A novel diagnosis scoring model to predict invasive pulmonary aspergillosis in the intensive care unit

Diagnostic Utility of Galactomannan Enzyme Immunoassay in Invasive Aspergillosis in Pediatric patients with Hematological Malignancy

OBJECTIVE

This study aims to determine the diagnostic utility of galactomannan enzyme immunoassay (GM EIA) in invasive aspergillosis (IA) in children with hematological malignancy (high risk population) in terms of sensitivity, specificity, negative predictive value (NPV) and positive predictive values (PPV) at various cut offs while validating the revised EORTC/MSG 2019 criteria in order to obtain the best cut-off.

MATERIAL AND METHODS

For 100 pediatric patients, serum and respiratory samples were collected. Clinical, mycological workup (potassium-hydroxide mount, fungal culture) and GM EIA was done to classify proven, probable, and possible IA as per EORTC-MSG guidelines,2019. Sensitivity, specificity, PPV and NPV were calculated of GM indices at cut-off 0.5, 0.7 and 1, and validated with revised EORTC -MSG, 2019.

RESULTS

Of 100 patients enrolled, 75 were diagnosed with ALL, 14 with AML, two with Hodgkin's, three had non-Hodgkin lymphoma, and six had undifferentiated leukemia. With routine mycological findings, 51 were classified as probable IA, 11 as possible IA, and 38 as no IA. Aspergillus flavus was the most prevalent on culture (56.9%, 29/51) followed by A. fumigatus (29%, 15/51) A. niger (7.8%, 4/51), A. terreus (3.9%, 2/51) and A. nidulans (2%, 1/51). GM EIA demonstrated sensitivity 82.3%, specificity 97.4%, PPV 98.1%, and NPV 77.1% at cut-off 0.67 when comparing probable/possible IA v/s no IA groups. The GM EIA had the best sensitivity (82.4%), specificity (81.8%), PPV (95.5%), and NPV (50%) at cut off 0.78 when the probable IA group was compared to the possible IA. Seven patients succumbed of whom 5 had GMI ≥ 2.

CONCLUSION

This study deduces the optimal cut-off for serum GM EIA to be 0.67 obtained by ROC analysis when comparing possible and probable IA versus no IA and reinforces the definition of probable category of EORTC-MSG criteria, 2019. At 0.5 ODI the sensitivity (87.1%) and NPV (80.5%) are high, thus making it the most suitable cut-off for detecting true positive and ruling out IA respectively, in pediatric patients with hematological malignancy. GM EIA when performed adjunctive to clinico-radiological findings can prove to be screening, diagnostic and prognostic test for IA in pediatric hematological malignancy patients.

Prototype early diagnostic model for invasive pulmonary aspergillosis based on deep learning and big data training

BACKGROUND

Currently, the diagnosis of invasive pulmonary aspergillosis (IPA) mainly depends on the integration of clinical, radiological and microbiological data. Artificial intelligence (AI) has shown great advantages in dealing with data-rich biological and medical challenges, but the literature on IPA diagnosis is rare.

OBJECTIVE

This study aimed to provide a non-invasive, objective and easy-to-use AI approach for the early diagnosis of IPA.

METHODS

We generated a prototype diagnostic deep learning model (IPA-NET) comprising three interrelated computation modules for the automatic diagnosis of IPA. First, IPA-NET was subjected to transfer learning using 300,000 CT images of non-fungal pneumonia from an online database. Second, training and internal test sets, including clinical features and chest CT images of patients with IPA and non-fungal pneumonia in the early stage of the disease, were independently constructed for model training and internal verification. Third, the model was further validated using an external test set.

RESULTS

IPA-NET showed a marked diagnostic performance for IPA as verified by the internal test set, with an accuracy of 96.8%, a sensitivity of 0.98, a specificity of 0.96 and an area under the curve (AUC) of 0.99. When further validated using the external test set, IPA-NET showed an accuracy of 89.7%, a sensitivity of 0.88, a specificity of 0.91 and an AUC of 0.95.

CONCLUSION

This novel deep learning model provides a non-invasive, objective and reliable method for the early diagnosis of IPA.

Analysis of Fungal and Bacterial Co-Infections in Mortality Cases among Hospitalized Patients with COVID-19 in Taipei, Taiwan

Fungal or bacterial co-infections in patients with H1N1 influenza have already been reported in many studies. However, information on the risk factors, complications, and prognosis of mortality cases with coronavirus disease 2019 (COVID-19) are limited. We aimed to assess 36 mortality cases of 178 hospitalized patients among 339 patients confirmed to have had SARS-CoV-2 infections in a medical center in the Wenshan District of Taipei, Taiwan, between January 2020 and September 2021. Of these 36 mortality cases, 20 (60%) were men, 28 (77.7%) were aged >65 years, and the median age was 76 (54–99) years. Comorbidities such as hypertension, coronary artery disease, and chronic kidney disease were more likely to be found in the group with length of stay (LOS) > 7 d. In addition, the laboratory data indicating elevated creatinine-phosphate-kinase (CPK) (p < 0.001) and lactic acid dehydrogenase (LDH) (p = 0.05), and low albumin (p < 0.01) levels were significantly related to poor prognosis and mortality. The respiratory pathogens of early co-infections (LOS < 7 d) in the rapid progression to death group (n = 7 patients) were two bacteria (22.2%) and seven Candida species (77.8.7%). In contrast, pathogens of late co-infections (LOS > 7 d) (n = 27 patients) were 20 bacterial (54.1%), 16 Candida (43.2%), and only 1 Aspergillus (2.7%) species. In conclusion, the risk factors related to COVID-19 mortality in the Wenshan District of Taipei, Taiwan, were old age, comorbidities, and abnormal biomarkers such as low albumin level and elevated CPK and LDH levels. Bacterial co-infections are more common with Gram-negative pathogens. However, fungal co-infections are relatively more common with Candida spp. than Aspergillus in mortality cases of COVID-19.

Serious fungal disease incidence and prevalence in Indonesia

BACKGROUND

Indonesia is a tropical country, warm and humid, with numerous environmental fungi. Data on fungal disease burden help policymakers and clinicians.

OBJECTIVES

We have estimated the incidence and prevalence of serious fungal diseases.

METHODS

We found all published and unpublished data and estimated the incidence and prevalence of fungal diseases based on populations at risk. HIV data were derived from UNAIDS (2017), pulmonary tuberculosis (PTB) data from 2013-2019, data on chronic pulmonary aspergillosis (CPA) were used to estimate CPA prevalence and likely deaths, COPD data from Hammond (2020), lung cancer incidence was from Globocan 2018, and fungal rhinosinusitis was estimated using community data from India.

RESULTS

Overall ~7.7 million Indonesians (2.89%) have a serious fungal infection each year. The annual incidence of cryptococcosis in AIDS was 7,540. Pneumocystis pneumonia incidence was estimated at 15,400 in HIV and an equal number in non-HIV patients. An estimated 1% and 0.2% of new AIDS patients have disseminated histoplasmosis or Talaromyces marneffei infection. The incidence of candidaemia is 26,710. The annual incidence of invasive aspergillosis was estimated at 49,500 and the prevalence of CPA is at 378,700 cases. Allergic bronchopulmonary aspergillosis prevalence in adults is estimated at 336,200, severe asthma with fungal sensitisation at 443,800, and fungal rhinosinusitis at 294,000. Recurrent vulvovaginal candidiasis is estimated at 5 million/year (15-50 years old). The incidence of fungal keratitis around 40,050. Tinea capitis prevalence in schoolchildren about 729,000.

CONCLUSIONS

Indonesia has a high burden of fungal infections.

Epidemiological Correlation of Pulmonary Aspergillus Infections with Ambient Pollutions and Influenza A (H1N1) in Southern Taiwan

An increase in fungal spores in ambient air is reported during a spike in particulate matter (PM_2.5 and PM₁₀) aerosols generated during dust or smog events. However, little is known about the impact of ambient bioaerosols on fungal infections in humans. To identify the correlation between the incidence of pulmonary aspergillosis and PM-associated bioaerosols (PM_2.5 and PM₁₀), we retrospectively analyzed data between 2015 and 2018 (first stage) and prospectively analyzed data in 2019 (second stage). Patient data were collected from patients in three medical institutions in Tainan, a city with a population of 1.88 million, located in southern Taiwan. PM data were obtained from the Taiwan Air Quality Monitoring Network. Overall, 544 non-repeated aspergillosis patients (first stage, n = 340; second stage, n = 204) were identified and enrolled for analysis. The trend of aspergillosis significantly increased from 2015 to 2019. Influenza A (H1N1) and ambient PMs (PM_2.5 and PM₁₀) levels had significant effects on aspergillosis from 2015 to 2018. However, ambient PMs and influenza A (H1N1) in Tainan were correlated with the occurrence of aspergillosis in 2018 and 2019, respectively. Overall (2015–2019), aspergillosis was significantly correlated with influenza (p = 0.002), influenza A (H1N1) (p < 0.001), and PM_2.5 (p = 0.040) in Tainan City. Using a stepwise regression model, influenza A (H1N1) (p < 0.0001) and Tainan PM₁₀ (p = 0.016) could significantly predict the occurrence of aspergillosis in Tainan. PM-related bioaerosols and influenza A (H1N1) contribute to the incidence of pulmonary aspergillosis.

Diagnostic Accuracy of Bronchoalveolar Lavage Fluid Galactomannan for Invasive Aspergillosis

Background

The pathogenesis of invasive aspergillosis (IA) is still unknown, but its progression is rapid and mortality rate remains high. Bronchoalveolar lavage fluid (BALF) galactomannan (GM) analysis has been used to diagnose IA. This study is aimed at making an accurate estimate of the whole accuracy of BALF-GM in diagnosing IA.

Methods

After a systematic review of the study, a bivariate meta-analysis was used to summarize the specificity (SPE), the sensitivity (SEN), the positive likelihood ratios (PLR), and the negative likelihood ratios (NLR) of BALF-GM in diagnosing IA. The overall test performance was summarized using a layered summary receiver operating characteristic (SROC) curve. Subgroup analysis was performed to explore the heterogeneity between studies.

Results

A total of 65 studies that are in line with the inclusion criteria were included. The summary estimates of BALF-GM analysis are divided into four categories. The first is the proven+probable vs. possible+no IA, with an SPE, 0.87 (95% CI, 0.85-0.98); SEN, 0.81 (95% CI, 0.76-0.84); PLR, 9.78 (5.78-16.56); and NLR, 0.20 (0.14-0.29). The AUC was 0.94. The BALF-GM test for proven+probable vs. no IA showed SPE, 0.88 (95% CI, 0.87-0.90); SEN, 0.82 (95% CI, 0.78-0.85); PLR, 6.56 (4.93-8.75); and NLR, 0.24 (0.17-0.33). The AUC was 0.93. The BALF-GM test for proven+probable+possible vs. no IA showed SPE, 0.82 (95% CI, 0.79-0.95); SEN, 0.59 (95% CI, 0.55-0.63); PLR, 3.60 (2.07-6.25); and NLR, 0.31 (0.15-0.61). The AUC was 0.86. The analyses for others showed SPE, 0.85 (95% CI, 0.83-0.87); SEN, 0.89 (95% CI, 0.86-0.91); PLR, 6.91 (4.67-10.22); and NLR, 0.18 (0.13-0.26). The AUC was 0.94.

Conclusions

The findings of this BALF-GM test resulted in some impact on the diagnosis of IA. The BALF-GM assay is considered a method for diagnosing IA with high SEN and SPE. However, the patients' underlying diseases may affect the accuracy of diagnosis. When the cutoff is greater than 1, the sensitivity will be higher.