COVID-19-associated pulmonary aspergillosis in Latin America: A systematic review of 575 cases

Abstract

Background

Although COVID-19-associated pulmonary aspergillosis (CAPA) has been well characterized in Europe and North America, data from Latin America remain scarce, despite the region's high COVID-19 burden.

Aims

This systematic review compiles all published CAPA cases from Latin America, providing a comprehensive analysis of clinical characteristics, diagnostic practices, treatment patterns, and mortality risk factors across the region.

Methods

A systematic review was conducted on 32 studies, encompassing a total of 575 CAPA cases from Latin America. Data were extracted from PubMed, Embase, and LILACS (search date: March 7, 2024), following Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Individual and aggregate patient data were used.

Results

Most cases originated from Argentina, Brazil, Chile, and Mexico. Patients were predominantly male (72.6%), with hypertension (39.8%), obesity (38.1%), and diabetes (35.3%) as leading comorbidities. Probable CAPA accounted for 81.9% of cases, with Aspergillus fumigatus identified in 58.5%. Antifungal treatment, mainly voriconazole (57.5%), was administered in 75.5% of cases. Overall mortality was 50.1%. In the multivariate analysis, higher galactomannan (GM) levels in bronchoalveolar lavage (BAL) were independently associated with mortality (OR: 1.40; 95% CI: 1.04–2.03; p = 0.045).

Conclusions

This is the most comprehensive analysis of CAPA in Latin America to date, revealing high mortality despite antifungal treatment and limited diagnostic access. The presence of GM in BAL was independently associated with death, reinforcing its prognostic value. These findings underscore the urgent need for improved diagnostic strategies and targeted interventions in resource-limited settings.

Resumen

Antecedentes

Aunque la aspergilosis pulmonar asociada a COVID-19 (CAPA) ha sido bien caracterizada en Europa y América del Norte, los datos de América Latina siguen siendo escasos, a pesar de la alta carga de COVID-19 en la región.

Objetivos

Esta revisión sistemática aúna los casos de CAPA en América Latina publicados, ofreciendo un análisis exhaustivo de las características clínicas, las prácticas diagnósticas, los patrones de tratamiento y los factores de riesgo de mortalidad en la región.

Métodos

Se realizó una revisión sistemática de 32 estudios que incluyeron 575 casos de CAPA en América Latina, de los que se obtuvieron datos individuales y agregados de pacientes. Los datos fueron extraídos de PubMed, Embase y LILACS (fecha de búsqueda: 7 de marzo de 2024), siguiendo las directrices PRISMA.

Resultados

La mayoría de los casos tuvieron lugar en Argentina, Brasil, Chile y México. Los pacientes fueron predominantemente hombres (72,6%), con hipertensión (39,8%), obesidad (38,1%) y diabetes (35,3%) como comorbilidades principales. El 81,9% de los casos correspondieron a un diagnóstico de CAPA probable, con Aspergillus fumigatus identificado en el 58,5% de los casos. El tratamiento antifúngico, principalmente voriconazol (57,5%), se administró en el 75,5% de los casos. La mortalidad global fue del 50,1%. En el análisis multivariado los niveles más altos de galactomanano (GM) en el lavado broncoalveolar (LBA) se asociaron independientemente con la mortalidad (OR: 1,40; IC 95%: 1,04–2,03; p = 0,045).

Conclusiones

Este es el análisis más completo de CAPA en América Latina hasta la fecha; a pesar del tratamiento antifúngico y el acceso limitado al diagnóstico queda en evidencia una alta mortalidad. La presencia de GM en LBA se asoció independientemente con óbito, lo que refuerza su valor pronóstico. Estos hallazgos resaltan la necesidad urgente de mejorar las estrategias de diagnóstico y las intervenciones dirigidas en entornos con recursos limitados.

The genus Aspergillus comprises ubiquitous filamentous fungi capable of causing a wide spectrum of clinical manifestations, depending on the host's immune status, with invasive pulmonary aspergillosis (IPA) being the most severe form.⁸³ Classical risk factors for IPA include neutropenia and bone marrow transplantation. However, novel risk factors have been associated with the disease, such as solid organ transplantation, intensive care unit (ICU) admission, use of broad-spectrum antibiotics, chronic obstructive pulmonary disease (COPD), cirrhosis, chronic kidney disease, non-hematological cancers, and influenza H1N1 infection.^{8,12,17,55,64,88} Other risk factors have been described for pulmonary aspergillosis, such as tuberculosis – particularly regarding chronic pulmonary aspergillosis (CPA)²⁷; nevertheless, its role in IPA remains uncertain.¹⁸

The emergence of coronavirus disease 2019 (COVID-19) has been accompanied by several reports of co-infection with IPA^5,15,43,91 leading to the identification of a new entity termed COVID-19-associated pulmonary aspergillosis (CAPA). The reported incidence of CAPA varies widely, ranging from 3% to 39% depending on the study analyzed.^76,78 Despite this variability, the mortality rate remains consistently high, reaching up to 55%,⁶⁰ even when appropriate treatment is administered.^1,4

However, significant challenges remain in understanding the global burden of the disease.²¹ To date, most published studies originate from European countries or the United States,^16,87 while only a limited number of case reports or case series have been documented from Latin American countries.^26,52,73,90 The incidence of invasive mold diseases in this region differs significantly from other parts of the world.⁶² Thus, our objective was to perform a systematic review of CAPA cases reported in Latin America, focusing on clinical characteristics, diagnostic approaches, radiological findings, treatment, and outcomes. Additionally, we aimed to identify the risk factors for mortality associated with CAPA.

Methods

We conducted a systematic review strictly adhering to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 checklist during all phases of the study: implementation, analysis, and reporting.⁶³ We included CAPA cohorts, case series and case report studies, including conference abstracts. The cases were sourced exclusively from Latin America. Possible, probable, and proven CAPA were defined according to a modified ECMM/ISHAM consensus criteria published in 2021,⁵¹ in which we accepted as criteria for possible CAPA not only galactomannan (GM) in non-bronchoalveolar lavage (NBL), but also in tracheal aspirate (TA), as detailed in Table S1 (Supplementary material). Non-bronchoscopically guided respiratory tract specimens were categorized as NBL, thereby excluding BAL, TA and sputum. Furthermore, the methodology for GM dosage used in each study is detailed in Table S2 (Supplementary material). Studies published as reviews or conducted on animals were excluded; however, their references were screened for additional relevant records.

We searched PubMed/MEDLINE, Embase, and LILACS on March 7th, 2024, using the search terms described in Supplementary data (Appendix 1). There was no language or date of publication restriction. We registered the protocol on PROSPERO (CRD 42024547876). Two qualified investigators independently assessed the eligibility of the identified publications, and discrepancies were resolved by a discussion. We first performed an initial screen of titles or abstracts to assess potential relevance (ICLVC and MMCM). Afterwards, we obtained relevant full-text articles, reevaluated their eligibility, and determined their final inclusion or exclusion (ICLVC and VFO).

A standardized, pilot-tested data collection form was utilized. The collected variables included the following: authors’ names; year of publication; country of study; study design; number of participants; participants’ age, sex, occupation, and comorbidities [systemic arterial hypertension (SAH), diabetes mellitus (DM), COPD, liver disease, obesity, smoking, chronic kidney disease, solid organ malignancy, haematological malignancies, hematopoietic stem-cell transplantation, solid organ transplantation, other immunosuppression, and acquired human immunodeficiency syndrome (AIDS)]; ICU admission; mechanical ventilation (MV); treatments for COVID-19 (corticosteroids, tocilizumab, remdesivir, ritonavir combined with lopinavir, baricitinib, and hydroxychloroquine); use of other immunosuppressants; prior use of antibiotics; prior antifungal use (within the last 2 weeks); neutropenia (<500 cells/mm³); lymphopenia (<1000 cells/mm³); CAPA symptoms (fever, pleuritic chest pain, pleural friction rub, ventilatory worsening, and dyspnea); GM testing [serum, BAL, NBL, and TA]; positive fungal cultures for Aspergillus spp. obtained from peripheral blood, BAL, NBL, TA, and sputum samples; histopathology findings; Aspergillus GM lateral flow assay (LFA) in BAL, NBL, TA or blood; Aspergillus polymerase chain reaction (PCR) in lung biopsy, serum, BAL, NBL or TA; beta-d-glucan (BDG); radiological features of CAPA (nodules, cavitation, halo sign, consolidation, ground glass opacities, pulmonary infiltrate, and air crescent sign); bronchoscopy findings indicative of tracheobronchitis; time from onset of COVID-19 symptoms to CAPA diagnosis; treatment for CAPA (specific antifungal used and duration); and outcome (death). The collection was performed by two investigators, independently (ICLVC and ALM). The extracted data were then checked again by another evaluator (VFO).

After the initial data collection and extraction, it was decided to email all corresponding authors of the included papers requesting additional data for individual data analysis. The emails were sent from July 29th to August 1st, 2024. The data were sent through Excel tables and included information on sex, age, comorbidities, ICU admission, MV, culture, GM, Aspergillus PCR, BDG, histopathological evidence of Aspergillus, antifungal treatment, and death.

Statistical analysis

Demographic, clinical, laboratory, and treatment data were assessed using total counts and percentages for categorical variables, and median with interquartile range (IQR) or mean with standard deviation for continuous variables. For all simple proportions, 95% confidence intervals (95% CI) were calculated using the Wilson score method.

For aggregate data analysis, all case reports were combined into a “single-case data” category. In order to minimize the impact of between-study heterogeneity, varied sample sizes, and the differing precision of individual study estimates, we elected to perform a meta-analysis of proportion (metaprop) or mean (metamean) whenever three or more case series were available. This methodology was utilized to estimate a summary proportion with 95% confidence intervals (CI) using the meta package in RStudio software version 1.4. For a comparison between simple proportions and metaprop, please refer to Supplementary Table S3.

For individual data analysis, comparisons were made between survival and non-survival groups. Categorical variables were assessed using Fisher's exact test or Pearson's chi-squared test, as appropriate, while continuous variables were evaluated using the Mann–Whitney test. A multivariate binary logistic regression model was employed to assess factors associated with mortality, with survival status (non-survivor vs. survivor) as the dependent variable. Adjusted odds ratios (ORs) and 95% confidence intervals (CI) were estimated by including clinically relevant covariates selected based on prior knowledge and univariate analysis. Statistical significance was defined as a p-value <0.05. For the multivariate analysis, variables showing high collinearity with mortality were excluded. Concerning the GM measurements, we decided that only GM in BAL would be included, given its most established diagnostic threshold in the literature.^8,51 Also, we opted to incorporate in our analysis the variables age and glucocorticoid use, based on prior associations with mortality in CAPA.⁷⁷

Results

A total of 480 records were identified during the initial search across selected databases, with 133 duplicates removed. After screening titles and abstracts, 277 articles were excluded based on relevance. Subsequently, 70 full-text articles were assessed for eligibility, and all were successfully retrieved. The primary reason for exclusion at this stage was the origin of the studies, as 18 were from countries outside Latin America. In total, 32 manuscripts were included in the review, representing 575 cases of CAPA reported in Latin America between 2020 and 2024.^{3,6,7,10,11,13,23,24,26,31,33,36,40,42,54,56–59,61,66,68–70,72–75,79,80,84,86,89} The study selection process is summarized in the flow diagram presented in Fig. 1.

Fig. 1.

Study selection for a systematic review of cases of COVID-associated pulmonary aspergillosis (CAPA) in Latin America from 2020 to 2024.

Seven of the studies included were case reports, while 20 were case series, four were cohorts and one was a cross-section study. Among the sample evaluated, eight were conference abstracts (Table S2). Notably, only one of the studies was multicentered. Among the countries evaluated, six studies were published in Argentina, nine in Brazil, three in Chile, and ten in Mexico and Colombia. Nicaragua, Peru, and the Dominican Republic had one study each. The total number of patients from each country can be seen in Fig. 2.

Fig. 2.

Distribution of COVID-19-associated pulmonary aspergillosis (CAPA) cases and frequency of Aspergillus species isolated across Latin American countries included in the study (2020–2024).

We summarized the patients’ demographics and clinical data in Table 1, utilizing meta-proportion. Seven case series described the total number of patients hospitalized with COVID-19, and the prevalence of CAPA was 3.6% (95% CI: 0.9–13.0). Most patients were male (72.6%; 95% CI: 68.3–76.6), and the most important comorbidities described were SAH (39.8%; 95% CI: 35.3–44.5), obesity (38.1%; 95% CI: 33.2–43.2), and DM (35.3%; 95% CI: 31.2–39.6) (Table 1). Furthermore, lymphopenia was evidenced in 80 of the 127 patients for whom this information was described (63%). Almost all the patients were in ICU and MV, and 84.9% (95% CI: 39.1–98) of them received glucocorticoids, while 17.3% (95% CI:3.7–53.0) were prescribed tocilizumab. In 56.9% (95% CI: 41.6–70.9) of the patients a bacterial co-infection was confirmed. The median time from onset of COVID-19 symptoms to diagnosis of CAPA was 12.9 days (95% CI: 6.5–19.4). With regard to the clinical presentation of CAPA, we only had data relating to 35 patients. The most prevalent symptom was a ventilatory worsening in 80% of cases, followed by fever in 42.9%. Pulmonary involvement was present in 99% (95% CI: 97.7–99.5) of patients, with only 0.5% (95% CI: 0.2–1.6) presenting evidence of Aspergillus tracheobronchitis associated with COVID-19 in bronchoscopy, and 0.2% (95% CI: 0.0–1.2) presenting a disseminated disease (Table 1).

Table 1.

Demographics, clinical characteristics, diagnosis and treatment of 575 patients with CAPA in Latin America, calculated as metaproportion (metaprop) when three or more case series were available or simple proportion when there were only one or two case series.

Variables	Number of cases(n/N)	Proportion(%)	95% CI
Prevalence of CAPA in hospitalized patients with COVID-19	255; 7832^a	3.6	0.9–13
Demographics
Male	316; 435^a	72.6	68.3–76.6
Mean age (years)	484; 575^a	60.8	57.7–63.9
Comorbidities and clinical condition
SAH	174; 437^a	39.8	35.3–44.5
Obesity	137; 368^a	38.1	33.2–43.2
Diabetes mellitus	174; 493^a	35.3	31.2–39.6
Smoking	49; 328^b	14.9	11.5––19.2
COPD	56; 435^a	12.9	10–16.4
Immunosuppression	32; 315^b	10.2	7.3–14
Haematological malignancies	7; 286^b	2.4	1.2–5
Solid organ malignancy	7; 302^b	2.3	1.1–4.7
SOT	5; 286^b	1.7	0.7–4
AIDS	4; 286^b	1.4	0.5–3.5
HSCT	1; 286^b	0.3	0.1–1.9
Other immunosuppression	8; 315^b	2.5	1.3–5
Chronic kidney disease	8; 286^b	2.8	1.4–5.4
Liver disease	6; 286^b	2.1	1–4.5
Lymphopenia	80; 127^b	63.0	54.3––70.9
Neutropenia	2; 36^b	5.6	1.5–18.1
ICU	436; 448^a	99.3	87.7–100
MV	420; 447^a	99.3	76.6–100
Bacterial co-infection	102; 195^a	56.9	41.6–70.9
Treatment of COVID-19
Glucocorticoids	310; 390^a	84.9	39.1–98
Tocilizumab	53; 390^a	17.3	3.7–53
Ritonavir and lopinavir	37; 390^a	2.9	0–56.4
Remdesivir	8; 390^a	2.1	1–4.1
Baricitinib	35; 390^a	0.6	0–40.8
Hydroxychloroquine	18; 390^a	0.6	0–51.2
Median time from COVID to CAPA, days	393; 575^a	12.9	6.5–19.4
CAPA symptoms and signs
Ventilatory worsening	28; 35^b	80	64.1–90
Fever	15; 35^b	42.9	28–59.1
Dyspnea	13; 35^b	37.1	23.2–53.7
Pleuritic chest pain	5; 35^b	14.3	6.3–29.4
Pleural friction rub	1; 35^b	2.9	0.5–14.5
Site of infection
Pulmonary	571; 575^a	99	97.7–99.5
Tracheobronchitis	3; 575^a	0.5	0.2–1.6
Disseminated	1; 575^a	0.2	0–1.2
Diagnosis
Positive direct microscopy	7; 51^b	13.7	6.8–25.7
Patients with positive culture	265; 403^a	47	9.2–88.5
Positive culture in TA	178; 221^b	80.5	74.8–85.2
Positive culture in BAL	46; 230^a	20	15.3–25.7
Positive culture in NBL	17; 235^a	7.2	4.5–11.3
Positive culture in sputum	6; 221^b	2.7	1.3–5.8
Species
Aspergillus fumigatus	155; 265^a	58.5	52.5–64.3
Aspergillus flavus	42; 265^b	15.8	11.9–20.8
Aspergillus niger	31; 265^a	11.7	8.4–16.2
Aspergillus terreus	7; 265^a	2.6	1.3–5.4
Aspergillus nidulans	5; 265^b	1.9	0.8–4.3
Aspergillus tamarii	3; 265^b	1.1	0.4–3.3
Aspergillus versicolor	3; 265^b	1.1	0.4–3.3
Aspergillus sclerotiorum	1; 265^b	0.4	0.1–2.1
Aspergillus lentulus	1; 265^b	0.4	0.1–2.1
Aspergillus spp.	34; 265^b	12.8	9.3 – 17.4
Galactomannan (GM)
Positive GM in NBL	2; 2^b	100	34.2–100
Positive GM in TA	63; 141^a	81.2	5.3–99.7
Positive GM in BAL	95; 146^a	63.6	50–75.2
Positive GM in Serum	107; 181^a	59.1	51.8–66
Positive PCR for Aspergillus spp.
PCR in BAL	32; 39^b	82.1	67.3 – 91
PCR in Serum	7; 39^b	17.9	9–32.5
PCR in TA	2; 39^b	5.1	1.4–16.9
PCR in Biopsy	2; 39^b	5.1	1.4–16.9
Lung biopsy	6; 6^b	100	61–100
Radiological features
Altered chest CT scan	197; 234^a	84.2	78.9–88.3
Ground glass opacities	97; 197^b	49.2	42.3–56.2
Pulmonary infiltrate	53; 197^b	26.9	21.2–33.5
Consolidation	47; 197^b	23.9	18.4–30.3
Cavitation	16; 197^b	8.1	5.1–12.8
Nodules	10; 197^b	5.1	2.8–9.1
Halo sign	2; 197^b	1	0.3–3.6
Air crescent sign	2; 197^b	1	0.3–3.6
Classification of CAPA
Proven	9; 575^a	1.6	0.8–3
Probable	337; 575^a	81.9	9.8–99.5
Possible	229; 575^a	7.3	0–94.4
Treatment
Use of antifungal drugs	294; 394^a	75.5	43.3–92.5
Voriconazole	169; 294^a	57.5	51.8–63
Amphotericin B	61; 294^a	20.8	16.5–25.8
Amphotericin B deoxycholate	35; 61^a	57.4	44.8–69.1
Amphotericin B liposomal	25; 61^a	41	29.4–53.6
Isavuconazole	57; 294^a	23.1	0.1–99
Echinocandin	13; 294^a	4.4	2.6–7.5
Itraconazole	2; 294^a	0.7	0.2–2.7
Change of antifungal	17; 294^a	5.8	3.6–9.1
Combination therapy	14; 294^a	4.8	2.8–7.9
Outcome
Overall mortality	286; 542^a	50.1	43–57.2

CI: confidence interval; CAPA: COVID-19-associated pulmonary aspergillosis; SAH: systemic arterial hypertension; COPD: chronic obstructive pulmonary disease; SOT: solid organ transplantation; AIDS: acquired immunodeficiency syndrome; HSCT: hematopoietic stem-cell transplantation; ICU: intensive care unit; MV: mechanical ventilation; TA: tracheal aspirate; BAL: bronchoalveolar lavage; NBL: non-bronchoalveolar lavage; PCR: protein chain reaction; CT: computed tomography; CI: confidence interval; SD: standard deviation; n: number of events; N: number of cases for which the information was available.

a

metaprop.

b

Simple proportion.

Concerning the modified ECMM/ISHAM criteria, 81.9%; (95% CI: 9.8–99.5) of cases were classified as probable CAPA, 7.3% (95% CI: 0.0–94.4) as possible CAPA and only nine patients (1.6%, 95% CI: 0.8–3) were categorized as having proven disease. Of the 403 patients who had at least one respiratory sample collected, the growth of an Aspergillus species was obtained from 47% of the cases (95% CI: 9.2–88.5). Most isolates were obtained from TA (80.5%), followed by BAL samples (20%; 95% CI: 15.3–25.7) (Table 1); eight patients had a positive culture in more than one site. Aspergillus fumigatus was the most frequently identified species, present in 58.5% (95% CI: 52.5–64.3) of the samples, followed by Aspergillus flavus (15.8%) and Aspergillus niger (11.7%; 95% CI: 8.4–16.2) (Table 1). There was no significant difference in species prevalence among the Latin American countries (Fig. 2). In 12.8% of cases, no species were identified (Table 1).

Concerning the GM test, we found that 81.2% (95% CI: 5.3–99.7) of TA samples, 63.6 (95% CI: 50.0–75.2) of BAL samples, and 59.1% (95% CI: 51.8–66.0) of serum samples were positive (Table 1). The highest median GM value was observed in TA [3.6 (IQR: 1.9–3.8)], followed by BAL GM [2.2 (IQR: 1–5)], and serum GM [1.2 (IQR: 0.8–2.4)] (Table 2). Among the 39 cases with PCR information, positivity rates were 82.1% in BAL, 17.9% in blood, and 5.1% in both biopsy and TA samples. Direct fungal testing was positive in 13.7% of patients, and only six patients underwent biopsy or autopsy confirming Aspergillus tissue invasion. The LFA test was performed in only 33 patients (12%), all of which were positive. BDG was not widely available in most Latin American countries, limiting its use in these cases, and none of our patients underwent this test.

Table 2.

Comparison of individual data between the survivors and non-survivors groups of 286 patients with CAPA in Latin America from 2020 to 2024, based on clinical, laboratory, and therapeutic features and variables associated with mortality by multivariate analyses in patients with CAPA.

Variables		Survivors (n = 119)	Non-survivors (n = 167)	p-Value
Demographics
Median age, years (IQR)	60 (49–69)	55 (47–64)	63 (52–72)	<0.001
Male	204/286	86 (72%)	104 (62%)	0.8
Comorbidities and clinical condition
SAH	113/286	44 (37%)	69 (41%)	0.5
Diabetes mellitus	104/286	39 (33%)	65 (39%)	0.3
Obesity	107/286	48 (40%)	59 (35%)	0.4
Smoking	47/286	20 (17%)	27 (16%)	0.9
COPD	32/286	14 (12%)	18 (11%)	0.8
CKD	8/286	3 (2.5%)	5 (3%)	>0.9
Liver disease	6/286	2 (1.7%)	4 (2.4%)	>0.9
Immunosuppression
Haematological malignancies	7/286	2 (1.8%)	5 (3%)	0.7
Other immunosuppression	7/286	4 (3.4%)	3 (1.8%)	0.3
SOT	5/286	1 (0.8%)	4 (2.4%)	0.4
AIDS	4/286	1 (0.8%)	3 (1.8%)	0.6
Solid organ malignancy	4/286	0	4 (2.4%)	0.14
HSCT	1/286	0	1 (0.6%)	>0.9
ICU	279/283	113 (97%)	166 (100%)	0.028
MV	270/282	104 (90%)	166 (100%)	<0.001
Bacterial co-infection	58/130	19 (41%)	32 (45%)	0.7
Treatment of COVID-19
Glucocorticoids	109/269	78 (75%)	119 (78%)	0.5
Tocilizumab	14/269	6 (5.9%)	5 (3.4%)	0.4
Remdesivir	6/269	1 (1%)	5 (3.4%)	0.4
Hydroxychloroquine	4/269	2 (2%)	3 (2%)	>0.9
Baricitinib	2/269	1 (1%)	1 (0.7%)	>0.9
Ritonavir and lopinavir	1/269	0	1 (0.7%)	>0.9
Median time from COVID-19 to CAPA, days (IQR)	17 (13–27)	16 (12–23)	20 (14–27)	0.3
CAPA symptoms and signs
Ventilatory worsening	28/35	9 (43%)	19 (70%)	0.055
Fever	15/35	5 (24%)	10 (37%)	0.3
Dyspnea	14/35	7 (58%)	6 (29%)	0.09
Pleuritic chest pain	5/35	1 (4%)	4 (15%)	0.4
Pleural friction rub	1/35	0	1 (3.7%)	>0.9
Diagnosis
Direct microscopy finding of filamentous fungi	7/49	5 (17%)	2 (2.6%)	0.2
Positive culture	221/240	90 (76%)	131 (78%)	0.6
Positive culture in TA	178/221	67 (56%)	111 (66.6%)	0.081
Positive culture in BAL	37/221	17 (14%)	20 (12%)	0.6
Positive culture in sputum	6/221	5 (4.2%)	1 (0.6%)	0.085
Positive culture in NBL	3/221	0	3 (1.8%)	0.3
Positive GM in Serum	73/118	30 (86%)	43 (96%)	0.2
Positive GM in BAL	37/45	17 (68%)	20 (87%)	0.2
Positive GM TA	10/82	0	10 (100%)	0.002
Median of positive GM Serum (IQR)	1.2 (0.8–2.4)	1.1 (0.7–1.8)	1.40 (0.8–2.8)	0.3
Median of positive GM BAL (IQR)	2.2 (1.0–5)	1.5 (0–3)	4.5 (1.2–6.1)	0.04
Median of positive GM TA (IQR)	3.6 (1.9–3.8)	2.2 (0–3.5)	3.7 (3.5–4.1)	<0.001
Radiological features
Ground glass opacities	71/129	34 (64%)	37 (67%)	0.4
Pulmonary infiltrate	48/129	20 (38%)	27 (42%)	0.7
Consolidation	47/129	11 (26%)	24 (39%)	0.2
Cavitation	16/129	10 (19%)	6 (9.2%)	0.13
Nodules	10/129	6 (11%)	4 (6.2%)	0.3
Halo sign	2/129	2 (3%)	0	0.2
Air crescent sign	2/129	2 (3.8%)	0	0.2
Treatment
Use of antifungals drugs	210/286	92 (77%)	118 (71%)	0.2
Voriconazole	147/210	74 (62%)	73 (44%)	0.002
Amphotericin B deoxycholate	35/210	14 (12%)	21 (13%)	0.8
Amphotericin B liposomal	16/210	3 (2.5%)	13 (7.8%)	0.056
Echinocandin	12/210	3 (2.5%)	9 (5.4%)	0.4
Isavuconazole	6/210	1 (0.8%)	5 (3%)	0.4
Itraconazole	2/210	2 (1.7%)	0	0.2
Combination therapy	13/210	6 (5%)	7 (4.2%)	0.7
Median time of treatment, days (IQR)	13 (5–25)	24 (14–38)	9 (3–14)	<0.001

Multivariate analysis	OR	95% CI	p-Value
Median age	1.02	0.97–1.09	0.4
Glucocorticoids	1.62	0.20–16.5	0.7
Median of positive GM BAL	1.40	1.04–2.03	0.045
Voriconazole	0.39	0.07–2.08	0.3

IQR: interquartile range; SAH: systemic arterial hypertension; COPD: chronic obstructive pulmonary disease; CKD: chronic kidney disease; SOT: solid organ transplantation; AIDS: acquired immunodeficiency syndrome; HSCT: hematopoietic stem-cell transplantation; ICU: intensive care unit; MV: mechanical ventilation; CAPA: COVID-19-associated pulmonary aspergillosis; TA: tracheal aspirate; BAL: bronchoalveolar lavage; NBL: non-bronchoalveolar lavage; GM: galactomannan.

Most of the patients had alterations in computed tomography (CT) scan (84.2%; 95% CI: 78.9–88.3), and the most common alteration was ground-glass opacities (49.2%) and pulmonary infiltrate (26.9%). Other radiological findings classically attributed to CAPA, such as new cavitations, nodules, halo sign, and air crescent sign, could be seen in 8.1%, 5.1%, 1%, and 1% of cases, respectively.

Of all the patients, 75.5% (95% CI: 43.3–92.5) received at least one antifungal drug for CAPA treatment. The most common drug was voriconazole (57.5%; 95% CI: 51.8–63), followed by isavuconazole (23.1%; 95% CI: 0.1–99) and amphotericin B (AmB) (20.8%; 95% CI: 16.5–25.8). Regarding AmB formulations, most patients received AmB deoxycholate (57.4%; 95% CI: 44.8–69.1). Only 5.8% (95% CI: 3.6–9.1) needed a change of antifungal, and a combination therapy was prescribed for 4.8% (95% CI: 2.8–7.9) of the cases (Table 1). The median time of antifungal treatment found was 13 days [IQR (5–25)] (Table 2).

The overall mortality rate was 50.1% (95% CI: 43–57.2) across all cases (Table 1). To identify factors associated with mortality, we compared survivors and non-survivors using individual data from 286 CAPA patients included in our study (Table 2). Our analysis revealed that median age, ICU admission, need for MV, median of positive GM in BAL, median of positive GM in TA, use of voriconazole, and median time of treatment showed statistically significant differences between the two groups (Table 2).

In the multivariate analysis of patients with available individual data, we excluded ICU admission and the need for MV, as these variables were present in 100% of non-survivors. It is important to note that we also excluded median treatment duration due to its high collinearity with mortality (since patients had a shorter treatment duration due to mortality). Only higher GM levels in BAL fluid were independently associated with increased mortality (OR: 1.40; 95% CI: 1.04–2.03; p = 0.045). Voriconazole use (OR: 0.39; 95% CI: 0.07–2.08; p = 0.3), glucocorticoid therapy (OR: 1.62; 95% CI: 0.2–16.5; p = 0.7), and age (OR: 1.02; 95% CI: 0.97–1.09; p = 0.4) were not significantly associated with mortality (Table 2).

Discussion

To the best of our knowledge, this is the first comprehensive review of CAPA in Latin America, gathering data from 575 patients reported in the published literature. Our study revealed a high mortality rate among CAPA cases in the region, approaching 50%, despite almost 75% of patients receiving antifungal therapy. Notably, most cases were reported from Mexico. Moreover, our individual data analysis identified GM in BAL as an independent prognostic factor for mortality.

The most common comorbidities (hypertension, obesity, and DM) and risk factors, such as MV and glucocorticoid use, were consistent with those reported in the literature.^4,21,39,87 However, other recently identified risk factors, including renal replacement therapy, COPD, hematologic malignancy, chronic liver disease, and the use of IL-6 inhibitors,^39,81 were not significant in our sample. The prevalence of lymphopenia that we found was similar to data reported in earlier COVID-19 studies (35%–63%).^19,44,92

The prevalence of CAPA was lower than that observed in some non-American countries, and distinctly lower when compared to various individual Latin American studies, where rates ranged from 10.7% to 22.6%.^{32,34,60,67,87} The findings mentioned above resemble the findings related to influenza-associated pulmonary aspergillosis (IAPA), in which most studies focus on Europe, Asia, the USA, and Canada, and only case reports or small case series are published in Latin American countries.^55,71 We hypothesize that diagnostic challenges, a potentially lower environmental burden of Aspergillus conidia, and the warm climate in the region may account for this difference in prevalence.⁸²

Even though we found a lower prevalence of CAPA in our region, the species distribution was very similar to what has been previously described in the literature, with the most common ones being A. fumigatus, A. flavus, and A. niger.⁴⁹ This distribution showed no changes among the countries evaluated. Conversely, we would like to emphasize that many possible and probable cases of CAPA lacked positive cultures or species identifications, as illustrated in Fig. 2, which may lead to an underestimation of the species distribution.

Diagnosing CAPA can also present a significant challenge, given that lung biopsy is widely regarded as the gold standard.⁴⁶ Typically, patients with severe COVID-19 exhibit substantial radiological changes on admission. However, new inflammatory infiltrates or cavitations may not always be attributable to a fungal infection.^28,37 Our study corroborates these findings, as only six patients underwent lung biopsy and less than half of them had BAL collected. This was probably due to the difficulty of performing bronchoscopy during the early COVID-19 pandemic because of the potential of virus aerosolization.⁴⁸

Although most studies utilized the ECMM/ISHAM criteria,⁵¹ eleven of them opted to include patients with GM from TA among their participants (see Table S2 in the Supplementary material). Our median GM value from TA was lower than the threshold recommended by the ECMM/ISHAM consensus for a single sample of GM in NBL. This finding highlights the need to establish an appropriate cutoff value for GM in NBL and TA, especially in critical care settings, which is particularly relevant in low- and middle-income countries where performing BAL on critically ill patients presents significant challenges.^22,30,47,77 Additionally, the value currently used for GM in respiratory samples in the literature is poorly validated, and previous systematic review studies have recommended a change in the cutoff, even for GM in BAL.^14,53

The most used drug for treating CAPA was voriconazole, in line with international recommendations.^51,87 However, there was a significant use of AmB deoxycholate, which is discouraged in intensive care settings^85,87; itraconazole and echinocandins in monotherapy were also prescribed, but their efficacy has not been established for IPA or CAPA.⁹ The treatment schedule for CAPA has been extrapolated based on the classical risk factors for IPA, a fact that reinforces the need to study and estimate a more appropriate therapy for this scenario. This highlights the challenges associated with the diagnosis of CAPA and its effective management, which may be attributed to delays in diagnosis or limited access to gold-standard medications.

Our overall mean treatment duration was 13 days, while the mean treatment duration for the survival group was 24 days, both of which are shorter than the 6–12 weeks recommended by the ECMM/ISHAM guidelines.⁵¹ Despite the shorter treatment period, our mortality rate was not higher than previously reported rates, which ranged from 42.6% to 74.0%.^21,35,45,50 This may suggest that a better response to IPA occurs when it is associated with viral infections, as the alterations in host immune response linked to these viruses are typically temporary.³⁵ Thus, a reduction in treatment duration has already been proposed by Peral et al.⁶⁵ Conversely, this finding may suggest that our cohort included some patients with only Aspergillus spp. colonization rather than confirmed cases of CAPA, as we lacked more precise tests to distinguish between the two conditions.² This only reinforces the difficulties in utilizing the current criteria for diagnosing CAPA,⁵¹ and even for IPA outside of classical contexts.^8,29,41,88

Regarding our multivariate analysis, none of the other two factors previously associated with death in a retrospective study of 162 patients across 23 centers in 15 countries conducted between March 2020 and December 2021⁷⁷ presented a statistical significance. Despite that, we found that the presence of GM in BAL was independently associated with mortality in cases of CAPA. This finding aligns with a previous post hoc analysis of 218 patients, which included 56 probable and proven cases of CAPA, conducted as part of a multicenter, multinational observational study. That study indicated an independent association between 90-day mortality and the combination of positive GM in BAL and the isolation of Aspergillus in culture from BAL.³⁸ However, in our study, this variable demonstrated its significance on its own, what is corroborated by earlier research in IPA. For instance, Chen et al. concluded from a three-year follow-up study of patients with probable, proven, and putative IPA that the initial GM value in BAL, as well as the maximum GM value in BAL or serum, were risk factors for mortality.²⁰ These values were also associated with an increased likelihood of developing renal failure and requiring mechanical ventilation, suggesting that higher GM levels indicate more severe disease. This underscores the critical role of GM in predicting outcomes in IPA, a relationship that has been previously identified, particularly regarding serum GM,²⁵ as the positivity of this biomarker typically signifies a more invasive disease.

Finally, this study has a few limitations, including a low number of cases described in Central America and the fact that most cases found in only a few countries (Mexico, Brazil, Chile, and Argentina), making it difficult to draw detailed conclusions concerning the whole continent. Another limitation involves potential misdiagnoses driven by a lack of clinical familiarity with the condition and restricted access to diagnostic tests. In addition, the information found in the evaluated studies was very variable, with few of them containing a description of clinical symptoms, radiological findings, and diagnostic methods, such as PCR, LFA, and histology. Lastly, many of the available studies were primarily presented as conference abstracts, which led to a high heterogeneity between studies.

Conclusion

This systematic review presents the most comprehensive synthesis of CAPA cases in Latin America to date. Despite high rates of antifungal use, mortality remained strikingly high, underscoring the need for earlier diagnosis and optimized therapeutic strategies. Notably, elevated GM levels in BAL fluid were independently associated with mortality, reinforcing its prognostic relevance. These findings highlight significant diagnostic and therapeutic gaps across the region and the urgent need to validate alternative diagnostic approaches, particularly in critically ill patients in whom bronchoscopy is often unfeasible. In the post-pandemic era, with persistent ICU-associated IPA and emerging viral epidemics, our results remain clinically relevant and call for better preparedness in fungal diagnostics and care across Latin America.

Authors’ contributions

I.C.L.V.C. (conceptualization, data curation, methodology, formal analysis, writing original draft, writing review and editing), A.L.M. (data curation, writing review and editing), V.H.A.T. (data curation), C.M.R.M. (data curation), M.F.G.L. (data curation, writing review and editing), T.A.C. (data curation, writing review and editing), F.A.R. (data curation, writing review and editing), M.L.B. (data curation, writing review and editing), M.L.C. (data curation, writing review and editing), R.A.R.F. (data curation, writing review and editing), L.S.D. (data curation, writing review and editing), M.O.X. (data curation, writing review and editing), M.E.F. (data curation, writing review and editing), M.B. (data curation, writing review and editing), M.T. (data curation), A.S.G.K.M. (writing review and editing), V.F.O. (conceptualization, data curation, methodology, formal analysis, writing original draft, writing review and editing), and M.M.C.M. (conceptualization, methodology, supervision, writing review and editing). All authors have approved the final manuscript draft. This article has not been published previously or has been submitted or subjected to consideration to any other publication.

Funding

This work received no funding.

Conflict of interest

M.M.C.M. has received support for attending educational meetings from Knight and Mundipharma. The remaining authors declare no conflict of interest.