SUMMARY
Objectives:
Severe and prolonged neutropenia is associated with poor outcomes of invasive pulmonary aspergillosis (IPA) in leukemia patients. Given the high frequency of IPA in patients with relapsed/refractory leukemia, we studied the association of peripheral blood blast burden (blastemia), IPA outcomes, and antifungal immune failure, even without neutropenia.
Methods:
We retrospectively reviewed adult patients with acute leukemia (AL) or myelodysplastic syndrome and culture-positive proven/probable IPA (2011–2022). Blast and neutropenia indices were calculated and incorporated into multi-variable prognostic models. The impact of blasts on immune cell-mediated fungal clearance was studied in vitro.
Results:
Among 74 patients, 69% had neutropenia and 57% had blastemia at IPA diagnosis. Blast index ≥90 at IPA diagnosis and ≥3 lines of prior chemotherapies were independent predictors of 42-day mortality and early antifungal treatment failure. Leukemic blasts had minimal immune activity against Aspergillus fumigatus and impaired fungal inhibition by peripheral blood mononuclear cells.
Conclusion:
Blastemia is common in contemporary leukemia patients with IPA and is a significant risk factor for poor IPA outcomes, possibly due to interference with fungal clearance by immune cells. Therefore, blastemia should be considered as a risk stratification parameter in future mycology trials and as an experimental variable in preclinical IPA models.
Keywords: Blastemia, Acute leukemia, Invasive pulmonary aspergillosis
Introduction
Severe and prolonged neutropenia is a well-known risk factor for invasive pulmonary aspergillosis (IPA) in patients with hematologic malignancies (HM), especially acute leukemia (AL), and is also associated with poor response to antifungal therapy, even in the era of potent mold-active triazoles.1 In view of their prognostic implications, active leukemia and/or neutropenia have been proposed for risk stratification and used in modern randomized control studies of IPA.2–4 However, it remains unclear whether active leukemia status, independent of neutropenia status, has prognostic ramifications in IPA patients.
Over the last two decades, modern chemotherapies for AL have evolved dramatically, enabling prolonged survival rates, even in patients with difficult-to-treat AL.5 Recent studies revealed a sharp increase in relapsed or refractory (R/R) leukemia patients developing fungal pneumonia, including IPA,6 even in those who received mold active antifungals.7,8 As the number AL patients with significant peripheral blood (PB) blast burden (hereinafter, blastemia), with or without neutropenia, is increasing, we sought to explore the prognostic implications of blastemia in patients with AL and IPA.
Although the interference of leukemic blasts with immune defense against opportunistic pathogens is poorly understood, leukemic blasts have been shown to evade immune destruction and contribute to dysregulated cytokine production, release of anti-inflammatory mediators, and suppression of cellular immunity.9 We therefore hypothesized that blastemia is an additional predictor of poor fungal control, suboptimal response to antifungal therapy (as it has been the case for neutropenia-related treatment failure), and excess mortality in IPA patients, even in the absence of neutropenia. To test this hypothesis, we studied the prognostic significance of blastemia in IPA patients with AL and performed proof-of-concept experiments to explore impairment of antifungal immunity against Aspergillus fumigatus by leukemic blasts in vitro.
Methods
Retrospective study
We retrospectively reviewed electronic medical records of all consecutive adult (≥18-year-old) patients with AL or myelodysplastic syndrome (MDS) diagnosed with culture-positive proven or probable IPA at MD Anderson Cancer Center between January 2011 and February 2022. Culture-documented proven or probable IPA was defined according to the European Organization for Research and Treatment of Cancer and Mycoses Study Group Education and Research Consortium criteria.10 Only the first IPA episode for each patient was evaluated. Demographic data, underlying diseases, leukemia status including prior leukemia therapies, and laboratory results were reviewed. Variables are detailed and defined in Supplemental Table 1.
Blastemia and neutropenia indices
To quantify blastemia and neutropenia over time, we calculated indices for blastemia (B-index) and neutropenia (D-index) from admission to IPA diagnosis and from IPA diagnosis to 42 days after IPA diagnosis (Fig. 1A). Time of IPA diagnosis was defined as collection date of the first respiratory culture yielding Aspergillus species. D-index was defined as the area under the curve (AUC) of absolute neutrophil counts (ANC) below 500/μL over time, as previously described.11 B-index was calculated using an adapted version of this AUC-based approach and was defined as the AUC of absolute PB blast counts over time. To adjust for different hospitalization periods (i.e., availability of daily complete blood cell [CBC] count and blast burden data) and early mortality, we used three different normalization approaches, (i) raw AUCs (B- and D-indices), (ii) AUCs divided by number of hospital days (h-B- and h-D-indices), and (iii) AUCs divided by total days of blastemia (d-B-index) or neutropenia (d-D-index), respectively (Fig. 1A).
Fig. 1.
Blastemia is associated with antifungal treatment failure and poor outcomes of IPA in patients with acute leukemia or myelodysplastic syndrome. (A) Approach for calculation and normalization of B- and D-indices from admission to IPA diagnosis (period 1) and from IPA diagnosis to day 42 after IPA diagnosis or earlier death (period 2). (B-C) Distributions of B-indices in period 1 (B) and h-D-indices in period 2 (C) according to 42-day survival outcome. Mann-Whitney U test. (D) Rank correlation matrix of B- and D-indices across review periods and normalization approaches. Color coding and asterisks indicate Spearman coefficients and statistical significance, respectively. (E-F) Forty-two-day survival curves according to B-index in period 1 (E) and the number of prior lines of leukemia chemotherapy (F). Error bands represent 95% confidence intervals. Mantel-Cox log-rank test. (G) Distributions of B-indices in period 1 according to response to antifungal therapy at day 14 after IPA diagnosis. Mann-Whitney U test. (B-G) * p < 0.05. ** p < 0.01, *** p < 0.001. Abbreviation: IPA = invasive pulmonary aspergillosis.
Outcomes
All-cause mortality was assessed for 42 days from IPA diagnosis. In addition, antifungal treatment response was evaluated as complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD) on day 14 after IPA diagnosis according to MSG/EORTC consensus criteria.12 Antifungal treatment failure was defined as SD, PD, or death.12
Enrichment and expansion of CD34+ blasts from MDS/AML patients
Bone marrow mononuclear cells (BMMNCs) were isolated from anonymized acute myeloid leukemia (AML) or MDS patient bone marrow aspirates using gradient separation with Ficoll-Paque Premium (Fisher Scientific, #45–001–752). CD34+ cells were sorted from BMMNCs by magnetic cell sorting using the CD34 Microbead Kit UltraPure (Miltenyi Biotec, #130-100-453). Purified CD34+ cells were then cultured for 7 days in stem cell expansion medium. StemSpan SFEM II (Stem Cell Technologies, #09605) was supplemented with 2 mM GlutaMAX (Thermo Fisher Scientific, #35050061), 50 ng/ml thrombopoietin (TPO) (R&D systems, #288-TP-025), 100 ng/ml SCF (R&D systems, #255-SC-010), and 40 ng/ml FLT-3 ligand (R&D systems, #308-FK-005).
In-vitro experiments
For direct co-culture experiments, 200,000 A. fumigatus AF-293 conidia were co-cultured with an equal number of enriched CD34+ cells from bone marrow aspirates of patients with AML or MDS (Supplemental Table 2), C1498 cells (murine AML blasts; American Type Culture Collection, #TIB-49), or human peripheral blood mononuclear cells (PBMCs; IQ Biosciences, #IQB-PBMC102) in 24-well plates (Corning/Falcon, #353047). The total culture volume was adjusted to 1 ml with Roswell Park Memorial Institute (RPMI)–1640 medium (Gibco, #A10491–01) supplemented with 10% fetal bovine serum (FCS; Atlas Biologicals, #F-0500-A). Plates were incubated for 20 h at 37 °C, 5% CO2. After washing the plates with ice-cold double-distilled water (ddH2O) thrice, fungal growth and metabolism were measured using a 2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide (XTT)-based colorimetric assay.13 Additionally, mycelial growth was assessed by live imaging with automated quantification of hyphal length (IncuCyte NeuroTrack)14 using co-cultures of 200 A. fumigatus AF-293 conidia expressing green fluorescent protein (GFP) and 20,000 (effector/target ratio [E:T] 100) or 100,000 (E:T 500) blasts or PBMCs in 96-well plates (Corning/Falcon, #353072). Plates were imaged hourly for 20 h at 37 °C, 5% CO2 with an IncuCyte ZOOM system (Sartorius).
To study the impact of AML/MDS blasts and their secreted mediators on PBMC-mediated inhibition of A. fumigatus, a Transwell assay was performed. The lower compartment contained 200,000 A. fumigatus AF-293 conidia and 200,000 PBMCs in 600 μL RPMI 1640 medium + 10% FCS in 24-well plates. The upper compartment contained either 100,000 blasts in 100 μL RPMI/FCS or cell-free medium in 0.4 μm-pore polyester membrane inserts (Corning, #3470). Inserts were removed after 6 h of incubation at 37 °C, 5% CO2 and 400 μL of fresh medium was added to each well. After incubation for another 14 h at 37 °C, 5% CO2 (total co-culture period, 20 h), plates were washed thrice with ice-cold ddH2O water and fungal metabolism was assessed by XTT assay.
Statistical analysis
Categorical variables were compared using chi-square or Fisher’s exact tests, as appropriate. Continuous variables were compared using Student’s t-test or Wilcoxon rank-sum test for two-group comparisons or one-way analysis of variance with Tukey’s or Dunnett’s post-test for multi-group comparisons. Correlation between raw and normalized B- and D-indices was analyzed using Spearman’s rank correlation coefficient. Receiver operating characteristic (ROC) curves and Youden index analyses were used to determine optimal cutoffs for B- and D-indices according to 42-day survival status. Cochran-Armitage trend test was performed to analyze associations between treatment response and mortality. For survival trend analysis, survival curves were analyzed using the Kaplan-Meier method and compared using Mantel-Cox log-rank test. Multivariable logistic regression analyses were used to identify independent predictors of mortality by day 42 and antifungal treatment failure by day 14 after IPA diagnosis, respectively. Data analysis was performed using SAS version 9.4 (SAS Institute Inc.) and Prism version 10 (GraphPad Software).
Ethics statement
This study was approved by MD Anderson’s Institutional Review Board (#2024–1612) and conducted in accordance with the Declaration of Helsinki. Written informed consent was waived for anonymized chart review. Bone marrow aspirates were collected under protocols Lab01–473 and PA19–0345.
Results
Study cohort, underlying malignancies, and comorbidities
After excluding repeated IPA episodes in the same patients (n=3), hematological diseases other than AL or MDS (n=73), and cases without complete blood cell counts on the day of IPA diagnosis (n=14), 74 AL/MDS patients were included in the final analysis.
Median age was 62 (range 18–89), and 57% (n=42) of patients were male (Table 1). AML was the most common underlying leukemia (n=44, 59%), followed by acute lymphocytic leukemia (ALL) (n=12), and MDS (n=10). Sixty-one percent of AML patients (27/44) had molecular profiles consistent with the adverse European LeukemiaNet (ELN) risk category. Twenty-four patients (32%) had a history of allogeneic hematopoietic cell transplantation (HCT). At the time of IPA diagnosis, 85% of patients were undergoing treatment for their underlying AL. Over a third of patients (37%) had received 3 or more lines of chemotherapy by the time of IPA diagnosis.
Table 1.
Clinical characteristics according to survival status at day 42 after IPA diagnosis.
| Characteristics | All patients (n=74) | 42-day survival (n=34)a | 42-day death (n=39) | p-value |
|---|---|---|---|---|
| Age (years), median (range) | 62 (18–89) | 58 (18–82) | 65 (23–78) | 0.027 |
| Gender, male, n (%) | 42 (57%) | 17 (50%) | 24 (62%) | 0.322 |
| Race, n (%) | 0.977 | |||
| White | 48 (65%) | 21 (62%) | 26 (67%) | |
| Hispanic | 11 (15%) | 6 (18%) | 5 (13%) | |
| Asian | 9 (12%) | 4 (12%) | 5 (13%) | |
| Black | 6 (8%) | 3 (9%) | 3 (8%) | |
| Type of hematological malignancy, n (%) | 0.365 | |||
| Myeloid | 60 (81%) | 29 (85%) | 30 (77%) | |
| Lymphoid | 14 (19%) | 5 (15%) | 9 (23%) | |
| Lines of chemotherapy ≥ 3, n (%) | 27/73 (37%) | 6/33 (18%) | 21 (54%) | 0.002 |
| Cancer status, n (%) | 0.011 | |||
| Active | 63 (85%) | 25 (74%) | 37 (95%) | |
| Remission | 11 (15%) | 9 (26%) | 2 (5%) | |
| ELN 2022 for AML patients, n (%)b | 0.819 | |||
| Favorable | 5/44 (11%) | 3/20 (15%) | 2/24 (8%) | |
| Intermediate | 12/44 (27%) | 5/20 (25%) | 7/24 (29%) | |
| Adverse | 27/44 (61%) | 12/20 (60%) | 15/24 (63%) | |
| Prior allo-HCT, n (%) | 24 (32%) | 10 (29%) | 14 (36%) | 0.556 |
| Significant GvHD at PIA diagnosis, n (%) | 13 (18%) | 7 (21%) | 6 (15%) | 0.562 |
| Diabetes mellitus, n (%) | 11 (15%) | 5 (15%) | 6 (15%) | 0.936 |
| Chronic kidney disease, n (%) | 5 (7%) | 1 (3%) | 4 (10%) | 0.364 |
| COPD, n (%) | 6 (8%) | 2 (6%) | 3 (8%) | > 0.999 |
| Liver disease, n (%) | 13 (18%) | 4 (12%) | 9 (23%) | 0.208 |
| Hypoalbuminemia (< 3 mg/dL) at IPA diagnosis, n (%) | 41 (55%) | 17 (50%) | 24 (62%) | 0.322 |
| Previous steroid use (≥ 3 weeks), n (%) | 12 (16%) | 7 (21%) | 5 (13%) | 0.372 |
| Previous antifungal exposure, n (%) | 64 (86%) | 27 (79%) | 36 (92%) | 0.173 |
| Type of IPA, n (%) | > 0.999 | |||
| Proven | 4 (5%) | 2 (6%) | 2 (5%) | |
| Probable | 70 (95%) | 32 (94%) | 37 (95%) | |
| Causative Aspergillus species, n (%)c | 0.195 | |||
| A. fumigatus | 32/72 (44%) | 15 (44%) | 17/37c (46%) | |
| A. flavus | 10/72 (14%) | 6 (18%) | 4/37 (11%) | |
| A. terreus | 10/72 (14%) | 6 (18%) | 4/37 (11%) | |
| A. niger | 13/72 (18%) | 7 (21%) | 5/37 (14%) | |
| Other rare Aspergillus spp. | 5/72 (7%) | 0 (0%) | 5/37 (14%) | |
| Mixed Aspergillus infectionsd | 2/72 (3%) | 0 (0%) | 2/37 (5%) | |
| Days from admission to IPA diagnosis, median (IQR) | 3 (2–13) | 2 (1–4) | 7 (3–20) | < 0.001 |
| Length of hospital stay (days), median (IQR) | 16 (8–33) | 13 (7–33) | 17 (11–33) | 0.407 |
Abbreviations: Allo-HCT = allogeneic hematopoietic cell transplantation, AML = acute myelogenous leukemia, ELN = European LeukemiaNet, COPD = chronic obstructive lung disease, GvHD = graft-versus-host disease, IPA = invasive pulmonary aspergillosis, IQR = interquartile range.
Follow up was lost in one patient on day 12 after invasive aspergillosis diagnosis.
Only for AML patients (n=44).
Not identified to species level in 2 patients.
A. flavus + A. terreus (n=1), A. fumigatus + A. terreus + other rare Aspergillus spp. (n=1).
The median length of hospital stay was 16 days (interquartile range [IQR], 8–33 days). Of note, IPA was the presumed reason for admission in most patients, with a median time from admission to IPA diagnosis of 3 days (IQR, 2–13 days). Other patient characteristics including comorbidities, causative Aspergillus species, and corticosteroid use are summarized in Table 1.
Univariate comparison of clinical variables according to 42-day outcome after IPA diagnosis
All-cause mortality at 42 days after IPA diagnosis was 53% (n=39/73, lost follow-up in one patient). Comparing 42-day survivors and patients deceased by day 42 after IPA diagnosis, median age was higher in non-survivors (65 vs. 58, p=0.027). Likewise, the proportion of patients not in remission of AL (95% vs. 74%, p=0.011) and those with 3 or more lines of chemotherapy (54% vs. 18%, p=0.002) were significantly higher among deceased patients compared to 42-day survivors (Table 1).
Association of blastemia and neutropenia with 42-day outcomes after IPA diagnosis
Next, we compared variables related to neutropenia and blastemia according to 42-day outcomes after IPA diagnosis (Table 2). Fifty-one patients (69%) had neutropenia and 42 (57%) had blastemia at IPA diagnosis. Among those, 37 patients had both neutropenia and blastemia, 5 had only blastemia, and 14 had only neutropenia. Blastemia and neutropenia status at IPA diagnosis was significantly associated with decreased 42-day survival after IPA diagnosis in both 4-group survival curve analysis (p=0.013, Fig. 2) and categorical analysis according to 42-day survival status (p=0.008, Table 2).
Table 2.
Comparison of variables related to blastemia and neutropenia according to 42-day survival status after IPA diagnosis.
| Characteristics | All patients (n=74) | 42-day survival (n=34)a | 42-day death (n=39)a | p-value |
|---|---|---|---|---|
| (A) Period 1: from admission to IPA diagnosis | ||||
| Groups by blastemia/neutropenia | 0.008 | |||
| Neither | 18 (24%) | 13 (38%) | 4 (10%) | |
| Neutropenia only | 14 (19%) | 8 (24%) | 6 (15%) | |
| Blastemia only | 5 (7%) | 2 (6%) | 3 (8%) | |
| Both blastemia and neutropenia | 37 (50%) | 11 (32%) | 26 (67%) | |
| Presence of blastemia, n (%) | 42 (57%) | 13 (38%) | 29 (74%) | 0.002 |
| B-index, median (IQR) | 82 (0–7581) | 0 (0–235) | 899 (0–18,444) | < 0.001 |
| d-B-index, median (IQR) | 25 (0–619) | 0 (0–116) | 180 (0–4538) | < 0.001 |
| h-B-index, median (IQR) | 6 (0–722) | 0 (0–28) | 145 (0–4538) | 0.001 |
| Days of blastemia, median (IQR) | 2 (0–4) | 0 (0–2) | 3 (0–6) | 0.002 |
| Neutropenia | ||||
| ANC ≤500/mm3 | 51 (69%) | 19 (56%) | 32 (82%) | 0.015 |
| ANC ≤100/mm3 | 32 (43%) | 10 (29%) | 22 (56%) | 0.020 |
| D-index, median (IQR) | 677 (0–2615) | 9 (0–1119) | 1338 (410–3848) | 0.001 |
| d-D-index, median (IQR) | 165 (0–373) | 9 (0–243) | 289 (130–405) | 0.002 |
| h-D-index, median (IQR) | 170 (0–386) | 5 (0–254) | 332 (27–421) | 0.003 |
| Days of neutropenia, median (IQR) | 3 (0–9) | 1 (0–4) | 5 (2–12) | 0.002 |
| (B) Period 2: from IPA diagnosis to day 42 | ||||
| Groups by blastemia/neutropenia, n (%) | 0.077 | |||
| Neither | 10 (14%) | 7 (21%) | 2 (5%) | |
| Neutropenia only | 13 (18%) | 6 (18%) | 7 (18%) | |
| Blastemia only | 9 (12%) | 6 (18%) | 3 (8%) | |
| Both blastemia and neutropenia | 42 (57%) | 15 (44%) | 27 (69%) | |
| Presence of blastemia, n (%) | 51 (69%) | 21 (62%) | 30 (77%) | 0.159 |
| B-index, median (IQR) | 50 (0–729) | 29 (0–457) | 102 (0–6074) | 0.321 |
| d-B-index, median (IQR) | 17 (0–234) | 13 (0–72) | 28 (0–675) | 0.278 |
| h-B-index, median (IQR) | 1 (0–45) | 1 (0–11) | 4 (0–190) | 0.136 |
| Days of blastemia, median (IQR) | 2 (0–5) | 2 (0–3) | 2 (0–6) | 0.436 |
| Blast kinetics after IPA diagnosis, n (%) | 0.022 | |||
| From No to No | 23 (31%) | 13 (38%) | 9 (23%) | |
| From Yes to No | 11 (15%) | 3 (9%) | 8 (21%) | |
| From No to Yes | 10 (14%) | 8 (24%) | 2 (5%) | |
| From Yes to Yes | 30 (41%) | 10 (29%) | 20 (51%) | |
| Existing neutropenia, n (%) | 55 (74%) | 21 (62%) | 34 (87%) | 0.012 |
| D-index, median (IQR) | 500 (0–2395) | 4 (0–3355) | 1310 (0–2355) | 0.529 |
| d-D-index, median (IQR) | 140 (0–435) | 4 (0–377) | 223 (0–462) | 0.140 |
| h-D-index, median (IQR) | 4 (0–14) | 0 (0–9) | 11 (0–51) | 0.006 |
| Days of neutropneia, median (IQR) | 2 (0–7) | 1 (0–8) | 4 (0–7) | 0.517 |
| Neutropenia kinetics after IPA diagnosis, n (%) | 0.077 | |||
| From No to No | 19 (26%) | 13 (38) | 5 (13) | |
| From Yes to No | 14 (19%) | 5 (15) | 9 (23) | |
| From No to Yes | 4 (5%) | 2 (6) | 2 (5) | |
| From Yes to Yes | 37 (50%) | 14 (41) | 23 (59) |
Abbreviations: ANC = absolute neutrophil count, IPA = invasive pulmonary aspergillosis, IQR = interquartile range.
Follow up was lost in one patient on day 10 after invasive aspergillosis diagnosis.
Fig. 2.
Comparison of survival trends in patients with acute leukemia or myelodysplastic syndrome and invasive pulmonary aspergillosis according to blastemia and neutropenia status. Mantel-Cox log-rank test. Censored subjects are indicated by tick marks.
Given this observation, we sought to obtain a more granular quantitative index of blastemia and neutropenia risk status over time by comparing B- and D-index according to 42-day survival after IPA diagnosis. Median B-index before IPA diagnosis (period 1) was significantly higher in deceased patients than in survivors (899; interquartile range [IQR] 0–18,444 vs. 0, IQR 0–235, p < 0.001) (Table 2, Fig. 1B). Likewise, median D-index before IPA diagnosis (period 1) was also significantly higher in deceased patients than in 42-day survivors (1338 [410−3848] vs. 9 [0−1119], p=0.001) (Table 2A). Significant differences were observed regardless of the normalization approach (i.e., raw AUCs, normalization by length of hospital stay, or normalization by duration of blastemia/neutropenia) (Table 2A). In contrast, for the period from IPA diagnosis until day 42 or earlier death (period 2), only persistent neutropenia at the end of follow-up and D-index normalized by length of hospital stay (h-D index), but not blastemia indices differed significantly between the two groups (Fig. 1C, Table 2B).
Multivariable models to identify independent predictors of 42-day outcomes after IPA diagnosis
Next, we selected the most informative indices for subsequent incorporation in multivariable models. Spearman correlation analyses revealed that all three B-indices and D-indices showed strong positive correlation with one another within each review period (period 1 and 2), with correlation coefficients ranging from 0.986 to 0.993 among the three B-indices, and 0.824 to 0.969 among the D-indices (Fig. 1D).
Given the multicollinearity of the indices, we focused on a single B- and D-index in each period for further prognostic modeling. Therefore, ROC analyses were performed to identify the normalization approach resulting in the highest AUC values for B- and D-indices in each review period to predict 42-day mortality. Based on this criterion, we selected the (raw) B- and D-indices in period 1 and h-B-/h-D-indices in period 2, respectively (Supplementary Fig. 1). The optimal cutoff values for these indices to distinguish between 42-day survival and death were determined using ROC curves and Youden Index analysis. Using these thresholds, the classification performance of the prioritized indices for predicting 42-day survival outcomes after IPA diagnosis was evaluated by univariate association analysis (Supplementary Table 3).
We then developed two multivariable logistic regression models to identify independent predictors of 42-day mortality, (i) a retroactive model considering all variables and B-/D- indices both before and after IPA diagnosis (Table 3A); and (ii) a model including only variables available at the time of IPA diagnosis for prognostic risk stratification (Table 3B). In the first model, ≥3 prior lines of chemotherapy (adjusted odds ratio [aOR] 4.15; 95% confidence interval [CI] 1.17–14.72, p=0.028), B-index > 90 between admission and IPA diagnosis (aOR 4.20; 95% CI 1.26–13.96, p=0.019), and h-D-Index > 12 between IPA diagnosis and day-42 (aOR 15.24; 95% CI 2.38–97.46, p=0.004) were significant independent predictors of 42-day mortality (Table 3A). Notably, the first two variables, i.e., ≥3 prior lines of chemotherapy (aOR 5.38; 95% CI 1.64–17.65, p=0.006) and B-index > 90 between admission and IPA diagnosis (aOR 6.13; 95% CI 2.03–18.49, p=0.001) were identical in the prognostic model (Table 3B). Significant associations of these variables with IPA outcomes were further corroborated by survival curve analysis, with highly significant differences in survival trends and non-overlapping 95% CIs by day 42 after IPA diagnosis (Fig. 1E–F).
Table 3.
Predictors of 42-day mortality (A-B) and antifungal treatment failure by day 14 (C) after IPA diagnosis.
| (A) Retroactive model for 42-day mortality | |||
|---|---|---|---|
| Independent predictors | aOR | 95% CI | p-value |
| Lines of chemotherapy ≥ 3 | 4.15 | 1.17–14.72 | 0.028 |
| B-Index before IPA diagnosis > 90 | 4.20 | 1.26–13.96 | 0.019 |
| h-D-Index from IPA diagnosis up to Day 42 > 12 | 15.24 | 2.38–97.46 | 0.004 |
| (B) Prognostic model for 42-day mortality | |||
| Independent predictors | aOR | 95% CI | p-value |
| Lines of chemotherapy ≥ 3 | 5.38 | 1.64–17.65 | 0.006 |
| B-Index before IPA diagnosis > 90 | 6.13 | 2.03–18.49 | 0.001 |
| (C) Antifungal treatment failure by day +14 | |||
| Independent predictor | aOR | 95% CI | p-value |
| Lines of chemotherapy ≥ 3 | 11.00 | 1.72 – 70.25 | 0.011 |
| B-Index before IPA diagnosis > 90 | 6.85 | 1.69 – 27.70 | 0.007 |
Abbreviations: aOR = adjusted odds ratio, CI = confidence interval, IPA = invasive pulmonary aspergillosis.
Collectively, these findings indicate that both refractory leukemia after multiple lines of prior chemotherapy and significant blastemia at IPA diagnosis strongly and significantly affect IPA outcomes, with neutropenia recovery during the follow-up period being an additional late modulator of IPA outcomes in the retroactive model.
Association of clinical variables with early antifungal treatment response
Because predictors of 42-day all-cause mortality might be confounded by deterioration of the underlying malignancy rather than IPA severity, we additionally evaluated the association of clinical variables, B-, and D-indices with early response to antifungal therapy by day 14 after IPA diagnosis. Notably, classification of response to antifungal therapy by day 14 was significantly associated with 42-day mortality outcomes (p < 0.001, Supplemental Table 4).
Interestingly, independent predictors of 14-day antifungal treatment failure (Table 3C), i.e., ≥3 prior lines of chemotherapy (aOR 11.00; 95% CI 1.72–70.25, p=0.011) and B-index > 90 between admission and IPA diagnosis (aOR 6.85; 95% CI 1.69–27.70, p=0.007), matched those identified in our prognostic model for 42-day mortality after IPA diagnosis. Consistent with this model, patients with early antifungal treatment failure had significantly higher B-indices before/at IPA diagnosis (period 1) than those with at least PR to antifungal therapy by day 14 after IPA diagnosis (p=0.005, Fig. 1G). These findings further underscore an association of refractory leukemia and significant blastemia at IPA diagnosis with impaired fungal control and poor early response to antifungal therapy.
Effect of blast cells on A. fumigatus proliferation and PBMC-mediated fungal inhibition in vitro
The adverse impact of blastemia on early fungal control in our clinical review supported our hypothesis that blasts might interfere with antifungal immunity. To further test this hypothesis, we examined ex vivo how blast cells or blast-derived mediators impact effector cell killing of A. fumigatus. While co-culture of A. fumigatus conidia with PBMCs led to considerable inhibition of fungal XTT metabolism compared to A. fumigatus alone (mean 0.17, p < 0.001), neither enriched CD34+ blasts from AML/MDS patients (mean 0.94) nor murine C1498 AML cells (mean 0.91) significantly altered fungal XTT metabolism (Fig. 3A). IncuCyte imaging confirmed that only PBMCs (mean 0.16, p=0.012 at E:T 100; mean 0.02, p < 0.001 at E:T 500) significantly reduced hyphal proliferation, whereas neither C1498 cells nor human blasts had notable effects (Fig. 3B). Likewise, addition of blasts to the upper compartment of a Transwell assay (Fig. 3C) did not inhibit fungal proliferation in the lower compartment without immune effector cells (Fig. 3D). However, presence of blasts in the upper compartment significantly attenuated PBMC-mediated inactivation of A. fumigatus (mean 0.63 vs. 0.16 compared to A. fumigatus alone = 1.00, p=0.019), as determined by XTT metabolism (Fig. 3D).
Fig. 3.
MDS/AML blasts have minimal immune activity against Aspergillus fumigatus and can interfere with the host’s antifungal immune response. (A-B) Fungal proliferation after 20 h of co-culture of A. fumigatus conidia with AML/MDS blasts and PBMCs as measured by XTT metabolism (A) and IncuCyte imaging (B). (C) Schematic summarizing the Transwell assay setup, as detailed in Materials & Methods. (D) Impact of blasts on PBMC-mediated inhibition of A. fumigatus in a Transwell assay, as determined by XTT metabolism. (A-B, D) The dashed line represents the “A. fumigatus only” control (= 1). Co-culture conditions were compared to “A. fumigatus only” using one-way ANOVA with Dunnett’s post-test. # p < 0.05, ## p < 0.01, ### p < 0.001. Different co-culture conditions were compared using one-way ANOVA with Tukey’s post-test. * p < 0.05. ** p < 0.01, *** p < 0.001. (E) Working model. Abbreviations: AML = acute myeloid leukemia, ANOVA = analysis of variance, MDS = myelodysplastic syndrome, PBMC(s) = peripheral blood mononuclear cell(s), UC = upper compartment, XTT = 2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide.
Collectively, these proof-of-concept experiments and our retrospective clinical data support a model whereby blasts, while not contributing to fungal clearance themselves, impair antifungal host immunity (Fig. 3E), which contributes to poor early response to antifungal therapy. Impaired early fungal clearance combined with sustained neutropenia in turn drives poor IPA outcomes.
Discussion
The modern era of cancer medicine poses new challenges in managing invasive mold infections, driven by an increasing proportion of patients with R/R leukemia and those undergoing multiple lines of (salvage) chemotherapy. These complexities challenge the traditional paradigms of patient stratification based solely on cytopenia and underscore the need for more refined risk models integrating leukemia activity as a critical factor. In this study, we demonstrated that blastemia is prevalent among leukemia patients with IPA, with 57% presenting with leukemic blast cells in peripheral blood at IPA diagnosis. Furthermore, blastemia emerged as a significant independent risk factor for poor IPA outcomes, irrespective of neutropenia status, and was an important predictor of early antifungal treatment failure following IPA diagnosis.
To quantify the effect of blastemia over time, we adapted an AUC method previously described as a longitudinal surrogate of neutropenia.15,16 In many cancer centers, blast count is readily obtained from daily CBC results, highlighting the clinical feasibility of such indices. Notably, B-indices differed significantly based on 42-day survival status, with higher median values observed in non-survivors, indicating a longer duration and greater burden of blastemia in patients with poor outcomes. Although further investigations are warranted to define “dose-response” relationship between blastemia and IPA outcomes in greater detail, the low cutoff value identified in our ROC analysis (i.e. B-index > 90 from admission to IPA diagnosis) suggests that even a relatively low blast burden (e.g., a white blood cell count of 1000 cells/mm3 with 9% PB blasts on a single CBC result) at IPA diagnosis can adversely affect response to antifungal therapy and IPA outcomes.
Timing of evaluation is critical for applying prognostic models effectively. While prognosis often centered around IPA diagnosis, it remains unclear whether pre- or post-diagnosis factors carry greater prognostic weight. To address this, we calculated the B- and D-index for two periods: admission to IPA diagnosis (period 1) and the 42-day follow-up period after IPA diagnosis (period 2). Interestingly, the blast burden in period 1 was more informative for risk assessment and remained as an independent predictor of 42-day mortality in the multivariable models considering both periods (Table 3A). If validated in future multicenter studies, this finding suggests potential suitability of the early B-index as a component of host-based risk models to guide IPA management. In contrast, neutropenia status (h-D-index > 12 [aOR 15.24]) proved to be more informative during the 42-day period following IPA diagnosis (period 2), consistent with prior literature indicating that prolonged neutropenia, as opposed to neutrophil recovery, is a key determinant of patient prognosis after invasive mold infections.17 Collectively, these findings suggest that different components of immune impairment (i.e., cytopenia and blastemia) might dynamically alter the net state of immunosuppression during and after IPA infection.
To further validate the link between blastemia and impairment of antifungal immunity, we conducted proof-of-concept experiments using a Transwell model. These experiments demonstrated reduced fungal inhibition by PBMCs in the presence of blasts, supporting our hypothesis that blastemia interferes with fungal clearance. This likely contributes to the significantly inferior antifungal treatment response observed in patients with B-index > 90 from admission to IPA diagnosis compared to those with no or negligible blast counts in PB.
Detailed mechanistic investigations were beyond the scope of this clinically focused manuscript, as they would require comprehensive in vitro studies incorporating multiple immune effector cell populations and blasts from various leukemia subtypes, along with evidence from pertinent preclinical models. Nevertheless, several plausible mechanisms may underpin our clinical and experimental observations. Notably, the impaired antifungal activity of PBMCs observed in the Transwell assay, despite the absence of direct cellular contact between blasts and PBMCs, suggests that secreted blast-derived mediators are a likely mechanistic link. For instance, several micro-RNAs released by AML and ALL progenitors have been associated with impaired natural killer cell functionality, T-cell apoptosis, altered T-cell polarization, and recruitment of myeloid-derived suppressor cells.18,19 Similarly, both pro- and anti-inflammatory cytokines produced by blast cells have been implicated in immune modulation in patients with myeloid malignancies.20,21 Particularly, elevated blood levels of Treg-associated cytokines such as IL-10 and TGF-β in AML patients may provide a potential link to the suppression of antifungal immunity.22,23 Additionally, enzymes such as arginase II, secreted by leukemic progenitor cells, have been associated with impaired T-cell proliferation and polarization of monocytes toward a suppressive M2 phenotype.24,25 Moreover, mechanisms involving direct intercellular crosstalk, such as overexpression of checkpoint markers, might contribute to modulation of antifungal immunity in vivo. However, these interactions would not be captured in the Transwell assay design, which was necessary due to the use of allogeneic PBSCs in combination with blast cells. Conversely, it is also conceivable that defective innate immune defenses are not only a consequence but also a driver of leukemic activity.26,27 Thus, blastemia might serve as both a surrogate marker for and a contributing factor to an overall dysfunctional host environment.
The receipt of 3 or more lines of chemotherapy was identified as an independent risk factor for 42-day mortality (aOR 4.15 in retroactive model and aOR 5.38 in prognostic model), even with the B-index included as a marker of leukemia activity. This reflects that refractoriness of underlying HMs, with cumulative toxicities, immunosuppression, and uncontrolled disease, may further compound impairment of antifungal immunity and might play a potential role in determining IPA outcome. Thus, our data affirm the need to incorporate assessment of both HM activity and refractoriness in emerging clinical and immunological prognostic risk models to guide individualized management of invasive mold infections, including host-targeted immunotherapy.28,29
Our study has limitations that may affect its generalizability. As GM has cross-reactivity with other Hyalohyphomycetes,30 we focused only on culture-positive IPA cases, limiting the applicability of our findings to IPA cases diagnosed solely based on GM positivity. Additionally, although we did not find an association between ELN classification of AML patients and IPA outcomes (Table 1), the limited number of cases and heterogeneity of both underlying hematologic malignancies and genetic aberrations precluded us from establishing detailed associations between specific cytogenetic markers or mutations, response to antifungal therapy, and 6-week survival. Likewise, although the leukemia treatment approach is not dictated by the presence of blastemia, provided that the patient has evidence of active leukemia in the bone marrow, the tremendous heterogeneity of clinical scenarios precluded us from evaluating the impact of chemotherapy regimens on IPA outcomes and potential drug-drug interactions with antifungal therapy. Furthermore, the single-center, retrospective nature of our study, the relatively small sample size limited to patients with AL, and our focus solely on IPA – the most common manifestation of invasive aspergillosis – might pose challenges in extrapolating our findings to other centers with different diagnostic and therapeutic approaches for IPA in patients with AL or other hematologic conditions (e.g., lymphoma).
In conclusion, blastemia is common in contemporary AL patients with IPA and is a significant independent risk factor for poor IPA outcome and failure of early antifungal therapy, possibly due to interference with fungal clearance by immune effector cells. In addition to prognostication of patients admitted with IPA, this observation could have significant implications for stratification of leukemia patients in future mycology trials. Antifungal therapy trials have historically focused on patients with less heavily pretreated hematologic diseases and have therefore been prone to under-representing the most severely immunosuppressed populations. To better evaluate antifungal efficacy in these higher-risk patients, factors such as blastemia and prior lines of chemotherapy should be considered as stratification variables. Likewise, given the potential interference of blasts with antifungal immune defense, blastemia might be a critical stratification variable for future trials of antifungal immunotherapies (e.g., cytokines or checkpoint inhibitors).29,31 As several investigational drugs are being introduced for the treatment of IPA,32 more granular understanding of the independent impact of blastemia (irrespective of neutropenia) for early risk stratification might be important. Finally, our data underscores the need to incorporate active leukemia and blastemia in preclinical IPA models to better explore the immunopathogenesis of Aspergillus and the impact of active or relapsing leukemia on response to antifungal therapy and investigational immunomodulators.
Supplementary Material
Appendix A. Supporting information
Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jinf.2025.106535.
Acknowledgements
This work was supported in part by the National Institute of Allergy and Infectious Diseases Grant #R03-AI-166285 (S.W., D.P.K.), the MD Anderson Cancer Center Institutional Research Grant #2021-00060204 (S.W.) and the Robert C. Hickey Chair endowment (D.P.K.).
Funding
The views expressed in this publication are those of the authors and not necessarily those of the NIHR, the Department of Health and Social Care or the UKHSA.
Footnotes
Declaration of Competing Interest
S.-Y.C. received consulting fees from Takeda and speaking fees from Pfizer and Merck. R.E.L. has served on advisory boards for Gilead Inc, F2G, and Basilea, and received speaking fees from Gilead and Avir. D.P.K. received research support from Gilead Sciences and Astellas Pharma; received consultant fees from Astellas Pharma, Matinas, Basilea, Knight, Inc., and Gilead Sciences; and is a member of the data review committee for Cidara Therapeutics, AbbVie, and the Mycoses Study Group.
References
- 1.Kontoyiannis DP, Selleslag D, Mullane K, Cornely OA, Hope W, Lortholary O, et al. Impact of unresolved neutropenia in patients with neutropenia and invasive aspergillosis: a post hoc analysis of the SECURE trial. J Antimicrob Chemother 2018;73(3):757–63. [DOI] [PubMed] [Google Scholar]
- 2.Herbrecht R, Bories P, Moulin JC, Ledoux MP, Letscher-Bru V. Risk stratification for invasive aspergillosis in immunocompromised patients. Ann N Y Acad Sci 2012;1272:23–30. [DOI] [PubMed] [Google Scholar]
- 3.Maertens JA, Raad II, Marr KA, Patterson TF, Kontoyiannis DP, Cornely OA, et al. Isavuconazole versus voriconazole for primary treatment of invasive mould disease caused by Aspergillus and other filamentous fungi (SECURE): a phase 3, randomised-controlled, non-inferiority trial. Lancet 2016;387(10020):760–9. [DOI] [PubMed] [Google Scholar]
- 4.Maertens JA, Rahav G, Lee DG, Ponce-de-León A, Ramírez Sánchez IC, Klimko N, et al. Posaconazole versus voriconazole for primary treatment of invasive aspergillosis: a phase 3, randomised, controlled, non-inferiority trial. Lancet 2021;397(10273):499–509. [DOI] [PubMed] [Google Scholar]
- 5.DiNardo CD, Wei AH. How I treat acute myeloid leukemia in the era of new drugs. Blood 2020;135(2):85–96. [DOI] [PubMed] [Google Scholar]
- 6.Aldoss I, Dadwal S, Zhang J, Tegtmeier B, Mei M, Arslan S, et al. Invasive fungal infections in acute myeloid leukemia treated with venetoclax and hypomethylating agents. Blood Adv 2019;3(23):4043–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Rausch CR, DiPippo AJ, Bose P, Kontoyiannis DP. Breakthrough fungal infections in patients with leukemia receiving isavuconazole. Clin Infect Dis 2018;67(10):1610–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Boutin CA, Durocher F, Beauchemin S, Ziegler D, Abou Chakra CN, Dufresne SF. Breakthrough invasive fungal infections in patients with high-risk hematological disorders receiving voriconazole and posaconazole prophylaxis: a systematic review. Clin Infect Dis 2024;79(1):151–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Khaldoyanidi S, Nagorsen D, Stein A, Ossenkoppele G, Subklewe M. Immune biology of acute myeloid leukemia: implications for immunotherapy. J Clin Oncol 2021;39(5):419–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Donnelly JP, Chen SC, Kauffman CA, Steinbach WJ, Baddley JW, Verweij PE, et al. Revision and update of the consensus definitions of invasive fungal disease from the European Organization for Research and Treatment of Cancer and the Mycoses Study Group Education and Research Consortium. Clin Infect Dis 2020;71(6):1367–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Portugal RD, Garnica M, Nucci M. Index to predict invasive mold infection in high-risk neutropenic patients based on the area over the neutrophil curve. J Clin Oncol 2009;27(23):3849–54. [DOI] [PubMed] [Google Scholar]
- 12.Segal BH, Herbrecht R, Stevens DA, Ostrosky-Zeichner L, Sobel J, Viscoli C, et al. Defining responses to therapy and study outcomes in clinical trials of invasive fungal diseases: Mycoses Study Group and European Organization for Research and Treatment of Cancer Consensus Criteria. Clin Infect Dis 2008;47(5):674–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kontoyiannis DP, Georgiadou SP, Wierda WG, Wright S, Albert ND, Ferrajoli A, et al. Impaired bactericidal but not fungicidal activity of polymorphonuclear neutrophils in patients with chronic lymphocytic leukemia. Leuk Lymphoma 2013;54(8):1730–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wurster S, Kumaresan PR, Albert ND, Hauser PJ, Lewis RE, Kontoyiannis DP. Live monitoring and analysis of fungal growth, viability, and mycelial morphology using the IncuCyte NeuroTrack processing module. mBio 2019;10(3):e00673–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kanda Y, Kimura SI, Iino M, Fukuda T, Sakaida E, Oyake T, et al. D-index–guided early antifungal therapy versus empiric antifungal therapy for persistent febrile neutropenia: a randomized controlled noninferiority trial. J Clin Oncol 2020;38(8):815–22. [DOI] [PubMed] [Google Scholar]
- 16.Rattanathammethee T, Munsamai K, Punnachet T, Hantrakun N, Piriyakhuntorn P, Hantrakool S, et al. D-index and invasive fungal infections (IFIs) in adult acute myeloid leukemia (AML) patients with the first episode of febrile neutropenia. PLoS One 2023;18(5):e0286089. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Pagano L, Caira M, Picardi M, Candoni A, Melillo L, Fianchi L, et al. Invasive aspergillosis in patients with acute leukemia: update on morbidity and mortality—SEIFEM-C report. Clin Infect Dis 2007;44(11):1524–5. [DOI] [PubMed] [Google Scholar]
- 18.Del Gaizo M, Sergio I, Lazzari S, Cialfi S, Pelullo M, Screpanti I, et al. MicroRNAs as modulators of the immune response in T-cell acute lymphoblastic leukemia. Int J Mol Sci 2022;23(2):829. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Otmani K, Rouas R, Lagneaux L, Krayem M, Duvillier H, Berehab M, et al. Acute myeloid leukemia-derived exosomes deliver miR-24–3p to hinder the T-cell immune response through DENN/MADD targeting in the NF-κB signaling pathways. Cell Commun Signal 2023;21(1):253. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Luciano M, Krenn PW, Horejs-Hoeck J. The cytokine network in acute myeloid leukemia. Front Immunol 2022;13:1000996. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Camacho V, Kuznetsova V, Welner RS. Inflammatory cytokines shape an altered immune response during myeloid malignancies. Front Immunol 2021;12:772408. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wu H, Li P, Shao N, Ma J, Ji M, Sun X, et al. Aberrant expression of Treg-associated cytokine IL-35 along with IL-10 and TGF-β in acute myeloid leukemia. Oncol Lett 2012;3:1119–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Cunha C, Gonçalves SM, Duarte-Oliveira C, Leite L, Lagrou K, Marques A, et al. IL-10 overexpression predisposes to invasive aspergillosis by suppressing antifungal immunity. J Allergy Clin Immunol 2017;140(3):867–70. [DOI] [PubMed] [Google Scholar]
- 24.Khaldoyanidi S, Nagorsen D, Stein A, Ossenkoppele G, Subklewe M. Immune biology of acute myeloid leukemia: implications for immunotherapy. J Clin Oncol 2021;39(5):419–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Mussai F, De Santo C, Abu-Dayyeh I, Booth S, Quek L, McEwen-Smith RM, et al. Acute myeloid leukemia creates an arginase-dependent immunosuppressive microenvironment. Blood 2013;122(5):749–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Gonçalves Silva I, Yasinska IM, Sakhnevych SS, Fiedler W, Wellbrock J, Bardelli M, et al. The Tim-3-galectin-9 secretory pathway is involved in the immune escape of human acute myeloid leukemia cells. EBioMedicine 2017;22:44–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Khaznadar Z, Boissel N, Agaugué S, Henry G, Cheok M, Vignon M, et al. Defective NK cells in acute myeloid leukemia patients at diagnosis are associated with blast transcriptional signatures of immune evasion. J Immunol 2015;195(6): 2580–2890. [DOI] [PubMed] [Google Scholar]
- 28.Stanzani M, Lewis RE. Development and applications of prognostic risk models in the management of invasive mold disease. J Fungi (Basel) 2018;4(4):141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lauruschkat CD, Einsele H, Loeffler J. Immunomodulation as a therapy for Aspergillus infection: current status and future perspectives. J Fungi (Basel) 2018;4(4):137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Tortorano AM, Esposto MC, Prigitano A, Grancini A, Ossi C, Cavanna C, et al. Cross-reactivity of Fusarium spp. in the Aspergillus Galactomannan enzyme-linked immunosorbent assay. J Clin Microbiol 2012;50(3):1051–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Sam QH, Yew WS, Seneviratne CJ, Chang MW, Chai LYA. Immunomodulation as therapy for fungal infection: are we closer? Front Microbiol 2018;9:1612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Lamoth F, Lewis RE, Kontoyiannis DP. Investigational antifungal agents for invasive mycoses: a clinical perspective. Clin Infect Dis 2022;75(3):534–44. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.