Analysis of Prognostic Risk Factors and Establishment of Prognostic Scoring System for Secondary Adult Hemophagocytic Syndrome

Introduction: The objective of this paper is to identify the prognostic risk factors of secondary adult hemophagocytic syndrome (HLH) in hospitalized patients and establish a simple and convenient prognostic scoring system. Method:We reviewed 162 adult patients secondary with HLH treated in Zhejiang Cancer Hospital and the First Affiliated Hospital of Medical College of Zhejiang University from January 2014 to December 2018 were enrolled to form the test group; from January 2019 to February 2021, 162 adult patients in the hospitals constituted the validation group. The HLH prognosis scoring system was constructed according to the risk factors, and the patients were divided into three risk groups: low risk, medium risk, and high risk. The scoring system was verified by Kaplan–Meier method and log rank test survival analysis. The discrimination ability was evaluated according to the receiver operating characteristic (ROC) curve. Results: Univariate and multivariate analysis showed that the independent risk factors for the prognosis of HLH were male sex, activated partial prothrombin time (APTT) greater than 36 s, lactate dehydrogenase (LDH) greater than 1000 U/L, and C-reactive protein (CRP) greater than 100 mg/L. The area under the ROC curve was 0.754 (95% Cl: 0.678–0.829). The patients were divided into a low-risk group (0–1), a medium-risk group (2–4), and a high-risk group (5–6). The 5-year overall survival (OS) rate were 87.5%, 41.8% and 12.8%, respectively (p < 0.001). The area under ROC curve was 0.736 (95% Cl: 0.660–0.813) in the validation group, and the 2-year OS of patients in low-risk, medium-risk and high-risk groups were 88.0%, 45.1% and 16.7%, respectively (p < 0.001). Conclusion:The new prognostic scoring system can accurately predict the prognosis of secondary adult HLH and can further provide basis for the accurate treatment of secondary adult HLH.


Introduction
Hemophagocytic syndrome, also known as hemophagocytic lymphohistiocytosis (HLH), is a clinical syndrome of immune overactivation, particularly of lymphocytes and histiocytes, with resultant hypercytokinemia [1]. HLH disease may be associated with specific genetic and/or environmental causes. HLH disease mimics are disorders that resemble HLH syndrome but are caused by other conditions. Historically, HLH has been divided into a primary form and secondary forms. Primary HLH is a heritable disease conferred by highly penetrant genetic mutations/variations impacting cytolytic functions, lymphocyte survival, or inflammasome activation. In contrast, secondary HLH is driven primarily by acquired factors such as chronic inflammation, infection, or malignancy. Uncontrolled and harmful immune activation results in excessive inflammation and tissue destruction.

Follow Up Time and Primary End Point
The main outcome measure was overall survival, that is, from the date of disease diagnosis to the date of death or the last follow-up (28 February 2021).

Statistical Analysis
Statistical analysis was performed using IBM SPSS 26 software. We compared clinical and laboratory data between test and validation groups, survival and death group using Chi square test. The independent risk factors of OS were analyzed by univariate and multivariate Cox risk regression model, grouped according to the independent risk factors, and survival analysis was carried out by Kaplan-Meier method and logrank test to preliminarily verify the value of independent risk factors. Then, the clinical prediction model scoring system was established by assigning independent risk factors. According to the score, the patients were divided into low-risk, medium-risk, and high-risk groups. Kaplan-Meier method and logrank test survival analysis were used to verify the application value of the scoring system, and ROC curve was used to evaluate the model discrimination ability. p < 0.05 was statistically significant.

Patient Characteristics
A total of 324 patients were included in the study. A total of 162 cases were included in the test group, including 97 males (59.8%), with an average age of 51 years. A total of 162 cases were included in the validation group, including 91 males (56.2%), with an average age of 49 years. As of the end point of follow-up, 181 patients died and 143 survived. There was no significant difference in basic values between the two groups (Tables 1 and 2).

Establishment of Prognostic Scoring System
Based on β coefficient and OR value, assign scores to each risk factor. Male sex, APTT > 36 s, LDH > 1000 U/L, and CRP > 100 mg/L are, respectively, assigned as 1 point, 2 points, 2 points, and 1 point, otherwise it is 0 point, and the total score of the four items is 6 points. The patients were divided into groups, including 12 cases with 0 points, 19 cases with 1 point, 24 cases with 2 points, 36 cases with 3 points, 28 cases with 4 points, 34 cases with 5 points, and 9 cases with 6 points. There was no significant difference in OS between patients with scores of 0 and 1 (p > 0.05), between patients with scores of 2, 3, and 4 (p> 0.05), and between patients with scores of 5 and 6 (p> 0.05). Therefore, we established a simple scoring system according to the score (Table 5). After this, we divided the system into three risk groups: low-risk (0~1 points), medium-risk (2~4 points), and high-risk (5~6 points).

Establishment of Prognostic Scoring System
Based on β coefficient and OR value, assign scores to each risk factor. Male sex, APTT > 36 s, LDH > 1000 U/L, and CRP > 100 mg/L are, respectively, assigned as 1 point, 2 points, 2 points, and 1 point, otherwise it is 0 point, and the total score of the four items is 6 points. The patients were divided into groups, including 12 cases with 0 points, 19 cases with 1 point, 24 cases with 2 points, 36 cases with 3 points, 28 cases with 4 points, 34 cases with 5 points, and 9 cases with 6 points. There was no significant difference in OS between patients with scores of 0 and 1 (p > 0.05), between patients with scores of 2, 3, and 4 (p> 0.05), and between patients with scores of 5 and 6 (p> 0.05). Therefore, we established a simple scoring system according to the score (Table 5). After this, we divided the system into three risk groups: low-risk (0~1 points), medium-risk (2~4 points), and high-risk (5~6 points).

Establishment of Prognostic Scoring System
Based on β coefficient and OR value, assign scores to each risk factor. Male sex, APTT > 36 s, LDH > 1000 U/L, and CRP > 100 mg/L are, respectively, assigned as 1 point, 2 points, 2 points, and 1 point, otherwise it is 0 point, and the total score of the four items is 6 points. The patients were divided into groups, including 12 cases with 0 points, 19 cases with 1 point, 24 cases with 2 points, 36 cases with 3 points, 28 cases with 4 points, 34 cases with 5 points, and 9 cases with 6 points. There was no significant difference in OS between patients with scores of 0 and 1 (p > 0.05), between patients with scores of 2, 3, and 4 (p> 0.05), and between patients with scores of 5 and 6 (p> 0.05). Therefore, we established a simple scoring system according to the score (Table 5). After this, we divided the system into three risk groups: low-risk (0~1 points), medium-risk (2~4 points), and high-risk (5~6 points).  Abbreviations: APTT = activated partial thromboplastin time; LDH = lactate dehydrogenase; CRP = C-reactive protein.

HLH Scoring System Verification
In order to further verify the prediction efficiency of HLH scoring system, the ROC curve was analyzed. In the test group, the area under the ROC curve of 162 cases was 0.792, p < 0.001, and the 95% confidence interval was 0.723~0.861 (Figure 8). In addition, in the validation group, the area under the ROC curve of 162 cases was 0.736, p < 0.001, Figure 6. Comparison of overall survival in patients in the low-risk group, medium-risk group, and high-risk group (test group, 71.7% vs. 41.8% vs. 12.8%, p < 0.001).

HLH Scoring System Verification
In order to further verify the prediction efficiency of HLH scoring system, the ROC curve was analyzed. In the test group, the area under the ROC curve of 162 cases was 0.792, p < 0.001, and the 95% confidence interval was 0.723~0.861 (Figure 8). In addition, in the validation group, the area under the ROC curve of 162 cases was 0.736, p < 0.001, Figure 7. Comparison of overall survival in patients in the low-risk group, medium-risk group, and high-risk group (validation group, 88.0% vs. 45.1% vs. 16.7%, p < 0.001).

HLH Scoring System Verification
In order to further verify the prediction efficiency of HLH scoring system, the ROC curve was analyzed. In the test group, the area under the ROC curve of 162 cases was 0.792, p < 0.001, and the 95% confidence interval was 0.723~0.861 (Figure 8). In addition, in the validation group, the area under the ROC curve of 162 cases was 0.736, p < 0.001, and the 95% confidence interval was 0.660~0.813 (Figure 9). This shows that it has good prediction ability. and the 95% confidence interval was 0.660~0.813 (Figure 9). This shows that it has good prediction ability.

Discussion
HLH includes two types: primary and secondary, but both primary and secondary include the activation of immune tissue cells in a superimposed state and the out-of-control regulation of the immune system. If not blocked, it can lead to continuous proliferation and activation coupled with blocked apoptosis, resulting in a high level of cytokines and a cytokine storm [5]. At present, the prognostic factors and survival time of HLH are Curr. Oncol. 2022, 29, FOR PEER REVIEW 10 and the 95% confidence interval was 0.660~0.813 (Figure 9). This shows that it has good prediction ability.

Discussion
HLH includes two types: primary and secondary, but both primary and secondary include the activation of immune tissue cells in a superimposed state and the out-of-control regulation of the immune system. If not blocked, it can lead to continuous proliferation and activation coupled with blocked apoptosis, resulting in a high level of cytokines and a cytokine storm [5]. At present, the prognostic factors and survival time of HLH are

Discussion
HLH includes two types: primary and secondary, but both primary and secondary include the activation of immune tissue cells in a superimposed state and the out-of-control regulation of the immune system. If not blocked, it can lead to continuous proliferation and activation coupled with blocked apoptosis, resulting in a high level of cytokines and a cytokine storm [5]. At present, the prognostic factors and survival time of HLH are uncertain, but understanding the adverse factors related to disease progression and prognosis is of great significance for the evaluation of the disease, the rational formulation of a treatment plan, and the improvement of the prognosis and survival rate of patients.
Shunichi et al. [11] reported that 116 patients with autoimmune-associated HLH, malesex (p < 0.01, HR = 6.47, 95% CI: 2.06~30.39) was identified as the factors associated with mortality. Coburn et al. [12] reported an incidence for systematically characterize HLH in moderate-to-severe inflammatory bowel disease. Additionally, found that HLH occurred more often in males (70.0%). Risk factors may include male sex, presence of Crohn's disease, and induction phase of treatment.
Coagulation disorders are common during HLH and play a key role both in the global severity of the disease and in the occurrence of hemorrhagic complications [13]. Coagulation disorders confer a higher risk of bleeding, and this complication can be severe. Raised D-dimer levels and coagulation disorders are also reported in 50% of the cases, and nearly half of the patients fulfill disseminated intravascular coagulation (DIC) criteria [14][15][16]. In a retrospective study, 16 of 29 patients (55%) with lymphoma-related HLH had DIC [17]. Coagulation impairment is strongly correlated to the prognosis in patients with HLH [13]. Chen et al. [18] put forward that prolonged APTT > 44.35 s is a strong predictive factor for mortality. Multivariate Cox regression analysis demonstrated that APTT (p = 0.045, HR = 1.03, 95% Cl: 1.00~1.10) was an independent risk factor for mortality. DIC caused by coagulation dysfunction is also one of the main causes of HLH death [13,19,20]. Coagulation disorders are often related to severe systemic inflammation, DIC, and coagulation factor defects caused by liver failure. A large number of IFN cytokines such asγ, TNF, and IL-1 release activate cytotoxic T cells and macrophages, make them proliferate and activate in large numbers, produce hypercytokinemia, enhance macrophage phagocytosis, and cause coagulation disorders [21].
LDH could sensitively and comprehensively reflect the organ index of tissue damage. Clinical risk factors related to HLH included maximum LDH [22]. A second case series by Leow et al. [23] described a cohort of pediatric patients with HLH admitted to the cardiac ICU and assessed for poor prognostic factors and mortality. Patients with a higher median peak serum LDH levels were associated with higher mortality. Furthermore, elevated LDH was demonstrated to be a poor prognostic factor for survival in lymphoma associated hemophilus syndrome (LAHS). Jia et al. [20] put forward that univariate analysis showed that patients of NK/T-cell lymphoma associated hemophilus syndrome (NK/T-LAHS) with LDH > 1000 U/L (p = 0.048) and DIC (p = 0.004) had shorter survival time. Other studies also demonstrated that elevation of LDH was associated with unfavorable outcomes in patients with NK/T-LAHS [8,9,24,25]. Li et al. [9] reported that the risk factors for NK/T-LAHS was elevated LDH level (> 314 U/L) (p = 0.038, HR = 6.293, 95%Cl: 1.108~35.735). However, the optimal value of LDH serving as a risk factor in LAHS was unclear. Different studies show that 1000 U/L or 2000 U/L was the threshold value for 10 prognoses [20,26,27]. Zhang et al. [28] noticed that a four times elevation of LDH (>1000 U/L) predicted poor prognosis in patients with NK/T-LAHS. We inferred that the prominent elevation of LDH might be a direct consequence of the cytokine storm and hyperinflammation.
The prognosis of Epstein-Barr-virus-associated hemophilus syndrome (EBV-HLH) patients was significantly correlated with CRP [29]. Bozkurt et al. [30] showed that CRP was detected as mortality predictors in the univariate analysis. Fukaya et al. [31] multivariate analysis showed that the presence of infections and CRP level (>50 mg/L) on HLH related with poor prognosis, thus high CRP level may relate to infection rather than to HLH itself. Taking these studies and our study together, infections seem to be the common risk factor in adult HLH patients. Interestingly, CRP also showed correlations with ferritin values which might indicate that these inflammatory parameters are part of the cytokine pattern of HLH [32].
In our study, the four indexes were assigned one by one according to their weight in Cox regression, and a more accurate clinical prediction equation was established. According to the score, the patients were divided into low-risk, medium-risk, and high-risk groups. The 5-year OS of the low-risk group was 71.7%, that of the medium-risk group was 41.8%, and that of the high-risk group was 12.8%. There was significant difference in survival among the low-risk, medium-risk, and high-risk groups. In addition, in clinical practice, we further verified the efficacy of the HLH prognosis scoring system in HLH patients. It was found that the 2-year OS of patients in the low-risk, medium-risk, and high-risk groups were 88.0%, 45.1%, and 16.7%, respectively. This study also has some limitations. First, the risk scoring model needs to be further verified in prospective studies. Secondly, due to the large difference of adjuvant therapy in patients, it is not included in the analysis of influencing factors, and its impact on the prognosis of HLH is still unclear. This work needs to be further carried out.
In conclusion, male sex, APTT > 36 s, LDH > 1000 U/L and CRP > 100 mg/L are risk factors for the prognosis of HLH patients. The prognostic scoring system established in this study can be used to predict the long-term survival of HLH patients.

Supplementary Materials:
The following supporting information can be downloaded at https: //figshare.com/s/a1bffb14f5291fbfbbe2.
Author Contributions: Q.Z. collected, analyzed, and interpreted the data and drafted the manuscript. Y.L. collected, analyzed, and interpreted the data and revised the manuscript. Y.B. and Y.J. performed statistical analyses, interpreted the data, and revised the manuscript. X.Y. obtained the samples, updated the clinical data, and reviewed the manuscript. Y.T. reviewed the paper and gave final approval for the submitted version. All authors have read and agreed to the published version of the manuscript.
Funding: This study was partly supported by Zhejiang Provincial Public Welfare Technology Application Research and Subsidy Project (LGF19H080003) and Zhejiang Medical and Health Science and Technology Program (2020ky1071).

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of Zhejiang Cancer Hospital (protocol code IRB-2022-55, 21 January 2021).
Informed Consent Statement: Additional informed consent was obtained from all individual participants for whom identifying information is included in this article (some patients received informed consent from their families because of death).

Data Availability Statement:
Supplementary data to this article can be found online at https:// figshare.com/s/a1bffb14f5291fbfbbe2.