1.57–11.47)
CFU <104/ml	12/63 (19)		Ref

^aGM+, galactomannan positive (≥0.5); GM−, galactomannan negative (<0.5).

^bRef, reference. Values in boldface are statistically significant on a confidence interval of 95%.

Non-albicans Candida species, especially C. glabrata, are known to occur more frequently in immunocompromised patients (13). In our study, GM positivity was particularly associated with C. glabrata; in contrast, other non-albicans yeasts such as I. orientalis, C. tropicalis, and rare yeast species were negatively associated with GM positivity. The use of a higher GM cutoff value for nonseverely immunosuppressed patients was consistently proposed to enhance specificity of the assay (14–16). Specificity was shown to be increased from 62.4% up to 89.2% by raising the cutoff index up to 1.19 (5). Interestingly, in our test series, only 1 of 26 GM values was within the range up to an index of 1. Thus, increasing the GM cutoff up to 1.0 would have had little impact on GM positivity results in our test series.

We tested batches of antimicrobials to rule out false positives by antimicrobials, which all remained GM negative except for liposomal amphotericin B (see Table S1 in the supplemental material). In addition, we tried to confirm our hypothesis in vitro, evaluating the GM positivity of Candida species exposed to NaCl 0.9% and RPMI 1640. Due to the variability and complexity of microbiota and drugs present in the respiratory tract of severely ill patients, the in vivo effects could be only partially reproduced. Only 3 out of 250 Candida samples (1.2%) tested positive for GM (see Table S2 in the supplemental material). Positive GM results were not reproducibly associated with dilution solvent, Candida species, CFU count, or incubation period.

The results of our study may be limited by the selection of our controls. This is mainly attributed to the high rate of Candida isolation from respiratory samples of ICU patients (79%) and the known unbalanced sex distribution toward male patients in the ICU (17). Moreover, numerous factors that are known to cause false-positive results of the serum GM assay, including underlying diseases (e.g., multiple myeloma), hydration, and blood product conditioning fluids, as well as nutritional supplements and parenteral nutrition (18–22), were not considered in our study and may partially contribute to GM positivity. However, positive GM results were not significantly associated with hospitalization (University Hospital of Innsbruck versus secondary hospitals) or treatment care (ICU versus non-ICU), indicating heterogeneous underlying diseases and treatment regimens within our study population, making a common factor of non-fungal-associated false positivity improbable.

Summarizing our results, we proved a statistical association between the abundance of Candida spp. and positive GM values in BAL fluid samples from nonhematological patients. Due to the frequent abundance of Candida spp. in respiratory samples of critically ill patients, the value of GM positivity of BAL fluids in the absence of other clinical, radiological, and microbiological criteria for IA might be limited.