A total of 285 patients were investigated, and 277 patients were enrolled and included for final analysis after excluding the patients < 18 years old and > 90 years old.
3.1 Demographic And Basic Clinical Information
There were 195 males and 82 females, with an average age of 62 (49, 70) years old. According to the qSOFA score, 162 hospitalized patients with infection were diagnosed with sepsis and 115 patients without sepsis. 270 patients with 30-day follow-up data showed that the mortality was 44.81% (121/270). Table 1 presented the basic information of the patients.
Table 1
Demographic and basic clinical characteristics of hospitalized patients with and without sepsis.
Basic information
|
Sepsis group (n = 162)
|
Non-sepsis group (n = 115)
|
P-value
|
Gender, male/female
|
114/48
|
81/34
|
0.991
|
Age, median (q1, q3)
|
63.0 (49.8, 69.0)
|
60 (43, 72)
|
0.451
|
Comorbidities [n (%)]
|
|
|
|
|
Pulmonary disease a
|
13 (8.02%)
|
12 (10.43%)
|
0.490
|
|
Congestive heart failure
|
11 (6.79%)
|
4 (3.48%)
|
0.230
|
|
Cerebrovascular disease
|
19 (11.73%)
|
11 (9.57%)
|
0.568
|
|
Diabetes
|
34 (20.99%)
|
14 (12.17%)
|
0.056
|
|
Hepatic cirrhosis
|
3 (1.85%)
|
0 (0%)
|
0.269
|
|
Acute/chronic renal failure
|
23 (14.20%)
|
15 (13.04%)
|
0.783
|
Smoking history [n (%)]
|
56 (34.57)
|
32 (27.83%)
|
0.235
|
Antibiotic use before enrollment b [n (%)]
|
144 (88.89%)
|
104 (92.04%)
|
0.679
|
Invasive procedures before onset of symptoms c [n (%)]
|
92 (56.79%)
|
55 (47.83%)
|
0.141
|
Corticosteroids/immunosuppressive drug/cytotoxic chemotherapy before onset [n (%)]
|
38 (23.46%)
|
48 (41.74%)
|
0.001
|
Recent surgery/trauma history d [n (%)]
|
54 (33.33%)
|
39 (33.91%)
|
0.920
|
ICU admission [n (%)]
|
141 (87.04%)
|
71 (61.74%)
|
< 0.001
|
a: including chronic obstructive pulmonary disease (COPD), asthma, interstitial lung disease, structural lung disease; b: antibiotics application within 2 weeks before enrollment; c: invasive procedures include punctures and drainages, tracheal intubation, urinary catheterization, arteriovenous catheterization, and superficial vein indwelling needle puncturing; d: surgery/trauma history within 3 months before enrollment. |
Abbreviations: ICU, intensive care unit. |
3.2 Characteristics Of Infection
The average time from symptom onset to enrollment was 8 (3, 17) days, with leukocyte count of 10.16 (6.70, 15.69) (×10^9/L) and neutrophil count of 9.03 (5.39, 13.38) (×10^9/L). The rest of data was shown in Table 2.
Table 2
The infection characteristics of patients in sepsis and non-sepsis group.
Infection characteristics
|
Sepsis group (n = 162)
|
Non-sepsis group (n = 115)
|
P-value
|
Time from symptom onset to enrollment (d), median (q1, q3)
|
5.5 (2, 15)
|
11 (5, 22)
|
0.008
|
Site of infection [n (%)]
|
|
|
|
Pulmonary infection
|
114 (80.28%)
|
70 (60.87%)
|
0.100
|
Extrapulmonary infection
|
18 (11.11%)
|
8 (6.96%)
|
0.247
|
APACHE II (means ± SD)
|
19.97 ± 8.33
|
14.87 ± 8.28
|
< 0.001
|
Shock index (means ± SD)
|
1.04 ± 0.36
|
0.81 ± 0.19
|
< 0.001
|
Fever [n (%)]
|
117 (72.22%)
|
82 (71.30%)
|
0.867
|
Altered mental status a [n (%)]
|
115 (70.99%)
|
20 (17.39%)
|
< 0.001
|
Death within 30 days [n (%)]
|
82, 51.57%
|
39, 35.14%
|
0.008
|
PCT (ug/L) [median (q1, q3)]
|
1.92(0.36, 11.17)
|
0.53(0.22, 3.82)
|
0.214
|
WBC count (×10^9/L) [median (q1, q3)]
|
10.83(6.93, 15.65)
|
9.79(6.10, 15.76)
|
0.410
|
Neutrophil count (×10^9/L) [median (q1, q3)]
|
9.07(5.73, 13.42)
|
8.84(5.00, 13.03)
|
0.475
|
a: including coma, delirium, ambiguity of consciousness, et al. |
Abbreviations: APACHE II, the Acute Physiologic and Chronic Health Evaluation II score; PCT, procalcitonin; WBC, white blood cell. |
Among the 148 patients with confirmed pathogens, bacterial infections were the most common (118 cases). Klebsiella pneumoniae (22 cases) counted for the first, followed by Acinetobacter baumannii (19 cases) and Pseudomonas aeruginosa (11cases). For patients with fungi infections (22 cases), the suspected pathogens were Pneumocystis (13cases), Aspergillus (7 cases), and Candida (2 cases). Adenovirus (3 cases) were the most common pathogens in viral infected patients (8 cases). 4 cases were diagnosed with Mycobacterium tuberculosis infection while their blood culture results were negative, and only 2 of them had positive mNGS results. Figure 1 listed the etiological examination results of 148 patients.
3.3 Performance Of Mngs Test As Compared To Culture
A total of 217 patients performed blood cultures, of which 50 were positive, and 167 were negative. Among 277 patients’ mNGS test results, 140 were positive, 135 were negative, and 2 were missing.
The percentage of patients with identified pathogen increased from 16.25%(45/277)when using only culture to 52.71% (146/277) when combined mNGS with culture. According to the final etiology result, the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of blood culture were 36.00% (45/112), 94.57% (87/105), 90.00% (45/50) and 52.10% (87/167), respectively, while the results of mNGS test were 90.54% (134/148), 94.53% (121/128), 95.71% (134/140) and 89.63% (121/135), respectively. The AUC value of blood culture and mNGS test, which were 0.653 and 0.929 (P<0.001), respectively.
It should be noted that among 22 patients diagnosed with fungi infections, only 1 patient had a positive blood culture for Candida. Therefore, for patients suspected of fungal infection, mNGS testing should be actively performed to avoid missing diagnosis.
Besides, the average turn-out time required for mNGS tests was 27.0 (26.0, 29.0) (hours), which was significantly shorter than the average time required for blood culture was 96.0 (72.0, 140.3) (hours) (P < 0.001).
3.4 Comparison of mNGS test and blood culture in different subgroups
We divided the patients into different subgroups according to various factors, and analyzed the diagnostic efficacy of mNGS test and blood culture in each subgroup. As results shown in Table 3, the sensitivity of mNGS test significantly preceded than that of blood culture, while the specificity remained not inferior to blood culture.
Table 3
The diagnosis efficacy comparison between mNGS and blood culture in different subgroups.
Subgroups
|
mNGS test
|
Blood culture
|
Sensitivity
|
Specificity
|
Sensitivity
|
Specificity
|
Age
|
|
|
|
|
> 65 years old
|
90.74%*
|
90.20%
|
37.78%
|
95.00%
|
≤ 65 years old
|
90.43%**
|
98.68%
|
35.00%
|
94.44%
|
Gender
|
|
|
|
|
Male
|
90.48%**
|
94.32%
|
36.05%
|
95.45%
|
female
|
90.70%**
|
97.44%
|
35.90%
|
92.31%
|
Shock index
|
|
|
|
|
> 1.0
|
89.09%**
|
100.00%
|
34.00%
|
96.30%
|
≤ 1.0
|
91.40%**
|
93.18%
|
37.33%
|
93.85%
|
Pulmonary infection
|
|
|
|
|
Yes
|
92%**
|
92.68%
|
32.50%
|
92.59%
|
No
|
87.50%**
|
100%
|
42.22%
|
97.36%
|
Bloodstream infection
|
|
|
|
|
Yes
|
96.15%*
|
92.86%
|
56.00%
|
90.91%
|
No
|
89.34%**
|
95.58%
|
31.00%
|
95.06%
|
Surgery/trauma history a
|
|
|
|
|
Yes
|
95.83%**
|
97.78%
|
34.88%
|
97.22%
|
No
|
88.88%**
|
93.90%
|
36.59%
|
92.86%
|
Time from infection onset to enrollment
|
|
|
|
|
≤ 5 days
|
91.23%**
|
96.30%
|
26.92%
|
95.45%
|
> 5 days
|
89.77%**
|
97.18%
|
41.67%
|
93.62%
|
Invasive operations before infection onset b
|
|
|
|
|
Yes
|
89.61%**
|
98.53%
|
42.19%
|
91.53%
|
No
|
91.54%**
|
91.53%
|
29.51%
|
97.67%
|
Fever
|
|
|
|
|
Yes
|
88.68%**
|
95.65%
|
40.45%
|
95.24%
|
No
|
95.24%**
|
94.29%
|
25.00%
|
93.10%
|
Altered mental status
|
|
|
|
|
Yes
|
86.67%**
|
95.00%
|
41.54%
|
91.89%
|
No
|
94.37%**
|
95.52%
|
29.31%
|
96.36%
|
Antibiotics use before enrollment c
|
|
|
|
|
Yes
|
90.70%**
|
94.96%
|
34.58%
|
94.05%
|
No
|
89.47%*
|
100%
|
44.44%
|
100%
|
ICU administration
|
|
|
|
|
Yes
|
90.43%**
|
97.89%
|
35.58%
|
94.37%
|
No
|
90.91%*
|
87.50%
|
38.10%
|
95.24%
|
Sepsis
|
|
|
|
|
Yes
|
89.13%**
|
98.57%
|
34.15%
|
93.88%
|
No
|
92.86%**
|
91.23%
|
39.53%
|
95.35%
|
a: surgery/trauma history within 3 months before enrollment; b: invasive procedures include punctures and drainages, tracheal intubation, urinary catheterization, arteriovenous catheterization, and superficial vein indwelling needle puncturing; c: antibiotics application within 2 weeks before enrollment. |
Abbreviations: mNGS, metagenomic next-generation sequencing; ICU, Intensive care unit. |
*P<0.05, mNGS vs blood culture; **P<0.001, mNGS vs blood culture. |
Although there was no statistical significance, the sensitivity of blood culture exhibited downward trends in subgroups of patients within 5 days of infection onset, patients without prior invasive operation, patients without fever, patients without altered mental status, and patients who applied antibiotics before enrollment. This result indicated that mNGS test might be more suitable for patients with mild symptoms or prior antibiotics application, and more prominent in early etiological diagnosis of infectious diseases.
3.5 Reads of responsible pathogens in mNGS reports related to the prognosis of patients and the level of inflammatory factors.
Among the mNGS-positive patients, we compared the reads of responsible pathogens in mNGS reports in patients with different prognosis (shown in Fig. 2). The results showed that reads in the non-survival group was also higher than that in the survival group, and this trend remained significant between non-survival and survival patients in sepsis group. It suggested that the reads of mNGS reports might be related to the prognosis of infected patients.
We detected the level of inflammatory factors in patients with positive mNGS results and analyzed the correlation between the level of inflammatory factors and the reads of pathogens. The results showed that there was a positive correlation between interleukin 1-β (IL-1β), eotaxin, interferon-α (IFN-α), tumor necrosis factor-α (TNF-α) and the reads of pathogen (shown in Table 4). The positive correlation between eotaxin, IFN-α, TNF-α and the reads of pathogen remained significant in sepsis patients (shown in Additional Table 1).
Table 4
Univariate linear regression of inflammatory factors level and reads of pathogens in mNGS reports.
Inflammatory factor
|
P-value
|
B (95% CI)
|
R-square
|
IL-1β
|
0.024
|
0.349 (0.051, 0.647)
|
0.182
|
Eotaxin
|
0.004
|
0.465 (0.157, 0.773)
|
0.247
|
IFN-α
|
< 0.001
|
2.629 (1.450, 3.809)
|
0.436
|
HGF
|
0.761
|
0.039 (-0.221, 0.299)
|
0.003
|
TNF-α
|
< 0.001
|
0.905 (0.503, 1.307)
|
0.432
|
IL-9
|
0.149
|
0.415 (-0.158, 0.988)
|
0.070
|
Abbreviations: mNGS, metagenomic next-generation sequencing; B, standardized linear regression coefficient; IL, interleukin; IFN, interferon; HGF, hepatocyte growth factor; TNF, tumor necrosis factor. |
3.6 Pathogenic test facilitated antimicrobial prescribing modification ameliorated prognosis of patients
Of all the patients, 151’s antibiotic regimen was modified within 2–7 days after the enrollment, while the rest remained the original antibiotic regimen. It was observed that patients who made modification of antibiotic regimen had higher 30-day survival rate (63.3% vs 45.5%, P = 0.005), and the trend was also obtained in sepsis group (58.5% vs 37.7%, P = 0.009).
Among the mNGS-negative patients (135 cases), 25 patients de-escalated antibiotic use with a 30-day mortality of 32.0%, while the other 110 patients did not reduce the administration of antibiotics, and the 30-day mortality was 40.0%. There was no statistically significant difference in mortality. The result indicated that negative mNGS results may help with de-escalating antibiotics without worsening prognosis.