Invasive fungal disease (IFD) is a common complication after hematopoietic stem cell transplantation (HSCT) and a leading cause of transplant-related mortality (Pagano et al., 2007; Sun et al., 2015). Patients experience conditioning regimens, agranulocytosis, impaired mucosal barrier, use of central venous catheterization, graft-versus-host disease (GVHD), immunosuppression, and delayed immune reconstitution, which all significantly increase the risk of IFD (Kuster et al., 2018; Sun et al., 2021). Advances in microbiological techniques and antifungal drugs have resulted in improvements in IFD diagnosis and treatment (Akan et al., 2013; Tissot et al., 2017; Chinese Association, H and Chinese Invasive Fungal Infection Working, G, 2020; Ruhnke et al., 2020; Chinese Society Of Hematology and Chinese Medical Association, Antimicrobial Infection Branch, 2023). The specific symptoms and adequate diagnostic procedures for IFD remain lacking, resulting in delayed diagnosis and treatment of IFD and a poor prognosis of patients. Early detection of IFD and prompt initiation of appropriate treatment are critical factors in the survival of patients following HSCT (Puerta-Alcalde and Garcia-Vidal, 2021)

Standard diagnostic techniques are clinically challenging for IFD diagnosis because of their invasiveness, long detection period, and lack of sensitivity, specificity, and species identification (Ostrosky-Zeichner, 2012). The gold standard for diagnosing IFD is culture-based testing; unfortunately, sterile specimens frequently necessitate potentially invasive procedures (Donnelly et al., 2020). Moreover, serum biomarkers, including galactomannan (GM) and 1,3-β-D-glucan (BDG), are the adjunct to clinical diagnosis of IFD (Guo et al., 2010; Lu et al., 2011a; Li et al., 2015; Donnelly et al., 2020). However, GM and BDG cannot detect all fungal pathogens, and their diagnostic efficacy in HSCT varies (Ullmann et al., 2018; Warris et al., 2019). According to the European Conference on Infections in Leukemia guidelines, serum BDG can help diagnose Pneumocystis jirovecii pneumonia (Maertens et al., 2016). However, GM performance remains significantly lower in non-neutropenic patients and/or those receiving prophylactic therapy (Wu et al., 2021). Radiographic findings may aid in the diagnosis of fungal lung lesions in patients with fever and neutropenia (FN). Nonspecific observations, on the other hand, often lead to overdiagnosis and incorrect treatment. Although PCR-based technologies have been shown to be effective in the identification of some fungal pathogens, false-positive and false-negative results limit their widespread application (Lu et al., 2011b; Trubiano et al., 2016; Ala-Houhala et al., 2018).

Metagenomic next-generation sequencing (mNGS) is a promising culture-independent technique that has been extensively used to diagnose infections (Gabaldón, 2019; Irinyi et al., 2020). mNGS may detect many types of microorganisms in a microbial sample both rapidly and concurrently (Barreda-García et al., 2018) while also recognizing non-culturable microbes (Yang et al., 2022). The application of mNGS to identify fungal infections has recently increased (Alonso et al., 2018; Shivaji et al., 2019; Chien et al., 2022). However, there have been relatively few investigations on the use of mNGS for diagnosis in patients with IFD following HSCT, and its clinical application has not been standardized. Owing to the severity and specificity of IFD after HSCT, efforts to optimize pathogen detection technologies are important for a favorable prognosis. This study examined the efficacy of standard mNGS technology in detecting infections in patients with IFD after HSCT by comparing the diagnostic performance of mNGS and conventional microbiological testing (CMT). In addition, we describe IFD diagnoses, survival, and risk factors for IFD following HSCT.

IFD is a common complication of HSCT that is associated with high morbidity and mortality rate (Neofytos et al., 2009; Sun et al., 2015). Although IFD treatment has advanced significantly, delayed diagnosis causes significant morbidity and mortality (Ostrosky-Zeichner, 2012). Substantial advances in the diagnostic approaches for IFD include radiographic findings, biomarkers (e.g., GM and BDG), and PCR assays (Greene et al., 2007; Arvanitis et al., 2015). However, CMT may not be appropriate for rare fungi, particularly those that are difficult to culture, resulting in delayed detection, whereas mNGS can directly execute DNA or RNA sequencing (Tsang et al., 2021). Jiang et al. (2022) report that mNGS is increasingly being used to treat fungal infections.

The predominant fungal species in our study were Aspergillus (14 of 46, 30.43%) and Rhizopus (8 of 46, 17.39%). Candida and Pneumocystis accounted for 15.22% (7 of 46) of the infections. Owing to the different characteristics of immune function damage, the pathogens of IFD also differ at different stages after transplantation. Twenty (43.48%) of the 46 patients developed IFD during the conditioning period to the engraftment period, five (10.87%) within 100 days after transplantation, and 21 (45.65%) cases occurred more than 100 days later.

In our study, mNGS technology had distinct advantages over CMT in diagnosing different types of fungal species. The detectability increased significantly from 36.96% to 84.78%. The detection time was significantly reduced using mNGS. mNGS also provides accurate identification of specific fungal pathogens and is more specific than CMT. The mNGS results for 15 patients were consistent with those of CMT. The quick and accurate diagnosis of fungal species is critical for selecting appropriate antifungal agents, avoiding resistance, and managing patients (Omrani and Almaghrabi, 2017; Wang et al., 2020; Stemler et al., 2022). Pathogens are distributed in a variety of tissues and organs and have the ability to detect variations in different types of samples. The sensitivity of peripheral blood samples to CMT, particularly blood cultures, was significantly lower than that of mNGS; however, for specimens from possible pathological tissues, the sensitivity of mNGS and CMT was comparable.

In two incidents the CMT results in this study were positive, but the NGS produced a false negative in two patients. The specific analysis was as follows: mNGS detected human beta-herpesvirus 5 and torque teno virus in P1, whereas blood culture detected Candida albicans. Further analysis of the original mNGS data revealed that mNGS detected Candida albicans using an updated database. The second incident involved P6, where a positive GM test in the BALF was accompanied by an abnormal chest CT scan, which improved after receiving antifungal therapy.

Sometimes, the CMT results can indicate that the patient experienced suspected IFD following HSCT; however, the identification of fungal species is nonspecific. mNGS has low sensitivity when: 1) the pathogen load is below the detection limit or is filtered owing to low-ranking readings; 2) the pathogen’s nucleic acid is easily degraded; 3) the pathogen is suspected to be a background microorganism; and 4) the dataset or database has limitations. Further reasonable interpretations of the results are then required. The mcfDNA-Seq assay and serum GM can generally provide complementary results for the diagnosis of invasive pulmonary mold infections following HSCT (Hill et al., 2021). As proposed for P. jirovecii infection following HSCT (Liu et al., 2021), mNGS and BDG may aid in distinguishing colonization from infection. By using CMT in conjunction with mNGS technologies, the vision of a “one-stop” for IFD diagnosis seems promising in the foreseeable future.

Despite a decline in IFD-related deaths over the past decade, IFD remains a significant limiting factor for effective HSCT (Neofytos et al., 2009; Kontoyiannis et al., 2010). As a result, patients with suspected IFD after HSCT are frequently given prophylactic or empirical antifungal treatments. Most of the patients (76.09%) in our study received prophylactic antifungal therapy, with voriconazole being the most widely used drug. Although most patients received antifungal prophylaxis, many patients developed GVHD or required corticosteroids and immunosuppressants following HSCT. Among the patients with Pneumocystis infection, 71.43% occurred withrow the sulfamethoxazole, and all patients were still successfully treated in the subsequent therapy with sulfamethoxazole in combination with or without other antifungal drugs. This suggests that prophylactic antifungal therapy lowers the risk of IFD. According to the NGS results, dose adjustments or medication changes on the basis on empiric antifungal treatment, occurred in 56.52% of patients with IFD. The estimated 4-year OS was 71.55%, whereas the mOS of patients was 59.07 months. Eight patients died because of IFD. In this investigation, corticosteroid use (HR, 8.008; 95% CI, 3.550–18.063) and hypoproteinemia (HR, 5.121; 95% CI, 2.245–11.681) were identified as independent risk factors for IFD using multivariate logistic regression.

This study has several limitations. First, being a retrospective study, there were some inherent limitations, such as information bias and selection bias. The sample size of our study was relatively small. In addition, mNGS was usually performed only once owing to its high cost. Finally, there was no information on antifungal susceptibility or resistance. Further prospective randomized controlled studies with larger sample sizes are needed to be designed to elucidate the diagnostic value of mNGS for IFD following HSCT according to STARD 2015.

In conclusion, our study demonstrated the feasibility of mNGS in patients with IFD following HSCT. We recommend that clinicians use mNGS technology more actively when there is a clinical suspicion of IFD after HSCT. For clinical infection related samples, the traditional microbiological detection should be improved first. Pathology and sterile specimen culture are still the gold standard for infection diagnosis. Pathogenic mNGS is a powerful supplement and extension, which may be a helpful adjunct to the clinical diagnosis of IFD, allowing for more rapid and focused antifungal treatment.

The datasets presented in this study can be found in online repositories. The name of the repository and accession number can be found below: NCBI: PRJNA1085906.

YZ and NW designed, supervised the clinical study, and revised the manuscript, who contributed equally to this work and should be considered corresponding authorship. XZ analyzed data and drafted the manuscript. LZ collected clinical data. YL was responsible for patient management. All authors contributed to the article and approved the submitted version.