MVIRA: A model based on Missing Value Imputation and Reliability Assessment for mortality risk prediction.

BACKGROUND: Mortality risk prediction is to predict whether a patient has the risk of death based on relevant diagnosis and treatment data. How to accurately predict patient mortality risk based on electronic health records (EHR) is currently a hot research topic in the healthcare field. In actual medical datasets, there are often many missing values, which can seriously interfere with the effect of model prediction. However, when missing values are interpolated, most existing methods do not take into account the fidelity or confidence of the interpolated values. Misestimation of missing variables can lead to modeling difficulties and performance degradation, while the reliability of the model may be compromised in clinical environments. MATERIALS AND METHODS: We propose a model based on Missing Value Imputation and Reliability Assessment for mortality risk prediction (MVIRA). The model uses a combination of variational autoencoder and recurrent neural networks to complete the interpolation of missing values and enhance the characterization ability of EHR data, thus improving the performance of mortality risk prediction. In addition, we also introduce the Monte Carlo Dropout method to calculate the uncertainty of the model prediction results and thus achieve the reliability assessment of the model. RESULTS: We perform performance validation of the model on the public datasets MIMIC-III and MIMIC-IV. The proposed model showed improved performance compared with competitive models in terms of overall specialties. CONCLUSION: The proposed model can effectively improve the accuracy of mortality risk prediction, and can help medical institutions assess the condition of patients.

A drug molecular classification model based on graph structure generation.

Molecular property prediction based on artificial intelligence technology has significant prospects in speeding up drug discovery and reducing drug discovery costs. Among them, molecular property prediction based on graph neural networks (GNNs) has received extensive attention in recent years. However, the existing graph neural networks still face the following challenges in node representation learning. First, the number of nodes increases exponentially with the expansion of the perception field, which limits the exploration ability of the model in the depth direction. Secondly, the large number of nodes in the perception field brings noise, which is not conducive to the model's representation learning of the key structures. Therefore, a graph neural network model based on structure generation is proposed in this paper. The model adopts the depth-first strategy to generate the key structures of the graph, to solve the problem of insufficient exploration ability of the graph neural network in the depth direction. A tendentious node selection method is designed to gradually select nodes and edges to generate the key structures of the graph, to solve the noise problem caused by the excessive number of nodes. In addition, the model skillfully realizes forward propagation and iterative optimization of structure generation by using an attention mechanism and random bias. Experimental results on public data sets show that the proposed model achieves better classification results than the existing best models.

Heterotopic uterus transplantation in a swine model.

BACKGROUND: The aim of our study was to examine the feasibility of allogeneic uterine transplantation in a large animal model. METHODS: We performed heterotopic uterine transplants in genetically defined mini-pigs. Immunosuppression was tacrolimus administered intravenously for the first 12 days posttransplantation followed by oral cyclosporine maintenance immunosuppression. The graft was transplanted heterotopically in the lower abdominal cavity of the recipient. The vaginal vault was exteriorized as a stoma in the lower right abdominal wall. The uterine grafts were followed with endoscopies and biopsies. RESULTS: Ten transplants were performed. Follow-up was until July 2008. At the end of the follow-up period, 5 animals were alive and healthy, 0.5 to 12 months posttransplantation. There were 5 deaths due to pneumonia (n=1), intussusception of the graft (n=1), cardiorespiratory arrest during anesthesia (n=1), and complications of the stoma (n=2). Acute rejections of the graft presented during the 2nd and 3rd month posttransplantation were treated successfully with increase of the maintenance immunosuppression and steroids. Other complications included prolapse and infections of the graft stoma. Pathological changes seen in the endometrial biopsies included acute rejection and acute endometritis. CONCLUSION: These findings demonstrate that successful uterus transplantation in a large animal model (miniature swine) is feasible using this heterotopic model, and it can be useful for the study of these transplants.

A novel approach to detect donor/recipent immune responses between HLA-identical pairs.

Renal transplant rejection and graft versus host reactions between HLA genetically-identical sibling (HLAgi) donor/recipient (D/R) pairs are thought to result from minor histocompatibility antigen (mHAg) disparities. We have compared two methods of measuring HLAgi D/R T lymphocyte responses to "matured" dendritic cells: 1.) a modified Cylex assay of CD4(+) ATP levels (MLDC-ATP) versus 2.) (3)H-thymidine uptake (MLDC-(3)H). The MLDC-ATP kinetics peaked at 48 hours versus the MLDC-(3)H at 7 days, and appeared more sensitive. We tested HLAgi (normal) volunteer siblings (NLs), and D/R sibling pairs before and after renal transplantation (pre-Tx and post-Tx). The overall frequencies of positive responses in the MLDC-ATP for HLAgi NLs, pre-Tx, and post-Tx D/R pairs were 63%, 50%, and 42%, respectively. The percentage with reciprocal responses was 37.5%, 20%, and 22.22%, respectively. In one set of three HLAgi (NLs) siblings (two males and one female), there was a nongender-associated differential response. There was no MLDC correlation with class I MHC-associated mHAg (SSP) incompatibility, nor could some MLDC positive reactive pairs theoretically process the necessary HLA-class I restriction molecules for presentation of known (nanomeric) mHAg peptides. Speculatively, the MLDC reflects class II MHC-restricted mHAg reactions (not yet definable), with possible effects of other polymorphic (nonhistocompatibility) immune response genes, and thereby may be a useful measurement of CD4(+) T-cell HLAgi transplantation immunity.

Induction of auto-reactive regulatory T cells by stimulation with immature autologous dendritic cells.

We have shown in ex vivo studies in donor bone marrow-infused kidney transplant recipients, that chimeric cells of either donor or recipient origin taken from the recipient's bone marrow down-regulated the recipient's cellular immune responses. In the present study, we have now induced regulatory T cells from peripheral blood mononuclear cells (PBMC) of renal transplant recipients or laboratory volunteers by multi-stimulation with autologous immature dendritic cell (iDC) enriched populations derived from either bone marrow cells (BMC) of the (immunosuppressed) kidney transplant recipients or PBMC of the laboratory volunteers (i.e., ibDC and ipDC, respectively). These regulatory T cells, induced by ibDC and ipDC, were autoreactive and designated as TAb and TAp with similar phenotypes and functional profiles. They were largely CD4 + CD25high, CD45RA low and CD45RO high, and uniformly expressed intracellular CTLA-4, and message of IL-4, IL-10, Foxp3, and differentially expressed TGFbeta. Their proliferative responses to autologous mature dendritic stimulating cells (mDC) were approximately two-fold stronger than to allogeneic mDC, and to allogeneic mDC were significantly lower than those of (control) autologous TPBL, suggesting an anergic state. TAb and TAp were not cytotoxic to autologous cells expressing Epstein-Barr virus (EBV) antigens, but were able to inhibit (regulate) the effector phase of this TPBL response to both autologous and allogeneic EBV lymphoblasts. This regulation appeared to require cell-to-cell contact.

Campath-1H does not alter bone marrow cell regulatory function.

We have previously reported in laboratory volunteers (in vitro) and renal transplant recipients (ex vivo) that bone marrow cells (BMC) are potent downregulators of the immune response. Also, the use of alemtuzumab (Campath-1H, C1H) for immunodepletion is associated with the most potent lasting effects yet seen on T-cell immunity in renal transplantation. We questioned whether the administration of C1H to kidney allograft recipients of donor bone marrow cell (DBMC) infusions would lead to stronger or weaker immunoregulatory effects. Human BMC depleted of T cells (nT-BMC) were either untreated or treated with C1H and rabbit complement and compared for their ability to downregulate autologous or allogeneic T-cell responses and to generate T regulatory (T reg) cells. The proliferative responses to anti-CD3 monoclonal antibody of T cells derived from cocultures with C1H-treated or untreated autologous nT-BMC were equally suppressed, i.e., an equivalent alteration in CD3 complex signaling, not regained by the addition of interleukin 2. Adenosine triphosphate levels were also markedly reduced in T cells both from C1H-treated and untreated nT-BMC cocultures. The ability of C1H-treated or untreated nT-BMC to suppress autologous T-cell cytotoxic function was also equivalent, with a marked, but equivalent, capacity to induce CD4/CD25(high) T regs from CD3(+) cells, which effectively downregulated cytotoxic T cells. To mimic the clinical infusion of DBMC into (allogeneic) recipients, peripheral blood mononuclear cells were also cultured with allogeneic C1H-treated and untreated nT-BMC. T cells derived from these cultures secondarily stimulated with the same-donor mature antigen-presenting cells exhibited suppressed cytotoxicity by 85% and 54%, respectively. These in vitro studies suggest that C1H does not abrogate BMC immunoregulation and thus may allow its lympho-depleting effect to be synergistic.

Antigen presentation and immune regulatory capacity of immature and mature-enriched antigen presenting (dendritic) cells derived from human bone marrow.

Several reports including those from this laboratory have demonstrated that bone marrow cells (BMC) downregulate in vitro both mixed leukocyte reaction and cytotoxic T lymphocyte reactions. We consequently hypothesized that a general property of immature cells of hematopoietic organs is their ability to suppress immune reactivity. As one of these suppressive activities, the lack of costimulatory molecules was proposed as a mechanism by which immature antigen presenting cells of the bone marrow might be involved. In the present report, we used two culture environments, each of which would regulate a different maturation pattern of human bone marrow-derived enriched dendritic antigen presenting cells (DC or APC) to determine the respective effects on in vitro immune regulatory function. Human BMC depleted of CD3+ cells were cultured with either: interleukin-4 (IL-4) and granulocyte macrophage-colony stimulating factor (GM-CSF), to maintain DC-enriched populations in an immature state (iAPC); or an interferon-gamma (IFNgamma), tumor necrosis factor alpha (TNF-alpha), GM-CSF, LPS, and IL-6 cocktail to promote the maturation of DC-enriched APC (mAPC). These iAPC and mAPC were, respectively, phenotypically characterized and also tested in vitro for the following: (1) both direct and indirect-antigen presentation functions; (2) immune regulatory functions on the response of autologous and allogeneic peripheral blood lymphocytes (PBL); and (3) Western blot analysis determining the levels of both major histocompatibility complex (MHC) class I related cytoplasmic transporter molecules associated with antigen processing (TAP1) and as well as proteasome activator molecules (PA28alpha). The iAPC population expressed fewer dendritic cell markers (CD83 and DCsign), and costimulator molecules (CD86 and CD40) than the mAPC, such that there was an approximate threefold increase in expression of CD83, 2.5-fold increase in DCsign, and a threefold increase in CD40 and CD86 on mAPC than on iAPC (p=0.005 for CD83; p=0.001 for DCsign; p=0.001 for CD86; and p=0.001 for CD40). In lymphoproliferative assays, indirect and direct alloantigen presentation by iAPC was weaker than by mAPC (p=0.05 and 0.04). In addition, iAPC were able to downregulate allogeneic CTL responses. Also, after pulsing with Epstein-Barr virus (EBV) protein antigens, the iAPC were less efficient in their presentation to autologous EBV-specific T-cell lines, and caused an inhibition of EBV-CTL generation. The expression of TAP1 and PA28alpha was reduced in iAPC in comparison to mAPC. These findings support the notion that a maturation state of BMC-derived APC correlates with their capacity to present antigen. The observed in vitro deficiency of this function by immature bone marrow cells may therefore contribute to the immune downregulatory capacity seen in the BMC compartment.

Donor bone marrow infusion in deceased and living donor renal transplantation.

The infusion and persistence in a transplant recipient of donor-derived bone marrow cells (DBMC) of multi-lineage can lead to a state of permanent chimerism. In solid vascular organ transplantation, the donor bone marrow lineage cells can even be derived from the transplant organ, and these cells can be detected in very small numbers in the recipient. This has been called microchimerism. Much controversy has developed with respect to the function of chimeric cells in organ transplantation. One idea is that the occurrence of these donor cells found in microchimerism in the recipient are coincidental and have no long-term beneficial effect on engraftment. A second and opposing view, is that these donor cells have immunoregulatory function that affect both the acute and chronic phases of the recipient anti-donor responses. It follows that detecting quantitative changes in chimerism might serve as an indication of the donor-specific alloimmune or regulatory response that could occur in concert with or independent of other adaptive immune responses. The latter, including autoimmune native disease, need to be controlled in the transplant organ. The safety and immune tolerance potential of DBMC infusion with deceased and living donor renal transplants was evaluated in a non-randomized trial at this center and compared with non-infused controls given identical immunosuppression. Overall DBMC infusions were well tolerated by the recipients. There were no complications from the infusion(s), no episodes of graft-vs-host disease (GVHD) and no increase infections or other complications. In the deceased DBMC-kidney trial, actuarial graft survival at 5 years was superior especially when graft survival was censored for recipient death. Acute rejections were significant reduced in patients given two DBMC infusions, and chronic rejection was dramatically reduced in all DBMC treated patients. The most interesting finding was that the degree of microchimerism slowly increased over the years the DBMC group that had exhibited no rejection episodes. In the DBMC-living related trial, the incidence of acute rejection did not differ between groups. However, DBMC chimerism in recipient iliac crest marrow had increased more rapidly than might be predicted from results previously seen in the cadaver group, despite four times fewer DBMC infused, with the generation of T- regulartory cells in-vitro assays.

Immune responses and their regulation by donor bone marrow cells in clinical organ transplantation.

Infusions of donor bone marrow derived cells (DBMC) continue to be tested in clinical protocols intended to induce specific immunologic tolerance of solid organ transplants based on the observations that donor-specific tolerance is induced this way in animal models. We studied the immunological effects of human DBMC infusions in renal transplantation using modifications in lymphoproliferation (MLR) and cytotoxicity (CML) assays. The salient observations and tentative conclusions are summarized in this review. Among many types of organs transplanted using DBMC at this center, it was found that the cadaver renal recipients (CAD) had significantly decreased chronic rejection and higher graft survival when compared to equivalent non-infused controls. DBMC infusion was also associated with a marginal and non-specific immune depression. It was also observed that the number of chimeric donor cells gradually increased in the iliac crest bone marrow compartment with a concomitant decrease in the peripheral blood and that the increase was more rapid in living-related donor (LRD)-kidney/DBMC recipients in spite of a lower number of DBMC infused (<25%) than in the CAD-kidney/DBMC group. In the LRD recipients with residual anti-donor responses, purified chimeric cells of either donor or recipient inhibited recipient immune responses to the donor significantly more strongly than the freshly obtained bone marrow from the specific donor or volunteer suggesting an active regulatory role for chimeric cells. A number of (non-chimeric) subpopulations of bone marrow cells including CD34(+) stem cells and the CD34(-) early progeny like CD38(+), CD2(+), CD5(+) and CD1(+) lymphoid cells as well as CD33(+) (but CD15(-)) myeloid cells down-regulated the MLR and CML responses of allogeneic PBMC stimulated with (autologous) donor spleen cells. These regulatory effects appeared to be refractory to the action of commonly used immunosuppressive drugs and occurred during the early phase of the immune response through cell-cell interactions. Most of these DBMC sub-populations had stimulatory capabilities, albeit markedly lower than donor spleen cells, but only through the indirect antigen presentation pathway. When co-cultured with allogeneic stimulators, purified CD34(+) cells were found to give rise both to CD3(-) TCRalphabeta(+), as well as CD3(+) TCRalphabeta(+) cells and, thereby, responded in MLR to allogeneic stimulation (but did not generate cytotoxic effector cells). Also, a number of DBMC subpopulations inhibited the CML and to a lesser extent the MLR, of autologous post-thymic responding T cells stimulated with allogeneic irradiated cells, mediated through soluble factors. Finally, non-chimeric DBMC also inhibited the proliferative and cytotoxic responses of autologous T cells to EBV antigens, inducing T suppressor cells, which in turn could inhibit autologous anti-EBV CTL generation and B cell anti-CMV antibody production. These studies all suggested a strong inhibitory property of a number of DBMC sub-populations in vitro and in vivo with the notion that they promote unresponsiveness.

Functional and phenotypic properties of peripheral T cells anergized by autologous CD3(+) depleted bone marrow cells.

We have previously demonstrated that bone marrow cells (BMC) inhibit the generation of autologous Epstein-Barr virus (EBV) -specific cytotoxic T lymphocytes (CTL). It was also observed that CD3(+) cells obtained after 7 days of culture in the presence of autologous BMC could be used as inhibitors of EBV-CTL generation. In the present study, we examined these BMC induced regulator CD3(+) T cells with respect to phenotype, function, and T-cell activation pathways. We also questioned if the CD3(+) regulatory cell function is mediated by their direct effect on peripheral T cells or on the ability of antigen presenting cells (APC) to stimulate peripheral T cells. To answer this, CD3(+) cells from peripheral blood lymphocytes (PBL) were cultured with either CD3-depleted BMC or with CD3-depleted PBL. The CD3(+) cells were then isolated with immunomagnetic beads, designated as T(BM) and T(PBL), and were compared in functional studies. There was an increase in the expression of CD25 on T(BM) cells. The T(BM) cells also expressed less CD122 and a decreased number of CD3 molecules per cell. Both T(BM) and T(PBL) cell populations responded to mitogen (PHA) to the same magnitude. However, when stimulated through the CD3 complex with anti-CD3 monoclonal antibody (mAb), the T(BM) cells had a significantly decreased response than did T(PBL). The addition of IL-2 to these latter cultures augmented, but could not fully restore, the response. Additionally, stimulation of T(BM) cells with allogeneic cells failed to produce cytotoxic T cells. These "anergized" T(BM) and "nonanergized" (control) T(PBL) cells were added as third-party cells to a CTL generating culture of autologous PBL stimulated with allogeneic cells. The T(BM) cells exhibited suppressor function and inhibited the generation of CTL, in contrast with T(PBL). The effect of T(BM) cells on direct and indirect antigen presentation pathways demonstrated that T(BM) primarily effected indirect, but not direct, alloantigen presentation. To further explore the cytoplasmic T-cell activation events that occurred after the coculture of the PBL T cells with BMC, the levels of zeta-associated protein 70 (ZAP70) and extracellular receptor-activated kinase (ERK) were determined. There was a decrease in ZAP70 levels in the T(BM), which correlated with its reduced expression of cell surface CD3 and the attenuated response to anti-CD3 mAb activation. However, the activity of ERK was equally expressed by T(BM) and T(PBL). It, therefore, appears that the culturing of peripheral T cells with (non-T) BMC anergizes these cells (which become refractory to stimulation through the T-cell receptors), and induces immune suppressor function. These in vitro observations may provide a mechanism by which infused donor BMC serve to downregulate T-cell immunity.