Machine learning meets partner matching: Predicting the future relationship quality based on personality traits.

To what extent is it possible to use machine learning to predict the outcome of a relationship, based on the personality of both partners? In the present study, relationship satisfaction, conflicts, and separation (intents) of 192 partners four years after the completion of questionnaires concerning their personality traits was predicted. A 10x10-fold cross-validation was used to ensure that the results of the linear regression models are reproducible. The findings indicate that machine learning techniques can improve the prediction of relationship quality (37% of variance explained), and that the perceived relationship quality of a partner is mostly dependent on his or her own individual personality traits. Additionally, the influences of different sets of variables on predictions are shown: partner and similarity effects did not incrementally predict relationship quality beyond actor effects and general personality traits predicted relationship quality less strongly than relationship-related personality.

Dynamic changes of the normal B lymphocyte repertoire in CLL in response to ibrutinib or FCR chemo-immunotherapy.

Using next-generation immunoglobulin (IGH) sequencing and flow cytometry, we characterized the composition, diversity and dynamics of non-malignant B cells in patients undergoing treatment with the Bruton tyrosine kinase (BTK) inhibitor ibrutinib or chemo-immunotherapy with fludarabine, cyclophosphamide, and rituximab (FCR). During ibrutinib therapy, non-malignant B cell numbers declined, but patients maintained stable IGH diversity and constant fractions of IGH-mutated B cells. This indicates partial preservation of antigen-experienced B cells during ibrutinib therapy, but impaired replenishment of the normal B cell pool with naïve B cells. In contrast, after FCR we noted a recovery of normal B cells with a marked predominance of B cells with unmutated IGH. This pattern is compatible with a deletion of pre-existing antigen-experienced B cells followed by repertoire renewal with antigen-naïve B cells. These opposite patterns in B cell dynamics may result in different responses towards neoantigens versus recall antigens, which need to be further defined.

Monitoring multiple myeloma by next-generation sequencing of V(D)J rearrangements from circulating myeloma cells and cell-free myeloma DNA.

Recent studies suggest that circulating tumor cells and cell-free DNA may represent powerful non-invasive tools for monitoring disease in patients with solid and hematologic malignancies. Here, we conducted a pilot study in 27 myeloma patients to explore the clonotypic V(D)J rearrangement for monitoring circulating myeloma cells and cell-free myeloma DNA. Next-generation sequencing was used to define the myeloma V(D)J rearrangement and for subsequent peripheral blood tracking after treatment initiation. Positivity for circulating myeloma cells/cell-free myeloma was associated with conventional remission status (P<0.001) and 91% of non-responders/progressors versus 41% of responders had evidence of persistent circulating myeloma cells/cell-free myeloma DNA (P<0.001). About half of the partial responders showed complete clearance of circulating myeloma cells/cell-free myeloma DNA despite persistent M-protein, suggesting that these markers are less inert than the M-protein, rely more on cell turnover and, therefore, decline more rapidly after initiation of effective treatment. Positivity for circulating myeloma cells and for cell-free myeloma DNA were associated with each other (P=0.042), but discordant in 30% of cases. This indicates that cell-free myeloma DNA may not be generated entirely by circulating myeloma cells and may reflect overall tumor burden. Prospective studies need to define the predictive potential of high-sensitivity determination of circulating myeloma cells and DNA in the monitoring of multiple myeloma.

T-cell diversification reflects antigen selection in the blood of patients on immune checkpoint inhibition and may be exploited as liquid biopsy biomarker.

Cancer immunotherapy with antibodies targeting immune checkpoints, such as programmed cell death protein 1 (PD-1), shows encouraging results, but reliable biomarkers predicting response to this costly and potentially toxic treatment approach are still lacking. To explore an immune signature predictive for response, we performed liquid biopsy immunoprofiling in 18 cancer patients undergoing PD-1 inhibition before and shortly after initiation of treatment by multicolor flow cytometry and next-generation T- and B-cell immunosequencing (TCRß/IGH). Findings were correlated with clinical outcomes. We found almost complete saturation of surface PD-1 on all T-cell subsets after the first dose of the antibody. Both T- and B-cell compartments quantitatively expanded during treatment. These expansions were mainly driven by an increase in the activated T-cell compartments, as well as of naïve B- and plasma cells. Deep immunosequencing revealed a clear diversification pattern of the clonal T-cell space indicative of antigenic selection in 47% of patients, while the remaining patients showed stable repertoires. 43% of the patients with a diversification pattern showed disease control in response to the PD-1 inhibitor. No disease stabilizations were observed without clonal T-cell space diversification. Our data show for the first time a clear impact of PD-1 targeting not only on circulating T-cells, but also on B-lineage cells, shedding light on the complexity of the anti-tumor immune response. Liquid biopsy T-cell next-generation immunosequencing should be prospectively evaluated as part of a composite response prediction biomarker panel in the context of clinical studies.

Dopamine gene predicts the brain's response to dopaminergic drug.

Dopamine is critical for reward-based decision making, yet dopaminergic drugs can have opposite effects in different individuals. This apparent discrepancy can be accounted for by hypothesizing an 'inverted-U' relationship, whereby the effect of dopamine agents depends on baseline dopamine system functioning. Here, we used functional MRI to test the hypothesis that genetic variation in the expression of dopamine D2 receptors in the human brain predicts opposing dopaminergic drug effects during reversal learning. We scanned 22 subjects while they engaged in a feedback-based reversal learning task. Ten subjects had an allele on the Taq1A DRD2 gene, which is associated with reduced dopamine receptor concentration and decreased neural responses to rewards (A1+ subjects). Subjects were scanned twice, once on placebo and once on cabergoline, a D2 receptor agonist. Consistent with an inverted-U relationship between the DRD2 polymorphism and drug effects, cabergoline increased neural reward responses in the medial orbitofrontal cortex, cingulate cortex and striatum for A1+ subjects but decreased reward responses in these regions for A1- subjects. In contrast, cabergoline decreased task performance and fronto-striatal connectivity in A1+ subjects but had the opposite effect in A1- subjects. Further, the drug effect on functional connectivity predicted the drug effect on feedback-guided learning. Thus, individual variability in how dopaminergic drugs affect the brain reflects genetic disposition. These findings may help to explain the link between genetic disposition and risk for addictive disorders.