Incidence of invasive fungal infection in acute lymphoblastic and acute myelogenous leukemia in the era of antimold prophylaxis

The incidence of invasive fungal infection (IFI) in patients with acute myeloid leukemia (AML) has decreased with the introduction of antimold prophylaxis. Although acute lymphoblastic leukemia (ALL) has a lower risk of IFI than does AML, the incidences of IFI in both AML and ALL in the era of antimold prophylaxis should be re-evaluated. We analyzed adults with AML or ALL who had undergone induction, re-induction, or consolidation chemotherapy from January 2017 to December 2019 at Seoul National University Hospital. Their clinical characteristics during each chemotherapy episode were reviewed, and cases with proven or probable diagnoses were regarded as positive for IFI. Of 552 episodes (393 in AML and 159 in ALL), 40 (7.2%) were IFI events. Of the IFI episodes, 8.1% (12/148) and 5.9% (13/220) (P = 0.856) occurred in cases of ALL without antimold prophylaxis and AML with antimold prophylaxis, respectively. After adjusting for clinical factors, a lack of antimold prophylaxis (adjusted odds ratio [aOR], 3.52; 95% confidence interval [CI], 1.35–9.22; P = 0.010) and a longer duration of neutropenia (per one day, aOR, 1.02; 95% CI, 1.01–1.04; P = 0.001) were independently associated with IFI. In conclusion, the incidence of IFI in ALL without antimold prophylaxis was not lower than that in AML. A lack of antimold prophylaxis and prolonged neutropenia were independent risk factors for IFI. Clinicians should be on guard for detecting IFI in patients with ALL, especially those with risk factors.

Comparing the clinical characteristics between episodes with and without antimold prophylaxis, there were significant differences in CCI score, rate of induction or re-induction chemotherapy, type of leukemia, and duration of neutropenia (Supplementary Table S1). We examined the incidence of IFI in AML or ALL episodes with or without antimold prophylaxis ( Table 2). The incidence of IFI in AML episodes with antimold prophylaxis was 5.9% (13/220) versus 8.7% (15/173) without. In ALL episodes, the incidence of IFI was 0% (0/11) with antimold prophylaxis and 8.1% (12/148) without. The difference in IFI incidence between ALL episodes without antimold prophylaxis and AML episodes with antimold prophylaxis was not significant (OR, 1.41; 95% CI, 0.62-3.17; P = 0.856). When the latest 2020 European Organization for Research and Treatment of Cancer and the Mycoses Study Group Education and Research Consortium (EORTC/MSGERC) definition was used (Supplementary Table S2), the incidences of IFI were 5.0% (11/220) in AML with prophylaxis and 7.5% (13/173) in ALL without prophylaxis.

Sensitivity analyses. Seventy-two episodes involving the therapeutic administration of antimold agents
during the episodes were excluded for the first sensitivity analysis. The incidence of IFI in AML episodes with antimold prophylaxis was 7.4% (13/176) versus 9.0% (12/134) in ALL episodes without antimold prophylaxis (Supplementary  Table S6).

Discussion
Although AML has been traditionally believed to be associated with a greater risk of IFI than ALL 2-5 , the recent introduction of antimold prophylaxis in induction and re-induction chemotherapy has considerably reduced the incidence of IFI in AML 1,6,7 . Since then, it was unknown whether ALL is still associated with a lower risk of IFI than AML in the era of antimold prophylaxis. The study was performed to address this, and we found that the incidence of IFI in ALL was not lower than that in AML.
In addition to previous observations that AML is associated with a higher incidence of IFI than is ALL 8,9 , there are some theoretical reasons for a higher risk of IFI in AML than in ALL. First, myeloid cell dysfunction poses a greater risk of IFI than does lymphoid cell dysfunction. Second, the duration of neutropenia after cytotoxic chemotherapy is generally longer in AML than in ALL 2,5,9 . In an effort to reduce IFI in AML, a multicenter randomized controlled trial of patients with AML or myelodysplastic syndrome was performed, which showed that posaconazole prophylaxis significantly reduced the incidence of IFI compared with fluconazole or itraconazole prophylaxis 1 . This led to the current recommendation for posaconazole prophylaxis in patients with AML undergoing induction or re-induction chemotherapy 6,10 .
No routine antimold prophylaxis is recommended for patients with ALL receiving cytotoxic chemotherapy, owing to the lower IFI incidence and possible neurotoxic interactions between azoles and vinca alkaloids 6,11 . Therefore, studies of the incidences of, and risk factors for IFI in AML or ALL that reflect the current situation are warranted to guide clinicians. Because the diagnosis of IFI is often complicated, estimating the pre-test probability of IFI in patients with acute leukemia is very important for interpreting the results of clinical and mycological tests [12][13][14] . We showed that the incidence of IFI in ALL without antimold prophylaxis was not lower than that in AML with antimold prophylaxis, indicating that the notion of a lower risk of IFI in ALL than in AML needs to be modified. Clinicians should remain highly suspicious of IFI in ALL patients who are not receiving antimold prophylaxis, especially those with prolonged neutropenia. In addition, antimold prophylaxis could be considered in patients with ALL who are expected to have a longer duration of neutropenia than usual occasion.
Studies suggest that an advanced age, prolonged neutropenia, and the relapse of hematological malignancies are risk factors for IFI in acute leukemia 2,5 . However, most studies were performed on AML cases, and few have examined the risk factors for IFI in patients with ALL. In a single-center, retrospective study, Nicolato et al. showed that a longer duration of neutropenia, the relapse of underlying diseases, a longer duration of Table 3. Uni-and multi-variable analysis for risk factors of invasive fungal infection. Data are shown as number (%), not otherwise specified.

Characteristics IFI episodes (n = 40) Non-IFI episodes (n = 512)
Univariable www.nature.com/scientificreports/ hospitalization before the first fever, and allogeneic HSCT were risk factors for IFI in ALL 9 , which is in accordance with our results. Notably, the type of acute leukemia itself (i.e., AML and ALL) was not independently associated with the occurrence of IFI in this study.
To increase the reliability of our results, sensitivity analyses were performed under two conditions. In the first, we excluded those who received a therapeutic antimold agent during an episode. In the second, we included only the first episodes of every patient in the study period, although we used a GEE to explore independent risk factors for IFI to lessen the effects of repeated inclusion of patients 15 . The main results, namely, that the incidence of IFI in ALL without prophylaxis was comparable to that in AML with prophylaxis, and a lack of antimold prophylaxis and a longer duration of neutropenia were independent risk factors for IFI, remained unchanged in these analyses (Supplementary Table S5-S8). In addition, the incidence of invasive mold infection did not differ significantly between these two groups (Supplementary Table S3).
There are some limitations to our study. First, it included a study population with a small sample size enrolled from a single university-affiliated hospital. Particularly, the analyses of the difference in IFI incidence according to the clinical scenario might have been under-powered. Second, because the diagnosis of IFI is often challenging and the majority of IFI cases in this study were only probable diagnoses, the retrospective nature of our study also needs to be considered. Lastly, we could not suggest a specific antimold prophylaxis regimen or schedule in ALL in this study. Further studies are warranted to validate our results and to examine clinical impact of antimold prophylaxis in selected patients with ALL. Therefore, a larger prospective study is warranted to validate our results.
In conclusion, the incidence of IFI in ALL episodes without antimold prophylaxis was not lower than that in AML episodes with antimold prophylaxis, and a lack of antimold prophylaxis and a longer duration of neutropenia were independent risk factors for IFI. Therefore, in the current era of antimold prophylaxis, clinicians should also be on guard for the occurrence of IFI in patients with ALL, especially those with characteristics that put them at higher risk.

Materials and methods
Study design and data collection. We retrospectively reviewed all adults (≥ 18 years) with AML or ALL who received induction, re-induction, or consolidation cytotoxic chemotherapy from January 2017 to December 2019 at Seoul National University Hospital. Cases of hematopoietic stem-cell transplantation (HSCT) were excluded, because antimold prophylaxis is routinely given regardless of the type of acute leukemia 6 . Episodes of palliative chemotherapy, such as that using decitabine or azacitidine, and research-based chemotherapy were also excluded.
Clinical data were reviewed using electronic medical records including those on age, gender, the type of acute leukemia, purpose of cytotoxic chemotherapy, specific agents administered, severity of underlying diseases measured by based on the Charlson's comorbidity-weighted index (CCI) score 16 , administration of antimold agent, duration of neutropenia, and occurrence and type of IFI.
The incidence of IFI was examined in AML or ALL episodes with or without antimold prophylaxis. The clinical characteristics of the IFI episodes were compared to reveal independent risk factors for IFI.
This study was approved by the Institutional Review Board at Seoul National University Hospital (IRB no. 2008-167-1151), and they waived obtaining the informed consent because of the retrospective nature of the study. All study procedures were performed in accordance with the Declaration of Helsinki.
Definition. Only cases with proven and probable diagnoses following the 2008 Revised EORTC/MSG guideline were regarded as positive for IFI in this study 17 . Although we used the 2008 version in main analysis to account for the study period preceding the guideline revision, and to retain a higher sensitivity for the clinical diagnosis of IFI 12 , we performed the same analyses and presented the results using 2020 version.
An episode was defined as the period between the date of starting chemotherapy and the date of discharge. If two or more rounds of chemotherapy were performed during one hospitalization, the period was divided according to the day the subsequent chemotherapy began 19 . Antimold prophylaxis included routine posaconazole prophylaxis for induction or re-induction chemotherapy in AML or antimold agents for secondary prophylaxis due to a history of invasive mold infection.
The duration of neutropenia was defined as the time between the first date of an absolute neutrophil count (ANC) lower than 500/mm 3 and the third day with an ANC higher than 500/mm 3 , as previously described 9,20,21 . Sensitivity analysis. Two sensitivity analyses were performed to account for potential confounders. First, the incidences of, and risk factors for IFI were re-examined after excluding all episodes with therapeutic administration of an antimold agent (e.g., liposomal amphotericin B, caspofungin, etc.) for prolonged neutropenic fever, because this could obscure the effect of a prophylactic antimold agent. Second, we analyzed the subgroup that included only the first episodes for every patient to exclude the effects from repeated inclusion of the patients.
Statistical analysis. The chi-square or Fisher's exact test was used to compare categorical variables. Based the normality of continuous variables assessed using the Shapiro-Wilk test, the Mann-Whitney U test or Students' t-test was used. To calculate odds ratio (OR) and associated 95% confidence interval (CI) for IFI in different clinical scenarios (AML or ALL with or without antimold prophylaxis) logistic regression was used. For the multivariable analysis, variables were analyzed using backward stepwise logistic regression. Variables were removed using a P value of 0.05 as the cut-off value. Where applicable, generalized estimating equations (GEEs) were used to account for the inclusion of multiple episodes from a patient 15