Endometriosis brain function connection identification method based on stage driving
By constructing a causal model and utilizing genome-wide association data and resting-state functional magnetic resonance imaging (fMRI) data, we identified brain functional connectivity characteristics at different stages of endometriosis, solving the problem that the rASRM staging system could not predict central nervous system manifestations. This enabled a systematic evaluation of the central nervous system regulatory mechanism and the identification of key regulatory nodes.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- FOSHAN MATERNAL & CHILD HEALTH CARE HOSPITAL
- Filing Date
- 2026-01-23
- Publication Date
- 2026-05-08
AI Technical Summary
The existing rASRM staging system cannot accurately predict central symptoms such as chronic pain, anxiety, and depression in patients with endometriosis, and lacks a systematic assessment of the impact of disease staging status on brain functional connectivity, thus failing to reveal the central network regulatory mechanisms during disease progression.
By acquiring genome-wide association data and resting-state functional magnetic resonance imaging (fMRI) data, we constructed a causal model, identified brain functional connectivity characteristics at different stages, used inverse variance weighting to assess the impact of genetic burden on brain connectivity structure, identified central feedback signals, and explored key regulatory nodes.
This study breaks through the limitations of traditional models, clarifies the central feedback signals at different stages, provides an explanatory basis for the clinical classification and central symptoms of endometriosis, identifies key regulatory nodes, and provides a theoretical foundation for individualized central intervention strategies.
Smart Images
Figure CN121997228A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical imaging and bioinformatics analysis technology, and in particular to a method for identifying brain functional connectivity in endometriosis based on staging. Background Technology
[0002] Endometriosis (EMs) is a chronic disease characterized by the ectopic growth of endometrial-like tissue. Clinically, the rASRM staging system is used to assess the extent of lesions (stages I–IV) to guide intraoperative staging and postoperative management. However, this staging system cannot accurately predict central nervous system manifestations such as chronic pain, anxiety, and depression, suggesting that it does not cover the systemic regulatory dimensions of the disease. Previous studies have indicated abnormal resting-state connectivity patterns in brain regions such as the insula, anterior cingulate cortex, precuneus, and thalamus in EMs patients. However, most current studies are based on cross-sectional observations and cannot yet clearly define whether the disease burden at different stages affects brain function, especially whether there are hierarchical regulation or structural remodeling mechanisms of the central nervous system during disease progression. Currently, there is still a lack of a systematic framework for evaluating whether EMs staging status can feedback-influence brain connectivity characteristics through genetic burden, and no studies have revealed whether central sensitization or inhibition pathways during disease progression are genetically cumulative or stage-dependent. Therefore, it is urgent to construct a causal model based on publicly available genetic and imaging data, identify potential feedback pathways of rASRM staging on brain functional connectivity, and provide a theoretical basis for explaining the central adaptation mechanism of EMs and individualized central intervention strategies. Summary of the Invention
[0003] To address the aforementioned problems, this invention proposes a stage-driven brain functional connectivity recognition method for endometriosis, which mainly solves the problems in the background technology.
[0004] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:
[0005] A stage-driven brain functional connectivity recognition method for endometriosis includes the following steps:
[0006] Genome-wide association data of the target population is obtained as the source of exposure variables, and the EMs staging results corresponding to the target population are obtained. After unifying the genome-wide association data and the EMs staging results to the same reference genome version, SNP site matching and allele alignment are performed.
[0007] Acquire resting-state functional magnetic resonance imaging (fMRI) data, extract brain functional connectivity phenotypes from the resting-state fMRI data, and standardize the brain functional connectivity phenotypes, wherein the brain functional connectivity phenotypes include at least inter-brain region connectivity features and brain region node activity indicators.
[0008] For each type of EMs stage, instrumental variables significantly associated with single nucleotide polymorphism sites are screened from the corresponding genome-wide association data. After preprocessing, the instrumental variables are used to construct an independent set of effective instrumental variables.
[0009] Using the EMs staging results as exposure variables and the brain functional connectivity phenotype as outcome variables, three independent causal models were constructed, each corresponding to an EMs staging stage. The inverse variance weighted method was used as the primary analysis strategy to assess whether the genetic burden of EMs significantly alters the activity level of specific brain connectivity structures or nodes. Significantly altered brain region connectivity features under different EMs staging stages were recorded, and the stability and differences of these brain region connectivity features across different EMs staging stages were compared to identify central feedback signals that are dependent on or continuously influenced by each staging stage.
[0010] By comparing the changes in brain functional connectivity phenotypes affected in different EMs stages, common change patterns and stage-specific pathway combinations in different stages are identified. Based on the common features and specific pathways, a disease-brain connectivity feedback network is constructed. Combining connectivity direction and brain region functional attributes, key regulatory nodes that are likely to participate in chronic pain regulation, emotion control, or central integration function are identified, forming a reserve of candidate intervention targets.
[0011] In some implementations, the standardization process for the brain functional connectivity phenotype includes:
[0012] A predefined brain network template is used to perform unified spatial registration of the inter-brain region connectivity features and the brain region node activity indicators. The predefined brain network template covers the default mode network, attention network, central executive network, sensorimotor system, cerebellar network, and limbic system.
[0013] In some implementations, the preprocessing of the instrumental variables includes:
[0014] The PLINK tool was used to perform linkage disequilibrium pruning on the instrumental variables. The pruning parameters were set to r² < 0.001 and the window width to 10 Mb. The single nucleotide polymorphism site with the highest significance in each linkage disequilibrium block was retained. After removing low-frequency variant sites with minor allele frequencies less than 0.01, the allele directions were uniformly referenced to obtain a set of mutually independent effective instrumental variables.
[0015] The beneficial effects of this invention are as follows: by using the brain functional feedback recognition path with rASRM staging status as the driving variable, it breaks through the limitation of traditional models that cannot characterize the central regulatory mechanism from the perspective of disease status. Furthermore, by systematically comparing the feedback connection patterns under different staging states, it clearly identifies the central feedback signals that depend on each staging stage, providing an explanatory basis for the association between the clinical classification of endometriosis and the manifestation of central symptoms. Attached Figure Description
[0016] Figure 1 A schematic diagram of the results of a two-sample Mendelian randomization analysis with EMs as the exposure variable and brain connectivity features of rsfMRI as the outcome variable;
[0017] Figure 2 A schematic diagram illustrating the results of Mendelian randomization analysis between rASRM stages I–II of endometriosis and rsfMRI brain region connectivity features;
[0018] Figure 3 This is a schematic diagram of the Mendelian randomization analysis results of the causal effects of rASRM stage III–IV endometriosis on the brain region connectivity features of rsfMRI. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the content of this invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to this invention are shown in the accompanying drawings, not all of them.
[0020] This embodiment proposes a stage-driven brain functional connectivity recognition method for endometriosis, including the following steps:
[0021] Step 1: Obtain genome-wide association data (GWAS) of the target population as the source of exposure variables, and obtain the EMs staging results corresponding to the target population. After unifying the GWAS and EMs staging results to the same reference genome version, perform SNP site matching and allele alignment.
[0022] Step 1 of this protocol first gathers the sources of exposure variables, namely, genome-wide association studies (GWAS) data for different stages of endometriosis (EMs). GWAS is an analytical method used to analyze the association between genomic variations and biological phenotypes, and is widely used in human disease mechanisms and microbial phenotypic research. In this protocol, the GWAS data is derived from the Finngen R11 project in the Finnish population. This project data includes the following three types of samples: overall EMs: 18,260 cases, 119,468 controls, total sample size 137,728; rASRM stages I–II (early stage): 7,061 cases, 247,557 controls; rASRM stages III–IV (late stage): 9,150 cases, 245,468 controls. All GWAS results were uniformly annotated using the hg19 / GRCh37 genome version as the genetic basis of the exposure variables, providing SNP loci and effect sizes for the subsequent causal inference process in step 4. Additionally, EMs staging results included overall EMs, rASRM I–II, and rASRM III–IV.
[0023] Step 2: Obtain resting-state functional magnetic resonance imaging (fMRI) data, extract brain functional connectivity phenotypes from the resting-state fMRI data, and standardize the brain functional connectivity phenotypes. The brain functional connectivity phenotypes include at least brain region connectivity features and brain region node activity indicators.
[0024] In this embodiment, the resting-state functional magnetic resonance imaging (rsfMRI) data is a type of medical imaging that studies spontaneous neural activity by detecting blood oxygen level dependent (BOLD) signals. It includes low-frequency oscillation signals of the human brain in a resting state. Brain functional connectivity phenotypes are extracted from this resting-state functional magnetic resonance imaging data. The brain functional connectivity phenotypes include at least inter-brain region connectivity features and brain region node activity indicators. The inter-brain region connectivity features are a quantitative description of the “synchronicity” or “functional correlation” of neural activity between two or more brain regions in a resting state, while the brain region node activity indicators are a quantitative description of the functional activity level of a single independent brain region (such as the prefrontal cortex, insula, cerebellum, posterior cingulate cortex, etc.) in a resting state. Before performing data processing in step 2, in order to eliminate individual differences in brain region location and improve the comparability of brain connectivity features among different subjects, it is necessary to standardize the brain functional connectivity phenotype. This standardization process specifically includes: using a predefined brain network template to uniformly register the inter-brain connectivity features and brain region node activity indicators. The predefined brain network template covers the default mode network, attention network, central executive network, sensorimotor system, cerebellar network, and limbic system.
[0025] In one example, step 2 of this embodiment will use resting-state functional magnetic resonance imaging (fMRI) data from over 34,000 European individuals in the UK Biobank project as the basic data to extract 191 brain functional connectivity phenotypes, including 111 pairs of brain region connectivity features (Edge) and 80 brain region node activity indicators (Node). The 191 features are uniformly registered using a predefined brain network template, covering common functional modules such as the default mode network (DMN), attention network, central executive network, sensorimotor system, cerebellar network, and limbic system.
[0026] In addition, to ensure that the brain functional connectivity phenotypes obtained in step 2 can be matched with the EMs staging results (which have been linked to GWAS data) in the subsequent step 4, the GWAS statistics corresponding to the rsfMRI phenotypes are also unified to the same reference genome version, and the SNP-phenotype related statistics are retained.
[0027] Step 3: For each EMs stage, select instrumental variables from the corresponding genome-wide association studies that are significantly associated with single nucleotide polymorphism (SNP, a type of genetic marker), such as those with a significance level of P < 1 × 10⁻. 5 The SNP sites were identified, and after instrumental variable preprocessing, an independent set of effective instrumental variables was constructed.
[0028] In one example, the preprocessing of the instrumental variables includes: using the PLINK tool to perform linkage disequilibrium pruning on the instrumental variables, setting the pruning parameters to r² < 0.001 and the window width to 10 Mb, retaining the single nucleotide polymorphism (SNP) site with the highest significance in each linkage disequilibrium block, removing low-frequency variant sites with minor allele frequencies less than 0.01, and uniformly referencing the allele direction to obtain a set of mutually independent and effective instrumental variables. In this step, an independent set of instrumental variables is constructed for each causal model (overall / I–II / III–IV) in step 4 to ensure that there are no shared SNPs between causal models and to avoid cross-interference.
[0029] Step 4: Using EMs staging results as exposure variables and brain functional connectivity phenotypes as outcome variables, three independent causal models were constructed, each corresponding to an EMs staging stage. The inverse variance weighted method was used as the primary analysis strategy to assess whether the genetic burden of EMs significantly alters the activity levels of specific brain connectivity structures or nodes. Significantly altered brain region connectivity characteristics under different EMs staging stages were recorded, and the stability and differences of brain region connectivity characteristics across different EMs staging stages were compared. This allowed for the identification of central feedback signals that have stage-dependent or cross-stage persistent effects.
[0030] Building upon the example above, after completing the docking of exposure and outcome variables and the screening of the effective instrumental variable set in steps 1-3, this example employs a two-sample Mendelian randomization (Two-sample MR) method to construct a three-class causal model. This systematically evaluates the genetic impact of overall endometriosis (EMs), rASRM stages I–II, and rASRM stages III–IV on 191 rsfMRI connectivity features. The main analysis method used inverse variance weighting (IVW), with a significance threshold of P < 0.05. MR-Egger and weighted median methods were used for stability verification. The results are as follows (i.e., each significant edge / node expanded from the central feedback signal in step 4):
[0031] (1) The overall EMs model identified five significant brain connectivity features that were genetically associated with EMs risk. All five of these features showed enhanced connectivity (OR>1), suggesting that the central changes associated with overall EMs risk are more likely to be functional mobilization / sensitization phenotypes of multiple networks, rather than inhibitory phenotypes mainly characterized by weakened connectivity. Figure 1 As shown:
[0032] Net100_Pair7_10 (frontal-occipital): OR = 1.023, P = 0.041, indicating activation of advanced vision – execution integration function;
[0033] Net100_Pair8_20 (visual-motor cortex): OR = 1.029, P = 0.013, indicating secondary feedback to the motor control area under chronic load;
[0034] Net100_Node52 (cerebellum–temporal lobe): OR = 1.026, P = 0.008, suggesting that the cerebellum may be involved in chronic pain modulation and emotion integration;
[0035] Net100_Node35 (motor cortex) (OR=1.024, P=0.03) and Net100_Node26 (prefrontal cortex) (OR=1.022, P=0.047): both showed an enhancing trend, which may reflect the mobilization status of the central executive and sensory integration system.
[0036] (2) The rASRM Phase I–II model (early EMs) identified seven significant rsfMRI brain region connectivity features associated with the risk of early EMs, mainly manifested as enhanced connectivity in motor, visual, and emotion regulation pathways, suggesting potential central activation or adaptive regulation mechanisms. The results are as follows: Figure 2 As shown;
[0037] Net100_Pair28_44 (supplementary motor area – anterior inferior frontal lobe): OR = 1.032, P = 0.008, indicating active higher motor planning and cognitive control networks;
[0038] Net100_Pair10_48 (occipital lobe–temporal lobe): OR = 1.025, P = 0.040, reflecting enhanced feedback in the posterior visual-auditory integration area;
[0039] Net100_Pair30_35 (anterior and posterior central gyrus – motor cortex): OR = 1.029, P = 0.015, representing activation of the somatic motor executive network;
[0040] Net100_Node1 (precuneus-occipital visual cortex): OR = 1.028, P = 0.022, indicating activity in the basal visual region;
[0041] Net100_Node4 (occipital lobe attention – visual cortex): OR = 1.029, P = 0.012, indicating an upregulation of the visual attention regulation mechanism;
[0042] Net100_Pair19_39 (precuneus–superior parietal lobe): OR = 0.960, P = 0.000359, was the only significantly weakened pathway, which may represent central suppression of attention-emotion integration function in early EMs patients;
[0043] Net100_Node52 (cerebellum–temporal lobe–thalamus): OR = 1.030, P = 0.003, suggesting that upregulation of the cerebellum–limbic relay network may be involved in pain or emotional feedback in early endometriosis.
[0044] (3) The rASRM stage III–IV model (late EMs) identified four significant connectivity features, exhibiting more pronounced decoupling and functional inhibition characteristics, such as Figure 3 As shown:
[0045] Net100_Pair10_33 (occipital lobe–temporal lobe): OR = 0.975, P = 0.005;
[0046] Net100_Pair29_39 (parietal lobe–precuneus): OR = 0.981, P = 0.035;
[0047] Meanwhile, nodes such as Net100_Pair12_13 (frontal lobe–cerebellum): OR = 1.023, P = 0.015 and Net100_Node39 (parietal lobe–precuneus): OR = 1.022, P = 0.025 showed compensatory enhancement, suggesting that the central nervous system attempted to reconstruct the functional integration network.
[0048] Step 5: By comparing the change patterns of affected brain functional connectivity phenotypes in different EMs stages, common change patterns and stage-specific pathway combinations in different stages are identified. Based on common features and specific pathways, a disease-brain connectivity feedback network is constructed. Combining connectivity direction and brain region functional attributes, key regulatory nodes that are likely to participate in chronic pain regulation, emotion control or central integration function are identified, forming a reserve of candidate intervention targets.
[0049] To summarize the impact patterns of different EMs stages on central connectivity, specifically, in one example, this step screened significant rsfMRI brain functional connectivity phenotypes from three causal models: overall EMs, rASRM stages I–II, and rASRM stages III–IV. Common change patterns and stage-specific pathway combinations were identified across different stages, and a "disease-brain connectivity feedback network" was constructed accordingly. This feedback network uses significant Edge / Node elements as node components, combining the direction of connectivity effects (enhancement / weakening) with corresponding brain region functional attributes to extract key regulatory nodes that may be involved in sensory-motor integration, cognitive control, and emotion / pain regulation, forming a reserve of candidate intervention targets. Based on this, the following four representative central feedback pathways were summarized to systematically characterize the central connectivity classification and regulatory patterns of different EMs stages:
[0050] (1) Overall-early common “cerebellar-related enhancement” pathway: In both the overall EMS model and the rASRM phase I–II model, Net100_Node52 reached significance and the effect direction was consistent as enhancement (overall: OR=1.026, P=0.008; phase I–II: OR=1.030, P=0.003), suggesting that the cerebellar-related network may be upregulated in both the overall risk and early stages, and can be used as a candidate central feedback node for overall-early common pathway.
[0051] (2) Overall risk-dominated “visual-executive / motor integration mobilization” pathway: All significant connectivity features in the overall EMS model were enhanced (5 / 5, OR>1), including Net100_Pair7_10 (frontal-occipital lobe), Net100_Pair8_20 (visual-motor cortex), Net100_Node35 (motor cortex), Net100_Node26 (prefrontal lobe), and Net100_Node52 (cerebellum-temporal lobe). This combination of features demonstrates a consistent direction of broad-based mobilization across multiple networks, suggesting that overall risk-related central changes are more inclined towards multi-system functional mobilization / sensitization phenotypes, and can serve as a core module of the “overall risk connectivity feature set” for subsequent modeling.
[0052] (3) Early-stage specific "motor planning-visual attention enhancement + single integrative pathway attenuation" pathway: The rASRM Phase I–II model identified a total of 7 significant connectivity features, of which 6 were enhancements (OR>1), covering motor planning / executive and visual attention-related pathways (Net100_Pair28_44, Net100_Pair10_48, Net100_Pair30_35, Net100_Node1, Net100_Node4, Net100_Node52). At the same time, only one significantly attenuated pathway, Net100_Pair19_39 (precuneus–upper parietal lobe: OR=0.960, P=0.000359), was present. Therefore, the early stage generally presents a mixed pattern of "predominant enhancement and secondary attenuation". Among them, Pair19_39 can be used as a distinguishable attenuated connectivity feature in the early stage for early stratification or specific pathway annotation.
[0053] (4) Late-stage “biphasic: decoupling weakened + compensation enhanced” pathway: The rASRM stage III–IV model identified four significant connectivity features, presenting a more typical “biphasic” pattern:
[0054] Attenuation: Net100_Pair10_33 (occipital-temporal lobe: OR=0.975, P=0.005), Net100_Pair29_39 (parietal-precuneus: OR=0.981, P=0.035); Enhancement: Net100_Pair12_13 (frontal-cerebellum: OR=1.023, P=0.015), Net100_Node39 (parietal-precuneus: OR=1.022, P=0.025). These results suggest that in the late stage, a pattern of "partially weakened / decoupled integration pathways" and "enhanced nodes or pathways" is more prominent. The enhanced terms can serve as candidate compensatory remodeling nodes, while the weakened terms can serve as late-stage specific decoupled marker pathways.
[0055] Even better, step 6, sensitivity analysis and robustness assessment of results, is also included:
[0056] To verify the robustness of the causal relationship between brain functional connectivity phenotype and EMs staging results, this embodiment further introduces the following sensitivity analysis procedure:
[0057] (1) The Cochran's Q test and I² value were used to assess whether there was heterogeneity in the effects among the instrumental variables. The results are shown in Table 1 below.
[0058] (2) The intercept term of MR-Egger regression was used to assess whether there is directional multiple effects bias in the model. The results are shown in Table 1 below.
[0059] (3) The MR-PRESSO framework was used to identify potential outlier IVs, and the causal effect was re-estimated after outliers were removed. The results are shown in Table 1 below.
[0060] (4) Perform a leave-one-out analysis to eliminate individual instrumental variables one by one and examine whether there is a risk of the outcome being driven by individual SNPs.
[0061] Table 1: Robustness Assessment of Causal Inference Models: Summary of Heterogeneity Test and Level Multiplicity Test Results
[0062]
[0063] The above multi-strategy sensitivity test results show that the results remain robust and reliable under various sensitivity analyses, indicating that the causal model constructed in this invention has good stability and credibility at the methodological level.
[0064] The above embodiments are merely illustrative of the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All equivalent changes or modifications made based on the essence of the content of the present invention should be covered within the scope of protection of the present invention.
Claims
1. A method for identifying brain functional connectivity in endometriosis based on stage-driven approaches, characterized in that, Includes the following steps: Genome-wide association data of the target population is obtained as the source of exposure variables, and the EMs staging results corresponding to the target population are obtained. After unifying the genome-wide association data and the EMs staging results to the same reference genome version, SNP site matching and allele alignment are performed. Acquire resting-state functional magnetic resonance imaging (fMRI) data, extract brain functional connectivity phenotypes from the resting-state fMRI data, and standardize the brain functional connectivity phenotypes, wherein the brain functional connectivity phenotypes include at least inter-brain region connectivity features and brain region node activity indicators. For each type of EMs stage, instrumental variables significantly associated with single nucleotide polymorphism sites are screened from the corresponding genome-wide association data. After preprocessing, the instrumental variables are used to construct an independent set of effective instrumental variables. Using the EMs staging results as exposure variables and the brain functional connectivity phenotype as outcome variables, three independent causal models were constructed, each corresponding to an EMs staging stage. The inverse variance weighted method was used as the primary analysis strategy to assess whether the genetic burden of EMs significantly alters the activity level of specific brain connectivity structures or nodes. Significantly altered brain region connectivity features under different EMs staging stages were recorded, and the stability and differences of these brain region connectivity features across different EMs staging stages were compared to identify central feedback signals that are dependent on or continuously influenced by each staging stage. By comparing the changes in brain functional connectivity phenotypes affected in different EMs stages, common change patterns and stage-specific pathway combinations in different stages are identified. Based on the common features and specific pathways, a disease-brain connectivity feedback network is constructed. Combining connectivity direction and brain region functional attributes, key regulatory nodes that are likely to participate in chronic pain regulation, emotion control, or central integration function are identified, forming a reserve of candidate intervention targets.
2. The method for identifying brain functional connectivity in endometriosis based on stage-driven approaches as described in claim 1, characterized in that, The standardization process for the brain functional connectivity phenotype includes: A predefined brain network template is used to perform unified spatial registration of the inter-brain region connectivity features and the brain region node activity indicators. The predefined brain network template covers the default mode network, attention network, central executive network, sensorimotor system, cerebellar network, and limbic system.
3. The method for identifying brain functional connectivity in endometriosis based on stage-driven approaches as described in claim 1, characterized in that... The preprocessing of the instrumental variables includes: The PLINK tool was used to perform linkage disequilibrium pruning on the instrumental variables. The pruning parameters were set to r² < 0.001 and the window width to 10 Mb. The single nucleotide polymorphism site with the highest significance in each linkage disequilibrium block was retained. After removing low-frequency variant sites with minor allele frequencies less than 0.01, the allele directions were uniformly referenced to obtain a set of mutually independent effective instrumental variables.