Machine-learning and mechanistic modeling of metastatic breast cancer after neoadjuvant treatment.

Clinical trials involving systemic neoadjuvant treatments in breast cancer aim to shrink tumors before surgery while simultaneously allowing for controlled evaluation of biomarkers, toxicity, and suppression of distant (occult) metastatic disease. Yet neoadjuvant clinical trials are rarely preceded by preclinical testing involving neoadjuvant treatment, surgery, and post-surgery monitoring of the disease. Here we used a mouse model of spontaneous metastasis occurring after surgical removal of orthotopically implanted primary tumors to develop a predictive mathematical model of neoadjuvant treatment response to sunitinib, a receptor tyrosine kinase inhibitor (RTKI). Treatment outcomes were used to validate a novel mathematical kinetics-pharmacodynamics model predictive of perioperative disease progression. Longitudinal measurements of presurgical primary tumor size and postsurgical metastatic burden were compiled using 128 mice receiving variable neoadjuvant treatment doses and schedules (released publicly at https://zenodo.org/records/10607753). A non-linear mixed-effects modeling approach quantified inter-animal variabilities in metastatic dynamics and survival, and machine-learning algorithms were applied to investigate the significance of several biomarkers at resection as predictors of individual kinetics. Biomarkers included circulating tumor- and immune-based cells (circulating tumor cells and myeloid-derived suppressor cells) as well as immunohistochemical tumor proteins (CD31 and Ki67). Our computational simulations show that neoadjuvant RTKI treatment inhibits primary tumor growth but has little efficacy in preventing (micro)-metastatic disease progression after surgery and treatment cessation. Machine learning algorithms that included support vector machines, random forests, and artificial neural networks, confirmed a lack of definitive biomarkers, which shows the value of preclinical modeling studies to identify potential failures that should be avoided clinically.

Accelerating Digitalization in Healthcare with the InSilicoTrials Cloud-Based Platform: Four Use Cases.

The use of in silico trials is expected to play an increasingly important role in the development and regulatory evaluation of new medical products. Among the advantages that in silico approaches offer, is that they permit testing of drug candidates and new medical devices using virtual patients or computational emulations of preclinical experiments, allowing to refine, reduce or even replace time-consuming and costly benchtop/in vitro/ex vivo experiments as well as the involvement of animals and humans in in vivo studies. To facilitate and widen the adoption of in silico trials, InSilicoTrials Technologies has developed a cloud-based platform, hosting healthcare simulation tools for different bench, preclinical and clinical evaluations, and for diverse disease areas. This paper discusses four use cases of in silico trials performed using the InSilicoTrials.com platform. The first application illustrates how in silico approaches can improve the early preclinical assessment of drug-induced cardiotoxicity risks. The second use case is a virtual reproduction of a bench test for the safety assessment of transcatheter heart valve substitutes. The third and fourth use cases are examples of virtual patients generation to evaluate treatment effects in multiple sclerosis and prostate cancer patients, respectively.

Machine Learning and Mechanistic Modeling for Prediction of Metastatic Relapse in Early-Stage Breast Cancer.

PURPOSE: For patients with early-stage breast cancer, predicting the risk of metastatic relapse is of crucial importance. Existing predictive models rely on agnostic survival analysis statistical tools (eg, Cox regression). Here we define and evaluate the predictive ability of a mechanistic model for time to distant metastatic relapse. METHODS: The data we used for our model consisted of 642 patients with 21 clinicopathologic variables. A mechanistic model was developed on the basis of two intrinsic mechanisms of metastatic progression: growth (parameter α) and dissemination (parameter µ). Population statistical distributions of the parameters were inferred using mixed-effects modeling. A random survival forest analysis was used to select a minimal set of five covariates with the best predictive power. These were further considered to individually predict the model parameters by using a backward selection approach. Predictive performances were compared with classic Cox regression and machine learning algorithms. RESULTS: The mechanistic model was able to accurately fit the data. Covariate analysis revealed statistically significant association of Ki67 expression with α (P = .001) and EGFR expression with µ (P = .009). The model achieved a c-index of 0.65 (95% CI, 0.60 to 0.71) in cross-validation and had predictive performance similar to that of random survival forest (95% CI, 0.66 to 0.69) and Cox regression (95% CI, 0.62 to 0.67) as well as machine learning classification algorithms. CONCLUSION: By providing informative estimates of the invisible metastatic burden at the time of diagnosis and forward simulations of metastatic growth, the proposed model could be used as a personalized prediction tool for routine management of patients with breast cancer.

Selective Inhibitors of T Cell Receptor Recognition of Antigen-MHC Complexes for Rheumatoid Arthritis.

Autoreactive T cells specific to human collagen type II have a crucial role in the development of rheumatoid arthritis (RA) in the context of MHC class II allele HLA-DRB1-*04. The protein-protein interactions between the T cell receptor (TCR) and the type II collagen bound to the allele MHC of class II may thus represent the target for the development of new drugs against RA. In this study, a structure-based pharmacophore model for potential small molecule inhibitors was developed from protein-protein interface structure. The 3D model obtained was used for a virtual screening workflow, which resulted in three hits for experimental follow up. Three compounds have been identified that interfere with the TCR/collagenII-MHCII (K i values below 10 µM) and open up new possibilities in the treatment of RA.

Collagen Specific T-Cell Repertoire and HLA-DR Alleles: Biomarkers of Active Refractory Rheumatoid Arthritis.

Rheumatoid arthritis (RA) is characterized by chronic joint inflammation and associates with HLA-DRB1*04. The Collagen IIp261-273-specific T cell repertoire in the peripheral blood of DR4 + patients at the onset of the disease shows a restricted TCR-beta chain usage among which the most frequent is TRBV25. To define whether this group of DR4-restricted collagen-specific shared T cell could represent markers of active-severe disease and response to therapy, 90 subjects affected by early-RA were enrolled in the study; peripheral blood mononuclear cells were cultured with or without the human collagen II peptide p261-273 and were examined by immunoscope analysis for the usage of the previously identified shared TCR-beta chains. We report that the presence of T cells carrying rearrangement TRBV25 associated with HLA-DR haplotype and disease activity. HLA-DRB1* haplotypes 04-04, 04-01 and 04-11 were significantly associated with usage of TRBV25, higher disease activity at the onset of disease and poor response to DMARDs. Finally, the HLA-DRB1* haplotype appeared complementary with current serologic tools to predict good and poor responders in a treat to target strategy. The data reported here offer clues to predict the course of the disease and to foresee personalized treatments in RA patients.

M tuberculosis in the adjuvant modulates time of appearance of CNS-specific effector T cells in the spleen through a polymorphic site of TLR2.

DC deliver information regulating trafficking of effector T cells along T-cell priming. However, the role of pathogen-derived motives in the regulation of movement of T cells has not been studied. We hereinafter report that amount of M tuberculosis in the adjuvant modulates relocation of PLP139-151 specific T cells. In the presence of a low dose of M tuberculosis in the adjuvant, T cells (detected by CDR3 BV-BJ spectratyping, the so-called "immunoscope") mostly reach the spleen by day 28 after immunization ("late relocation") in the SJL strain, whereas T cells reach the spleen by d 14 with a high dose of M tuberculosis ("early relocation"). The C57Bl/6 background confers a dominant "early relocation" phenotype to F1 (SJL×C57Bl/6) mice, allowing early relocation of T cells in the presence of low dose M tuberculosis. A single non-synonymous polymorphism of TLR2 is responsible for "early/late" relocation phenotype. Egress of T lymphocytes is regulated by TLR2 expressed on T cells. Thus, pathogens engaging TLR2 on T cells regulate directly T-cell trafficking, and polymorphisms of TLR2 condition T-cell trafficking upon a limiting concentration of ligand.

Intracellular bacteria can cause EAE in SJL mice or modify self-specific T cell repertoire.

Environment and genetic are both relevant in determining development of Multiple Sclerosis. Many epidemiological observations converge on indicating EBV infection and Vitamin D levels as major players among the environmental factors. Bacteria and bacterial products are however potent triggers of immune responses, and recent work from several laboratories indicates that the microbiota plays a prominent role in "priming" or protecting individuals for development of experimental autoimmune diseases. Here we report our recent work dealing with the role of non-pathogenic mycobacteria and their innate receptors in relapsing-remitting experimental autoimmune encephalomyelitis in the SJL mouse and in mobilization of CNS-reactive T cells. We finally discuss how bacteria are likely involved in the pathogenesis of Multiple Sclerosis, expecially with regard to their role in driving the recurring acute episodes of disease.

Surface expression of MPT64 as a fusion with the PE domain of PE_PGRS33 enhances Mycobacterium bovis BCG protective activity against Mycobacterium tuberculosis in mice.

To improve the current vaccine against tuberculosis, a recombinant strain of Mycobacterium bovis bacillus Calmette-Guérin (rBCG) expressing a Mycobacterium tuberculosis vaccine candidate antigen (MPT64) in strong association with the mycobacterial cell wall was developed. To deliver the candidate antigen on the surface, we fused the mpt64 gene to the sequence encoding the PE domain of the PE_PGRS33 protein of M. tuberculosis (to create strain (H)PE-ΔMPT64-BCG), which we have previously shown to transport proteins to the bacterial surface. In a series of protection experiments in the mouse model of tuberculosis, we showed that (i) immunization of mice with (H)PE-ΔMPT64-BCG provides levels of protection significantly higher than those afforded by the parental BCG strain, as assessed by bacterial colonization in lungs and spleens and by lung involvement (at both 28 and 70 days postchallenge), (ii) rBCG strains expressing MPT64 provide better protection than the parental BCG strain only when this antigen is surface expressed, and (iii) the (H)PE-ΔMPT64-BCG-induced MPT64-specific T cell repertoire when characterized by ß chain variable region-ß chain joining region (BV-BJ) spectratyping indicates that protection is correlated with the ability to recruit gamma interferon (IFN-γ)-secreting T cells carrying the BV8.3-BJ1.5 (172 bp) shared rearrangement. These results demonstrate that (H)PE-ΔMPT64-BCG is one of the most effective new vaccines tested so far in the mouse model of tuberculosis and underscore the impact of antigen cellular localization on the induction of the specific immune response induced by rBCG.

Erbb2 DNA vaccine combined with regulatory T cell deletion enhances antibody response and reveals latent low-avidity T cells: potential and limits of its therapeutic efficacy.

Rat (r)Erbb2 transgenic BALB-neuT mice genetically predestined to develop multiple invasive carcinomas allow an assessment of the potential of a vaccine against the stages of cancer progression. Because of rErbb2 expression in the thymus and its overexpression in the mammary gland, CD8(+) T cell clones reacting at high avidity with dominant rErbb2 epitopes are deleted in these mice. In BALB-neuT mice with diffuse and invasive in situ lesions and almost palpable carcinomas, a temporary regulatory T cells depletion combined with anti-rErbb2 vaccine markedly enhanced the anti-rErbb2 Ab response and allowed the expansion of latent pools of low-avidity CD8(+) T cells bearing TCRs repertoire reacting with the rErbb2 dominant peptide. This combination of a higher Ab response and activation of a low-avidity cytotoxic response persistently blocked tumor progression at stages in which the vaccine alone was ineffective. However, when diffuse and invasive microscopic cancers become almost palpable, this combination was no longer able to secure a significant extension of mice survival.

Mycobacterium smegmatis expressing a chimeric protein MPT64-proteolipid protein (PLP) 139-151 reorganizes the PLP-specific T cell repertoire favoring a CD8-mediated response and induces a relapsing experimental autoimmune encephalomyelitis.

We infected SJL mice with a recombinant Mycobacterium smegmatis expressing a chimeric protein containing the self-epitope of proteolipid protein 139-151 (p139) fused to MPT64, a secreted protein of Mycobacterium tuberculosis (rMS(p139)). Infected mice developed a relapsing experimental autoimmune encephalomyelitis (EAE), showing a prevailing demyelination of the CNS, and disease severity was significantly lower in comparison with the one that follows immunization with p139. rMS(p139) was not detected in lymph node or spleen in the course of clinical disease development or in the CNS during relapse. Infection with rMS(p139) modified the p139-specific T cell repertoire, recruiting the spontaneous p139-specific repertoire and activating CD4(+) T cells carrying the BV4 semiprivate rearrangement. T cells carrying the public BV10 rearrangement that are consistently found in the CNS during flares of disease were not activated by infection with rMS(p139) because lymph node APCs infected with rMS(p139) selectively fail to present the epitope for which BV10 cells are specific. Simultaneously, rMS(p139) expanded p139-specific CD8(+) cells more efficiently than immunization with peptide in adjuvant. SJL mice vaccinated against the CDR3 sequence of the BV10 public rearrangement reduced usage of the BV10 cells and displayed reduced symptoms during bouts of EAE. Thus, transient peripheral infection with a CNS-cross-reactive nonpathogenic Mycobacterium induces a relapsing EAE that continues long after clearance of the infectious agent. The composition of the self-reactive repertoire activated determines severity and histology of the resulting disease.

Collagen-specific T-cell repertoire in blood and synovial fluid varies with disease activity in early rheumatoid arthritis.

INTRODUCTION: Type II collagen is a DR4/DR1 restricted target of self-reactive T cells that sustain rheumatoid arthritis. The aim of the present study was to analyze the T-cell receptor repertoire at the onset of and at different phases in rheumatoid arthritis. METHODS: We used the CDR3 BV-BJ spectratyping to study the response to human collagen peptide 261-273 in 12 patients with DR4+ rheumatoid arthritis (six at the onset of disease and six during the course of disease) and in five healthy DR4+ relatives. RESULTS: The collagen-specific T-cell repertoire is quite restricted at the onset of disease, involving approximately 10 rearrangements. Within the studied collagen-specific rearrangements, nearly 75% is shared among patients. Although the size of the repertoire used by control individuals is comparable to that of patients, it is characterized by different T-cell receptors. Part of the antigen-specific T-cell repertoire is spontaneously enriched in synovial fluid. The specific T-cell repertoire in the periphery was modulated by therapy and decreased with the remission of the disease. Failure of immunoscopy to detect this repertoire was not due to suppression of collagen-driven proliferation in vitro by CD4+ CD25+ T cells. Clinical relapse of the disease was associated with the appearance of the original collagen-specific T cells. CONCLUSIONS: The collagen-specific T-cell receptor repertoire in peripheral blood and synovial fluid is restricted to a limited number of rearrangements in rheumatoid arthritis. The majority of the repertoire is shared between patients with early rheumatoid arthritis and it is modulated by therapy.

Administration of PLP139-151 primes T cells distinct from those spontaneously responsive in vitro to this antigen.

We examined the TCR repertoire used by naive SJL mice in their in vitro spontaneous response to proteolipid protein (PLP) 139-151 by Vbeta-Jbeta spectratyping and compared it to that used after immunization with the peptide. T cells from immunized mice use the public rearrangement Vbeta10-Jbeta1.1, but naive mice do not; in contrast, TCR CDR3-beta rearrangements of Vbeta18-Jbeta1.2 and Vbeta19-Jbeta1.2 consistently are associated with the spontaneous response. T cells involved in spontaneous and induced responses can each recognize PLP(139-151) presented in vivo, but its s.c. administration has different consequences for the two repertoires. Four days after immunization, T cells associated with spontaneous responsiveness appear in the draining lymph nodes but disappear by day 10 and never appear elsewhere. Simultaneously, Vbeta10-Jbeta1.1 T cells are likewise activated in the lymph nodes by day 4 and spread to the spleen by day 10. Eight- to 10-wk-old naive mice use a narrower repertoire of TCRs than do immunized age-matched mice. Induced Vbeta10-Jbeta1.1 T cells home to the CNS during experimental autoimmune encephalomyelitis, whereas we failed to detect Vbeta18-Jbeta1.2 and Vbeta19-Jbeta1.2 TCR rearrangements in the CNS. Thus, we observe that administration of PLP(139-151) primes a T cell repertoire distinct from the one responsible for spontaneous responsiveness. This "immunized" repertoire substitutes for the naive one and becomes dominant at the time of disease onset.

Distinct and non-overlapping T cell receptor repertoires expanded by DNA vaccination in wild-type and HER-2 transgenic BALB/c mice.

Central tolerance to tumor-associated Ags is an immune-escape mechanism that significantly limits the TCR repertoires available for tumor eradication. The repertoires expanded in wild-type BALB/c and rat-HER-2/neu (rHER-2) transgenic BALB-neuT mice following DNA immunization against rHER-2 were compared by spectratyping the variable (V)beta and the joining (J)beta CDR 3. Following immunization, BALB/c mice raised a strong response. Every mouse used one or more CD8+ T cell rearrangements of the Vbeta9-Jbeta1.2 segments characterized by distinct length of the CDR3 and specific for 63-71 or 1206-1214 rHER-2 peptides. In addition, two CD4+ T cell rearrangements recurred in >50% of mice. Instead, BALB-neuT mice displayed a limited response to rHER-2. Their repertoire is smaller and uses different rearrangements confined to CD4+ T cells. Thus, central tolerance in BALB-neuT mice acts by silencing the BALB/c mice self-reactive repertoire and reducing the size of the CD8+ T cell component. CD8+ and CD4+ T cells from both wild-type and transgenic mice home to tumors. This definition of the T cell repertoires available is critical to the designing of immunological maneuvers able to elicit an effective immune reaction against HER-2-driven carcinogenesis.

Mycobacterium tuberculosis in the adjuvant modulates the balance of Th immune response to self-antigen of the CNS without influencing a "core" repertoire of specific T cells.

In the present study, we use modified CDR3 beta-chain spectratyping (immunoscope) to dissect the effect of Mycobacterium tuberculosis (MT)-derived proteins on individual PLP139-151-specific cells in the SJL mouse strain. In this model, the immunoscope technique allows the characterization of a public TCR that involves rearrangement of Vbeta10 and Jbeta1.1 and a semi-private TCR characterized by rearrangement of Vbeta4 and Jbeta1.6. Both rearrangements are specific for PLP139-151 and sequences of the CDR3 region of the two beta-chains show a conserved motif for the public rearrangement and related but more variable sequences for the semi-private rearrangement. MT-derived proteins promote increase of IFN-gamma-secreting cells. However, we observe that the presence and amount of MT used during immunization have no effect on the frequency of usage, polarization and in vivo expansion of cells carrying the studied rearrangements. Rather, the strong Th1-promoting effect of adjuvant is possibly due to recruitment toward Th1 of a wider spectrum of TCR repertoires. Therefore, instead of having a comprehensive effect on the entire repertoire, MT modulates the immune response by affecting a subset of antigen-specific T cells whose polarization can be adapted to the environment. This step establishes the final balance between Th1 and Th2 and may be essential for the enhancement or protection of disease.

Calnexin suppresses GD3 synthase-induced apoptosis.

An accelerated activity of the GD3 synthase (ST8), with consequent GD3 accumulation, is part of the response to environmental stressors in different cell types. Depending on specific, yet largely undefined, cellular settings, this can be followed by adaptation or apoptosis, the latter mostly due to GD3-induced mitochondrial damage. Here we show that subcellular localization of ST8 could significantly affect the biological outcome of GD3 accumulation. Binding to the molecular chaperone calnexin causes the retention of ST8 within the endoplasmic reticulum (ER) and prevents its relocalization to the Golgi. Calnexin-dependent ER retention does not affect the activity of ST8; yet the de novo synthesized GD3 largely fails to reach the mitochondria. Accordingly, overexpression of calnexin suppresses the pro-apoptotic activity of ST8, while the loss of calnexin sensitizes the cells to ST8-induced apoptosis. Reconstitution of calnexin confers protection to deficient cells. Thus, calnexin controls the biological outcome of GD3 accumulation and reveals a novel role in the stress response.

p21(Waf1/Cip1/Sdi1) mediates shear stress-dependent antiapoptotic function.

OBJECTIVE: The antiapoptotic effect of p21(Waf1/Cip1/Sdi1) (p21) was examined in human umbilical vein endothelial cells (HUVEC) exposed to laminar shear stress (SS) or to the nitric oxide donor sodium nitroprusside (SNP) and in a mouse model of hindlimb ischemia. METHODS: In vitro: Cells were cultured without serum and in the presence of cobalt chloride to simulate hypoxia for 12 h (T0). Shear stress was applied to endothelial cells for additional 12 h. In vivo: Hindlimb ischemia was realized in mice by femoral artery ligation. SNP was acutely administered by subcutaneous injection or by Alzet osmotic pumps for a longer treatment. RESULTS: At T0, HUVEC were either exposed to SS (15 dyn/cm2/s(-1)), treated with SNP or kept in static condition (ST) for 1-12 h; after additional 12 h in ST, 30-35% of cells still alive at T0 had died. In this condition, both SS and SNP treatments markedly increased p21 levels and reduced apoptosis in HUVEC. Recombinant adenoviruses carrying p21 (AdCMV.p21) or antisense p21 (AdCMV.ASp21) cDNA revealed that AdCMV.p21-infected HUVEC were protected from death while AdCMV.ASp21 reduced SS- and SNP-dependent protection from apoptosis. In mice, apoptosis was detected in endothelial cells of ischemic hindlimbs as early as 8 h after femoral artery ligation. Treatment with SNP enhanced p21 expression and protected ischemic tissue from damage. Remarkably, direct in vivo injection of AdCMV.p21 significantly reduced the number of apoptotic nuclei in the presence of ischemia. CONCLUSIONS: The present study establishes that, under our experimental conditions, (a) p21 plays an important role in SS and nitric oxide antiapoptotic effect in vitro, and (b) p21 gene transfer prevents apoptosis in vitro and in vivo, following acute interruption of blood flow.

A caspaselike activity is triggered by LPS and is required for survival of human dendritic cells.

Bacterial endotoxin (lipopolysaccharide [LPS]) is a potent inducer of human dendritic cell (DC) maturation and survival. Here we show that immature DCs exposed to LPS trigger an early and sustained caspase-like activity, which can be blocked by zVAD (z-Val-Ala-Asp), in the absence of detectable caspase 8 and caspase 10 activation, or poly(ADP-ribose) polymerase (PARP)-cleaving activity. Preventing LPS-induced caspase-like activation in DC results in massive cell death. Importantly, triggering of the caspase-like activity is required for LPS-induced activation of extracellular signal-regulated kinases (ERKs) and for LPS-induced up-regulation of cFLIP (Fas-associating protein with death domain-like interleukin-1 beta-converting enzyme [FLICE]-like inhibitory protein). Therefore, a caspase-dependent pathway initiated by LPS controls survival of human DCs.