Baseline characteristics
Out of 259 samples obtained, 20% (53/259) were classified as probable IPA according to 2020 EORTC/MSG criteria and 2024 FUNDICU criteria, while the remaining 80% (206/259) were classified as having no IPA. No patient fulfilled the criteria for proven IPA. Demographic breakdown of the patients included 149/259 females (58%) and 110/259 males (42%), with a median age of 65 years, ranging from 20 to 89 years (Table 1). Nine percent (22/259) had hematological diseases, 1% (3/259) were SOT recipients, 30% (77/259) required ICU admission, 29% (75/259) had solid organ malignancies and 31% (82/259) presented with other underlying conditions. Among all samples, 9% (22/259) were collected from patients with ongoing mold-active antifungal prophylaxis at time of BALF sampling.
Table 1 Study cohort characteristics for BALF samples
Baseline characteristics for BALF samples
|
Probable IPA
(n=53)
|
No IPA
(n=206)
|
Total (n=259)
|
Sex, n (%)
Female Male
|
20 (38%)
33 (62%)
|
90 (44%)
116 (56%)
|
149 (58%)
110 (42%)
|
Median age in years (range)
|
66 (23-82)
|
65 (20-89)
|
65 (20-89)
|
Underlying diseases/conditions, n (%)
Hematological malignancy
Solid organ transplantation
Need for intensive care treatment
Solid organ malignancy
Other
|
3 (6%)
2 (4%)
24 (45%)
13 (25%)
11 (21%)
|
19 (9%)
1 (1%)
53 (26%)
62 (30%)
71 (35%)
|
22 (9%)
3 (1%)
77 (30%)
75 (29%)
82 (31%)
|
Antifungal prophylaxis at time of sampling, n (%)
|
13 (25%)
|
9 (4%)
|
22 (9%)
|
Mycological test results from BALF, n (%)
Aspergillus growth in BALF culture
Aspergillus spp. BALF PCR§
GM ≥0.5 ODI Platelia
GM ≥1.0 ODI Platelia
GM ≥0.20 ODI clarus
|
14 (26%)
2/6 (33%)
49 (92 %)
45 (85%)
51 (96%)
|
13 (6%)
1/47 (2%)
51 (25%)
19 (9%)
53 (26%)
|
27 (10%)
3/53 (6%)
100 (39%)
64 (25%)
104 (40%)
|
Abbreviations: GM= Galactomannan; BALF = bronchoalveolar lavage fluid; IPA = invasive pulmonary aspergillosis; PCR = polymerase chain reaction; ODI= optical density index
§ AsperGenius (PathoNostics, Maastricht, The Netherlands)
|
Analytic performances
Spearman's correlation analysis demonstrated a strong positive correlation between the clarus GM and Platelia GM results (rho = 0. 727, p < 0.001) (Figure 1) in the overall study cohort.
When distinguishing between probable IPA and no IPA, the clarus GM performance (Table 2) with the manufacturer-recommended cutoff of ≥0.20 ODI achieved a sensitivity of 96% (51/53) and a specificity of 74% (153/206). Optimizing diagnostic discriminatory power using Youden’s index resulted in a cutoff value of ≥0.26 ODI for clarus GM. With this adjustment, sensitivity remained 96% (51/53), while specificity increased to 81% (167/206). The specificity of clarus GM with ≥0.26 ODI was significantly lower compared to Platelia GM with ≥1.0 ODI (p=0.003), while the difference in sensitivity was not statistically significant (p=0.09).
For the Platelia GM comparator test (used for IPA classification), sensitivity was 85% (45/53) and specificity was 91% (188/206) when utilizing a cutoff of ≥1.0 ODI, while sensitivity was 92% (49/53) and specificity was 75% (155/206) when using the ≥0.5 ODI cutoff.
When combining clarus GM ≥0.26 with Platelia GM ≥1.0, test performance sensitivity remained unchanged with 96% (51/53) with either/or both tests resulting positive, while specificity increased to 94% (194/206) when positive results of both tests were required.
ROC curve analysis demonstrated an AUC of 0.936 (95% CI of 0.901-0.971) for clarus GM and 0.918 (95% CI of 0.876-0.959) for the Platelia GM assay (Figure 2a).
Nine percent (22/259) of the samples were obtained from patients undergoing mold-active antifungal prophylaxis. Among these, 13 had probable IPA. Using for clarus GM a cutoff value of ≥0.2 ODI, the sensitivity was 92% (12/13), with a specificity of 56% (5/9). Increasing the cutoff value to ≥0.26 ODI maintained the sensitivity at 92%, while the specificity increased to 67% (6/9). The AUC for clarus GM was numerically but not significantly lower with antifungal prophylaxis (0.863 (95% CI: 0.711 – 1) compared to patients without prophylaxis 0.951 (95% CI: 0.920 – 0.981; p=0.27). The AUC for the Platelia GM in patients with ongoing mold-active antifungals was 0.940 (95%CI 0.835 – 1), and not significantly different from the performance of the clarus GM in this cohort (p=0.42).
When classifying probable IPA vs. no IPA based solely on Aspergillus culture and PCR (Table 2), only 19/53 probable IPA cases remained probable IPA cases. The AUCs for Platelia GM and clarus GM were 0.948 (95% CI: 0.882-1) and 0.945 (95% CI: 0.884-1), respectively (Figure 2b).
The Platelia GM had a sensitivity of 93% (14/15) and a specificity of 92% (190/206) at a cutoff value of ODI ≥1.0. Lowering the cutoff to ≥0.5 ODI maintained sensitivity at 93% (14/15), with specificity decreasing to 76% (157/206). At the manufacturer-recommended cutoff of ODI ≥0.2, the clarus GM had a sensitivity of 93% (14/15) and specificity of 74% (152/206). Using the calculated optimal cutoff of ODI ≥0.26, sensitivity remained 93% (14/15), while specificity increased to 81% (167/206).
Table 2 Sensitivity and specificity comparison of Platelia GM and clarus GM for diagnosing probable IPA in BALF, with and without Platelia GM test results as a mycological criterion
Probable IPA vs. no IPA classification
|
|
Sensitivity (n=53)
|
Specificity (n=206)
|
GM ≥0.2 ODI Clarus
|
96% (51)
|
74 % (153)
|
GM ≥0.26 corrected ODI Clarus
|
96% (51)
|
81% (167)
|
GM ≥0.5 ODI Platelia
|
92% (49)
|
75% (155)
|
GM ≥1.0 ODI Platelia
|
85% (45)
|
91% (188)
|
Probable IPA vs. no IPA classification after exclusion of GM as mycological criterion
|
|
Sensitivity (n=15)
|
Specificity (n=206)
|
GM ≥0.2 ODI Clarus
|
93% (14)
|
74 % (152)
|
GM ≥0.26 ODI Clarus
|
93% (14)
|
81% (167)
|
GM ≥0.5 ODI Platelia
|
93% (14)
|
76% (157)
|
GM ≥1.0 ODI Platelia
|
93% (14)
|
92% (190)
|
Abbreviations: GM= Galactomannan; BALF = bronchoalveolar lavage fluid; IPA = invasive pulmonary aspergillosis; ODI= optical density index
|