Effect of involved Aspergillus species on galactomannan in bronchoalveolar lavage of patients with invasive aspergillosis

Abstract

Purpose

The detection of galactomannan (GM) in bronchoalveolar lavage (BAL) fluid is an important surrogate marker for the early diagnosis and therapeutic monitoring of invasive aspergillosis (IA), regardless of the involved species of Aspergillus. Here, we utilized the Platelia Aspergillus GM enzyme immunoassay (Bio-Rad) to evaluate the GM index in BAL fluid samples from patients with proven, probable or putative IA due to Aspergillusflavus versus Aspergillusfumigatus.

Methodology

In a prospective study between 2009 and 2015, 116 BAL samples were collected from suspected IA patients referred to two university hospitals in Tehran, Iran.

Key findings

According to European Organization for Research and Treatment of Cancer and Mycoses Study Group and Blot criteria, 35 patients were classified as IA patients, of which 33 cases tested positive for GM above 0.5 and, among these patients, 22 had a GM index ≥1. Twenty-eight were culture positive for A. flavus and seven for A. fumigatus. The GM index for A. flavus cases was between 0.5–6.5 and those of A. fumigatus ranged from 1 to 6.5. The sensitivity and specificity of a GM index ≥0.5 in cases with A. flavus were 86 and 88 % and for A. fumigatus patients were 100 and 73 %, respectively.

Conclusion

Overall, the mean GM index in patients with A. fumigatus (3.1) was significantly higher than those of A. flavus (1.6; P-value=0.031) and the sensitivity of GM lower for A. flavus when compared to A. fumigatus. This finding has implications for diagnosis in hospitals and countries with a high proportion of A. flavus infections.

Abbreviations

AUC, area under the receiver operating characteristics curve; BAL, bronchoalveolar lavage; EIA, enzyme immunoassay; EORTC/MSG, European Organization for Research and Treatment of Cancer and Mycoses Study Group; GM, galactomannan; IA, invasive aspergillosis; NPV, negative predictive value; PPV, positive predictive value; ROC, receiver operating characteristics curve.

Introduction

Invasive aspergillosis (IA) is one of the most significant and challenging invasive fungal infections, and is a major cause of morbidity and mortality in high-risk patients [1, 2]. This disease occurs mainly in patients with a markedly impaired immune system. After A. fumigatus, A. flavus is the second leading cause of invasive and non-invasive aspergillosis [3, 4]. In many tropical and sub-tropical countries, including India, Saudi Arabia, Pakistan, Iraq, Turkey and Iran, A. flavus has most frequently been reported as the predominant etiological agent causing mainly cerebral or sino-orbital aspergillosis, or ocular infections [5, 6]. Early diagnosis of IA remains a challenge, but it is important for better therapeutic outcomes [7]. Galactomannan (GM) antigen, a polysaccharide secreted from the Aspergillus spp., correlates with the growth phase of the fungus [8–11]. Aspergillus GM antigen detection in serum and bronchoalveolar lavage (BAL) fluid by the Bio-Rad Platelia sandwich enzyme immunoassay (EIA, Bio-Rad) has been studied extensively [12, 13]. Several previous studies have also evaluated the GM assay in BAL samples in various patient groups and conditions; they reported a higher parallel sensitivity and specificity in comparison to serum specimens [14–16]. According to the revised European Organization for Research and Treatment of Cancer and Mycoses Study Group (EORTC/MSG) diagnostic criteria, a positive BAL GM test is a mycological criterion of IA in association with host and clinical or radiological findings [17]. The GM assay has gained widespread acceptance as a sensitive method for the early diagnosis and therapeutic monitoring of IA by applying it directly to specimens from the target organ [18, 19]. In the present study, we evaluated the GM index in BAL samples from patients with proven, probable or putative IA due to A. flavus versus A. fumigatus using Platelia Aspergillus EIA.

Methods

Patient

In a prospective study, 116 BAL fluid samples were collected from patients with pulmonary disorders who were referred to the bronchoscopy section of the Shariyati and Imam-Khomeini Hospitals, Tehran University of Medical Sciences, Iran, from June 2009 to October 2015. Written informed consent was taken from all participants. We included all data on suspected patients of IA (host factors, clinical features and mycological data) as defined by EORTC/MSG [17]. In addition, we used the clinical algorithm suggested by Blot et al. [20], which includes the endotracheal aspirate culture to demonstrate proven or putative IA in the existence of matchable symptoms, unusual thoracic medical imaging and either host factors or BAL fluid positive for Aspergillus on direct examination and culture.

BAL sampling

Fiberoptic bronchoscopy (Olympus BF20D) with BAL was performed if feasible. The sampling area was selected based on the infiltrate location on the chest radiograph. Then, 50 ml sterile normal saline was injected through the device. The suction channel of the bronchoscope was used to aspirate 25–30 ml fluid, yielding clinical samples, which were collected into sterile tubes and immediately transferred to the Medical Mycology Laboratory at Mazandaran University of Medical Sciences.

Processing of BAL fluid specimens

All BAL samples were centrifuged for 5 min at 5000 r.p.m. and then the supernatants were stored at −80 °C. The remaining sediment was used for direct microscopic examination and culture. The smear from each sample was mounted with 20 % potassium hydroxide (KOH) and Calcofluor White to detect the hyphal elements. Sabouraud dextrose agar supplemented with chloramphenicol (40 µg ml⁻¹) was used for the isolation of fungi. The plates were incubated at 30 °C for 5 days. Aspergillus species were identified by morphological characteristics and molecular sequencing of beta tubulin [21].

GM Platelia Aspergillus assay

The Platelia Aspergillus GM EIA (Bio-Rad Laboratories, Marnes-la-Coquette, France) was used to detect GM on centrifuged BAL fluid specimens, according to the manufacturer's manual, using 300 µl of the BAL sample. Positive and negative control ELISA kits were included in each test. The results were recorded as GM indexes ≥0.5 and ≥1.

Data analysis

The mean age, standard deviation, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), receiver operating characteristic curve (ROC) and interactive dot diagram were calculated for each value using MedCalc software version 14.8. Positive and negative likelihood ratios were computed by an online website (http://araw.mede.uic.edu/cgi-bin/testcalc.pl). The t-test was applied to compare the GM index of A. flavus and A. fumigatus among positive IA patients. The estimated prevalence of invasive pulmonary aspergillosis (IPA) was determined by the following formula: (proven IPA+probable IPA)/(proven IPA+probable IPA+patients without IPA).

Results

The mean age of the 116 studied patients was 46±15.4 ranging from 12 to 78 years. Seventy-three (62.9 %) patients were male. Table 1 shows clinical characteristics of all studied patients in the current study. Of the 116 BAL samples, 26/116 (22.4 %) and 38/116 (32.7 %) were positive by direct microscopy and culture, respectively. Out of 116 patients, 35 cases were classified as IA. According to EORTC/MSG criteria [17], two cases were classified as proven IA and 16 cases as probable IA. In addition, using the clinical algorithm suggested by Blot et al. [20], 17 cases were categorized as putative IA (Table S1, available with the online Supplementary Material). Three cases were also defined as possible IA based on the underlying host factor and radiological features, however these cases were excluded from the final analysis since mycological criteria were lacking.

Table 1. Clinical characteristics of the 116 patients studied.

Characteristics	All patients	Patients with IA			Patients without IA
No. of patients	116	35			81
Men/women	73/43	23/12			50/31
Age, mean±sd, year	46±15.4	44.7±16.2			46.6±15.2

IA classification	No. of patient	EORTC/MSG criteria		Blot et al. criteria	Non-IA
Underlying disease		No. of proven IA	No. of probable IA	No. of putative IA	No. of non-IA
Haematological malignancies	29	0	10	0	19
COPD	16	0	0	5	11
Lung cancer	10	0	5	1	4
Pulmonary and respiratory disorders	21	0	1	2	18
Organ transplantation	16	2	0	0	14
Autoimmune diseases	3	0	0	2	1
Renal failure	4	0	0	2	2
Tuberculosis with cavity	3	0	0	3	0
Breast cancer	2	0	0	1	1
ICU patients	6	0	0	0	6
CGD	2	0	0	1	1
Diabetes mellitus	4	0	0	0	4
Total cases	116	2	16	17	81

IA, invasive aspergillosis; No., number; sd, standard deviation; COPD, chronic obstructive pulmonary disease; ICU, intensive care unit; CGD, chronic granulomatous disease.

Using polymerase chain reaction (PCR)-sequencing targeting the beta-tubulin region of rDNA, A. flavus (28, 80 %) and A. fumigatus (7, 20 %) were identified. Out of 35 patients with IA, 10 (28.6 %) had underlying haematologic malignancy. Seventeen patients had evidence of invasive airway infection with nodules, ulceration, pseudomembrane and mass obstruction of the airway; therefore invasive Aspergillus tracheobronchitis is an alternative or additional disease classification to IPA for these patients. There were three patients with cavities and prior tuberculosis or a lung injury from chemical warfare who may be better classified as having chronic pulmonary aspergillosis (CPA), but Aspergillus IgG detection was not done to confirm the diagnosis. Five patients with lung cancer probably had IA, although this is not one of the host groups in the EORTC/MSG criteria. Likewise, there were five patients with COPD, two of whom had Aspergillus tracheobronchitis, in whom the precise and complete Aspergillus diagnosis was unclear.

At a GM index ≥0.5, of 116 patients, 33 (28.4 %) with IPA and five (4.3 %) cases without IPA had positive results. However, at a GM index ≥1, of these studied patients, 22 (19.0 %) cases had positive results. Ten patients with IA had a GM index ≥2, including two cases with proven IA. In IA patients with positive GM, the GM index ranged from 0.5 to 6.5. In positive GM cases with a positive BAL fluid culture, the mean GM index (1.9) was higher than those with negative cases with a negative culture (0.34). The mean of the GM index in IA patients with A. fumigatus (3.1) was significantly higher than those of A. flavus (1.6; P=0.031).

Table 2 shows the sensitivity and specificity, positive and negative predictive values of GM detection in all BAL samples for GM indexes ≥0.5 and ≥1. The sensitivity and specificity of a GM index ≥0.5 in cases with A. flavus were 86 and 88 % and for A. fumigatus patients were 100 and 73 %, respectively.

Table 2. Sensitivity, specificity and predictive values for two cut-off GM indexes ≥0.5 and ≥1 for 116 collected BAL samples.

Criteria	GM index	Sensitivity	Specificity	PPV	NPV	95 % CI	AUC value	Cost value*	P-value
A. flavus	≥0.5	86	88	52	97	91.0–99.5	0.89	0.137	P<0.05
	≥1	57	89	47	92	94.2–96.3	0.82	0.157	P<0.05
A. fumigatus	≥0.5	100	73	40	100	95.1–100	0.94	0.226	P<0.05
	≥1	86	88	56	97	91–99.5	0.92	0.123	P<0.05
Total	≥0.5	91	82	47	98	92.1–99.8	0.91	0.166	P<0.05
	≥1	57	86	44	92	94.5–96.6	0.86	0.172	P<0.05

PPV, positive predictive value; NPV, negative predictive value; CI, confidence interval; AUC, area under the ROC curve.

*The average cost resulting from the use of the diagnostic test at that decision level. Note that the cost reported here excludes the ‘overhead cost’, i.e. the cost of doing the test, which is constant at all decision levels.

The area under the ROC curve (AUC) for a GM index ≥0.5 in all BAL samples was 0.91 (95 % CI, 92.1–99.8, P<0.0001) while at a GM index ≥1 it was 0.86 (95 % CI, 94.5–96.6, P<0.0001). On the other hand, the AUC using the standard cut-off GM index ≥0.5 for A. flavus culture-positive BAL samples was 0.89 (95 % CI, 91.0–99.5, P<0.0001) and for A. fumigatus it was 0.94 (95 % CI, 95.1–100, P<0.0001).

Fig. 1 shows the GM index based on culture results in four different groups of patients, including all culture-negative BAL samples, A. flavus culture-positive BAL samples, A. fumigatus culture-positive BAL samples and other Aspergillus species culture-positive BAL samples.

Fig. 1.

Log₁₀ transformation of GM index based on BAL fluid culture results in all patients. (a) Culture-negative BAL samples. (b) A. flavus culture-positive BAL samples. (c) A. fumigatus culture-positive BAL samples. (d) Other Aspergillus species culture-positive BAL samples.

According to our results on the prevalence of IA, the positive and negative likelihood ratios were calculated as 15 and 0.06, respectively (Fig. S1). The value cost of 0.166 at a GM index ≥0.5 was the best cut-off point for diagnosis of IA, while this value at a GM index ≥1 was 0.172.

Discussion

In clinical practice, GM index evaluation in BAL has been encouraged as a suitable assessment for early diagnosis of IA in patients with various underlying disorders [22–25].

Several studies reported the value of GM antigen detection in BAL fluid as a promising tool for diagnosis of IA due to A. fumigatus [26–29]. Two GM indexes ≥0.5 [16, 30, 31] and ≥1.0 [7, 28, 32] for BAL fluid have been proposed for the assessment of GM in diagnosis of IA due to A. fumigatus; however, to the best of our knowledge, data on A. flavus is scarce.

In the present study, we therefore compared the Platelia GM index in BAL fluid from patients with proven, probable or putative pulmonary aspergillosis due to A. flavus versus A. fumigatus. Importantly, we found an increased index of GM in seven cases with A. fumigatus with critical host factors in comparison to patients with A. flavus. A GM index of ≥0.5 has shown reliable sensitivity and specificity for all tests. On the other hand, applying a GM index ≥1, 11 cases with positive A. flavus in BAL fluid culture had negative GM test results. Therefore, it seems that a GM index ≥0.5 would be an appropriate option for early diagnosis of IA, especially due to A. flavus and subsequently to proceed the rapid antifungal therapy in patients at high risk for IA. The ROC curve analysis indicated AUC values of 0.91 (95 % CI, 92.1–99.8) for GM testing using a cut-off ≥0.5, which is in agreement with the previous report by He et al. [33] who reported a similar ROC value for the GM index in BAL fluid (AUC=0.91; 95 % CI).

In the current study, the overall sensitivity and specificity of the GM index considering a cut-off value of ≥0.5 were 91 and 82 %, respectively. The sensitivity results were in line with previous reports of Maertens et al. [7], D’Haese et al. [28] and Luaong et al. [34], and specificity was similar to the findings of Maertens et al. [7], Desai et al. [35] and Clancy et al. [36].

However, in contrast to our results, [32], Luong et al. [37] and Reinwald et al. [38] reported sensitivities of 100, 100, 67 % and specificities of 73, 78, 96 %, respectively [32, 37, 38]. In our study, the overall sensitivity and specificity were 57 and 86 % for a GM index of ≥1.0, respectively, which is somewhat different from the previous studies by Husain et al. [39], Musher et al. [40] and Khodavaysi et al. [41]. These differences might be due to different underlying conditions of the IA patients tested, various sizes of investigated populations, antifungal treatment, the different guidelines or diagnostic criteria for definition of proven or probable IA, the applied GM index as the diagnostic criteria and the species of Aspergillus involved in IA.

Notably, we also found that five of the non-IA patients who were receiving beta-lactam antibiotics, had a GM index ≥0.5, which is comparable with the Meersseman et al. [42] study which reported four of 33 intensive care unit patients with a false-positive GM index in BAL samples. The mechanism of beta-lactam giving a false positive BAL GM test has not been well determined, however certain domains of the β-lactam drugs show cross-reactivity to Aspergillus GM molecules at different concentrations. The interfering molecules in serum probably diffuse into the epithelial lining fluid to exceed the cut-off threshold, leading to a false positive BAL GM test [43, 44].

In our study, the high frequency of positive GM tests in IA patients due to A. flavus is similar to results from Tunisia [45] and India [46]. However, the GM result in BAL fluid for A. flavus in our study (28/116, 24.1 %) is not comparable to the results from The Netherlands (3.44 %) [47], Italy (4.5 %) [48], USA (0.8 and 1.2 %) [36, 39], Belgium (2.3 %) [7], France (2 %) [49], UK (0.6 %) [50], China (4 %) [33], Japan (0.6 %) [22], Czech and Slovak Republics (4.2 %) [51], Turkey (3.3 %) [52] and India (11.6 %) [46]. To the best of our knowledge, none of the previous studies reported a twofold mean of the GM index in IA patients with A. fumigatus in comparison to IA patients with A. flavus. Xavier et al. [53] in an in vitro study found that the quantity of scattered GM varies according to the species of Aspergillus. They demonstrated that among different species of Aspergillus, Aspergillus terreus has the maximum GM index, closely followed by A. fumigatus in comparison to Aspergillus niger and A. flavus [53]. Another study showed a higher secretion of GM by A. niger and A. flavus in comparison to A. fumigatus [54]. In general, the reasons for differences in GM indexes between different strains of Aspergillus are not yet well known. However, the higher amount of in vivo GM released by A. fumigatus strains, in comparison to A. flavus, may be due to a higher germination rate (logarithmic phase) at 37 °C by A. fumigatus. On the other hand, the high β-d-galactofuranosidase produced by A. fumigatus might respond more avidly in the GM assay than A. flavus [55]. Further study is needed to clarify the reason for the differences in GM indexes in IA patients with different species of Aspergillus.

Although A. fumigatus is reported as the most prevalent etiologic agent of IA and CPA compared with other Aspergillus species [56, 57], in the present study A. flavus was the most common isolated species among Aspergillus spp. In Table 3 we also summarized the reported etiologic agents of IA in different countries and highlighted the GM index and mycological criteria used. Of note, A. flavus has been frequently reported as the leading cause of invasive aspergillosis in certain tropical and sub-tropical countries [58]. A limitation of the current study is that we did not study invasive rhinosinusitis, which is more often caused by A. flavus than other species.

Table 3. Summarized data on IA in different underlying patients due to A. flavus and A. fumigatus in different studies from 2003 to 2017.

Country	Host factors	Total number of patients	No. positive A. flavus/ A. fumigatus	Kit used (laboratory)	Cut-off Gm index value	Sn (%)	Estimated prevalence IA (%)	Reference
The Netherlands	Haematological malignancies	29	1/42	Sanofi	1.0	100	28	[47]
Italy	Haematology patients	44	2/4	Bio-Rad	nd	nd	45	[48]
USA	Solid-organ transplant recipients	81	1/4	Bio-Rad	1	100	6.2	[36]
USA	Solid organ transplants	116	1/3	Bio-Rad	0.5	67	5	[39]
India	Bronchogenic carcinoma	69	8/10	Bio-Rad	1	100	41	[46]
Belgium	Haematological malignancies	128	3/23	Bio-Rad	ns	97	45	[7]
France	Haematologic malignancies	101	2/9	Bio-Rad	0.5	57.6	33	[49]
Italy	High-risk patients (haematology and intensive care unit)	158	1/6	Bio-Rad	1	94.1	11	[50]
China	Chronic obstructive pulmonary disease	50	2/6	Sanofi	0.5	88.9	18	[33]
Japan	Chronic pulmonary aspergillosis	144	1/11	nd	0.4	77.2	12.5	[22]
Czech and Slovak	Haematological malignancies	176	1/19	nd	0.5	nd	6.8	[51]
Turkey	Neutropenic and non-neutropenic patients	839	28/80	Bio-Rad	1	nd	nd	[52]
Iran	Patients with various underlying disorders	116	28/7	BioRad	0.5	91	30	Present study

nd, not-determined; ns, not-specified; Sn, sensitivity.

Overall, our findings emphasized that a GM index ≥0.5 is the best cut-off value for screening patients suspected of IA. On the other hand, we observed that using a GM index ≥0.5 in patients with positive BAL fluid culture with A. flavus can help to avoid missing IA cases. In a patient group selected for a high likelihood of IPA, the sensitivity of GM was lower for A. flavus compared with A. fumigatus, which has implications for diagnosis in hospitals and countries with a high proportion of A. flavus infections.