Early prognosis evaluation method for hemophagocytic lymphocytosis

Abstract

The invention discloses an early prognosis evaluation method for hemophagocytic lymphocytosis, which belongs to the technical field of medical examination and comprises the following steps: researching, designing and determining patients: completing random grouping of the patients according to clear diagnosis of secondary HLH (human leukocyte lymphocytosis); after clinical diagnosis data retrieval and clinical confirmation, clinical cause grouping of each secondary HLH patient is completed, and mNGS detection of a peripheral blood sample of each patient is further completed; dNA extraction, library preparation and NGS; pathogen detection and identification; performing abnormal CNV feature analysis; and statistical analysis. According to the early prognosis evaluation method for the hemophagocytic lymphocytosis, early evaluation of treatment prognosis can be carried out on clinically diagnosed secondary HLH patients, the economic cost and the time cost of disease diagnosis are greatly reduced, and clinical precise diagnosis and treatment and patient benefits are facilitated.

Description

Technical Field

[0001] The invention belongs to the technical field of medical testing, and in particular relates to an early prognosis evaluation method for hemophagocytic lymphohistiocytosis. Background Art

[0002] Hemophagocytic lymphohistiocytosis (HLH), also known as hemophagocytic syndrome, is a life-threatening hyperinflammatory syndrome caused by abnormal activation of cytotoxic T lymphocytes, natural killer cells, and macrophages, resulting in hypercytemia and immune-mediated damage to multiple organs. As a syndrome, the clinical manifestations of HLH vary, including fever, organomegaly, pancytopenia, consumptive coagulopathy, hypertriglyceridemia, and elevated acute phase reactants. Subspecialists in different fields such as hematology / oncology, infectious diseases, and rheumatology / clinical immunology are challenged by this rare, multifaceted syndrome, as patients often present with recurrent fever, pancytopenia, abnormal liver function, and sepsis-like syndromes, and may rapidly develop end-stage multi-organ failure. Therefore, despite being a rare disease, clinicians should be aware of HLH, as early recognition can avoid irreversible organ damage and subsequent death.

[0003] Because the etiology of HLH is complex, patients suspected of HLH need to be comprehensively evaluated and systematically diagnosed. In clinical practice, once a patient is suspected of having HLH, multiple evaluations should be started simultaneously. Laboratory and imaging tests should be performed to collect supportive evidence for the diagnosis of HLH syndrome and to assess which organ systems are affected and the severity of the involvement. All patients should also be evaluated for infection and malignancy, including laboratory tests, bone marrow evaluation, systemic imaging, and biopsy of any suspicious results.

[0004] Since HLH is an aggressive and fatal syndrome, early diagnosis and etiology identification are crucial to initiating appropriate treatment and improving patients' quality of life and survival rate. However, it is challenging to quickly identify the underlying infectious etiology of HLH because traditional etiology diagnosis takes a long time and sometimes the pathogen cannot be identified. In addition, traditional malignant tumor diagnosis is also time-consuming, has a low positive rate, and greatly increases the psychological and economic burden on patients due to invasive measures. Summary of the invention

[0005] The object of the present invention is to provide a method for early prognosis assessment of hemophagocytic lymphohistiocytosis to solve the problem that the traditional etiology diagnosis proposed in the above background technology is time-consuming and sometimes the pathogen cannot be identified, so it is challenging to quickly determine the potential infectious cause of HLH; in addition, the traditional diagnosis of malignant tumors is also time-consuming and has a low positive rate, and due to the invasive measures taken, it greatly increases the psychological and economic burden on patients.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for early prognosis assessment of hemophagocytic lymphohistiocytosis, comprising the following steps:

[0007] Step 1. Study design and patient identification: The random enrollment of patients was completed based on the clear diagnosis of secondary HLH. Then, the etiology clinical grouping of each secondary HLH patient was completed after clinical diagnostic data retrieval and clinical confirmation. The mNGS test of each patient's peripheral blood sample was further completed. Finally, based on the obtained CNV parameters and characteristic microbial data, combined with clinical indicators, the construction and effect evaluation of random forest binary or ternary classifiers for early prognosis evaluation of secondary HLH were completed.

[0008] The relevant data came from peripheral blood samples collected from 92 patients at Beijing Friendship Hospital affiliated to Capital Medical University;

[0009] Step 2, DNA extraction, library preparation, and NGS: DNA extraction and sequencing library preparation were performed on peripheral blood samples. Before sequencing, the quality of DNA was assessed using a BioAnalyzer 2100 (Agilent Technologies; Santa Clara, CA, United States) combined with quantitative PCR to measure adapters. The library concentration was then adjusted (with a target of 20 million (M) reads) and NGS sequencing was performed on an Illumina Next Seq 550 Dx system using a single-end (SE) 50 base pair (bp) sequencing strategy.

[0010] Step 3, pathogen detection and identification: A total of 10 to 20 million sequences of sequencing data were obtained for each sample (clinical sample or parallel NTC (no template control));

[0011] Step 4: Abnormal CNV feature analysis: Sequencing reads were mapped to the human reference genome (hg19) in the NCBI database, and only uniquely aligned sequencing reads were selected for subsequent analysis;

[0012] Step 5. Statistical analysis: All collected data were statistically analyzed using the R software package.

[0013] In a further embodiment, the patient conditions in step 1 are as follows: according to the HLH-2004 diagnostic criteria, all patients were diagnosed with HLH and divided into four subtypes, including EBV-induced HLH (EBV-HLH, 43 people, accounting for 46.74%), HLH caused by other pathogens (nEBV-HLH, 25 people, accounting for 27.17%), HLH caused by malignant tumors (M-HLH, 13 people, accounting for 14.13%), and HLH caused by rheumatic diseases (Rh-HLH, 11 people, accounting for 11.96%). All peripheral blood samples were tested by mNGS, and the data were used for subsequent identification of microorganisms and infectious pathogens, as well as chromosome copy number variation analysis based on human genome sequences.

[0014] In a further embodiment, the clean sequences (reads) obtained after filtering the raw data in step 3 are first compared with the NCBI database to clarify the microbial species present in each sample. Subsequently, all microorganisms detected in the clinical samples are compared and screened with the microorganisms detected in the parallel NTC (no template control) (background microorganisms) according to the parameters of unique reads per million (RPM), wherein the microorganisms not detected in the parallel NTC and the microorganisms with RPM (clinical samples): RPM (parallel NTC)>10 are defined as the microbial population that actually exists in the clinical samples. Finally, all microbial species of the microbial population that actually exists in the clinical samples are searched on PubMed, and all microorganisms that have been reported in the literature to cause infection in humans or mammals are defined as infectious pathogens.

[0015] In a further embodiment, the specific operation of step 4 is as follows: the reference genome is divided into continuous windows of fixed length, the reading depth of each window is calculated, and then normalized to the total number of reads for each sample, the normalized reading depth is divided by the average reading depth of the reference data set to obtain the copy number ratio of each window, and then the copy number is Log2, adjacent windows with similar ratios are merged into fragments, and the chromosome position and average ratio are marked, the copy number of each fragment is calculated based on the average ratio and the normal copy number of the corresponding chromosome, and then compared with the preset threshold to verify CNV.

[0016] In a further embodiment, the specific operation of step 5 is as follows: the continuous measurement data of normal distribution are expressed as mean (standard deviation) or mean (standard error), the data of non-normal distribution are expressed as median (range), the differences and significance between groups are calculated by Student's t-test (normal distribution data) and Wilcoxon rank sum test or Kruskal-Walli test (non-normal distribution data), data visualization is performed in R (version 4.2.3), alpha diversity index includes Shannon index, richness index, Simpson index and inverse Simpson index, t-test is used to evaluate the differences between groups, and Beta diversity is described by unconstrained PCoA (principal coordinate analysis). As described above, similarity analysis (Anosim) was used to test the differences between groups, and linear discriminant analysis effect size (LEfSe) analysis was performed to identify microorganisms with significant differences in abundance between groups. The threshold was log10LDA (linear discriminant analysis) score ≥ 2 and P value ≤ 0.05. Before analyzing the relative abundance of microorganisms, the RPM values ​​of microorganisms were logarithmically transformed. In this study, a bilateral P value less than 0.05 was considered statistically significant. The random forest method was used to construct a prediction model, and ten-fold cross validation was used to evaluate whether specific microorganisms and clinical indicators could be used as biomarkers to distinguish different patient groups. The predictive performance of the classifier was analyzed by ROC curve.

[0017] Technical effects and advantages of the present invention:

[0018] This method for early prognosis assessment of hemophagocytic lymphohistiocytosis has established an innovative analysis strategy and method based on mNGS data. It can establish a clinical model for early assessment of treatment prognosis for HLH patients through integrated analysis of patient chromosome copy number variation (CNV) data and microbial data generated by mNGS and screening of key evaluation parameters. It can significantly reduce the economic and time costs of disease diagnosis, contribute to clinical precision diagnosis and treatment and benefit patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 It is a schematic diagram of the process of the present invention;

[0021] Figure 2 A data graph for clinical diagnosis classification of the present invention;

[0022] Figure 3For the CNV data analysis based on Onco-mNGS of the present invention;

[0023] Figure 4 A schematic diagram of the distribution of pathogens identified by mNGS of the present invention;

[0024] Figure 5 is a schematic diagram of the pathogen composition in each sample of the present invention;

[0025] Figure 6 It is a data graph of the positive rate and diagnostic performance of mNGS of the present invention;

[0026] Figure 7 It is a data graph of the detection rate and load of EBV and CMV in the blood of patients with different HLH subtypes of the present invention;

[0027] Figure 8 The present invention is based on the analysis of microbiome data from Onco-mNGS and clinical examination data;

[0028] Fig. 9 A schematic diagram of the composition and diversity of the blood microbiome of HLH patients with different etiologies according to the present invention;

[0029] Fig.10 This is a diagram of the treatment effect of predicting EBV-HLH patients based on CNVs and blood microbiome characteristics of the present invention;

[0030] Fig.11 A schematic diagram of the relationship between the treatment outcomes of HLH patients of the present invention and CNV characteristics and blood microbiome;

[0031] Fig.12 A schematic diagram of the relationship between the blood microbiome and treatment outcomes of EBV-HLH patients of the present invention;

[0032] Fig.13 Schematic diagram of the relationship between the blood microbiome and treatment outcomes of nEBV-HLH patients of the present invention. DETAILED DESCRIPTION

[0033] In the following description, a large number of specific details are provided to provide a more thorough understanding of the present invention. However, it is apparent to those skilled in the art that the present invention can be implemented without one or more of these details. In other examples, in order to avoid confusion with the present invention, some technical features well known in the art are not described.

[0034] Unless otherwise defined, the up, down, left, right, front, back, inside and outside directions involved in this document are based on the up, down, left, right, front, back, inside and outside directions in the figures shown in the present invention, and are explained here together.

[0035] See also Figure 1-13 The present invention provides a method for early prognosis assessment of hemophagocytic lymphohistiocytosis, the method comprising the following steps, please refer to Figure 1 The flowchart shown includes the following steps:

[0036] Step 1. Study design and patient identification: The randomization of patients was completed based on the clear diagnosis of secondary HLH. Then, the etiology of each secondary HLH patient was clinically grouped after clinical diagnostic data retrieval and clinical confirmation. The mNGS test of each patient's peripheral blood sample was further completed. Finally, based on the obtained CNV parameters and characteristic microbial data, combined with clinical indicators, the construction and effect evaluation of random forest binary or ternary classifiers for early evaluation of the prognosis of secondary HLH treatment were completed.

[0037] The relevant data were obtained from peripheral blood samples collected from 92 patients at Beijing Friendship Hospital affiliated to Capital Medical University. According to the HLH-2004 diagnostic criteria, all patients were diagnosed with HLH and divided into four subtypes, including HLH caused by EBV (EBV-HLH, 43 patients, accounting for 46.74%), HLH caused by other pathogens (nEBV-HLH, 25 patients, accounting for 27.17%), HLH caused by malignant tumors (M-HLH, 13 patients, accounting for 14.13%), and HLH caused by rheumatic diseases (Rh-HLH, 11 patients, accounting for 11.96%). All peripheral blood samples were tested by mNGS, and their data were used for subsequent identification of microorganisms and infectious pathogens, as well as chromosome copy number variation analysis based on human genome sequences.

[0038] Step 2, DNA extraction, library preparation, and NGS: DNA extraction and sequencing library preparation were performed on peripheral blood samples. Before sequencing, the quality of DNA was assessed using a BioAnalyzer 2100 (Agilent Technologies; Santa Clara, CA, United States) combined with quantitative PCR to measure adapters. The library concentration was then adjusted (with a target of 20 million (M) reads) and NGS sequencing was performed on an Illumina Next Seq 550 Dx system using a single-end (SE) 50 base pair (bp) sequencing strategy.

[0039] Step 3, pathogen detection and identification: A total of 10 to 20 million sequencing data were obtained for each sample (clinical sample or parallel NTC (no template control)). The clean sequences (reads) obtained after filtering the raw data were first compared with the NCBI database to clarify the microbial species present in each sample. Subsequently, all microorganisms detected in clinical samples were compared and screened with the microorganisms detected in parallel NTC (no template control) (background microorganisms) according to the parameter of unique reads per million (RPM). Among them, microorganisms not detected in parallel NTC and microorganisms with RPM (clinical sample): RPM (parallel NTC)>10 were defined as the real microbial population in clinical samples. Finally, all microbial species of the real microbial population in clinical samples were searched on PubMed. All microorganisms reported in the literature to cause human or mammalian infection were defined as infectious pathogens.

[0040] Step 4, abnormal CNV feature analysis: Sequencing reads were mapped to the human reference genome (hg19) in the NCBI database, and only uniquely aligned sequencing reads were selected for subsequent analysis. The reference genome was divided into continuous windows of fixed length, and the reading depth of each window was calculated and then normalized to the total number of reads for each sample. The normalized reading depth was divided by the average reading depth of the reference data set to obtain the copy number ratio of each window, and then the copy number was Log2. Adjacent windows with similar ratios were merged into fragments, and the chromosome position and average ratio were annotated. The copy number of each fragment was calculated based on the average ratio and the normal copy number of the corresponding chromosome, and then compared with the preset threshold to verify the CNV;

[0041] Step 5. Statistical analysis: All collected data were statistically analyzed using the R software package. Normally distributed continuous measurement data were expressed as mean (standard deviation) or mean (standard error), and non-normally distributed data were expressed as median (range). The differences and significance between groups were calculated using Student's t-test (normally distributed data) and Wilcoxon rank sum test or Kruskal-Walli test (non-normally distributed data). Data visualization was performed in R (version 4.2.3). Alpha diversity indices included Shannon index, richness index, Simpson index and inverse Simpson index. The t-test was used to evaluate the differences between groups. Beta diversity was analyzed using unconstrained PCoA (main covariate correlation coefficient, P<0.05). Similarity analysis (Anosim) was used to test the differences between groups. Linear discriminant analysis effect size (LEfSe) analysis was performed to identify microorganisms with significant differences in abundance between groups. The threshold was log10LDA (linear discriminant analysis) score ≥ 2 and P value ≤ 0.05. Before analyzing the relative abundance of microorganisms, the RPM values ​​of microorganisms were logarithmically transformed. In this study, a bilateral P value less than 0.05 was considered statistically significant. The random forest method was used to construct a prediction model. Ten-fold cross validation was used to evaluate whether specific microorganisms and clinical indicators could be used as biomarkers to distinguish different patient groups. The predictive performance of the classifier was analyzed by ROC curve.

[0042] The specific operations are as follows:

[0043] 1) Differentiate EBV-HLH and M-HLH from other HLH subtypes by random forest analysis of CNV signatures:

[0044] Differentiation of EBV-HLH and M-HLH from other HLH subtypes by random forest analysis of CNV signatures Human DNA sequences extracted from mNGS for chromosome copy number analysis, Figure 3 A analyzed the positive rate of CNV occurrence in each subtype. There are two subtypes with the most unstable genomes, namely EBV-HLH (43.14%) and M-HLH (36.36%). CNV amplification and deletion were observed in different HLH subtypes. However, in EBV-HLH subtypes and M-HLH, the distribution of CNV on different chromosomes is different ( Figure 3B), except for chromosome 21, the M-HLH subtype had a relatively high incidence of CNVs on all chromosomes, while in the EBV-HLH subtype, CNVs on chromosome 19 were the most common genomic events. The types of CNVs on some chromosomes were also different between the M-HLH subtype and the EBV-HLH subtype. On chromosomes 8 and 13, only CNV deletions occurred in the M-HLH subtype, while CNV deletions and duplications occurred simultaneously in the EBV-HLH subtype. The opposite results were observed on chromosomes 16, 18, and 19. In addition, the frequency of CNV deletions on chromosome 19 exceeded 40% in the EBV-HLH subtype, while it was only 25% in the M-HLH subtype.

[0045] 2) Characteristics of pathogens found by Onco-mNGS in patients diagnosed with different HLH subtypes:

[0046] A total of 18 pathogens were detected in all blood samples, including 11 viruses, 4 bacteria, and 3 fungi ( Figure 4 A), human gammaherpesvirus 4 and human betaherpesvirus 5 were the most common viruses in all samples. In addition, Staphylococcus aureus was also the most common bacterial species. According to Onco-mNGS sequencing, most samples had at least one pathogen identified. Among the EBV subtypes, the pathogen burden of each sample was relatively high. Figure 4 The stacked bar chart in B shows the relative abundance of all pathogens in different subtypes, as well as the main pathogen species in each subtype. As expected, human gammaherpesvirus 4 is the most abundant pathogen in the EBV-HLH subtype and is almost undetectable in the nEBV-HLH and Rh-HLH subtypes, while human betaherpesvirus 5 is prevalent in the three HLH subtypes other than the EBV-HLH subtype. In addition to human betaherpesvirus 5, human alphaherpesvirus 1 and Torqueteno virus are also common in the nEBV-HLH subtype. Human gammaherpesvirus 4 and human betaherpesvirus 6B are also the main pathogens of the M-HLH subtype, while human alphaherpesvirus 3 is also common in the Rh-HLH subtype. The pathogen composition in samples collected from patients with different HLH subtypes is detailed in Figure 5 AD;

[0047] Among the four HLH subtypes, the two subtypes caused by pathogen infection, namely EBV-HLH and nEBV-HLH, had a relatively high mNGS sequencing positivity rate, which is consistent with the clinical consensus ( Figure 6 A), this study identified 144 microorganisms in blood samples of HLH patients by mNGS sequencing, of which bacteria accounted for 79%, fungi accounted for 11%, and viruses accounted for 10% ( Figure 6), but for further analysis, clinicians were also involved in determining the species identified by mNGS sequencing, with the goal of distinguishing pathogens from non-pathogenic microorganisms. In addition, the diagnostic performance of Onco-mNGS in identifying pathogens in patients with unknown etiology was evaluated. Compared with the gold standard method, the sensitivity of mNGS in distinguishing EBV-HLH and nEBV-HLH subtypes was 86.7% and 100%, respectively ( Figure 6 C), Figure 7 AB further illustrated the positive rate and load of two common pathogens (human gammaherpesvirus 4 and human betaherpesvirus 5) in the blood of patients with different HLH subtypes. In the EBV-HLH subtype, the positive rate of human gammaherpesvirus 4 was as high as 88.24%, and the pathogen load was significantly higher than that of other subtypes (P<0.001). The positive rate of human betaherpesvirus 5 was the highest in the nEBV-HLH subtype, but there was no significant difference in the pathogen load of each subtype (P>0.05).

[0048] 3) Correlation analysis between microbiological data and patient clinical data derived from onco-mNGS:

[0049] Microbial ecology is increasingly being implicated in the development and progression of disease; therefore, the diversity and dynamics of the microbiome of each HLH subtype suggest different etiological mechanisms. The nEBV-HLH subtype has the most complex and diverse microbiome, with the most microorganisms identified, of which 32 species are unique ( Figure 8 A), we counted all microorganisms in each sample and summarized the corresponding RPM values ​​detected by mNGS, verifying the complexity of HLH subtypes as a whole ( Figure 8 B), the number of microorganisms in the two infection-induced subtypes, EBV-HLH and nEBV-HLH, was significantly increased compared with the Rh-HLH subtype (P<0.05), but there was no significant difference compared with the M-HLH subtype. In addition, the total RPM value of all microorganisms in each sample of EBV-HLH was significantly increased compared with the other subtypes (P<0.01). The overall situation of the four HLH subtypes is summarized by heat map, see Fig. 9 A, To compare the overall composition of blood microbial signatures across HLH subtypes, we analyzed the diversity within and between each subtype using α-diversity and β-diversity ( Fig. 9B, 9C), the results showed that nEBV-HLH and M-HLH subtypes showed higher diversity by variance analysis (P<0.05), unconstrained PCoA (principal coordinates PCoA1 and PCoA2) and Bray-Curtis distance showed that the microorganisms detected in the EBV-HLH subtype were separated from other HLH subtypes on the first axis (P<0.05), indicating that the EBV-HLH subtype had a unique microbial community composition. In order to further explore the differences in microbial species between different HLH subtypes, the researchers used the linear discriminant analysis effect size (LEfSe) method to determine the microorganisms that were significantly enriched in each subtype. A total of 21 microorganisms were screened out that were differentially enriched in the HLH subtypes, most of which were significantly enriched in the M-HLH subtype ( Figure 8 C) Among these differential species, several species such as oral Streptococcus, Roseburia gingivalis, Pseudomonas, human alphaherpesvirus 3, human betaherpesvirus 5, and Bacillus cereus were enriched in the Rh-HLH subtype, but only human gammaherpesvirus 4 was significantly enriched in the EBV-HLH subtype. Interestingly, the average turnover number of human betaherpesvirus 5 in Rh-HLH was lower than that of the other three subtypes, and was therefore classified as an enriched microorganism. These different microorganisms provide evidence for the unique microbial community composition of different HLH subtypes.

[0050] 4) Predictive evaluation of the treatment prognosis of secondary HLH based on microbiological data derived from onco-mNGS and patient clinical data:

[0051] A variety of treatments have been shown to be effective for specific subtypes of HLH, but the treatment effects of each subtype vary from person to person. Further analysis is needed to investigate the association between CNVs and microbial communities and the treatment course and outcomes of different HLH subtypes (especially the two infection-induced subtypes, EBV-HLH and nEBV-HLH). Based on the treatment outcomes, all cases were divided into a remission group (R) and a non-remission group (NR). Fig.10 The CNV-positive rates of patients with different results were shown. The results showed that most samples (61 / 79) in the remission group were CNV-negative, while in the non-remission group, CNV-positive samples (10 / 20) and CNV-negative samples (10 / 20) each accounted for half ( Fig.10 A), which shows that the CNV-positive rate in the non-remission group samples is generally higher, which is also observed in the EBV-HLH subtype and M-HLH subtype ( Fig.10 B), the distribution of CNVs on different chromosomes was different between the remission group and the non-remission group ( Fig.11A) The remission group had a common amplification on chromosome 1 with a frequency of more than 40%, whereas the most common variant in the non-remission group was a deletion of a DNA fragment on chromosome 19. These findings suggest that CNV distribution can also lead to different treatment outcomes.

[0052] To analyze the association between blood microbiome and treatment outcomes, two parameters were described, namely, the number of all microorganisms detected in each sample and the total RPM value of all microorganisms in each sample. Microorganisms were divided into two groups, pathogenic microorganisms and non-pathogenic microorganisms, according to their pathogenic effects. The number of pathogenic microorganisms in each sample (P = 0.009) and the total RPM value (P = 0.001) were significantly different between the remission group and the non-remission group, indicating that the diversity and abundance of pathogenic microorganisms are related to treatment outcomes ( Fig.10 C), while the number of non-pathogenic microorganisms (P = 0.420) and the total RPM value (P = 0.560) were not significantly different between the remission group and the non-remission group ( Fig.11 B), The microbial community landscapes of the EBV-HLH subtype remission and non-remission groups showed different patterns, and the number of non-pathogenic microorganisms was significantly different (P = 0.009) Fig.12 ), the same analysis strategy was used for nEBV-HLH subtypes to compare the differences between the remission and non-remission groups at the pathogen and non-pathogen microbial levels, but the results did not show a consistent trend with the EBV-HLH subtypes ( Fig.13 ), for the EBV-HLH subtype, in addition to EBV infection being the main pathogen, there are other co-infecting pathogens, but they do not seem to be significantly associated with treatment outcomes ( Fig.11 C), however, the heat map of the entire microbiome showed that bacteria and fungi were enriched in the non-remitted group, suggesting that non-viral species play an important role in the therapeutic response of the nEBV-HLH subtype ( Fig.10 D)

[0053] Although the above studies have used CNV and characteristic microorganisms as key parameters for early prediction and evaluation of clinical outcomes of secondary HLH, when we constructed a random forest binary classification model based on the above biomarkers to distinguish the treatment effect of secondary HLH (Rvs.NR), we found that perhaps only CNV-related key parameters can be used for early prognostic evaluation of secondary HLH treatment. When we set the ratio of training set to test set to 3:1 and used the ten-fold cross-validation method to determine the number of important variables for classification, the optimal point of the cross-validation error determined by the number of biomarkers was 13, which means that according to the average decreasing accuracy, the first 13 variables can be selected as potential markers for distinguishing the treatment effect of secondary HLH (Rvs.NR) ( Fig.10E), Interestingly, the first 13 potential markers mentioned above are all CNV-related parameters, which indicates that the changes of peripheral blood CNV of patients may be closely related to the prognosis of secondary HLH treatment ( Fig.10 E), finally, we further understood the diagnostic performance of the binary classifier based on the above 13 biomarkers through the ROC curve ( Fig.10 F), the results showed that the classifier model performed well in the early evaluation of the prognosis of secondary HLH treatment, with an AUC value of 0.750 ( Fig.10 F).

[0054] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0055] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the essence and scope of the technical solution of the present invention.

Claims

1. A method for early prognosis assessment of hemophagocytic lymphohistiocytosis, characterized in that: The steps include: Step 1. Study design and patient identification: The random enrollment of patients was completed based on the clear diagnosis of secondary HLH. Then, the etiology clinical grouping of each secondary HLH patient was completed after clinical diagnostic data retrieval and clinical confirmation. The mNGS test of each patient's peripheral blood sample was further completed. Finally, based on the obtained CNV parameters and characteristic microbial data, combined with clinical indicators, the construction and effect evaluation of random forest binary or ternary classifiers for early prognosis evaluation of secondary HLH were completed. The relevant data came from peripheral blood samples collected from 92 patients at Beijing Friendship Hospital affiliated to Capital Medical University; Step 2, DNA extraction, library preparation, and NGS: DNA extraction and sequencing library preparation were performed on peripheral blood samples. Before sequencing, the quality of DNA was assessed using a BioAnalyzer 2100 (Agilent Technologies; Santa Clara, CA, United States) combined with quantitative PCR to measure adapters. The library concentration was then adjusted (with a target of 20 million (M) reads) and NGS sequencing was performed on an Illumina Next Seq 550 Dx system using a single-end (SE) 50 base pair (bp) sequencing strategy. Step 3, pathogen detection and identification: A total of 10 to 20 million sequences of sequencing data were obtained for each sample (clinical sample or parallel NTC (no template control)); Step 4: Abnormal CNV feature analysis: Sequencing reads were mapped to the human reference genome (hg19) in the NCBI database, and only uniquely aligned sequencing reads were selected for subsequent analysis; Step 5. Statistical analysis: All collected data were statistically analyzed using the R software package.

2. The method for early prognosis assessment of hemophagocytic lymphohistiocytosis according to claim 1, characterized in that: The patient conditions in step 1 are as follows: According to the HLH-2004 diagnostic criteria, all patients were diagnosed with HLH and divided into four subtypes, including EBV-induced HLH (EBV-HLH, 43 patients, accounting for 46.74%), HLH caused by other pathogens (nEBV-HLH, 25 patients, accounting for 27.17%), HLH caused by malignant tumors (M-HLH, 13 patients, accounting for 14.13%), and HLH caused by rheumatic diseases (Rh-HLH, 11 patients, accounting for 11.96%). All peripheral blood samples were tested by mNGS, and the data were used for subsequent identification of microorganisms and infectious pathogens, as well as chromosome copy number variation analysis based on human genome sequences.

3. The method for early prognosis assessment of hemophagocytic lymphohistiocytosis according to claim 1, characterized in that: The clean sequences (reads) obtained after filtering the raw data in step 3 were first compared with the NCBI database to clarify the microbial species present in each sample. Subsequently, all microorganisms detected in clinical samples were compared and screened with the microorganisms detected in the parallel NTC (no template control) (background microorganisms) according to the parameter of unique reads per million (RPM). Microorganisms not detected in the parallel NTC and microorganisms with RPM (clinical samples): RPM (parallel NTC)>10 were defined as the real microbial population in clinical samples. Finally, all microbial species of the real microbial population in clinical samples were searched on PubMed, and all microorganisms reported in the literature to cause infection in humans or mammals were defined as infectious pathogens.

4. The method for early prognosis assessment of hemophagocytic lymphohistiocytosis according to claim 1, characterized in that: The specific operation of step 4 is as follows: the reference genome is divided into continuous windows of fixed length, the reading depth of each window is calculated, and then normalized to the total number of reads for each sample, the normalized reading depth is divided by the average reading depth of the reference data set to obtain the copy number ratio of each window, and then the copy number is Log2, adjacent windows with similar ratios are merged into fragments, and the chromosome position and average ratio are marked, the copy number of each fragment is calculated based on the average ratio and the normal copy number of the corresponding chromosome, and then compared with the preset threshold to verify CNV.

5. The method for early prognosis assessment of hemophagocytic lymphohistiocytosis according to claim 1, characterized in that: The specific operations of step 5 are as follows: the continuous measurement data of normal distribution are expressed as mean (standard deviation) or mean (standard error), the data of non-normal distribution are expressed as median (range), the differences and significance between groups are calculated by Student's t-test (normal distribution data) and Wilcoxon rank sum test or Kruskal-Walli test (non-normal distribution data), data visualization is performed in R (version 4.2.3), alpha diversity index includes Shannon index, richness index, Simpson index and inverse Simpson index, t-test is used to evaluate the differences between groups, beta diversity is described by unconstrained PCoA (principal coordinate analysis), similarity Anosim was used to test the differences between groups, and linear discriminant analysis effect size (LEfSe) analysis was performed to identify microorganisms with significant differences in abundance between groups. The threshold was log10LDA (linear discriminant analysis) score ≥ 2 and P value ≤ 0.

05. Before analyzing the relative abundance of microorganisms, the RPM values ​​of microorganisms were logarithmically transformed. In this study, a bilateral P value less than 0.05 was considered statistically significant. The random forest method was used to construct a prediction model, and ten-fold cross-validation was used to evaluate whether specific microorganisms and clinical indicators could be used as biomarkers to distinguish different patient groups. The predictive performance of the classifier was analyzed by ROC curve.

Images