Liquid-based cytopathology test as a novel method to identify Aspergillus in patients with pulmonary aspergillosis

Conventional cytopathology examination of respiratory samples can aid in identifying Aspergillus but with poor sensitivity, so this study aimed to assess the potential of the liquid-based cytopathology test (LCT) for improving the identification of Aspergillus in respiratory samples following Papanicolaou’s or Special staining with Grocott’s methenamine silver or periodic acid-Schiff staining. Paired bronchial brushing samples (n = 54) and sputum samples (n = 117) from 171 patients with pulmonary aspergillosis were prepared as slides using either conventional cytopathology or SurePath LCT. LCT slides were generally superior to conventional slides, showing smaller cell monolayer surface area, clearer background and more distinct stereoscopic cytological features. For Papanicolaou’s staining, LCT-prepared slides allowed a higher positive rate of Aspergillus identification than conventional slides for bronchial brushing samples (59.25% vs. 20.37%, P < 0.05) and sputum samples (29.05% vs. 8.55%, P < 0.05). Similarly, Special staining of LCT-prepared slides showed a higher positive rate of Aspergillus identification for bronchial brushing samples (83.33% vs. 57.41%, P < 0.05) and sputum samples (43.59% vs. 19.66%, P < 0.05). This preliminary study suggests that LCT may be better than conventional slide preparation for identifying Aspergillus in respiratory samples from patients with pulmonary aspergillosis.

periodic acid-Schiff (PAS) staining 7,8 . Unfortunately, this conventional cytopathology shows poor sensitivity due to the presence of air-drying artifacts, overlapping clumps of cells and frequent presence of obscuring materials, such as necrotic tissue, mucus, inflammatory cells and blood 6,7 .
A liquid-based cytopathological test (LCT), already commonly used to screen malignant cells in a variety of cancers, including cervical, thyroid, breast and lung cancers [9][10][11][12][13] , may provide benefits over the conventional cytopathological approach. LCT typically results in cleaner background and more evenly distributed cells, and LCT samples can be equally divided among multiple slides. Studies comparing LCT with traditional cytopathology indicate that LCT can show equivalent test accuracy and cytopathology features, while offering more satisfactory results 14,15 . The present study aimed to compare LCT and conventional slide preparation techniques for identifying Aspergillus following Papanicolaou's or Special staining of respiratory samples from patients with pulmonary aspergillosis.

Patients and Methods
Patient selection. This study was conducted according to the latest version of the Declaration of Helsinki, and ethical approval was obtained from the Ethics Committee of West China Hospital of Sichuan University. Upon admission, all patients in the study provided written informed consent for samples to be collected and analysed.
This study involved respiratory samples from 171 consecutive patients diagnosed with pulmonary aspergillosis in our hospital between March 1, 2012 and December 31, 2015. IPA patients were diagnosed with proven or probable based on the criteria of the European Organisation for Research and Treatment of Cancer/Mycoses Study Group (EORTC/MSG) 16 . CPA, ABPA and aspergilloma were diagnosed based on the expert consensus of the Chinese Thoracic Society and on guidelines of the European Society for Clinical Microbiology and Infectious Diseases as well as the European Respiratory Society 17,18 . A total of 117 pairs of sputum samples were obtained from 117 patients, while 54 pairs of bronchial brushing samples were obtained from the remaining 54 patients.
Papanicolaou's staining. The 117 pairs of sputum specimens were divided into two sets; one was processed conventionally and the other was processed using LCT. For conventional processing, direct smears were prepared from the fresh samples, slides were fixed immediately in 95% alcohol and stained using Papanicolaou's method. LCT processing was carried out using the SurePath LCT kit (Tripath PREP, BD SurePath, Burlington, NC, USA), one of the most commonly used LCT platforms 17,18 . Samples were mixed with 10 mL of CytoRich TM Red and 1 mL of mucolytic agent per 10 mL of sputum. The mixture was incubated at room temperature for 30 min, vortexed for 10 s, and centrifuged at 600 rpm for 10 min. The supernatant was removed and the pellet was resuspended in 10 mL of distilled water. This suspension was vortexed again and centrifuged at 600 rpm for 5 min. The supernatant was removed and the pellet was vortexed and transferred to the AutoCyte PREP system (Tripath Imaging) for automated slide preparation and staining.
The 54 pairs of bronchial brushing samples, obtained via fiberoptic bronchoscopy, were also divided into two sets: one was processed conventionally and the other was processed using LCT. Conventional preparation was as described for sputum samples. LCT preparation began with 30-min incubation of bronchial washing specimens in CytoClear TM medium, followed by centrifugation at 600 rpm for 5 min. The supernatant was removed and the pellet was vortexed and transferred to the AutoCyte PREP system (Tripath Imaging) for automated slide preparation and staining.
Special staining with GMA and PAS. Samples were prepared as described above using conventional or LCT procedures and stained with GMS or PAS instead of Papanicolaou's stain. A Special Stains Automated Slide Stainer (NexEs, USA) was used with commercial kits (Roche Diagnostics, USA).
Pathological examination. Two experienced cytopathologists blinded to patient status analysed the samples independently. Slides stained with Papanicolaou's staining were considered positive if hyaline, septate mycelium filaments 2-4 μm in diameter with occasional branching at acute angles were observed. Slides stained with PAS were considered positive if magenta Aspergillus mycelium was observed; slides stained with GMS were defined as positive if brown-black Aspergillus mycelium was observed. Patients were defined as positive by Special staining only if both GMS-and PAS-stained samples were positive. Disagreements were resolved by discussion.
Statistical analysis. Data were analysed using SPSS 18.0 for Windows (IBM, Chicago, IL, USA). Categorical data were expressed as absolute or relative frequencies, while continuous data were expressed as mean ± SD. Sensitivities of the conventional and LCT methods were compared using the χ 2 test. The threshold of significance for all statistical tests was defined as a two-sided P < 0.05.

Results
Clinical characteristics of patients. The  Cytomorphological characteristics of Aspergillus. The morphological characteristics of Aspergillus were generally clearer and more easily distinguished in LCT samples than in conventional samples. In sputum and bronchial brushing samples analysed by LCT with Papanicolaou's staining, Aspergillus filaments showed a constant diameter, parallel edges, and dichotomous branching at acute angles ( Fig. 1). The microscopic fields were generally clearer in LCT samples than in conventional ones, showing less necrosis, mucus, inflammatory cells and blood. Similar results were observed for LCT-prepared slides following Special staining (Figs 2 and 3).
In Papanicolaou's staining, a proportion of bronchial brushing samples showed distinctive cytopathology that we termed "necrosis-containing branch profile" or "branched ashes-like necrosis". This feature was observed in 28 pairs of samples prepared conventionally and 10 pairs of samples prepared by LCT. All but one of the 10 paired samples that showed "necrosis-containing branch profile" after LCT also showed this result after conventional processing. This feature appeared under the microscope as patchy necrosis (mainly in LCT samples) or as lamellar necrosis (mainly in conventional samples), with light blue granular staining (Fig. 4). Areas of lamellar necrosis showed outlines of branching shapes with nearly the same size, like the ashes of burned branches, without tumor pyknosis or apoptosis.
Aspergillus identification using Papanicolaou's staining. The LCT platform led to identification of Aspergillus in 34 of 117 sputum samples, corresponding to a sensitivity of 29.05% (Table 2). This was significantly higher than conventional cytopathology (10 of 117, 8.55%, P < 0.001). The LCT platform led to identification of Aspergillus in 32 of 54 bronchial brushing samples, corresponding to a sensitivity of 59.25%. This was significantly higher than conventional cytopathology (11 of 54, 20.37%, P < 0.001).
Next we investigated the performance of LCT in subgroups of patients with pulmonary aspergillosis. IPA patients showed the highest rate of positive Aspergillus identifications with LCT-prepared slides: 64.00% (16/25) in the case of bronchial brushing samples, 36.50% (23/63) in the case of sputum samples.
Comparison of conventional or LCT slide preparation followed by Special staining showed that LCT was associated with a higher rate of positive Aspergillus identification in bronchial brushing samples (83.33% vs 57.41%, P < 0.05) and sputum samples (43.59% vs 19.66%, P < 0.05). Among subgroups of patients with pulmonary aspergillosis, IPA patients showed the highest rates of positive identification (Table 3).

Discussion
Identification of Aspergillus from stained cytological slides plays an important role in the diagnosis and management of pulmonary aspergillosis, but conventional slide preparation has several disadvantages that make Aspergillus identification in respiratory samples challenging 19,20 . In this study, we compared the ability of LCT and conventional cytopathology to allow Aspergillus identification in sputum and bronchial brushing samples following Papanicolaou's or Special staining. We found that LCT allowed significantly better pathogen detection in both types of samples, suggesting that this method may help improve timely and accurate diagnosis of pulmonary aspergillosis.
To identify Aspergillus, conventional culture-based methods on respiratory tract samples show poor sensitivity of approximately 40% for diagnosing IPA and 17.9% for diagnosing CPA 21 . Meanwhile, biomarker tests are limited and there is no consensus about which sampling sites (blood, broncho-alveolar fluid or tissue biopsy) are more practical and result in more sensitive detection. For example, the serum galactomannan assay shows relatively low sensitivity in patients who have no haematological conditions or who are on antifungal therapy 22 . Tests based on the pan-fungal marker β-d-glucan do not discriminate between invasive aspergillosis and other invasive fungal infections [23][24][25] . EORTC guidelines do not include Aspergillus PCR as a mycological criterion for diagnosing probable IPA 16,26 . Lung biopsy, which is considered a gold standard procedure, is highly invasive and available only at specialised institutions. These disadvantages of existing methods have helped drive efforts to develop new methods for rapidly identifying Aspergillus in order to facilitate early, accurate diagnosis of pulmonary aspergillosis.
Our results suggest that the LCT platform offers advantages over conventional cytopathology for both Papanicolaou's and Special staining of sputum and bronchial brushing samples. Our results support the idea that LCT improves visualisation of cells: the cytopathology slides have uniform thickness; cellular structure is better preserved; and air-drying artifacts, obscuring blood and inflammatory exudate are absent. In addition, LCT-processed samples contain approximately 5-fold more sputum than conventional cytopathology samples, improving the chances of detecting Aspergillus. The fact that LCT slides are automatically prepared and stained in the AutoCyte PREP system helps ensure uniform quality of results, regardless of the practitioner's experience. We found that the rate of positive identification with LCT-prepared slides was higher in bronchial brushing samples than in sputum samples (59.25% vs 29.05%), and that the positive rate on LCT-prepared slides was higher for IPA patients than for other subgroups with pulmonary aspergillosis. These results suggest that the diagnostic Bronchial brushing samples from the lower respiratory tract can aid in the diagnosis of pulmonary aspergillosis 27 . Our analysis of 54 such samples revealed a potential novel type of pathological necrosis different from caseous necrosis or tumor necrosis: in some samples, when we observed necrosis and subsequently darkened the light source, we observed that the necrosis presented with wither-like ashes, which we named "necrosis-containing branch profile" or "branched ashes-like necrosis". This level of cytomorphological detail into necrosis illustrates the greater clarity of subcellular structures in LCT samples than in conventional samples, since necrosis often masks Aspergillus in conventional smears. We suggest that "branched ashes-like necrosis" is a cytomorphological feature of Aspergillus in bronchial brushing samples, and that its detection can aid in the diagnosis of pulmonary aspergillosis. Specifically, "branched ashes-like necrosis" observed after Papanicolaou's staining suggests the possibility of pulmonary aspergillosis, which should be investigated with further tests. Neutrophil karyopyknosis can also appear in cytological samples from patients with pulmonary aspergillosis, but such changes may reflect general diapyetic inflammation and are useful for differential diagnosis 28 .
Several studies recommend using Special staining such as PAS and GMS stain on all available samples, especially for immunodeficient patients [6][7][8] . We found that using Special staining increased the chance of identifying Aspergillus in sputum and bronchial brushing samples over Papanicolaou's staining. We also found that positive Aspergillus identification rates were higher with LCT-prepared slides than with conventionally prepared ones. Our results suggest the benefit of using LCT with Papanicolaou's or Special staining for the identification of Aspergillus in respiratory samples, which may improve diagnosis of pulmonary aspergillosis.
Taken together, our results suggest that LCT may be a novel method for identifying Aspergillus in respiratory samples. However, several considerations suggest that it cannot provide definitive diagnosis of pulmonary aspergillosis on its own. First, LCT cannot accurately type Aspergillus species, which may affect the selection of antifungal drugs. Second, international guidelines dictate that aspergillosis be confirmed through histology and isolation of Aspergillus from samples obtained through sterile procedures. Therefore, results of LCT analysis may provide information valuable for subsequent culture and invasive biopsy tests. Third, although LCT leads to higher rates of positive identification following Papanicolaou's staining, the overall positive rate for sputum is still low (29.05%). This low sensitivity likely reflects that Papanicolaou's stain is not specific 29 and that it can give false negative results when small amounts of fungus are present, when sputum is of low quality or when the site of infection is not connected to the bronchus. Despite this low sensitivity, we suggest that Special staining of LCT-prepared slides should be performed on all available samples in order to maximise the possibility of identifying Aspergillus. In this study, LCT-based Special staining on sputum samples increased the positive rate to 43.59%, significantly higher than the rate with Papanicolaou's staining (29.05%). Our results indicate that LCT-based Papanicolaou's staining and Special staining can be useful for screening respiratory samples for the presence of Aspergillus in order to guide further biopsy and culture for definitive diagnosis. At the same time, this diagnostic approach may be useful in many critical ill patients who cannot tolerate invasive biopsy examination.
Our results with this small sample should be verified in larger, prospective diagnostic studies. The retrospective findings described here are based only on patients diagnosed with pulmonary aspergillosis, allowing us to calculate only the rate of positive identification (sensitivity) as a partial measure of diagnostic performance.

Conclusion
This preliminary study suggests that LCT-based Papanicolaou's staining and Special staining may be helpful for the identification of Aspergillus in sputum and bronchial brushing samples from patients with pulmonary aspergillosis. Our results justify further work, particularly prospective trials, to develop this assay for clinical use.  Table 3. Results of conventional and liquid-based cytopathology for diagnosing different types of pulmonary aspergillosis IPA: invasive pulmonary aspergillosis; CPA: chronic pulmonary aspergillosis; ABPA: allergic bronchopulmonary aspergillosis.