At this moment, databanks worldwide contain brain images of previously unimaginable numbers. Combined with developments in data science, these massive data provide the potential to better understand the genetic underpinnings of brain diseases. However, different datasets, which are stored at different institutions, cannot always be shared directly due to privacy and legal concerns, thus limiting the full exploitation of big data in the study of brain disorders. Here we propose a federated learning framework for securely accessing and meta-analyzing any biomedical data without sharing individual information. We illustrate our framework by investigating brain structural relationships across diseases and clinical cohorts. The framework is first tested on synthetic data and then applied to multi-centric, multi-database studies including ADNI, PPMI, MIRIAD and UK Biobank, showing the potential of the approach for further applications in distributed analysis of multi-centric cohorts.

Overexpression of c-Myc in tumors is usually associated with cell proliferation and increased susceptibility to apoptosis. Concomitantly, c-Myc contributes to tumorigenesis by its ability to destabilize the cellular genome. Here, we examined whether c-Myc induces genomic instability and apoptosis in c-Myc-activated cells. Wild-type Myc (wt-Myc) and two mutated Myc myc box II proteins (mt-Myc) were overexpressed in IL3-dependent murine Ba/F3 cells. As expected, wt-Myc triggered apoptosis in absence of IL3. Standard karyotyping, spectral karyotyping, and fluorescent in situ hybridization (FISH) were performed before and after c-Myc activation. Structural and numerical genomic instability was detected 48 h after wt-Myc activation and included gene amplification, the formation of extrachromosomal elements (EEs), chromosome breakage, deletions, increased aneuploidy, and polyploidization. Interestingly, some cells simultaneously displayed genomic instability and apoptosis. Both wt- and mt-Myc proteins were equally potent promoters of genomic instability. However, only wt-Myc simultaneously induced genomic instability and apoptosis. Mt-Myc proteins failed to induce apoptosis, thereby generating a strong imbalance towards the survival of genomically unstable cells.

While data in healthcare is produced in quantities never imagined before, the feasibility of clinical studies is often hindered by the problem of data access and transfer, especially regarding privacy concerns. Federated learning allows privacy-preserving data analyses using decentralized optimization approaches keeping data securely decentralized. There are currently initiatives providing federated learning frameworks, which are however tailored to specific hardware and modeling approaches, and do not provide natively a deployable production-ready environment. To tackle this issue, herein we propose an open-source federated learning frontend framework with application in healthcare. Our framework is based on a general architecture accommodating for different models and optimization methods. We present software components for clients and central node, and we illustrate the workflow for deploying learning models. We finally provide a real-world application to the federated analysis of multi-centric brain imaging data.

The lymphomagenic action of myc genes in conjunction with Epstein-Barr virus nuclear antigen-1 (EBNA-1) have been examined using transgenic mice in several separate tests. Synergy between Myc and EBNA-1 in lymphomagenesis was revealed in a cross breed study where co-expression of transgenic myc and EBNA-1 led to a tumor latency period reduced significantly in some crosses. In the resulting bitransgenic tumors, expression of the E-myc genes was not af- The virus was discovered in cell cultures from endemic Burkitt's lymphoma (BL) samples 2 and has since been found to have a very high association (Ͼ95%) with this disease. The more rare, sporadic form of BL (occurring worldwide) shows 15-25% association with EBV. Similarly, only a proportion (30 -40%) of the AIDS form of BL is positive for EBV genomes. Nevertheless, in all EBV-associated cases, the viral episomal genome is thought to be clonal, indicating that the virus was present in the tumor progenitor cell. This would argue that the probability of evolving into a tumor is higher for an EBV carrying than an EBV negative B-lymphocyte suggesting that the virus is causal in the genesis of the tumor. 3 Notably, all BL tumors carry a reciprocal chromosomal translocation between chromosome 8, in the proximity of the c-myc gene and one of chromosomes 14, 2, or 22 at the immunoglobulin heavy (IgH) or light (IgL) chain loci. This leads to deregulation of c-myc expression. 4 -6 Several transgenic mouse models to examine myc induced tumorigenesis have been developed. Overexpression of c-myc or N-myc in B cells using the IgH (E) intronic enhancer predisposes transgenic mice to B cell lymphoma. [7][8][9][10][11][12][13] The exact pathology of the c-myc induced tumors and the timing of onset varies according to the precise transgene construct used. Using an IgL locus enhancer instead to drive c-myc overexpression slightly later in B cell development results in a transgenic lymphoma with a pathology startlingly similar to BL. 14 In EBV-associated BL tumors a restricted subset of the viral latent genes is expressed. Of 6 EBV nuclear antigens (EBNA), EBNA-1 is the only one that has been detected consistently. 15,16 In addition the 2 small, polymerase III, untranslated RNAs (EBER-1 and -2) are expressed. LMP2A transcripts and the BamHI-A rightward transcripts have also been detected in BL biopsies. 17 EBV is a potent B cell transforming virus, nevertheless the role it plays in the genesis of BL remains unclear. The theory that EBV continues to contribute to the malignant phenotype of established tumor cells is suggested by the stable retention of the viral genome in cultured BL cell lines and by the loss of tumorigenicity of sub clones of BL cell lines that have lost the viral genome. 18,19 In this manner, the EBERs have been shown to contribute to the transformed phenotype in Akata cells, 20 and EBER expression can substitute for the presence of viral genomes in EBV positive Akata and Mutu I cell lines. This was not the case, however, for all the BL cell lines examined, suggest...

No abstract

Trisomy 15 is the most common chromosomal aberration in murine T-cell lymphomas. The relevant chromosomal region responsible for the growth advantage of the 15-trisomic cell has not been defined. In order to map this region, we have induced thymic lymphomas by chemical carcinogens (DMBA or MNU) in mice with 2 different constitutional translocations, T(7;15)9H homozygotes and [T(7;15)9H X T(5;15)4Ad] FI hybrids. Twenty-two tumors developed in 90 carcinogen-treated mice. Among the 14 cytogenetically analyzed thymic lymphomas, 4 were diploid and 5 were aneuploid, with no chromosome-15-associated changes. Five lymphomas showed partial duplication of chromosome 15. Four of them have duplicated the segment distal to the C/DI breakpoint of T9H, while the 5th carried an interstitial duplication of the D2 sub-band of the T(7;15) translocation chromosome. These findings suggest that the duplication of the D 2/3 region, known to contain the c-myc and the pvt-I genes (Banerjee et al., 1985), rather than other regions of chromosome 15, contributes to the development and/or progression of murine T-cell leukemias.

E mu-myc transgenic mice were back-crossed to BALB/c mice up to back-cross generation 3. The offspring that included transgene-carrying and -negative mice in approximately equal proportions were randomly divided into 2 groups. Thirty-four mice (group I) were treated with pristane, followed by A-MuLV, and 40 (group II) were injected with A-MuLV alone. Altogether, 16 lymphoid tumors developed in group I and 17 in group II. Nine of the tumors in group I and 4 in group II appeared as ascitic tumors. The ascites contained lymphoblasts and 10 to 45% plasmacytoid cells. These tumors were designated as plasmablastic lymphomas (PLs). All tumors except one were transgene-positive and did not carry translocations. An exceptional tumor in group I carried a variant 6;15 translocation but not the transgene. It obviously corresponds to the regular Abelson + pristane-induced plasmacytoma. Among 11 tested PLs, 10 had a single retroviral insertion site, while one tumor showed 3. Among 18 untreated transgenic descendants (group III), chosen randomly during serial back-crosses, 15 (83%) developed lymphomas, with no sign of plasmacytoid differentiation. The incidence was comparable in all 3 groups, assuming 50% of the mice in groups I and II to be transgenic. The time distribution of tumor development was also similar. Spleen cells from transgene-carrying mice with no clinical sign of lymphoma were infected in vitro with A-MuLV and transplanted i.p. into BALB/c recipients. PLs developed in 26 of 31 pristane-treated recipients, but in only one of 18 untreated recipients. One of 6 PLs tested was monoclonal, whereas the remaining 5 were oligoclonal. They all expressed v-abl. These results show that some of the preneoplastic B-cells that expressed constitutively active myc transgene turned into plasmablasts after infection with A-MuLV. Full development of their neoplastic potential was facilitated by the presence of pristane-granuloma.

Reciprocal chimeras were generated between BALB/c and DBA/2 mice by inoculating newborn recipients of either strain with bone-marrow (BM) cells of the other through the periorbital vein. DBA/2 mice inoculated with the BALB/c with proven chimerism will be referred to as C----D, the reciprocal as D----C. The BALB/c cells carried a Robertsonian 6;15 (Rb6;15) chromosome marker to facilitate identification. The chimeric mice contained between 5% and 70% of donor cells when examined at 4 to 5 weeks of age. Six of 10 C----D developed plasmacytomas (MPC) after 3 x 0.5 ml monthly pristane treatment (incidence 60%) and 8 of 25 (incidence 32%) after 2 to 3 x 0.5 ml pristane followed by Abelson virus (A-MuLV) infection. Seven of 15 D----C developed MPC after pristane treatment (incidence 47%) and 4 of 17 after pristane + A-MuLV (incidence 24%). All tumors that have arisen in both reciprocal chimeras originated from BALB/c cells independently of the degree of chimerism. All tumors contained an Ig/myc translocation. Among the C----D chimeras, 5 carried t(12;15) and I t(6;15) in the pristane-treated group, while 4 carried t(12;15), I t(6;15) and 3 t(15;16) in the pristane + A-MuLV. Among the D----C chimeras 6 carried t(12;15) and I t(6;15) in the pristane-treated group, while 3 t(12;15) and I t(6;15) in the pristane + A-MuLV. No tumors developed in 18 pristane- and 22 pristane + A-MuLV-treated DBA/2 mice nor in 15 pristane- and 17 pristane + A-MuLV-treated (BALB/c x DBA/2)F1 mice. The data indicate that BALB/c and DBA/2 cells differ in their propensity to transform into plasmacytoma in identical host environments after both pristane and pristane + A-MuLV treatment. They also show that the oil granuloma can support MPC development in either type of chimeric host.