A total of 100 pediatric patients, 63 male 37 female with median age of 6 years were enrolled in the study and taking into consideration the host factors, radiological findings, and mycological findings, they were categorised into proven, probable, possible, and no evidence of IA according to EORTC/MSG 2019 guidelines.
Out of 100, 61 patients were diagnosed cases of B-cell ALL (most common), followed by T-cell ALL 14, AML 14, 3 cases of non-Hodgkin’s lymphoma, 2 of Hodgkin’s lymphoma and 6 as unclassified/undifferentiated leukemia as shown in Table1. Forty six percent of patients with B-cell ALL (28/61), 57% (8/14) T-cell ALL, 42% (6/14) AML,33% (1/3) cases of NHL, were categorized as probable IA, satisfying the EORTC/MSG guidelines. There was no significant association (Fisher’s exact test) between the type of hematological malignancy in this study cohort and occurrence of IA(p value = 0.304)
Table 1
Distribution of hematological malignancy in the patients enrolled in the study
Diagnosis | Frequency | Percentage | 95%CI |
AML | 14 | 14.0% | 8.1%-22.7% |
B-cellALL | 61 | 61.0% | 50.7%-70.4% |
T-cellALL | 14 | 14.0% | 8.1%-22.7% |
NHL | 3 | 3.0% | 0.8%-9.2% |
HL | 2 | 2.0% | 0.3%-7.7% |
Undifferentiated/unclassified | 6 | 6.0% | 2.5%-13.1% |
Fever was the most common presenting symptom (82%), followed by cough (40%), pleuritic chest pain (24%), dyspnoea (16%), sore throat (7%), hemoptysis (5%), symptoms of malignancy like weight loss (16%), lymphadenopathy (13%) and other clinical features like loss of appetite, poor growth, diarrhoea, vomiting, epistaxis, gum bleeding, pain abdomen, bone pain, rash, in 36% patients. None of the clinical features were significantly associated with IA cases at the time of presentation.
Chest imaging by X-ray and/or HRCT scan was done for all the patients and 58 (58%) patients had significant findings, most common being reticular alveolar infiltrates (52% of the positive findings). The classical radiological features of IA, i.e., halo sign and cavity on HRCT scan was present in 3 and 4 cases, respectively. Other radiology findings were non-specific consolidations (15), ground glass opacities (12), pleural effusion (2), perihilar infiltrates (2) and mass lesion (1). Of the 58, 49 patients were diagnosed as probable IA and 9 as possible IA. In these probable cases the radiological finding was non-specific. However, there was a statistically significant association of HRCT scan abnormality with true positives for GM (p < 0.001).
In the routine fungal cultures 51% (51/100) yielded growth positive for Aspergillus and 49% (49/100) of the patients were culture negative. In the aetiological profile Aspergillus flavus was most common 56.8% (29/51) followed by A. fumigatus 29.4% (15/21) A. niger 7.8% (4/51) A. terreus 3.9% (2/51) and A. nidulans 1.9% (1/51)
Serum GM was measured in all (n = 100) patients. As per Platelia™ Aspergillus antigen kit ELISA literature the significant cut-off ODI ≥ 0.5 was taken. 59 (59%) patients were GM positive and 41 (41%) seronegative at this cut-off.
Out of the 59 positive cases, 49 (83%) were diagnosed as probable IA and 5 (8.4%) as possible IA. False positive GM ≥ 0.5 was found in 5 (8.4%) patients in association with Candida sepsis in 1 patient, grade III/IV mucositis in 2 patients and 2 patient had no identifiable cause.
Performance of GM EIA at different ODI cut-offs ≥ 0.5, ≥ 0.7 and ≥ 1.0 was compared with fungal culture from nonsterile sites and sensitivity, specificity, PPV and NPV for GM EIA were calculated. The findings observed are mentioned in Table 2. The best cut-off observed was ≥ 0.7 at which sensitivity, specificity, PPV and NPV were 88.2%, 91.8%, 91.8 and 88.2, respectively.
Table 2
Performance of GMEIA at various cut-offs in the study population
| Cutoffvaluesofgalactomannanassay |
0.5 | 0.7 | 1.0 |
Sensitivity(%) | 96.0 | 88.2 | 58.8 |
Specificity(%) | 79.6 | 91.8 | 95.9 |
PPV | 83.0 | 91.8 | 93.7 |
NPV | 95.1 | 88.2 | 69.1 |
Serum GMEIA results were benchmarked against the revised EORTC/MSG 2019 definitions for IA. [6] Taking these definitions are reference standard,[10] sensitivity, specificity, PPV and NPV of GM EIA calculated at cut-off ODI ≥ 0.5 were 87.1%, 86.8%, 91.5 and 80.5 for proven / probable / possible IA. The specificity and PPV increased to100% at cut-off ≥ 0.7 and ≥ 1 shown in Table 3. ROC curve analysis was done to determine the best cut-off and it was found to be 0.67 ODI depicted in Fig. 1.
Table 3
Performance of GMEIA for predicting EORTC diagnosis: Proven / Probable / Possible IA vs No IA
Variable | Sensitivity(CI) | Specificity(CI) | PPV(CI) | NPV(CI) | DiagnosticAccuracy(CI) |
GMI(Cutoff:0.67byROC) | 82.3%(70–91) | 97.4%(86–100) | 98.1%(90–100) | 77.1%(63–88) | 88.0%(80–94) |
GMI(Cutoff:≥0.5) | 87.1%(76–94) | 86.8%(72–96) | 91.5%(81–97) | 80.5%(65–91) | 87.0%(79–93) |
GMI(Cutoff:≥0.7) | 79.0%(67–88) | 100.0%(91–100) | 100.0%(93–100) | 74.5%(60–86) | 87.0%(79–93) |
GMI(Cutoff:≥1) | 51.6%(39–65) | 100.0%(91–100) | 100.0%(89–100) | 55.9%(43–68) | 70.0%(60–79) |
GMI(Cutoff:≥1.5) | 30.6%(20–44) | 100.0%(91–100) | 100.0%(82–100) | 46.9%(36–58) | 57.0%(47–67) |
While comparing probable IA versus possible IA groups sensitivity, specificity, PPV and NPV detected at cut-off ≥ 0.5 were 96.1%, 54.5%, 90.7 and 75 respectively. On increasing the cut-off ODI to ≥ 0.7 and ≥ 1, specificity and PPV increased to 63.6% and 91.8 respectively at ODI ≥ 0.7 and 81.8% and 93.8 respectively at ODI ≥ 1 represented in Table 4. Sensitivity although, decreased drastically. ROC curve analysis was done to determine the best cut-off and it was found to be 0.78 ODI and depicted in Fig. 2.
Table 4
Performance of GMEIA for predicting EORTC diagnosis: Probable IA vs Possible IA
Variable | Sensitivity(CI) | Specificity(CI) | PPV(CI) | NPV(CI) | DiagnosticAccuracy(CI) |
GMI(Cutoff:0.78byROC) | 82.4%(69–92) | 81.8%(48–98) | 95.5%(85–99) | 50.0%(26–74) | 82.3%(70–91) |
GMI(Cutoff:>0.5) | 96.1%(87–100) | 54.5%(23–83) | 90.7%(80–97) | 75.0%(35–97) | 88.7%(78–95) |
GMI(Cutoff:>0.7) | 88.2%(76–96) | 63.6%(31–89) | 91.8%(80–98) | 53.8%(25–81) | 83.9%(72–92) |
GMI(Cutoff:>1) | 58.8%(44–72) | 81.8%(48–98) | 93.8%(79–99) | 30.0%(15–49) | 62.9%(50–75) |
GMI(Cutoff:>1.5) | 37.3%(24–52) | 100.0%(72–100) | 100.0%(82–100) | 25.6%(14–41) | 48.4%(35–61) |
As per revised EORTC definitions for IA, the ideal cut-off ODI of 0.725 (average of 0.67 and 0.78) was obtained in this study for the classification of IA into probable, possible and no IA in the pediatric patients with hematological malignancies.
The outcome of the patients was determined in terms of the number of patients (n) who were discharged after a successful chemotherapy and/or antifungal therapy (87), had in-hospital mortality (7) or developed complication of respiratory failure and required ICU admission (6). Of the 87 patients who were discharged, 43 were diagnosed as probable IA, 9 as possible IA and 37 with no evidence of IA. GMI ranged from 0.5 to 3.9 in the probable category. Of the patients that required intensive care 3 were diagnosed as probable IA, 2 as possible IA and 1 with no evidence of IA. In the possible and probable IA category, GMI ranged from 0.4 to 1.5. Out of the 7 patients that succumbed, 5 were diagnosed as probable IA and 2 had no evidence of IA. In probable IA patients who died GM index was ≥ 2. The patients in the no IA category had associated Candida sepsis.