1. Introduction
Aspergillus (
A.) species (spp.) are ubiquitous, opportunistic fungal pathogens that, when inhaled, are readily eliminated from the lung of immunocompetent individuals but can lead to the highly lethal infection invasive pulmonary aspergillosis (IPA) in immunocompromised individuals. Specifically, patients undergoing immunosuppressive therapies for stem cell and solid-organ transplants, hematologic malignancies (HM), and the use of immunomodulating drugs such as corticosteroids are at the most significant risk of developing IPA [
1,
2,
3,
4].
A. fumigatus is the species most frequently attributed to IPA; however, additional species are occasionally identified as the cause of IPA, including
A. niger,
A. flavus, and azole-resistant
A. terreus [
5,
6,
7]. Dissemination of these fungi most frequently occurs via the hematogenous spread of the fungus from the lungs to subsequent organs [
8,
9,
10]. The central nervous system (CNS) is often reported as the most frequent site of
Aspergillus dissemination from the lung resulting in cerebral aspergillosis (CA); this is particularly true in immunocompromised populations [
7,
10,
11]. In addition to being amongst the most common organs of dissemination,
Aspergillus infection in the CNS is also regarded as one of the most lethal [
12,
13,
14].
IPA is diagnosed in more than 300,000 immunocompromised patients annually and is associated with a 30–80% mortality rate [
2,
15,
16]. On average, 20–50% of IPA infections result in disseminated disease, with 10–20% reported to result in CA [
17,
18]. However, this is likely a conservative estimate as the population of CA is greatly under-reported, owning partially to the fact that CA is notably difficult to diagnose, with some cases not being diagnosed until autopsy [
19,
20,
21,
22]. Further, as the number of immunosuppressed patients continues to increase, the CA population is likely larger than reported [
23,
24,
25]. Disseminated CA is associated with a particularly poor prognosis, resulting in death in up to 70–100% of patients [
23,
25,
26]. The difficult diagnosis of CA additionally contributes to the high mortality as the symptoms, including fever, headache, mental alteration, or lethargy, etc., are non-specific [
5,
27]. The difficulty of diagnosis is also due, in part, to the methods which are often invasive and have variable sensitivity and specificity [
28]. Latency to diagnose combined with poor therapeutic tools often result in a fatal infection. Within the immunocompromised population, patients with hematologic malignancies (HMs) (i.e., cancers that affect the blood, bone marrow, and lymph nodes) are considered to be one of the most prevalent populations to be diagnosed with disseminated CA [
1,
29]. HM patient subgroups include various leukemias (acute lymphocytic (ALL), chronic lymphocytic (CLL), acute myeloid (AML), chronic myeloid (CML)), myeloma, and lymphoma (Hodgkin′s and non-Hodgkin′s (NHL)). Additionally, the therapeutics and treatments associated with HMs, such as chemotherapy and stem cell transplants (SCT), leave the patients in a highly immunocompromised state, elevating the risk for opportunistic infections. Given the high proportion of patients reported to have disseminated CA also having HMs, it is essential to identify the most at-risk HM patient subgroups and characteristics for CA so prophylactic measures and therapeutic considerations can be taken.
Up-to-date reports of disseminated CA specifically related to HMs are relatively limited. The risk factors of disseminated CA related to HM patient subgroups and their therapeutics, including cytotoxic drugs, steroids, SCTs, and targeted agents, are also not well documented. This systematic review clarifies the evidence base available around the relationship between HM patients undergoing therapy related to HMs and CA. Additionally, this review aims to identify any relationships between HM patient subgroups and the prevalence of CA, thus, potentially identifying patient subgroups with increased risk. Further, this review aims to identify the post-infection characteristics of CA patients. Lastly, this review addresses any relationships between various patient characteristics, disseminated CA, and mortality. We systematically reviewed the literature using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to address these objectives. A systematic review was performed on the selected study population comparing patients receiving chemotherapy, SCT, corticosteroids, and/or targeted therapies before diagnosis with IPA with disseminated CA. Post-infection characteristics were also compared, including anti-fungal treatment, surgical intervention, Aspergillus spp., and mortality. Studies included in this review were published on or before 18 May 2021, and were accessed through four online databases.
2. Materials and Methods
The systematic review was conducted according to PRISMA guidelines. The study protocol for this systematic review was registered with the PROSPERO database with the registration number CRD42021288469 [
30].
2.1. Search Strategy and Study Selections
Data sources used for this systematic review were PubMed/Medline, Embase, Cumulative Index to Nursing and Allied Health Literature (CINAHL) Plus, Web of Science, and GreyLit. All databases were searched from inception to 18 May 2021, and all relevant peer-reviewed studies published were included for systematic review. No limits were placed for language, publication type, etc., in the initial search. The literature search strategy combined all synonyms for the disease “cerebral aspergillosis” combined with “disseminated,” combined with all synonyms for diseases resulting from “hematologic malignancies,” and all synonyms describing “chemotherapy” and related “immunosuppression.”
Supplemental Table S1 contains all MeSH terms and keywords that comprise the search strategies used for each database.
All articles retrieved through database searching were imported into Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia), where duplicate records were automatically removed. In Covidence, studies were screened first by title and abstract review and second by full-text review by two independent reviewers (BNS and MAB) in duplicate. Following the first and second reviews, those studies kept and rejected were compared between reviewers, and any discrepancies were resolved by consensus.
2.2. Eligibility Criteria
Research articles that met the following defined inclusion criteria were selected for systematic review. Eligible studies needed to include patients with leukemia or lymphoma (any age/gender), diagnosed IPA + CA (proven, any Aspergillus spp.) with the primary infection being in the lung and without any concurrent infections by other yeasts, viruses, and/or bacteria, etc., and the outcome of fungal infection. Studies additionally needed to include one or all of the following regarding whether or not patient(s) included in the study were receiving chemotherapy, SCT, and/or another therapeutic for HM. Studies without patients with CA were excluded, and patients without CA in studies included were excluded from the analysis. Study design: any full-text peer-reviewed reports available in English containing original clinical data were considered; this primarily included case reports and series. Further, preprint articles and articles with no full text available were not included. The primary outcomes of interest were mortality and prevalence of CA within HM patient subgroups. Comparison or control groups were not applicable.
2.3. Data Abstraction
Data were abstracted independently and in duplicate by two reviewers using standardized data extraction criteria for case reports and studies. The Covidence systematic review software was utilized for data abstraction for case reports. For case series, Google Sheets was used. For case series in which cohort data were additionally available, individual patient data were preferentially used as they provided more-detailed information about underlying HMs, treatment, and outcomes. Two independent investigators (BNS, MAB) abstracted the following data, when available, from eligible articles: general study information (including title, authors, PMID, country study was conducted in, year of infection diagnosis, and year of publication), study characteristics (case study versus series), participant characteristics (including age, gender, type of leukemia or lymphoma, neutropenic status, absolute neutrophil count (ANC) or white blood cell count (WBC), and additional sites of dissemination if any), information about the interventions (including chemotherapy regimen, SCT, any non-cytotoxic therapeutics pre- or post-IPA, prophylactic anti-fungal regimens, therapeutic interventions for aspergillosis, and surgical interventions), type of Aspergillus, and outcome measures (survival). Abstracted data were compared between the two reviewers, and any discrepancies were resolved by consensus. Upon resolving discrepancies, data were synthesized into a single form that was maintained on Google Sheets.
2.4. Assessment of Study Quality
The included publications were assessed for risk of bias for selection, ascertainment, causality, and reporting based on the modified Pearson Case Report Quality scale proposed by Murad and colleagues [
31]. For each bias domain, levels of bias were rated as high, low, or unclear, based upon the response of no, yes, or unclear, respectively, to the prompting questions. The overall risk of bias of a study was deemed low if the study had a low risk of bias for all domains. The overall risk of bias was considered unclear if a study had an unclear risk of bias for at least one domain. Lastly, the overall risk of bias was deemed high if a study had a high risk of bias for at least one domain. All responses were recorded through Covidence systematic review software (
Supplemental Table S2). The consensus of quality was reached by two independent researchers (BNS, MAB) for each study.
2.5. Data Synthesis and Analysis
A narrative summary approach was used to detail the key study characteristics and systematic review findings. As each study represented an individual patient or patients, data were synthesized and described in this way. The data were pooled to determine the prevalence of underlying HM patient subgroups, treatments, outcome, and other pertinent variables in the patient population. In some analyses, studies were excluded if relevant data were not available. For this reason, the number of patients varies in each analysis. Further, due to heterogeneity in study design, statistical analysis of the data collected from the 47 studies was not undertaken.
4. Discussion
To our knowledge, this is the first comprehensive systematic review of the literature focusing on disseminated CA following IPA in patients with HMs. In this study, we examined the characteristics of a large number of disseminated CA cases following IPA in HM patients published as single case reports, case series, or as a part of larger observational studies. All cases included had a proven Aspergillus spp. infection. The mortality rate due to CA was 53.95% overall for patients included in this systematic review.
Like studies focusing on IPA in patients with HMs, we found the predominant HM patient subgroup diagnosed with disseminated CA to be AML, closely followed by ALL [
78,
79,
80,
81]. Interestingly, estimates of the global incidence and prevalence of HMs have demonstrated the top reported HM patient subgroups to be NHL and CLL; however, in studies of IPA, and disclosed within this systematic review, have demonstrated those to be of mid-level prevalence. Conversely, the most prevalent HM patient subgroups for IPA and disseminated CA, AML and ALL, are globally regarded as mid-to low-level prevalence [
82,
83]. This suggests that the high prevalence of AML and ALL patients diagnosed with IPA or IPA coupled with disseminated CA is potentially due, at least in part, to the anti-cancer therapeutic regimen(s) given to those patients. In fact, several reports have linked the prevalence of IFIs and the chemotherapeutic regimens used in the acute leukemia populations [
84,
85]. Indeed, the top two cytotoxic drugs reported in the cases included in this systematic review were cytarabine and daunorubicin, longstanding chemotherapeutics for acute leukemia. However, in our systematic review of the literature, we did not find anti-fungal prophylaxis to be more prevalent in the acute leukemia population compared to other HM patient subgroups; rather, it was less than other populations, like those with NHL (
Table 4). Altogether, more aggressive monitoring prevention, and implementation of anti-fungal drugs into the therapeutic regimen of HM patients, particularly those with acute leukemias, is likely required for the prevention and/or reduction of the highly fatal CA.
Immunosuppression related to the treatment of HMs has long been considered a primary risk factor for IPA. Historically, immunosuppression in patients with HMs has been related to (i) prolonged neutropenia, primarily resulting from the use of chemotherapeutic agents; (ii) immunosuppressive drugs for the prevention and/or treatment of graft versus host disease (GvHD) following allogeneic hematopoietic-SCT; and (iii) corticosteroids prescribed for a range of indications during cancer care, including the reduction of chemotherapy side-effects, anticancer effects, and as a non-specific immunosuppressant following SCT. On average, 50–90% of IPA patients with underlying HMs received chemotherapy before infection [
79,
86]. Likewise, 88.06% and 78.43% of IPA patients with disseminated CA patients included in this systematic review of the literature received chemotherapeutic agents and were neutropenic, respectively, prior to infection. SCT is conducted in about 20–35% of HM patients with IPA, with allogenic being more prevalent than autologous [
18,
79,
80]. In the population of HM patients with disseminated CA following IPA, similar results were found, with about 36% of patients receiving SCT prior to infection, most of whom received allogenic SCTs. In studies of IPA, approximately 25–45% of patients with underlying HMs were reported to be receiving corticosteroids at the time of infection [
18,
80]. Interestingly, greater than 60% of the CA patients included herein were prescribed corticosteroids at the time of infection. The elevated prevalence of corticosteroids in HM patients with CA disseminated from IPA compared to HM patients with IPA alone points to a potential factor that increased the susceptibility of developing disseminated CA. Although, it is difficult to draw definitive conclusions related to specific treatments and their impact on developing CA as there is no way to account for all potential variables. The data presented here indicate that HM patients with corticosteroids included in their anti-cancer therapy should be closely monitored and receive prophylactic anti-fungal drugs to prevent the development of this severe disease.
Recently, targeted anti-cancer therapies have become more frequently attributed to increased risk of IFIs, including IPA [
87]. One of the most prominent targeted therapies is ibrutinib, a bruton tyrosine kinase (BTK) inhibitor, primarily prescribed to CLL and NHL patients. Ibrutinib is used as a single-agent therapy or as a part of combination therapy with other anti-cancer drugs such as rituximab, an anti-CD20 monoclonal antibody. Although CLL and NHL have not historically been considered as high-risk for developing IPA, the addition of ibrutinib and/or rituximab has been associated with increased prevalence of IPA in these patients [
88]. Herein, we report that 85.70% of CLL and 36.36% of NHL patients were given ibrutinib, frequently given in combination with rituximab and/or corticosteroids. Most patients receiving ibrutinib at the time of infection had received chemotherapy prior to initiating ibrutinib [
33,
42,
45,
56,
71,
75]. Only two patients were treatment naïve prior to ibrutinib therapy, and one patient began ibrutinib co-currently with chemotherapy [
33,
42,
72]. Recently, there have been several reports of CA in patients receiving ibrutinib [
89,
90,
91,
92,
93,
94]. While the number of reports at this time is relatively small, it bears noting as historically, the number of CLL patients diagnosed with CA has been relatively low in comparison to patients with other HMs. Thus, this indicates the importance of investigating the incidence of CA in patients receiving ibrutinib therapy to potentially identify an at-risk population.
Other targeted therapies, immunotherapies, and biologics given to HM patients prior to fungal infection, such as L-asparaginase, immunoglobulins, and venetoclax, among others, were administered to only a few patients, which does not permit us to make any inferences regarding their influence for the susceptibility for disseminated CA in this population. However, given the mechanisms of action of some of these drugs and previously published reports, it is reasonable that they could have impacted the immune status of patients [
87,
95,
96]. More reports on non-chemotherapeutic drugs given to HM patients are required to draw any substantial conclusions.
Anti-fungal prophylaxis has become a common addition to the treatment regimen of HM patients, with and without SCTs [
97,
98,
99]. The addition of anti-fungal prophylaxis is thought to contribute to the overall reduction of IPA cases amongst immunocompromised individuals [
97,
98]. Further, anti-fungal prophylaxis in HM patients is a positive predictor of survival in breakthrough cases of IPA [
97,
100]. The number of HM patients prescribed antifungal drugs prophylactically typically ranges from ~15–45% [
101,
102]. It should be noted, however, that the prevalence of antifungal prophylaxis in HM patients is on the rise with the development of new antifungal drugs and repeated demonstration of the efficacy of using these drugs prophylactically [
86,
103,
104]. Here, we report that 47.91% of HM patients had breakthrough IPA with disseminated CA. Although, data for this were only retrievable from two-thirds of the studies, which is likely attributable to the age of some studies and the lack of antifungal prophylaxis.
Posaconazole is a prophylactic antifungal that has been consistently found to be the most effective at preventing IPA, with as little as 1% of neutropenic patients on prophylactic posaconazole with breakthrough IPA [
86,
97]. No patients included in this systematic review received posaconazole prophylactically. Instead, the majority of patients received AmB, followed by fluconazole (
Table 3), both of which have been found to be less efficacious in preventing IPA, notably fluconazole [
86]. However, it is unknown whether the infection was due to wrong anti-fungal drug choice, insufficient drug levels in the CNS or host, or fungus-specific issues and thus warrants further investigation.
Historically, AmB has been considered as the standard of care for patients with IPA. However, one study compared voriconazole with AmB as primary treatment for IPA infections and overall exhibited improved survival and response rates [
105]. Improved response with voriconazole was further demonstrated through higher successful outcomes in HM patients, patients with extrapulmonary involvement, and others suggesting voriconazole to be superior to AmB at ameliorating
Aspergillus driven infections. Further still, treatment with voriconazole resulted in significantly fewer adverse events. Another study examining the inclusion or exclusion of voriconazole in the treatment of IPA in HM patients found the overall mortality of those receiving voriconazole to be 5%, significantly lower than the 49% mortality rate associated without voriconazole [
100]. By and large, in the population of disseminated CA disclosed herein, AmB was the number one therapeutic prescribed, whether singularly or in combination. The second most prescribed in the patients included within this systematic review was voriconazole, which was often given in combination with AmB and/or other anti-fungal drugs such as caspofungin and posaconazole. Generally, the inclusion of voriconazole reduced the overall mortality of disseminated CA. The inclusion of voriconazole with or without AmB was associated with ~30% mortality, while AmB in the absence of voriconazole was associated with ~75% mortality. While it is difficult to draw conclusions due to the inability to exclude confounding factors, the reduction of mortality associated with voriconazole suggests its therapeutic potential for CA and warrants further investigation.
In agreement with previously published cases of IPA in HM patients,
A. fumigatus was the most common isolate identified in this systematic review [
80,
81,
106,
107]. Here,
A. flavus was the second most common isolate identified in HM patients with IPA disseminated to CA. However, the trends observed in reports of HM patients with IPA are inconsistent, with some reporting
A. flavus or
A. terreus as the second most common isolate identified in HM patients with IPA [
18,
80,
100,
106]. In the cases included in this systematic review, only one report of infection by
A. terreus was identified. Of note, while the population of
A. flavus was less than half of the
A. fumigatus population, infection by
A. flavus was associated with a 90% mortality, approximately double that of
A. fumigatus (
Table 9).
Limitations
One of the primary limitations of the study was missing data. While most single case series provided adequate detail about patient history, treatment regimens, and outcome, this was not always the case for the case series and observational studies. Additionally, while many studies detailed whether, for example, chemotherapy and corticosteroids were included in patient treatment regimens, details on the type, dosing, and duration were often excluded. More to this point, disclosure of ANC levels was frequently neglected, despite neutropenia being a well-established risk factor for infection that is often resultant from chemotherapy. Further, several studies included were published over a decade ago, and thus missing data could not be retrieved. A limitation of the wide range of dates in which the studies were conducted is that therapeutic standards have changed vastly with advancements in modern medicine, thus often making it difficult to make direct or meaningful comparisons. An additional limitation pertaining to the range of dates of the studies included is that the tools and criteria for diagnosing proven Aspergillus infection have changed throughout time, thus we had to rely on standards appropriate for the time of diagnosis and best judgment to determine whether a case met our stringent criteria for inclusion. In doing so, it is possible that articles were excluded or included when others would not have made that judgment, thus introducing potential bias.
Further, during the screening process, many studies with CA patients had to be excluded because they did not provide adequate information about patients included with CA. Rather, the characteristics provided were for all patients, and thus no population data specific to CA could be retrieved from those articles. For this reason, single case reports, case series, and observational studies in which individual data could be retrieved were preferentially used. Ultimately, the lack of patient data at that level severely limited the number and types of articles that could be included, and thus it is possible some important information was excluded. Further, as we were unable to retrieve CA cohort data from larger studies investigating aspergillosis, a meta-analysis was not able to be conducted. Due to this, we were unable to potentially identify distinguishing factors amongst patient cohorts with disseminated CA that may have provided critical information to better identify high-risk populations. Therefore, for improved analysis of this population and potential identification of critical risk factors, more articles are required that distinguish and detail cohort characteristics for those with disseminated CA in the invasive aspergillosis population.