Torque Teno Virus or Herpesviruses Detection By Metagenomic Next‐Generation Sequencing Predicts In‐Hospital Major Adverse Events in Critically Ill Patients With Severe Infections

ABSTRACT

Viral detection occurs frequently in critically ill patients. Patients with multiple viremic events had a higher ICU mortality. The highly sensitive mNGS technology has significantly enhanced viral pathogen detection rates. We enrolled 134 critically ill patients with severe infections who underwent mNGS testing during January 2019 to December 2021, at Qilu Hospital (Qingdao) of Shandong University. Viral pathogens were identified in 78 cases (58.2%). Torque teno virus (TTV) or herpesviruses (HVs) showed the highest detection rates (23.1% and 29.9%, respectively). The incidence of major adverse events (MAEs) in the hospital was 53.0%. Patients with TTV or HVs detection had more secondary nosocomial infections and stress ulcers, and the incidence of MAEs showed an increasing trend. Multivariate Logistic regression analysis showed that APACHE II score (OR: 1.10, 95%CI: 1.02–1.19, p = 0.018) and TTV or HVs detection by mNGS (OR: 2.40, 95% CI: 1.05–5.50, p = 0.038) were independent risk factors for MAEs. This study advocates the use of mNGS for detecting viruses in critically ill patients with severe infections, as it serves as a predictor for heightened risk of in‐hospital MAE.

1. Background

Viral reactivation in critically ill patients has been of concern to clinicians. Herpesviruses (HVs) remain dormant in various types of human cells after primary infection and thus may reactivate at a later time when the host is in an immunosuppressed state [1]. The reactivation of cytomegalovirus (CMV) occurs frequently in critically ill immunocompetent patients and is associated with prolonged hospitalization or death [2]. Epstein–Barr virus (EBV) DNA is also detected in the peripheral blood of ICU patients and suggested that EBV reactivation is associated with the length of ICU stay and the length of mechanical ventilation [3]. However, viremia by other types of HV, such as human herpesvirus 6 (HHV‐6) and herpes simplex virus (HSV), has also been described [4, 5, 6].

Torque teno virus (TTV), a nonenveloped single‐strand DNA virus of the genus Anellovirus, has been similarly detected in adult [4] and pediatric septic patients [7], suggesting that it might also provide information on the immune status of these patients [8]. The TTV DNA load in polymorphonuclear leukocytes was negatively associated with NK cell activity in Italian elderly subjects [9]. Furthermore, 34%–38% of patients with septic shock were reported to have multiple concurrent viremia events [1]. In a prospective study, concurrent CMV and EBV reactivation remained independently associated with increased mortality. Patients with multiple viremic events had a higher ICU mortality than those with viremia from a single virus [10].

Metagenomic next‐generation sequencing (mNGS) enables the detection of nearly all known pathogens simultaneously from clinical samples. Miao et al. reported that mNGS had higher sensitivity and specificity than microbial culture, especially for the detection of Mycobacterium tuberculosis, viruses, anaerobic bacteria, and fungi [11]. Jing‐Wen Ai et al. reported that mNGS has the potential to identify more viruses that were not detected by conventional methods. CMV, herpes simplex virus‐1 (HSV‐1), and EBV detected in the lungs of patients with severe pneumonia are associated with increased mortality [12]. However, the TTV identified using mNGS has not been reported, and the co‐detection of multiple viruses was not considered in the previous studies.

Therefore, the primary aim of the present study was to evaluate the prevalence of TTV or HVs detected using mNGS and their contribution to the adverse outcome of critically ill patients with severe infections. The effect of multiple viral co‐detection on in‐hospital outcomes was also accounted for. Furthermore, we explored the possible immune mechanisms underlying detection of TTV or HVs.

2. Materials and Methods

2.1. Study Design and Participants

This retrospective cohort study enrolled critically ill patients with severe infections who underwent mNGS to identify potential pathogens. These patients admitted to Qilu Hospital (Qingdao) of Shandong University from January 2019 to December 2021 were included. Severe infections were diagnosed according to the Surviving Sepsis Campaign guidelines for sepsis and septic shock [13]. Patients who did not meet this definition but might deteriorate were also further included. The exclusion criteria were as follows: age < 18 years; outpatients; and emergency patients. This study was approved by the Ethics Committee of Qilu Hospital (Qingdao) of Shandong University (KYLL‐KS‐2022053). Written informed consent was waived owing to the use of retrospective data.

2.2. Primary Endpoint

For data acquisition, the following variables were recorded from the hospital information system: age, sex, surgical, or medical cause of admittance, comorbidities, first‐day Acute Physiology, Age and Chronic Health Evaluation II (APACHE II) score, first‐day sequential organ failure assessment (SOFA) score, time on mechanical ventilation (MV), renal replacement therapy (RRT), length of stay (LOS) in the hospital, and total hospitalization costs. The results of standard blood tests performed at ICU admission were registered from the laboratory information system and included white blood cell (WBC) count, platelet levels, alanine transaminase (ALT) levels, total bilirubin (TBIL) levels, blood urea nitrogen (BUN) levels, creatinine (Cr) levels, blood gas, prothrombin time (PT), fibrinogen levels, and D‐dimer levels. Comorbidities were obtained from the medical records and included hypertension, diabetes mellitus, coronary artery disease (CAD), chronic kidney disease (CKD), liver cirrhosis, and state of immunosuppression. Immunocompromised patients included patients receiving long‐term (> 3 months) or high‐dose (> 0.5 mg/kg/day) steroids or other immunosuppressant drugs, solid‐organ transplant recipients, patients with solid tumors requiring chemotherapy in the last 5 years or with hematological malignancy regardless of the time since diagnosis and receiving treatments, and patients with primary immune deficiency [14].

The primary endpoint in this study was major adverse events (MAE) in the hospital, which was defined as a composite outcome event including in‐hospital death, nosocomial infection, deep vein thrombosis, stress ulcer with bleeding requiring blood transfusion, and other cardiovascular and cerebrovascular accidents. Nosocomial infection was diagnosed according to guidelines. MAEs were recorded from medical records. If the above events occurred, the occurrence time of the events was recorded for analysis.

2.3. mNGS Workflow

Samples for mNGS tests were transferred to the same professional genomic laboratory (Vision Medical Co.) independently by cold‐chain transportation. Pathogens and human cells were separated from 1 mL samples by centrifuging it at 12 000 g for 5 min. The host nucleic acid was then removed from the precipitate using 1 U Benzonase (Sigma) and 0.5% Tween 20 (Sigma), which were incubated at 37°C for 5 min. A 400 µL dose of terminal buffer was then added to halt the reaction. A Minilys Personal TGrinder H24 Homogenizer (catalog number: OSE‐TH‐01, Tiangen, China) was used to beat beads after transferring a total of 600 µL mixture into new tubes containing 500 µL of ceramic beads. DNA was then extracted and eluted from 400 µL of pretreatment samples using a QIAamp UCP Pathogen Mini Kit in 60 µL elution buffer (catalog number: 50214, Qiagen, Germany). Using a Qubit dsDNA HS Assay Kit (catalog number: Q32854, Invitrogen, USA), the isolated DNA was quantified. Total RNA was extracted with a QIAamp® Viral RNA Kit (Qiagen), and ribosomal RNA was removed by a Ribo‐Zero rRNA Removal Kit (Illumina). cDNA was generated using reverse transcriptase and dNTPs (Thermo Fisher). Libraries were constructed for the DNA and cDNA samples using the KAPA low‐throughput library construction kit (KAPA Biosystems, USA) following the manufacturer's instructions. The Library was quality assessed by Qubit dsDNA HS Assay kit, followed by High Sensitivity DNA kit (Agilent) on an Agilent 2100 Bioanalyzer. Library pools were then loaded onto an Illumina Nextseq CN500 sequencer for 75 cycles of single‐end sequencing to generate approximately 20 million reads for each library.

2.4. Bioinformatic Analysis

Trimmomatic was used to eliminate low‐quality reads, duplicate reads, adapter contamination, and those shorter than 70 bp [15]. Low‐complexity reads were removed by Kcomplexity's default settings were used to eliminate low‐complexity reads. By utilizing SNAP v1.0beta.18 to match the human sequence data to the hg38 reference genome, the human sequence data were located and eliminated [16]. The Kraken 2 criteria for choosing representative assemblies for microorganisms (bacteria, viruses, fungi, protozoa, and other multicellular eukaryotic pathogens) from the NCBI Assembly and Genome databases (https://benlangmead.github.io/aws-indexes/k2) were used to select pathogens and their genomes or assemblies for the creation of the microbial genome database. Microbial reads were aligned to the database using the Burrows–Wheeler Aligner software. The reads with 90% identity of reference were defined as mapped reads. In addition, reads with multiple locus alignments within the same genus were excluded from the secondary analysis. Only reads mapped to the genome within the same species were considered. The clinical reportable range (CRR) for pathogens was established according to the following three references indicated in a previous study [17]. The definition of TTV or HVs detection was at least one reads of TTV or HV detected by mNGS.

2.5. Statistical Analyses

Descriptive variables are presented as medians with interquartile ranges (IQRs) or as n (proportions). Variables were compared via Student's t‐test, Mann–Whitney U‐test, Kruskal–Wallis test, or chi‐squared tests, as appropriate. Trend tests were analyzed by the chi‐squared test for trends or ANOVA for linear trends, with the results displayed as p‐values for trends. Logistic regression models were used for the prediction of MAEs. In the univariate regression models, variables with p < 0.1 were further included in multivariate logistic regression analyses.

Analysis of the data were performed using SPSS 23.0 for Windows (SPSS Inc., Chicago, IL, USA) and R version 2.14.1 (R Development Core Team, Vienna, Austria). Charts were generated using GraphPad Prism (version 10.1.2). p‐values < 0.05 were considered statistically significant.

3. Results

During the study period, a total of 205 patients who met the enrollment criteria were screened, of which 65 were excluded due to “outpatients and emergency patients who did not receive hospitalization” and 6 were excluded due to “age < 18 years.” Finally, a total of 134 patients were included. The average age of the patients included in the study was 59.27 ± 16.03 years old, with 76 male patients, accounting for 56.7%. The average APACHE II score and SOFA score were 15.99 ± 9.70 and 5.96 ± 4.79, respectively. The main comorbidities included immunosuppression (44.8%), hypertension (32.8%), diabetes (30.6%), coronary atherosclerotic heart disease (26.1%), chronic kidney disease (11.9%), and cirrhosis (2.2%). Immunosuppression mainly included hematological malignancies in 36 patients (3 stem cell transplantation patients), solid tumors treated with chemoradiation and chemotherapy in 14 patients, and long‐term glucocorticoid therapy in 17 patients (rheumatoid arthritis in 5 patients, ANCA‐related vasculitis in 3 patients, systemic lupus erythematosus in 3 patients, primary immune thrombocytopenia in 2 patients, and other immune system diseases in 4 patients) (Table 1).

Table 1.

Baseline demographic characteristics and infection sites.

Variable	All (n = 134)
Age, y	59.27 ± 16.03
Sex (Male), n (%)	76 (56.7)
APACHE II score	15.99 ± 9.70
SOFA score	5.96 ± 4.79
Sepsis	92 (68.7%)
Septic shock	51 (38.1%)
Comorbidity
Immunosuppression	60 (44.8%)
Hypertension	44 (32.8%)
Diabetes Mellitus	41 (30.6%)
CAD	35 (26.1%)
CKD	16 (11.9%)
Liver cirrhosis	3 (2.2%)
Infection sites
Lung infection	104
Bloodstream infection	27
Intra‐abdominal infection	16
Urinary tract infection	12
Central nervous system infection	8
Skin and soft tissue infection	7
others	2
Sample sources (n = 166)
Blood	82 (49.4%)
BALF	50 (30.1%)
Cerebrospinal fluid	13 (7.8%)
Sputum	8 (4.8%)
Others	13 (7.8%)

Note: Descriptive variables are presented as medians with interquartile ranges (IQRs) or as n (proportions).

Abbreviations: CAD, coronary artery disease; CKD, chronic kidney disease; BALF, bronchoalveolar lavage fluid.

In terms of infection sites, there were 104 cases of lung infection, 27 cases of bloodstream infection, 16 cases of intra‐abdominal infection, 12 cases of urinary tract infection, 8 cases of central nervous system infection, 7 cases of skin and soft tissue infection, and 2 cases of other infections. Among them, 50 patients had more than 2 sites of infections. Fifty‐one patients (38.1%) met the diagnostic criteria for septic shock. As a part of the patients had multiple samples collected, 166 samples from 134 patients were tested by mNGS. There were 82 blood samples (49.4%), 50 alveolar lavages (30.1%), 13 cerebrospinal fluid samples (7.8%), 8 sputum samples (4.8%), and 13 other samples (7.8%) (Table 1). Other samples included tissue specimens in seven cases, pleural fluid in four cases, ascites in one case, and pus in one case.

A total of 146 samples (88.0%) were positive for bacteria, spanning 56 species, with the most prevalent being Klebsiella pneumoniae (patients, n = 17), Acinetobacter baumannii (n = 16), Pseudomonas aeruginosa (n = 15), Stenotrophomonas maltophilia (n = 9), and Haemophilus parainfluenzae (n = 9) (Figure 1A). Concurrently, 13 fungal species were detected in 34 samples, predominantly Candida albicans (n = 9), Pneumocystis jirovecii (n = 7), Candida tropicalis (n = 6), Aspergillus fumigatus (n = 4), and Aspergillus niger (n = 2) (Figure 1C). Additionally, 18 viruses were found across 100 samples, most commonly TTV, CMV, EBV, HSV‐1, and HHV‐6 (Figure 1B). Among the 70 samples positive for TTV or HVs, 34 (48.5%) showed bacterial co‑detection—primarily with K. pneumoniae, P. aeruginosa, and A. baumannii—while 17 (24.2%) had fungal co‑detection, mainly with P. jirovecii, C. albicans, and A. fumigatus; 11 samples exhibited both bacterial and fungal co‑detection. The majority of bacterial and fungal pathogens, including those from cases with co‑detection, were derived from bronchoalveolar lavage fluid (BALF) samples. Patients with concurrent bacterial or fungal co‑detection among those positive for TTV/HVs (n = 36) showed a lower rate of clinical improvement (36.1% vs. 75.8%) and a higher mortality rate (19.4% vs. 10.3%) compared to those without such co‑detection, with 13 and 7 patients experiencing improvement or death, respectively.

Figure 1.

Distribution of bacteria (A), virus (B), and fungi (C) detected by the mNGS technique. (A) Distribution of the top 20 most frequently detected bacteria. CMV, human cytomegalovirus; EBV, Epstein–Barr virus; HHV‐6, human herpesvirus 6; HHV‐7, human herpesvirus 7; HSV‐1, Herpes simplex virus type 1; TTV, Torque teno virus.

According to the results of mNGS, a total of 65 patients (48.5%) showed TTV or HVs detection, including 31 cases of TTV, 17 cases of CMV, 15 cases of EBV, 11 cases of HSV‐1, 6 cases of HHV‐6, and 3 cases of HHV‐7 (Table 2). Co‐detection with two or more viruses occurred in 12 patients (8.9%). The number of sequencing reads detected for these viruses ranged from 1 to 100 000, with the majority below 100 (Table 3).

Table 2.

Clinical characteristics according to the TTV or HVs detection.

Variables	Non‐detection (n = 69)	Single‐detection (n = 53)	Multi‐detection (n = 12)	P for trend
APACHE II score	15.71 ± 9.70	16.02 ± 10.21	17.42 ± 7.86	0.577
SOFA score	5.75 ± 4.58	6.15 ± 5.40	6.25 ± 2.99	0.743
Sepsis, n (%)	49 (71.0)	33 (62.3)	10 (83.3)	0.969
Septic shock, n (%)	24 (34.8)	21 (39.6)	6 (50.0)	0.314
MV, n (%)	33 (47.8)	28 (52.8)	7 (58.3)	0.439
CRRT, n (%)	19 (27.5)	13 (24.5)	2 (16.7)	0.441
Primary endpoint
In‐hospital MAEs, n (%)	31 (44.9)	31 (58.5)	9 (75.0)	0.03
In‐hospital mortality, n (%)	10 (14.5)	8 (15.1)	2 (16.7)	0.851
Nosocomial infection, n (%)	14 (20.3)	18 (34.0)	6 (50.0)	0.017
DVT, n (%)	18 (26.1)	17 (32.1)	4 (33.3)	0.451
Stress ulcer, n (%)	4 (5.8)	6 (11.3)	3 (25.0)	0.043
Others, n (%)	4 (5.8)	8 (15.1)	1 (8.3)	0.258
Second endpoint
ICU LOS, d	7.81 (0.00, 13.14)	4.81 (0.00, 22.79)	14.38 (0.00, 24.78)	0.149
Hospital LOS, d	15.67 (8.93, 23.94)	17.78 (8.48, 27.24)	23.89 (12.20,27.96)	0.612
Total hospitalization costs, kRMB	109.30 (34.46, 216.19)	103.32 (28.28, 259.44)	141.17 (73.93, 320.44)	0.276

Note: Trend tests were analyzed by the chi‐squared test for trends or ANOVA for linear trends, with results displayed as p‐values for trends. Major adverse events (MAEs) in the hospital was defined as a composite outcome event, which included in‐hospital death, nosocomial infection, deep vein thrombosis, stress ulcers, and other cardiovascular and cerebrovascular accidents. Bold values indicate statistically significant at p < 0.05.

Abbreviations: CRRT, continuous renal replacement therapy; DVT, deep vein thrombosis; MV, mechanical ventilation.

Table 3.

Distribution of TTV or HVs sequence reads detected by mNGS.

Reads	TTV (n = 31)	CMV (n = 17)	EBV (n = 15)	HSV‐1 (n = 11)	HHV‐6 (n = 6)	HHV‐7 (n = 3)
Median (IQR) [Range]	7 (17) [1–793]	8 (57) [2–3234]	11 (19) [1–1107]	13 (1936) [1–47 971]	8.5 (285) [3–935]	4 (NA) [4–5]
1 ~ 1 × 102	28	14	12	7	5	3
1 × 102 ~ 1 × 103	3	2	2	1	1	0
1 × 103 ~ 1 × 104	0	1	1	1	0	0
1 × 104 ~ 1 × 105	0	0	0	2	0	0

The patients were stratified into three groups based on the number of unique TTV or HVs species detected: the single‐detection group (a single virus species), the multidetection group (≥ 2 species), and the nondetection group (no virus detected) (Table 2). There was no difference between the three groups in APACHE II score, SOFA score, proportion of septic shock, proportion of patients receiving mechanical ventilation, and patients receiving CRRT. The incidence of in‐hospital MAEs was 53.0%, including 20 cases of in‐hospital death (14.9%), 38 cases of secondary nosocomial infection (28.4%), 39 cases of deep vein thrombosis (29.1%), 13 cases of stress ulcers (9.7%), and 13 cases of cardiovascular and cerebrovascular accidents (9.7%). The incidence of MAEs was 44.9% in the nondetection group, 58.5% in the single‐detection group, and 75.0% in the multidetection group. The incidence of secondary nosocomial infection in the three groups was 20.3% in the nondetection group, 34.0% in the single‐detection group, and 50.0% in the multidetection group. The incidence of stress ulcers in the three groups was 5.8% in the nondetection group, 11.3% in the single‐detection group, and 25.0% in the multidetection group. In‐hospital mortality and the incidence of deep vein thrombosis also showed an increasing trend among the groups, but there was no significant difference. In terms of secondary endpoints, the length of hospital stay and the total hospitalization costs gradually increased among the three groups, but there was no significant difference (Table 2).

To determine whether TTV or HVs detection by mNGS was a risk factor for MAEs, we further constructed a logistic regression model (Table 4). In the univariate analysis, APACHE II score, SOFA score, and virus reactivation were risk factors for MAEs. The APACHE II score, SOFA score, and virus reactivation were further included in the multivariate analysis. The APACHE II score (OR: 1.10, 95% CI: 1.02–1.19, p = 0.018) and TTV or HVs detection (OR: 2.40, 95% CI: 1.05–5.50, p = 0.038) were independent risk factors for MAEs (Figure 2).

Table 4.

Univariate and multivariate logistic regression for prediction of in‐hospital MAEs.

	Univariate analysis		Multivariate analysis
Variable	OR	p‐value	OR	p‐value
APACHE II score	1.15 (1.09, 1.21)	0.000	1.10 (1.02, 1.19)	0.018
SOFA score	1.29 (1.17, 1.42)	0.000	1.12 (0.97, 1.30)	0.125
Immunosuppression	1.67 (0.84, 3.33)	0.144	NA	NA
TTV or HVs detection	1.96 (0.99, 3.91)	0.055	2.40 (1.05, 5.50)	0.038

Note: In the univariate regression models, variables with p < 0.1 were further included in multivariate logistic regression analyses. Bold values indicate statistically significant at p < 0.05.

Figure 2.

Forest plot for multivariate logistic analysis. The APACHE II score and TTV or HVs detection were independent risk factors for in‐hospital MAEs.

To investigate the causes of adverse hospital events in patients with TTV or HVs detection, we further reviewed the results of cellular immunity. A total of 47 patients (35.1%) were tested for lymphocyte subsets. A lower proportion of CD3+CD4+ cells (41.49% vs. 35.51% vs. 27.37%, p = 0.015) and an increased proportion of CD3+CD8+ cells (30.14% vs. 31.15% vs. 51.06%, p = 0.001) were observed in the case group with TTV or HVs detection. The ratio of CD3+CD4+/CD3+CD8+ cells showed a decreasing trend (1.78 vs. 1.36 vs. 0.81, p = 0.033) (Table 5).

Table 5.

Lymphocyte subsets according to the TTV or HVs detection.

Lymphocyte subsets	Non‐detection (n = 18)	Single‐detection (n = 20)	Multi‐detection (n = 9)	P for trend
CD19+ cells	15.33 ± 10.88	17.33 ± 12.75	8.88 ± 7.94	0.178
CD(16 + 56) cells	11.11 ± 9.37	14.72 ± 9.07	12.03 ± 7.75	0.805
CD3+ cells	72.59 ± 11.65	67.28 ± 16.02	78.54 ± 9.03	0.281
CD3+CD4+ cells	41.49 ± 14.00	35.51 ± 13.03	27.37 ± 14.53	0.015
CD3+CD8+ cells	30.14 ± 12.02	31.15 ± 14.07	51.06 ± 20.95	0.001
CD3+CD4+/CD3+CD8+ ratio	1.78 ± 1.46	1.36 ± 0.67	0.81 ± 0.90	0.033

Note: According to the cluster of differentiation, CD19+ cells, CD(16 + 56) cells, CD3+ cells, CD3+CD4+ cells, CD3+CD8+ cells refers to B cells, NK cells, total T cells, help T cells, and cytotoxic T cells. Trend tests were analyzed by the chi‐squared test for trends or ANOVA for linear trends, with results displayed as p‐values for trends. Bold values indicate statistically significant at p < 0.05.

4. Discussion

In patients with sepsis, the bacterial profile distribution in BALF obtained by mNGS testing was similar to that yielded by conventional culture methods. The top three detected pathogens were K. pneumoniae, A. baumannii, and P. aeruginosa by both methods. However, the overall clinical sample positivity rate for bacteria was significantly higher with mNGS (49%) [18]. Global epidemiology indicates that common causative agents of fungal sepsis are Candida species, P. jirovecii, and Aspergillus species, which aligns with the fungal spectrum identified in this study. Risk factors for these opportunistic pathogens include immunosuppression, indwelling catheters, prolonged neutropenia, environmental exposure, and chronic lung disease [19]. In this study, when patients positive for TTV/HVs had concurrent bacterial or fungal co‐detection, their clinical outcomes were worse, presenting a significant challenge in the diagnosis and treatment of sepsis patients. This finding is consistent with previous research showing that viral‐bacterial co‐infections result in poorer treatment outcomes compared to viral infections alone. Furthermore, the proportions of patients requiring renal replacement therapy, mechanical ventilation, management of shock, and extracorporeal membrane oxygenation were substantially higher [20]. Patients with community‐acquired pneumonia due to viral‐bacterial mixed infections also have higher rates of severe illness and mortality than those infected with a single pathogen [21].

This study confirmed that TTV or HVs detected by mNGS in critical ill patients with severe infections was associated with higher incidence of major in‐hospital adverse events, especially secondary nosocomial infection and stress ulcer. In the multivariate analysis, the APACHE II score (OR: 1.10, 95% CI: 1.02–1.19, p = 0.018) and TTV or HVs detection (OR: 2.40, 95% CI: 1.05–5.50, p = 0.038) were independent risk factors for in‐hospital MAEs. The decreased cellular immune function in TTV or HVs detection group may be the reason for the higher incidence of in‐hospital MAEs.

In our study, a total of 65 patients (48.5%) showed TTV or HVs detection, including 31 cases of TTV, 17 cases of CMV, 15 cases of EBV, 11 cases of HSV‐1, 6 cases of HHV‐6, and 3 cases of HHV‐7, of which 12 patients showed multiple viral detection. Previous studies have demonstrated herpesvirus reactivations were frequently occurred in septic shock patients without prior immunodeficiency. Concurrent reactivations of herpesvirus could be independently associated with mortality [1]. More specifically, our study showed TTV was also frequently detected in critical ill patients. In recent years, studies have revealed that TTV exhibits characteristics such as a high carriage rate in the population, nonpathogenicity, stable viral load, detectability, and the ability to reflect immune function [22]. In solid organ transplant recipients, a low or decreasing TTV load is associated with the development of acute rejection, while a high or increasing TTV load precedes the occurrence of infectious complications [23]. Our study confirmed that the TTV load has the potential to serve as a marker reflecting the states of immunosuppression in critical ill patients.

A prior study demonstrated through mNGS technology the critical link between lung viral reactivation and increased mortality in severe pneumonia patients [10]. Recently, a large number of studies have focused on the differences in the etiological detection efficiency between mNGS and traditional methods. A study that compared mNGS with traditional methods in pneumonia patients revealed that, apart from common respiratory viruses, mNGS was also capable of detecting additional pathogens such as CMV, HSV, EBV, and so on [12]. However, these viruses are often not within the scope of traditional respiratory tract infection screening. In another study on pneumonia in immunodeficient patients, it was proposed that in some underdeveloped regions, traditional etiological examination methods may be insufficiently comprehensive. Under such circumstances, mNGS has its unique advantages [24]. Therefore, mNGS is more comprehensive and is less likely to be affected by clinicians or geographical regions.

In this study, both the proportion of CD3+CD4+ cells and the ratio of CD3+CD4+/CD3+CD8+ cells in TTV or HVs detection group showed a decreasing trend. The results are similar to those of previous studies. Numerous cohort studies from healthy elderly populations have confirmed that CD4/CD8 < 1 is associated with CMV infection [25, 26], EBV infection [27], or coinfection with CMV and EBV [28]. The monitoring results of TTV from healthy people also reached a similar conclusion, and elderly people with a TTV load ≥ 4 log copies/mL were more likely to have CD4/CD8 < 1 [9]. In patients with sepsis, the number of T lymphocytes is also an important indicator of the diagnosis of sepsis‐related immunosuppression. The increase in T lymphocyte apoptosis and the high expression of inhibitory immune molecules are the main reasons for the decrease in T lymphocyte number during the occurrence of sepsis‐related immunosuppression. So, it is possible that HVs and TTV detected by mNGS represent a generic marker of a suppressed immune system.

This study has several limitations. First, being a single‐center retrospective investigation, it suffers from a relatively limited sample size. The statistical analysis of cellular immunity is solely grounded in retrospective data. Such a research design inherently gives rise to the potential for selection bias in the resultant outcomes. Second, due to the retrospective nature of this study, the specimens submitted for mNGS lack uniformity in their types. Third, this study refrained from conducting a further evaluation of the efficacy of antiviral therapy. The question of whether antiviral therapy can enhance the prognosis of patients with severe infections remains open‐ended and necessitates further verification through prospective studies.

5. Conclusions

This study confirms that metagenomic second‐generation sequencing can efficiently identify TTV or HVs in critically ill patients with severe infections, which is an independent risk factor for MAE during hospitalization. The decreased immune function in critically ill patients with TTV or HVs detection may be the reason for the higher incidence of adverse events in the hospital.

Author Contributions

Yanting Sun: investigation, formal analysis and data curation, visualization, writing – original draft. Xinyan Shuai: investigation, formal analysis and data curation, visualization. Qiping Sheng: investigation. Yan Lu: investigation. Zhiyang Wu: formal analysis and data curation. Yunbo Sun: resources, supervision. Dawei Wu: resources. Xi Guo: conceptualization, formal analysis and data curation, writing – original draft, writing – review and editing.

Ethics Statement

This study was approved by the Ethics Committee of Qilu Hospital (Qingdao) of Shandong University (KYLL‐KS‐2022053).

Conflicts of Interest

The authors declare no conflicts of interest.

Sun Y., Shuai X., Sheng Q., et al., “Torque Teno Virus or Herpesviruses Detection By Metagenomic Next‐Generation Sequencing Predicts In‐Hospital Major Adverse Events in Critically Ill Patients With Severe Infections,” Journal of Medical Virology 98 (2026): e70840. 10.1002/jmv.70840.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.