Background
Necrotizing soft-tissue infection (NSTI) is a rare, severe and fast-progressing bacterial infection [
1]. The incidence rate of NSTI is approximately two per 100,000 inhabitants/year in Denmark [
2], but varies markedly across countries [
3‐
6]. NSTI can be caused by a myriad of aerobic, anaerobic and facultative anaerobic bacteria [
7], but predominantly by Group A Streptococcus in monomicrobial infections [
8]. The clinical condition can rapidly progress into septic shock, multiple organ failure and death. Early recognition is key in NSTI, however timely diagnosis may be rendered as the initial symptoms such as swelling, pain and erythema are difficult to discriminate from less severe skin infections, consequently an initially misdiagnosis of 71% has been reported [
9]. Despite early, radical and intensive multidisciplinary care the mortality remains high and largely unaltered in several years [
2,
4].
Biomarkers may provide valuable information to the treating physician in treatment guidance and decision-making in NSTI patients, potentially improving morbidity and mortality. However, biomarkers have only been limited investigated in NSTI. No reliable diagnostic biomarkers exist [
10,
11], hence surgical exploration persists the diagnostic golden standard although some imaging modalities may provide valuable diagnostic information [
12‐
14]. Prognostic biomarkers have been sparsely assessed in patients with NSTI and none have provided sufficient robustness to have clinical impact [
15‐
20]. Therefore, standard biochemistry and disease severity scores, such as Simplified Acute Physiology Score II (SAPS II), continues to provide the best level of prognostic risk-stratification and disease monitoring in the intensive care unit. Consequently, novel biomarkers are warranted in effort of increasing early diagnostic, therapeutic and prognostic risk-stratification in patients with NSTI.
YKL-40, also called chitinase-3-like-1 protein, may be an attractive prognostic biomarker in NSTI. YKL-40 is an acute phase protein secreted by various of immune cells, including macrophages, neutrophils and endothelial cells [
21]. Of notice, proteomic analysis has indicated YKL-40 as a promising biomarker in patients with severe sepsis and septic shock [
22]. Studies have suggested that high YKL-40 levels could be prognostic in patients with sepsis [
23,
24]. However, the prognostic yields of YKL-40 remain unknown in NSTI. Extrapolation of YKL-40 results from patients with sepsis to NSTI may be questionable as the inflammatory response differ due to the combination of extensive tissue damage in NSTI and different pathogenesis in general. It would be of interests if high plasma YKL-40 levels could identify patients with NSTI with an increased risk of death as these patients might benefit from a more aggressive treatment protocol.
Therefore, we aimed to assess plasma YKL-40 as a prognostic biomarker in patients with NSTI and evaluate its accuracy as marker of disease severity and mortality. We hypothesized that high plasma levels of YKL-40 are associated with disease severity and 30-day mortality.
Discussion
In this study, assessing the prognostic yields of plasma YKL-40 in patients with NSTI, we observed high plasma YKL-40 levels to be associated with increased risk of death and severity of the disease. However, the association was not statistically significant when adjusted for SAPS II. We found the highest plasma YKL-40 levels at admission among patients in septic shock, patients requiring RRT and patients infected with Group A Streptococcus.
We compared the prognostic accuracy of 30-day mortality of plasma YKL-40 with plasma lactate, known to be associated with increased mortality [
8,
15], and with SAPS II which is among the most common disease severity score used in the intensive care unit. We found a lower ROC-AUC for YKL-40 compared to both plasma lactate and SAPS II, indicating that the prognostic information of plasma YKL-40 is not be better than the existing evaluation tools. Yet, caution is required when exclusively comparing AUC between prognostic indicators, as AUC reflects the overall performance of all possible cut-off values including those that will never be clinically applied. Therefore, we addressed the prognostic accuracy as sensitivity, specificity, PPV and NPV according to the optimal cut-off point. Here we observed a low sensitivity but moderate good specificity and NPV indicating patients with plasma YKL-40 levels below 1840 ng/mL to have an 89% chance of surviving the first 30 days. We observed a reduction in YKL-40 over time potentially reflecting a generally improved clinical condition among patients, but such result must be cautiously interpreted as a risk of dilution occurs when measuring across time, as the most critically ill patients who have highest YKL-40 levels at admission, are probably also the patients who first decease.
To our knowledge, no prognostic biomarkers have yet provided enough robustness to provide a clinical impact in NSTI treatment. Pentraxin-3, a molecule released by various immune cells at the onset of inflammation, has earlier demonstrated a ROC-AUC of 0.66 in predicting 180-day mortality with a sensitivity of 0.69 and a specificity of 0.56 in patients with NSTI [
18]. Additionally, the prognostic accuracy of suPAR in predicting 90-day mortality has shown a ROC-AUC of 0.77, but the combination of suPAR and SAPS II was not superior compared to SAPS II alone (ROC-AUC 0.87) [
17]. The burden of disease, the unaltered mortality rates and the lack of robust biomarkers all highlight the need for studies evaluating novel biomarkers potentially capable of reducing disease morbidity and mortality.
We included age, sex, comorbidity and disease severity as covariates in logistic regression analyses, as these are among the most documented prognostic risk factors of not surviving NSTI [
2,
4,
30‐
36]. Other covariates such as septic shock [
34,
37] need of inotropes and RRT [
35,
38] could have been included as covariates, but according to the ‘one-in-ten rule’ on the number of covariates in logistic analyses, only three covariates entered multivariate analyses [
39,
40]. Furthermore, these are included in the SAPS II. We observed high plasma YKL-40 at admission to be associated with increased mortality in univariate and age-sex adjusted analyses, but not after adjusting for SAPS II indicating that YKL-40 is not an independent predictor of mortality. Yet, plasma YKL-40 could provide an early prognostic information for the treating physician as SAPS II needs 24 h of admission in the intensive care unit before a complete score can be calculated, and some variables may be missing rendering the calculation. Although no quick/bed-side YKL-40 assays exist, well-performing and rapid multiplex assays (≈75 min) have been developed [
41].
Severe infection such as patients with septic shock and patients requiring RRT are associated with mortality [
34,
37,
38]. Therefore, we compared differences in plasma YKL-40 according to presence of septic shock or requiring RRT. We found patients with septic shock to have higher plasma YKL-40 levels at admission and the following three days compared to the levels in non-shock patients. Similar patterns were observed in patients requiring RRT and patients not receiving RRT. These results may reflect a more pronounced immune cell activity and a higher burden of disease in these patients. Additionally, patients with Group A
Streptococcus were found to have significantly higher plasma YKL-40 levels at admission compared to patients with other types of NSTI, potentially reflecting a higher rate of septic shock in these individuals as earlier demonstrated [
8]. These observations suggest that burden of disease is increased in patients with high plasma YKL-40 and is underlined by the positive correlations between baseline YKL-40 and SAPS II, SOFA score and plasma lactate in the present study.
In healthy individuals, plasma YKL-40 is approximately 40 ng/mL (2.5–97.5% CI 14–155) [
28] which is comparable to our observation of 40 ng/mL (IQR 29–87). We demonstrated approximately 30 times higher plasma levels in patients with NSTI compared to controls, indicating that severe bacterial infections influence YKL-40 secretion from inflammatory cells. Others have earlier found plasma YKL-40 levels in patients with sepsis and septic shock to be approximately 700–1000 and 1200–2200 ng/mL, respectively [
23]. These results are in accordance with the present plasma YKL-40 levels of 720 ng/mL in non-shock NSTI patients and 1942 ng/mL in NSTI patients with septic shock. Additionally, in a cohort of 89 patients with
Streptococcus pneumoniae bacteremia the median YKL-40 was 342 ng/mL and was associated with in-hospital mortality [
42], whereas in 289 hospitalized patients with community-acquired pneumonia, plasma YKL-40 was higher in patients with severe infection and associated with short- and long-term mortality [
43]. Interestingly, a study including both experimental and human data has revealed that YKL-40 may be able to discriminate been sepsis-induced acute kidney injury and sepsis without acute kidney injury [
44]. In that context, patients with Puumala hantavirus infection—a virus infection typically inducing acute kidney injury—YKL-40 was more pronounced during the acute phase, and correlated with the degree of inflammation and severity of acute kidney injury [
45]. These finding are in agreement with the present study demonstrating significantly higher levels plasma YKL-40 in those receiving RRT within 7 days from admission.
Furthermore, plasma YKL-40 has recently been demonstrated to be a biomarker of future infectious diseases in the general population with adjusted hazard ratio of future episodes of skin infections and sepsis of 1.76 and 1.90, respectively [
46]. These findings emphasize plasma YKL-40 as a promising biomarker for further research in infectious diseases including NSTI.
YKL-40 is not a sole biomarker for infectious disease [
21]. Plasma YKL-40 has been studied in patients with different types of cancer and high levels are associated with short overall survival [
47]. Furthermore, high plasma YKL-40 levels are also associated with poor prognosis in non-cancerous diseases including cardiovascular, respiratory, digestive, endocrine, nervous, urinary and skeletal diseases [
21]. In patients with infectious disease plasma YKL-40 may reflect the ongoing pathological inflammatory processes more accurately than traditionally biomarkers such as CRP (produced by hepatocytes), since YKL-40 is secreted by activated immune cells including macrophages, neutrophils and endothelial cells [
21]. YKL-40 is considered an acute-phase protein and is stimulated by INF-γ and IL-6 [
48,
49]. We have earlier demonstrated plasma IL-6 to be associated with NSTI disease severity and particular elevated in patients with septic shock and β-hemolytic streptococcal infection, potentially indicating a relationship between YKL-40/IL6 and burden of disease in NSTI [
20]. Despite some regulators of YKL-40 have been defined during the last decades [
21], a great knowledge gap exists on its pathophysiological role and functional effects in infectious diseases including sepsis and NSTI [
50,
51]. Future research may include evaluation of YKL-40 in non-necrotic soft-tissue lesions such as erysipelas, cellulitis or cutaneous abscesses in effort of investigating the diagnostic yields of YKL-40.
To our knowledge this is the first study to report plasma YKL-40 levels in patients with NSTI. A strength of the present study is the sample size and that no patients were lost to follow-up. We had broad inclusion criteria and only few exclusion criteria. Additionally, the clinical data was prospectively collected by dedicated personnel using a predefined and standardized protocol [
27]. Lastly, the laboratory conducting the plasma YKL-40 analyses was blinded for patient type and outcome. However, some limitations exist: First, since this is an observational study there is a potential risk of missing unknown confounders. Second, although our study center has a national NSTI service treatment and thereby increasing the heterogeneity, we may be missing the patients who are too ill for transportation consequently introducing an important selection bias into our analyses. Third, several patients were either on immunosuppressing drugs or steroid treatment at admission theoretically impacting YKL-40 concentration. Last, no predefined protocol regarding biomarker analyses was published.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.