Impact of next-generation sequencing on antimicrobial treatment in immunocompromised adults with suspected infections

Abstract

BACKGROUND:

Prompt pathogen identification can have a substantial impact on the optimization of antimicrobial treatment. The objective of the study was to assess the diagnostic value of next-generation sequencing (NGS) for identifying pathogen and its clinical impact on antimicrobial intervention in immunocompromised patients with suspected infections.

METHODS:

This was a retrospective study. Between January and August 2020, 47 adult immunocompromised patients underwent NGS testing under the following clinical conditions: 1) prolonged fever and negative conventional cultures; 2) new-onset fever despite empiric antimicrobial treatment; and 3) afebrile with suspected infections on imaging. Clinical data, including conventional microbial test results and antimicrobial treatment before and after NGS, were collected. Data were analyzed according to documented changes in antimicrobial treatment (escalated, no change, or de-escalated) after the NGS results.

RESULTS:

The median time from hospitalization to NGS sampling was 19 d. Clinically relevant pathogens were detected via NGS in 61.7% of patients (29/47), more than half of whom suffered from fungemia (n=17), resulting in an antimicrobial escalation in 53.2% of patients (25/47) and antimicrobial de-escalation in 0.2% of patients (1/47). Antimicrobial changes were mostly due to the identification of fastidious organisms such as Legionella, Pneumocystis jirovecii, and Candida. In the remaining three cases, NGS detected clinically relevant pathogens also detected by conventional cultures a few days later. The antimicrobial treatment was subsequently adjusted according to the susceptibility test results. Overall, NGS changed antimicrobial management in 55.3% (26/47) of patientst, and conventional culture detected clinically relevant pathogens in only 14.9% of patients (7/47).

CONCLUSION:

With its rapid identification and high sensitivity, NGS could be a promising tool for identifying relevant pathogens and enabling rapid appropriate treatment in immunocompromised patients with suspected infections.

INTRODUCTION

Infections are the main challenge for hospitalized immunocompromised adults, especially for patients with hematological malignancies receiving chemotherapy or peripheral bone stem cell transplantation.[1,2] Early initiation of empiric antibiotics is recommended because infections can progress rapidly in immune-incompetent patients.[3] Although conventional cultures (for instance, blood cultures) are the major diagnostic standard for infections, its low detection rate and long incubation time hinder the optimization of antimicrobial treatment.[4] Next-generation sequencing (NGS) can analyze all microorganisms within one sample and is time-saving because it is culture-free. Previous studies have reported the application of NGS for identifying pathogens in immunocompromised children with bloodstream infections and septic patients with bacteremia.[5,6] However, the application of NGS for the identification of pathogens in adult immunocompromised patients with suspected cases is still lacking. In the present study, we demonstrated the application of NGS in immunocompromised adults with suspected infections.

METHODS

Study participants

This study reviewed data from immunocompromised adults with suspected infections from Tongji Hospital between January and August 2020. Immunocompromised patients were defined as those who had received a solid organ or stem cell transplant or those who had an underlying disease that compromised their immune system (such as long-term use of corticosteroids and hematological malignancy).

Blood samples were sent for NGS tests under the following clinical circumstances: 1) prolonged fever (> 7 d) with negative conventional microbial tests; 2) new-onset fever despite broad-spectrum empiric antimicrobial treatment (> 7 d); and 3) afebrile patients with suspected infections on chest X-ray or computed tomography (CT) (radiological reports with “suspected infections”). Conventional microbial tests included sputum, blood, urine, and fecal cultures. Fever was defined as an axillary temperature ≥ 37.5 °C. The study flowchart is shown in Figure 1.

Figure 1.

Study flowchart. NGS: next-generation sequencing.

Study design

This retrospective study was approved by the hospital’s Institutional Review Board (TJ-IRB20220404). Due to its retrospective nature, patient consent for inclusion was waived. Clinical data, including conventional microbial test results and antimicrobial treatment before and after NGS, were obtained from the electronic chart.

Data were analyzed according to documented changes in antimicrobial treatment (escalated, no change, or de-escalated) after the NGS results. Antimicrobial escalation was defined as either a change to treatment or prolonged treatment due to the NGS results. Antimicrobial de-escalation was defined as narrowing the antimicrobial spectrum or antimicrobial discontinuation. No change was defined as either no change in treatment or change due to the results of conventional tests other than NGS. This study evaluated the real-time impact of NGS on antimicrobial management; thus, only each patient’s attending physician made clinical decisions according to the pathogens identified.

NGS of infectious pathogens

In this study, the Karius test (Karius, Inc., USA) was applied to detect microbial cell-free DNA in blood samples from all patients. Blood samples from each patient were collected in BD Vacutainer plasma preparation tubes (BD Bioscience, USA) and delivered to Beijing Genomics Institute laboratory for processing. NGS libraries were performed on the Illumina NexSeq500 (Illumina, USA). Infectious disease experts (two clinicians specializing in intensive care and clinical microbiology) were involved in the interpretation of all NGS results.

RESULTS

Patient characteristics

The demographic and clinical characteristics of the patients are summarized in Table 1. All patients had negative conventional blood or fluid culture results at the time of NGS sampling. Among the 57 patients screened, 47 patients with NGS results met the inclusion criteria. The patients had a median age of 48 years (interquartile range [IQR] 32–58 years), and 35 were males (74.5%). Three were post-renal transplantation patients, two were nephrotic syndrome patients receiving prednisolone treatment, and the remaining 42 patients had hematological diseases (18 leukemia patients, 18 lymphoma patients, three multiple myeloma patients, two myelodysplastic syndrome patients, and one aplastic anemia patient receiving reduced-intensity conditioning for intended peripheral blood stem cell transplant [PBSCT]).

Table 1.

The clinical characteristics of all patients

The median time between hospitalization and NGS sampling was 19 d (IQR 5–31 d). NGS was performed for 38 cases (80.9%) because of prolonged fever with negative conventional culture results, and the other 9 fever-free patients (19.1%) were treated due to suspected infections on chest imaging. Laboratory tests revealed that these patients had neutropenia, thrombocytopenia, and a severe inflammatory response. Advanced antibiotics are commonly used against gram-negative and gram-positive bacteria.

After a long hospital stay of more than one month, 43 patients survived at discharge, and the remaining four patients progressed to septic shock and eventually died.

NGS yielded more positive results than conventional culture tests

In our study, 31.9% (15/47) had positive results with conventional culture tests (CMTs). In contrast, microbes were found in all NGS samples, although 61.7% (29/47) of the results were considered clinically relevant (Supplementary Tables 1 and 2).

Of the 15 positive CMTs with NGS results, five (33.3%) were positive for typical skin commensals, such as coagulase-negative Staphylococci or Micrococcus luteus. Four (26.7%) patients were found to suffer from fungemia caused by Candida parapsilosis, Candida tropicalis, or Pneumocystis jirovecii. The predominant species in positive CMTs was Stenotrophomonas maltophilia (20%; n=3). Ten species were uniquely identified by CMTs; three were fungi, and the remaining seven were bacteria (42.9% gram-positive and 57.1% gram-negative).

Thirteen species identified by NGS were considered clinically relevant. Over half of the 29 patients with positive NGS results were found to suffer from fungemia (58.6%; n=17), which was caused predominantly by Candida parapsilosis or Pneumocystis jirovecii. Seven of the 13 species identified by NGS were fungi, and the remaining six were gram-negative bacteria. Typical skin commensals, such as coagulase-negative Staphylococci, were not identified by NGS (Figure 2).

Figure 2.

Comparison of positive results from conventional culture tests (CMT) and NGS.

Interventions based on the NGS results

The NGS test results and antimicrobial interventions based on these results are summarized in Table 2 and Supplementary Table 3. Clinically relevant pathogens were detected via NGS in 29 patients, which resulted in 25 cases of antimicrobial escalation (53.2%) and one case of antimicrobial de-escalation (antifungal treatment). In the 18 cases in which NGS failed to detect clinically relevant pathogens, no changes were made in antimicrobial treatment.

Table 2.

Antimicrobial intervention based on NGS results vs. conventional tests

The NGS results of one patient confirmed that Candida parapsilosi infection was de-escalated. In this patient, the serum G test for detecting (1, 3)-β-D-glucan was positive, which suggested a fungal infection. Empiric antifungal treatment involves a combination of two antifungal drugs. Monotherapy was used for this patient since NGS identified clinically relevant pathogens. In addition, in this patient, Micrococcus luteus, which was detected by blood culture, was considered a contaminant.

In the remaining three patients, NGS detected clinically relevant pathogens that were also detected by conventional culture a few days later. In Patient 23, NGS confirmed Stenotrophomonas maltophilia, and antibiotics were not adjusted since sulfonamide was already in use. Blood culture revealed the same pathogen, and levofloxacin was added based on the susceptibility results. NGS of the other two patients revealed bacteria (Klebsiella pneumoniae in Patient 38 and Pseudomonas aeruginosa in Patient 40). Because of the high suspicion of drug-resistant bacterial infections, the antibiotics were adjusted until the susceptibility results of the cultures were available.

Moreover, NGS plays an important role in the identification of opportunistic pathogens and fastidious organisms. For example, NGS identified five patients with Pneumocystis jirovecii infections, which resulted in antimicrobial escalation. Of these patients, one had nephrotic syndrome and was receiving prednisolone treatment, and the other four had lymphoma and were receiving various treatments; only one lymphoma patient had a positive sputum-polymerase chain reaction (PCR) test for Pneumocystis jirovecii. In addition, two leukemia patients with low reads (<3) of Pneumocystis jirovecii of NGS did not receive anti-Pneumocystis jirovecii treatment. Legionella infections were detected by NGS in three patients, which triggered anti-Legionella treatment. The first patient (Patient 1) was a lymphoma patient who received PBSCT one year prior and suffered from diarrhea and fever for more than 7 d. The second patient (Patient 36) was a kidney recipient who suffered from fever for more than two weeks and had suspected infections on chest CT. The last patient (Patient 45) had leukemia and underwent chemotherapy, who suffered from neutropenia and had fever for more than 5 d and suspicious infections on chest X-ray. In the BAL analysis of the last patient, Stenotrophomonas maltophilia were detected; however, no antimicrobial adjustment was made based on these results. Candida parapsilosis was detected by NGS in six patients and resulted in antimicrobial escalation in four patients, two of whom also had Candida parapsilosis detected in blood cultures.

NGS also identified cytomegalovirus (CMV), Epstein-Barr virus (EBV), and herpesvirus. At the time of NGS testing, a few patients were already receiving CMV therapy. Only three cases started CMV treatment after a positive NGS result. Specifically, in one case whose NGS results was EBV and Candida parapsilosis positive, the patient had leukemia with PBSCT one year ago and suffered from oral ulcers and fever for 2 d. In this patient, Candida parapsilosis explained his clinical presentation; thus, anti-Candida parapsilosis treatment was started.

Supplementary Table 3 lists the NGS and conventional testing results, and the antimicrobial intervention used in all the 47 patients.

DISCUSSION

Our study was consistent with other studies that have shown the application of NGS in identifying clinically relevant pathogens in immunocompromised patients.[5,7] In our study, despite the use of current diagnostic techniques, only 14.9% (7/47) of the cultures from hospitalized immunocompromised patients with suspected infections were able to identify an pathogen. Another study of hospitalized patients with suspected community-acquired pneumonia showed that approximately 60% of cultures failed to detect a pathogen.[8] A total of 40% to 50% of critically ill patients with sepsis had negative culture results, which was attributed to the presence of fastidious organisms or prior antibiotic therapy.[9] In our study, the NGS test for the assessment of immunocompromised patients with suspected infections resulted in antimicrobial adjustment in 55.3% of the patients. Moreover, NGS is crucial for identifying fastidious organisms such as pneumocystis, Legionella, and fungi. Our findings were consistent with those of a prospective observational study of NGS testing in hematological malignancy patients with neutropenic fever and negative conventional cultures,[10] which suggested that antimicrobial therapy should be adjusted for approximately 70% of the patients if the NGS results were applied to real clinical decisions. However, the clinical value of NGS results was challenged by another study, which found that only 7.3% of NGS tests had an impact on antimicrobial intervention.[11] The research population, study design, and definition of clinical impact are the variations between the two studies that could account for the disparate outcomes.

NGS has been observed a higher positivity rate for infectious pathogens than conventional microbial tests. A study of 78 critically ill patients showed that the diagnostic sensitivity was significantly increased from 12.82% (10/78) by blood culture alone to 30.77% (24/78) for NGS alone.[12] A retrospective cohort study of immunocompromised pediatric patients demonstrated that NGS had a higher sensitivity (93%) than conventional tests (76%).[13] In a prospective study with a small sample size, NGS results demonstrated positive agreement with those of conventional microbial tests in 70% (7/10) of immunocompromised patients with febrile neutropenia, pneumonia, or intra-abdominal infection.[7] In our study, the positivity rate of NGS for detecting clinically relevant pathogens was 61.7% (29/47). This high positivity rate and rapid detection suggest that NGS can help diagnose opportunistic infections in immunocompromised adults. In contrast to its high sensitivity, the specificity of NGS appears to be relatively low. One study demonstrated that NGS had lower specificity than conventional tests (59% vs. 92%).[13]

Our study showed that NGS performed better at detecting fungi in immunocompromised adults than conventional cultures (for instance, Aspergillus, 5/47 vs. 0/47). The low positivity in cultures for fungi, may have been due to several factors. First, the growth requirements for fungi in blood cultures vary between strains; for example, yeasts usually grow best at 37 °C, while most filamentous fungi grow best at lower temperatures (27–30 °C). In addition, molds often grow poorly in blood culture systems. A retrospective study of the blood cultures of 1,453 patients who underwent hematopoietic stem cell transplantation showed a low detection rate of Aspergillus (1.3%, 19/1,453); these positive results had ambiguous clinical significance because only one of the 19 patients had invasive aspergillosis.[14] In contrast, NGS showed a much higher detection rate and great clinical significance. A study of 40 patients from the high-risk group for invasive fungal diseases (IFDs) showed that NGS detected fungal pathogens in seven patients (17.5%), and the results were identical in four of the six proven IFD cases.[15] Another study showed that plasma NGS could be used to detect the same fungus in the biopsy tissue from seven of nine patients with proven IFD.[16] Although exact data on the NGS’s detection rate of fungi in at-risk patients with fungal infection are still lacking, a specific protocol for detecting fungi via NGS has been published. The protocol introduced a newly designed “fungal SIQ score”, i.e., the fungal sepsis indicating quantifier score, which allows unambiguous identification of fungus when interpreting NGS results in the plasma of septic patients, which would also help to distinguish invasive fungal infections from colonization in culture-positive fungus results.[17]

Blood samples from three patients were recovered with Staphylococcus hominins. The results were considered contamination since only one bottle was positive for this pathogen. Staphylococcus hominins were not detected through NGS in the same patients, which suggest that NGS may be less contaminated by gram-positive skin colonizing bacteria.

There are some limitations of our study. Although NGS may be more sensitive and specific than blood culture, additional clinical trials are needed to determine the positive and negative predictive values, as this study was limited by its single-center nature and small sample size.

CONCLUSION

This study indicated an almost two-fold higher positivity rate of NGS over conventional culture for identifying pathogens in immunocompromised patients. The NGS results led to a change in antimicrobial therapy in more than half of the study population. NGS excelled in detecting fungi compared with conventional culture, optimizing antifungal therapy. Therefore, this study will provide valuable data to support physicians in treating immunocompromised patients with suspected infections.