In the present study, serum SFN levels were significantly higher in the AE-ILD group than in the CS-ILD group. The high level of serum SFN at the onset of AE was an independent prognostic factor for mortality in the AE-ILD group. The cut-off level of SFN that best predict 90-day and 1-year survival rates in the AE-ILD group was 6.6 ng/mL. The serum SFN could be a valuable prognostic index as well as a biomarker for AE-ILD.
SFN (14-3-3σ) belongs to the 14-3-3 family, together with six other isoforms, namely, β, γ, ε, ζ, η, and τ [14]. This family constitutes a group of highly evolutionarily conserved proteins with a molecular mass of 25–30 kDa, expressed in all eukaryotes and involved in the modulation of various cellular processes via binding to phosphorylated proteins [11]. While most 14-3-3 isoforms are expressed in various normal tissues in a ubiquitous manner, SFN is expressed specifically in stratified epithelia [15]. On the other hand, SFN is weakly expressed in the normal respiratory alveolar epithelium, and the expression of SFN is reportedly associated with lung cancer [12, 16]. Recently, we found that serum and bronchoalveolar lavage fluid (BALF) levels of SFN functioned as biomarkers reflecting DAD patterns in drug-induced ILD. Serum SFN levels specifically increased in patients with DAD and could be used as a prospective biomarker in drug-induced ILD [13]. The performance of SFN as a biomarker in drug-induced DAD could be superior to existing biomarkers currently used in clinical settings, such as KL-6 and SP-D.
The intracellular function and biological activity of SFN have been investigated, although no definitive conclusions have been made. SFN expression is induced in a p53-dependent manner in response to DNA damage, and the progression of the G2/M phase of the cell cycle is arrested [17]. The alveolar epithelial damage may trigger an increase in the extracellular SFN concentration via the p53 pathway, thereby increasing the serum SFN concentration. We suggested that BALF and serum SFN concentrations were correlated [13]. In addition, SFN is thought to be a marker of the differentiation of epidermal keratinocytes. Keratinocyte-released SFN is thought to be associated with skin wound healing and suppression of skin fibrosis [18]. Perhaps, alveolar epithelia-produced SFN is associated with pulmonary damage and its healing. In this study, we demonstrated that serum SFN level may be a diagnostic biomarker for not only drug-induced ILD but also AE-ILD. The application of SFN as a biomarker of drug-induced lung injury and AEs in various types of ILD in clinical settings is under investigation.
In the present study, patients with high levels of serum SFN had a poorer prognosis. We divided patients with AE-ILD into two groups according to their serum SFN levels to determine whether the level established using the ROC curve reflected the prognosis of patients with AE-ILD. We examined 90-day and 1-year survival, and we demonstrated that the group with high SFN values had a significantly worse prognosis (p = 0.0017 and p < 0.0001, respectively). The occurrence of an AE is a clinically significant event because of its correlation with increased mortality. Despite the efforts of many researchers, the exact pathogenesis of AEs has not been elucidated, and AEs remain the leading cause of death in patients with ILD in Japan [19]. Serum biomarkers that are useful for predicting the risk of AEs in patients with ILD have been explored, and KL-6, interleukin-6, monomeric periostin, decorin and syndecan-4 have been proposed as candidate predictive biomarkers [20–24]. Some biomarkers are related to AE and prognosis; however, a biomarker associated with DAD has not been developed. In many studies, inflammatory cytokines have been used as biomarkers for AE-ILD [25–27]. However, SFN, derived from the respiratory alveolar epithelium, might be more relevant to alveolar epithelial damage since it is only mildly elevated in cases of lung infection. KL-6 is also a protein derived from the alveolar epithelium, but it is not useful for prognostic determination [9]. We believe that SFN is reflective of the disease state and may be a useful biomarker that can predict AE, disease course, and prognosis.
The present study has several limitations. First, it was a single-centre cohort study, and we used surplus serum samples collected during previous studies. Therefore, the sample size was small. Second, a comparison between stable and exacerbation periods in the same patient could not be performed, since such a sample was not obtained. Third, the arterial blood gas analysis data could not be gathered to differentiate between clinically stable condition and exacerbation; therefore, we used SpO2 instead. Lastly, we could not follow the course of SFN levels in patients with AE-ILD. Future studies are needed to confirm the pathophysiological significance of serum SFN levels.