A biomarker for bacteremia in pregnant women with acute pyelonephritis: soluble suppressor of tumorigenicity 2 or sST2

Abstract Objective Sepsis is a leading cause of maternal death, and its diagnosis during the golden hour is critical to improve survival. Acute pyelonephritis in pregnancy is a risk factor for obstetrical and medical complications, and it is a major cause of sepsis, as bacteremia complicates 15–20% of pyelonephritis episodes in pregnancy. The diagnosis of bacteremia currently relies on blood cultures, whereas a rapid test could allow timely management and improved outcomes. Soluble suppression of tumorigenicity 2 (sST2) was previously proposed as a biomarker for sepsis in non-pregnant adults and children. This study was designed to determine whether maternal plasma concentrations of sST2 in pregnant patients with pyelonephritis can help to identify those at risk for bacteremia. Study design This cross-sectional study included women with normal pregnancy (n = 131) and pregnant women with acute pyelonephritis (n = 36). Acute pyelonephritis was diagnosed based on a combination of clinical findings and a positive urine culture. Patients were further classified according to the results of blood cultures into those with and without bacteremia. Plasma concentrations of sST2 were determined by a sensitive immunoassay. Non-parametric statistics were used for analysis. Results The maternal plasma sST2 concentration increased with gestational age in normal pregnancies. Pregnant patients with acute pyelonephritis had a higher median (interquartile range) plasma sST2 concentration than those with a normal pregnancy [85 (47–239) ng/mL vs. 31 (14–52) ng/mL, p < .001]. Among patients with pyelonephritis, those with a positive blood culture had a median plasma concentration of sST2 higher than that of patients with a negative blood culture [258 (IQR: 75–305) ng/mL vs. 83 (IQR: 46–153) ng/mL; p = .03]. An elevated plasma concentration of sST2 ≥ 215 ng/mL had a sensitivity of 73% and a specificity of 95% (area under the receiver operating characteristic curve, 0.74; p = .003) with a positive likelihood ratio of 13.8 and a negative likelihood ratio of 0.3 for the identification of patients who had a positive blood culture. Conclusion sST2 is a candidate biomarker to identify bacteremia in pregnant women with pyelonephritis. Rapid identification of these patients may optimize patient care.

It is unknown whether a maternal plasma sST2 concentration has value in the assessment of pregnant patients with acute pyelonephritis. The objective of this study was to evaluate whether maternal plasma concentrations of sST2 could be of value in identifying pregnant patients with acute pyelonephritis and bacteremia because patients with bloodstream infections are at a greater risk for severe complications.

Study design and sample collection
This retrospective cross-sectional study included normal pregnant women (n ¼ 131) and pregnant patients with acute pyelonephritis (n ¼ 36). All women were enrolled at Hutzel Women's Hospital of the Detroit Medical Center, (Detroit, Michigan, USA).
Patients were considered to have a normal pregnancy if they met the following criteria: (1) no medical, obstetrical, and surgical complications, (2) absence of labor at the time of venipuncture, and (3) delivery of a term (!37 weeks of gestation) infant with a birthweight between the 10th and 90th percentiles for gestational age and without congenital anomalies. Patients with a multiple gestation and major fetal anomalies were excluded. Acute pyelonephritis was diagnosed when the patient presented with a fever (temperature !38 C), clinical signs of upper urinary tract infection, pyuria, and a positive urine culture. Blood cultures were performed in 30 of the patients with pyelonephritis.
All patients provided written informed consent for the collection and use of the samples under protocols approved by the Institutional Review Boards of Wayne State University and the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, U. S. Department of Health and Human Services (NICHD/NIH/DHHS). Many of these samples were previously used in other studies at our institution, including previous past studies on pyelonephritis.

Soluble ST2 immunoassay
Blood samples were collected in tubes containing EDTA, centrifuged at 1300 g for 10 min at 4 C, and stored at À70 C until analysis. Maternal plasma concentrations of sST2 were determined with a sensitive and specific enzyme-linked immunoassay (R&D Systems, Minneapolis, MN, USA), validated for determination of plasma sST2 concentration. The inter-and intra-assay coefficients of variation were 4.6% and 3.9% for sST2, respectively. The sensitivity of the assay was 17.5 pg/mL.

Statistical analysis
The sST2 concentration in each group did not follow a normal distribution after testing with the Kolmogorov-Smirnov test (even after logarithmic transformation). Non-parametric statistics were used for subsequent analysis. Mann-Whitney U and Kruskal-Wallis tests were applied to determine differences between and among continuous variables. Categorical variables were compared by using a contingency table or Chisquared test. A Receiver Operating Characteristic (ROC) curve was generated to assess the diagnostic performance of sST2 concentration for the identification of patients with bacteremia. A threshold of sST2 concentration was selected to maximize sensitivity while preserving a specificity of at least 95%. A p value of < .05 was considered significant. Statistical analyses were performed with the IBM SPSS version 19.0 (IBM Corporation., Armonk, NY).

Results
Demographic and clinical characteristics of the study groups are displayed in Table 1. There were no significant differences in all the variables evaluated except for gestational age at time of maternal venipuncture (acute pyelonephritis: 31 weeks, interquartile range (IQR) 25-36 weeks vs. normal pregnancy: 38 weeks, IQR 31-39 weeks; p ¼ .001).
An elevated plasma concentration of sST2 > 215 ng/mL had a sensitivity of 73% and a specificity of 95% (area under the receiver operating characteristic curve, 0.74, 95% confidence interval 0.50 À 0.97; p ¼ .003 compared to area 0.5) with a positive likelihood ratio of 13.8 and a negative likelihood ratio of 0.3 for the identification of patients who had a positive blood culture ( Figure 4).

Discussion
Principal findings of the study (1) Pregnant women with acute pyelonephritis had a significantly higher median plasma sST2 concentration than women with a normal pregnancy; (2) among patients with acute pyelonephritis, those with bacteremia had a higher median plasma concentration of sST2 than patients with a negative blood culture; and (3) maternal plasma sST2 concentration ! 215 ng/mL is predictive of positive blood cultures in pregnant patients with acute pyelonephritis with a positive likelihood ratio of 13.8 and a negative likelihood ratio of 0.3. This is the first study to report on sST2 plasma concentrations in pregnant women with acute pyelonephritis.

Pregnant patients with acute pyelonephritis have elevated sST2 plasma concentrations
Our data demonstrate that maternal plasma sST2 concentrations increase as a function of gestational age in normal pregnancy, and this evidence is consistent with studies previously published [65,66], reporting a shift toward a Th1 response before the onset of labor [61][62][63][64]118,119]. In our study population, pregnant patients with acute pyelonephritis, regardless of their blood culture results, had a higher median plasma sST2 concentration than women with a normal pregnancy. This observation is consistent with previous reports that the peripheral whole blood transcriptome of patients with pyelonephritis during pregnancy showed upregulation of genes involved in cytokinecytokine interactions as well as complement and coagulation cascades [120,121,]. We previously described that pyelonephritis during pregnancy is associated with changes in the profile of cytokines [122] such as CXCL10 [123] and TRAIL [124] as well as with changes in angiogenic factors [125], complement [121], coagulation [126], and adipokines [127][128][129].
sST2 as a biomarker for bacteremia in pregnant patients with acute pyelonephritis The findings that pregnant patients with acute pyelonephritis and bacteremia have a higher plasma sST2 concentration than those with a negative blood culture suggests that these patients had a more intense engagement of the innate immune system. In nonpregnant adults and in children, sST2 has been proposed as a biomarker for sepsis because elevated concentrations correlate with worse outcomes and inhospital mortality [51][52][53][54]. For pregnant patients, sepsis risk stratification has been dependent on several bedside scoring systems, including the Modified Obstetric Early Warning System (MOEWS), the Sepsis in Obstetrics Score (SOS), and the Obstetric Modified Quick Sepsis Related Organ Failure Assessment (omqSOFA) [130][131][132]. The only scoring system that employs biomarkers (leukocyte count, immature neutrophils, and lactic acid) is the SOS [130]. None of the currently proposed biomarkers for sepsis in pregnant women includes procalcitonin or C-reactive protein, frequently used in non-pregnant populations [133][134][135]. Similarly, none of the currently used scoring systems uses cytokine profiles [136], which are predictive of the outcome of patients with suspected sepsis. Further research is warranted to determine the utility of maternal plasma sST2 and other cytokine profiles in conjunction with existing sepsis scoring systems in pregnancy.
Given that a maternal plasma sST2 concentration !215 ng/mL is associated with bacteremia, plasma sST2 may be a useful method to select patients for Figure 2. Plasma sST2 concentrations in normal pregnancy compared to pregnancy complicated by acute pyelonephritis. In pregnancies complicated by acute pyelonephritis, the median plasma sST2 concentration was significantly higher than that in normal pregnancies [median 85 ng/mL (IQR 47-239) vs. median 31 ng/mL (IQR 14-52); p < .001].
blood culture collection. Although there is debate about the utility of performing blood cultures, the detection of bacteremia is likely to identify patients at risk for sepsis and multiple organ failure [137][138][139][140][141][142][143]. In pregnant patients with acute pyelonephritis who are not clinically ill, an sST2 concentration <215 ng/mL may be an alternative to forego blood culture collection. The implementation of plasma sST2 in clinical practice is easier given the availability of rapid immunoassays [144,145]. However, future studies are required to determine whether plasma sST2 provides any additional value over models incorporating standard risk factors for predicting patients who do or do not need blood culture collection in pregnant women with acute pyelonephritis.

Clinical and research implications
The standard treatment of pyelonephritis during pregnancy consists of the administration of parenteral broad-spectrum antimicrobial agents. The goal is to administer adequate concentrations of antibiotics to the site of infection (i.e. renal tissue, blood, or both) and not only the urine [146]. The first choice is broad-spectrum beta-lactam antibiotics and third-generation cephalosporins (e.g. ceftriaxone) are favored over first-(e.g. cefazolin) and second-generation (e.g. ceforanide). Parenteral antibiotics are administered until the patient is afebrile for 48 h and then switched to oral therapy, which consists of antimicrobial agents guided by the culture susceptibility results for 10-14 days [147][148][149][150][151]. Oral antimicrobials recommended are betalactams. Trimethoprim-sulfamethoxazole has been considered an alternative in the second trimester. Nitrofurantoin and fosfomycin achieve adequate concentrations only in the urine and are not considered appropriate for the treatment of pyelonephritis [146,152]. The treatment of patients with bacteremia during pregnancy is less clear. The standard recommendation has been to administer a course of antibiotics for 14 days; however, a recent randomized clinical trial in non-pregnant subjects reported that 7day treatment of patients with Gram-negative bacteremia was not inferior to 14 days [150,151]. A trial is in progress [153] [Bacteremia Antibiotic Length Actually Needed for Clinical Effectiveness (BALANCE)] to address this question in non-pregnant subjects. Biomarkers to monitor response to therapy could be of great value. sST2, alone or in combination with other cytokines or markers such as procalcitonin, could be useful to guide antibiotic treatment. A recent patient-level meta-analysis has shown that procalcitonin-guided antibiotic treatment in non-pregnant subjects with bacteremia can decrease antibiotic exposure without increasing mortality [154] We envision that sST2, or a biomarker panel, could be performed at the time of admission to determine if blood cultures should be performed to identify bacteremia [137,141,142].

Strengths and limitations
This is the first study to report on sST2 plasma concentrations in pregnant patients with acute pyelonephritis and to determine that concentrations of 215 ng/mL or greater are predictive of bacteremia. In our study population, there was evidence of bacteremia in 36.7% (11/30) of women with pyelonephritis, which is higher than that previously reported (ranging from 10% to 20%) [16,137,140]. A limitation of this study is that six patients with pyelonephritis did not have blood cultures obtained.

Conclusion
We report that plasma sST2 concentrations are elevated in pregnant patients with pyelonephritis and, in particular, those with bacteremia. These findings are consistent with those observed in non-pregnant patients with sepsis and other inflammatory disorders. Based on these observations, we propose sST2 as a candidate biomarker to identify the patient at risk for bacteremia.

Disclosure statement
No potential conflict of interest was reported by the author(s).