Usefulness of the modied hemophagocytic syndrome diagnostic score as a prognostic factor in lung transplantation patients

Background: Lung transplantation (LTX) is an established treatment for end-stage lung disease; however, the post-LTX mortality rate remains high. This study aimed to evaluate the prognostic value of the modied reactive hemophagocytic syndrome diagnostic score (mHScore) and its individual components on mortality after LTX. Methods: We retrospectively analyzed 294 patients who underwent LTX at Severance Hospital, Yonsei University, Korea, from January 2012 and December 2020, and classied them into high (n=114, mHScore > 104.0) and low mHScore (n=180, mHScore ≤ 104.0) groups. Triglyceride, ferritin, serum glutamic oxaloacetic transaminase, brinogen, and cytopenia were used to calculate the mHScore. We compared baseline characteristics and mortality rates as LTX prognostic factors. Results: The high mHScore group had signicantly more cytopenia and higher ferritin, triglyceride, lactate dehydrogenase, and C-reactive protein levels than the low mHScore group. The mortality rate was signicantly higher in the high than in the low mHScore group (hazard ratio, 4.429, p < 0.001). Multivariate regression analysis revealed that a high mHScore was signicantly associated with postoperative mortality, even after adjusting for other confounding factors. A high mHScore was also associated with postoperative complications. Conclusions: The mHScore can be used to estimate post-LTX prognosis and predict postoperative mortality. Receiver operating characteristic curve analysis for determining the cut-off value of the modied hemophagocytic syndrome score (mHScore) that best predicts overall survival in lung transplantation (LTX) patients. The optimal cut-off value for


Background
Lung transplantation (LTX) is the most effective therapeutic option for patients with end-stage lung disease [1,2]. Although the introduction of more effective immunosuppression regimens, advancements in surgical techniques, improved management of postoperative and infectious complications, and adoption of evidence-based guidelines for donor and recipient selection have led to improved outcomes after LTX, the outcomes for LTX remain suboptimal compared to those of other solid-organ transplants [3][4][5][6][7]. Various factors have been evaluated for predicting prognosis in patients who have undergone LTX, including age, underlying disease, bilateral LTX, body mass index (BMI), and malnutrition [8][9][10][11]. The immune system plays a critical role in prognosis at all time-points after LTX, because the degree of immunosuppression may increase the risk of infection, malignancy, and rejection [12,13].
Infection and allograft failure due to primary graft dysfunction (PGD) are the leading causes of death early after LTX, and chronic lung allograft dysfunction is the primary cause of death beyond the rst year after LTX. These postoperative complications are generally associated with the immune system, in ammatory processes, and recurrent infections, and directly increase the risk of death [12]. Indeed, the inability to accurately monitor the immune status in immunocompromised patients to predict prognosis remains a signi cant challenge in clinical practice.
The hyperin ammatory condition caused by infection, malignancy, autoimmune disorders, immunocompromised states, and solid organ transplant may be related to dysregulated and ineffective immune responses and may be a leading cause of poor prognosis. Numerous indicators have been used as markers re ecting immune status, including serum C-reactive protein (CRP), procalcitonin, lactate dehydrogenase (LDH), ferritin, brinogen, triglycerides (TG), cytopenia, and cytokines [14][15][16]. However, there are limits to using these biomarkers independently to predict immune system abnormalities and patient prognosis after LTX. Additionally, although biomarkers have been used to predict infection, there have been no reports on post-LTX survival outcomes according to biomarkers. If prognosis can be predicted using a scoring system that includes individual biomarkers, this may lead to earlier diagnosis of complications during the LTX follow-up period and post-LTX better outcomes.
Fardet et al. constructed and validated the HScore for diagnosis of reactive hemophagocytic syndrome, which is a hyperin ammatory condition caused by highly stimulated, but dysregulated and often ineffective immune responses [17]. The HScore is determined by biological (i.e., TG, ferritin, serum glutamic oxaloacetic transaminase [SGOT], brinogen levels, and cytopenia), cytological (i.e., hemophagocytosis features in aspirated bone marrow), and clinical (i.e., known underlying immunosuppression, high temperature, and organomegaly) features. Such a hyperin ammatory assessment tool may provide prognostic indicators for patients who have undergone LTX.
In the current study, we investigated whether the postoperative prognosis of LTX patients could be predicted using a modi ed HScore (mHScore), calculated using only the ve biological features (e.g., cytopenia, ferritin, brinogen, TG, and SGOT) in peripheral blood. Considering the relationship between a prolonged hyper-in ammatory status and the prognosis of LTX patients, we hypothesized that a high mHScore after LTX may be associated with poor prognosis.

Study Population and Data Source
This study included 294 patients who underwent LTX at Severance Hospital, Yonsei University, Korea, between January 2012 and December 2020. We retrospectively reviewed the medical data of all patients and extracted information pertaining to demographic characteristics and pre-and postoperative parameters to evaluate the relationships between these variables and survival after LTX. The included patients were divided into low and high mHScore groups, based on their mHScore calculated from 3month post-LTX laboratory data. We analyzed the following pre-LTX variables: demographic data, including age, sex, and BMI; comorbidities, such as diabetes mellitus and hypertension; primary lung disease; preoperative mechanical ventilation and extracorporeal membrane oxygenation (ECMO); donorrecipient total lung capacity (TLC) ratio, and the ratio of arterial oxygen concentration to the fraction of inspired oxygen (P/F) in the donor.
Modi ed Hemophagocytic Syndrome Diagnostic Score as a Prognostic Factor In this study, the mHScore was used to evaluate postoperative prognosis. Five biological variables (i.e., TG, ferritin, SGOT, brinogen levels, and cytopenia) were used to calculate the mHScore, using the scoring system at http://saintantonie,aphp.fr/score/. Supplemental Table 1 shows the parameters and criteria for scoring the mHScore. An optimal mHScore cut-off value of 104 for predicting postoperative overall survival was determined using receiver operating characteristic (ROC) curve analysis. When we analyzed serial mHScores (before LTX and 1, 3, 6, and 12 months after LTX), the mHScore at 3 months post-LTX was identi ed as most appropriate ( Figure 1). Consequently, the two groups were classi ed by an mHScore > 104 (high mHScore) and ≤ 104 (low mHScore) at 3 months after LTX.

Postoperative Complications
We compared the incidence of postoperative complications between patients with low and high mHScores. Acute kidney injury was de ned as an increase in the creatinine level to 0.3 mg/dL within 48 hours, or an increase in the creatinine level by 30% from the baseline value during the rst 3 months after LTX [18]. Pneumonia was diagnosed according to positive sputum cultures or bronchioalveolar lavage cultures, with corresponding radiological ndings, requiring treatment. Any positive blood culture and qSOFA score of ≥ 2 was de ned as sepsis. Hemophagocytic lymphohistiocytosis (HLH) was diagnosed if at least ve of the eight following features were present: fever, splenomegaly, cytopenia, hypertriglyceridemia, hypo brinogenemia, evidence of hemophagocytosis in bone marrow aspirates, hyperferritinemia, low natural killer cell activity, or elevated soluble CD25 [19]. Thrombotic thrombocytopenic purpura was de ned by the presence of hemolysis, such as thrombocytopenia, anemia, elevated reticulocyte count, elevated serum LDH, and bilirubin levels, fragmented red blood cells on the blood smear, negative direct antiglobulin test, normal coagulation pro le, and ADAMTS 13 < 10%, requiring treatment [20,21]. Hemolytic uremic syndrome was de ned by the presence of microangiopahtic hemolytic anemia, thrombocytopenia, and acute renal dysfunction [22]. PGD was graded based on the criteria from the International Society for Heart and Lung Transplantation which is in turnbased on the partial pressure of oxygen/fraction of inspired oxygen ratio and the presence of diffuse parenchymal in ltrates in the allograft on a chest radiograph [23][24][25]. Acute rejection was de ned as a medical condition with clinical evidence of suspected rejection or pathological ndings that necessitated steroid pulse therapy. All major postoperative complications and infections during the follow-up period were analyzed.

Statistical Analysis
All statistical analyses were performed using IBM SPSS Statistics software (version 25.0; IBM Corp., Armonk, NY). Data are expressed as means and standard deviations or medians and interquartile ranges, as appropriate. Continuous and categorical variables were analyzed by Student's t test and the chisquared or Fisher's exact test, respectively. The optimal cut-off mHScore was determined using ROC curve analysis. A multivariate logistic regression analysis was performed with variables with p < 0.05 in the bivariate analysis to identify independent risk factors. Hazard ratio (HRs) and 95% con dence intervals (CIs) were calculated. Survival data were estimated using the Kaplan-Meier method, and statistical differences in survival were determined using the log rank test. In all analyses, statistical signi cance was set at p < 0.05.

Cut-off Level of mHScore and Baseline Characteristics
The ROC curves for the mHScore at various perioperative times for prediction of post-LTX prognosis are shown in Figure 1. The median cut-off level of mHScore for the prediction of postoperative prognosis based on the 3-months post-LTX data was 104.0 (area under the curve = 0.71; 95% CI: 0.645-0.779; p < 0.001; Fig. 1) and the cut-off level at 3, 6, and 12-months were similar. This cut-off value corresponded to a sensitivity of 65.2% and speci city of 75.3%. Based on this cut-off value, 180 (61.2%) and 114 (38.8%) patients were classi ed into the low mHScore (≤104.0) and high mHScore (> 104.0) groups, respectively.
The baseline characteristics of all patients who underwent LTX are shown in Table 1. The most common reason for LTX was idiopathic pulmonary brosis in both groups. There was no signi cant difference between low and high HScore groups in the following preoperative parameters: age, sex, BMI, comorbidities, primary lung disease, preoperative mechanical ventilation, preoperative ECMO, donorrecipient TLC ratio, and donor P/F ratio. Table 2 presents the explanatory variables of in ammatory marker levels in LTX patients categorized by mHScore. LDH and CRP levels were signi cantly higher in the high than in the low mHScore group.  The Kaplan-Meier plots for the low and high mHScore groups are shown in Figure 2. The overall survival rate was signi cantly higher in the high mHScore group than in the low mHScore group (log-rank test; p < 0.001).

Relationship between Postoperative mHScore and Causative Postoperative Complications
We also analyzed the correlation between the mHScore and the incidence of postoperative complications ( Table 4). The proportions of patients requiring renal replacement therapy and with postoperative bleeding were signi cantly higher for the high mHScore group than in the low mHScore group (64.9% vs. 30.0%, p < 0.001; 27.4% vs. 8.9%, p < 0.001, respectively). The incidence of PGD ≥ grade 3 was signi cantly higher in the high mHScore group than in the low mHScore group (59.6% vs. 46.7%, p = 0.032). Although not statistically signi cant, the incidence of acute rejection was higher in the high mHScore group (7.9% vs. 2.8%, p = 0.053). In addition, the duration of intensive care unit admission after LTX and the duration of time on ECMO after LTX were signi cantly longer in the high mHScore group than the low mHScore group. Values are expressed as means (standard deviations). mHScore, modi ed hemophagocytic syndrome score; AKI, acute kidney injury; RRT, renal replacement therapy; TTP, thrombotic thrombocytopenic purpura; HUS, hemolytic uremic syndrome; HLH, hemophagocytic lymphohistiocytosis; BPF, bronchopleural stula; GI, gastrointestinal; PGD, primary graft dysfunction; LOS, length of stay; ICU, intensive care unit; ECMO, extracorporeal membrane oxygenation.

Discussion
In the present study, we investigated the association between postoperative mHScore and LTX prognosis, including overall survival and complications among LTX recipients. We showed that mortality and postoperative complication rates were higher in patients with a high mHScore, even after adjusting for other confounding factors. These ndings suggest that a high mHScore is an independent predictor of a poor survival rate among patients undergoing LTX.
The   [9][10][11]. However, no previous study has reported the association of biomarkers with post-LTX prognosis. Our study suggests that a high HScore is an independent predictor of overall survival in LTX recipients. To the best of our knowledge, no previous study has evaluated the association between high in ammation status (as assessed using the mHScore) and survival in LTX recipients.
The mHScore is based on ve biological features: cytopenia, ferritin, brinogen, triglyceride, and SGOT levels. The HScore was originally developed to diagnose HLH, and includes one cytological and three clinical features, in addition to the ve biological features. In this study, we modi ed the score to represent high in ammatory status in LTX patients, by using only the laboratory values that can be easily measured.
Cytopenia, including leukopenia, anemia, and thrombocytopenia, occurs after a solid organ transplant. The etiology of cytopenia is multifactorial and includes drugs, infections, and post-transplant lymphoproliferative disorders [27]. It also increases the risk of developing further complications, such as opportunistic infections, and may contribute to poor prognosis in patients undergoing LTX. Several studies have reported that leukopenia is associated with decreased survival and increased rates of infection in LTX recipients [28,29].
Iron metabolism is intrinsically linked to innate immunity by regulation of iron availability to pathogens. A high iron status is related to many infectious diseases and in ammatory responses [15]. Maher et al.
suggested that an excess of iron stored in ferritin may increase the risk of exposure to iron radicals, reactive oxygen species, and subsequent brosis, leading to a poor prognosis [30].
Plasma brinogen is an acute phase protein that increases with in ammation or tissue necrosis. It is also related to coagulopathy, which may decrease in response to infection, massive blood loss, and to a dilution effect caused by uid replacement. Coagulation disorders are common in HLH patients, and several retrospective studies have reported their prognostic role in adult HLH [31,32]. Fibrinogen, as a risk factor for bleeding complications in LTX patients, has been reported in a retrospective cohort study. It is also a biomarker of disseminated intravascular coagulation [33].
Lipid abnormality, including hypertriglyceridemia, is caused by high levels of in ammatory cytokines resulting from a dysregulated immune system [34,35].
An abnormal liver function was used a marker of acute liver failure in this study. LDH can act as a prognostic marker of diseases, including hemolytic anemia and infections. In our study, LDH and CRP levels, which were not included as components of the mHScore, correlated with overall survival in patients undergoing LTX.
Each of these values is an indicator of infection and a dysregulated immune response in immunocompromised patients. This study is meaningful in revealing biomarkers that correlated with poor prognosis post-LTX.
Nevertheless, this study had several limitations worth noting. First, it represents the clinical experience from a single center. Second, this study had a retrospective design, with the possibility of bias in the selection of parameters. Third, the HScore is a scoring system originally developed for the diagnosis of HLH, and we applied a subsection of this score in LTX recipients, without validation of its utility in this patient population; therefore, we cannot exclude the possibility that the weight of the score was over-or under-applied in this population. However, the strength of this study is that it showed the predictive value of the mHScore and its component biomarkers in LTX patients. In the future, we will construct a prognostic scoring model speci cally for LTX patients using the in ammatory markers included in the mHScore.

Conclusion
The current results showed that overall survival was signi cantly longer for the high mHScore group than for the low mHScore group. These ndings suggest that the mHScore is a useful marker for predicting prognosis in post-LTX patients. By alerting clinicians to the patient's likely prognosis, this nding may facilitate the development of both preventive and early intervention strategies that can improve the overall survival of LTX patients.

Availability of data and materials
Individual participant data will not be made available.

Competing interests
The authors declare that they have no competing interests.

Funding
This research did not receive any speci c grants from funding agencies in the public, commercial, or notfor-pro t sectors. M.S.P designed the concept and nally approved the paper. All authors have taken due care to ensure the integrity of this work. The nal manuscript has been read and approved by all authors.