Drug-induced autoimmunity (DIA) is an idiosyncratic adverse drug reaction. Although first reported in the mid-1940’s, the mechanisms underlying DIA remain unclear, and there is little understanding of why it is only associated with some drugs. Because it only occurs in a small number of patients, DIA is not normally detected until a drug has reached the market. We describe an ensemble machine learning approach using transcriptional data to predict DIA. The genes comprising the signature implicate dysregulation of cell cycling or proliferation as part of the mechanism of DIA. This approach could be adapted by pharmaceutical companies as an additional preclinical safety screen, reducing the risk of drugs with the potential to cause autoimmunity reaching the market.

Precision medicine holds great promise to improve outcomes in cancer, including haematological malignancies. However, there are few biomarkers that influence choice of chemotherapy in clinical practice. In particular, multiple myeloma requires an individualized approach as there exist several active therapies, but little agreement on how and when they should be used and combined. We have previously shown that a transcriptomic signature can identify specific bortezomib- and lenalidomide-sensitivity. However, gene expression signatures are challenging to implement clinically. We reasoned that signatures based on the presence or absence of gene mutations would be more tractable in the clinical setting, though examples of such signatures are rare. We performed whole exome sequencing as part of the CARDAMON trial, which employed carfilzomib-based therapy. We applied advanced machine learning approaches to discover mutational patterns predictive of treatment outcome. The resulting model accurately predicted progression-free survival (PFS) both in CARDAMON patients and in an external validation set of patients from the CoMMpass study who had received carfilzomib. The signature was specific for carfilzomib therapy and was strongly driven by genes on chromosome 1p36. Importantly, patients predicted to be carfilzomib-sensitive had a longer PFS when treated with carfilzomib/lenalidomide/dexamethasone than with bortezomib/carfilzomib/dexamethasone. However, in those predicted to be carfilzomib-insensitive, the latter therapy may have been capable of eradicating carfilzomib-resistant clones. We propose that the signature can be used to make rational therapeutic decisions and could be incorporated into future clinical trials. ### Competing Interest Statement Walker: Abbvie & Janssen: Honoraria. Popat: Takeda: Research Funding; GSK: Honoraria, Research Funding; Janssen, Takeda, Celgene, and GSK: Honoraria; Janssen, Takeda, GSK: Other: Travel expenses from Janssen, Takeda GSK; Roche:Honoraria; BMS: Honoraria; Janssen: Honoraria; Takeda, AbbVie, GlaxoSmithKline, and Celgene: Consultancy. Benjamin:Bristol Myers Squibb/Celgene: Research Funding; Amgen: Research Funding. Clifton-Hadley: Astra Zeneca, GSK, Pfizer, MSD, BMS, Amgen, Millennium Takeda: Other: CRUK and UCL CTC have received research funding in the past 24 months, Research Funding. Owen: Astra Zeneca: Honoraria; Janssen: Honoraria Membership on an entity's Board of Directors or advisory committees; Beigene: Honoraria ;Membership on an entity's Board of Directors or advisory committees. Chapman:Sanofi: Honoraria. ### Clinical Protocols ### Funding Statement IGW is supported by the Kay Kendall Leukaemia Fund KKL1442 also received support from the UK Myeloma Society UKMS, VdA has received support from the UKMS. MAC is supported by the Medical Research Council Toxicology Unit MC UU 00025 10. KY and RP are supported by the National Institute for Health Research University College London Hospitals Biomedical Research Centre. The Cardamon trial was funded by Amgen and endorsed by Cancer Research UK C9203 A17750. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The London City and East Research Ethics Committee London, UK gave full approval of study. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes All data produced in the present study are available upon reasonable request to the authors. Data and code will be made publicly available on acceptance in a peer reviewed journal

The accessibility of cell surface proteins makes them tractable for targeting by cancer immunotherapy, but identifying suitable targets remains challenging. Here we describe plasma membrane profiling of primary human myeloma cells to identify an unprecedented number of cell surface proteins of a primary cancer. We used a novel approach to prioritize immunotherapy targets and identified a cell surface protein not previously implicated in myeloma, semaphorin-4A (SEMA4A). Using knock-down by short-hairpin RNA and CRISPR/nuclease-dead Cas9 (dCas9), we show that expression of SEMA4A is essential for normal myeloma cell growth in vitro, indicating that myeloma cells cannot downregulate the protein to avoid detection. We further show that SEMA4A would not be identified as a myeloma therapeutic target by standard CRISPR/Cas9 knockout screens because of exon skipping. Finally, we potently and selectively targeted SEMA4A with a novel antibody-drug conjugate in vitro and in vivo.

Introduction: Multiple myeloma remains an incurable disease with a significant variation in therapeutic response. Choice of treatment is largely determined by prior lines of therapy and the patient's fitness. However, myeloma is a highly heterogeneous disease and this approach risks failing to deliver the right drug at the right time to an individual patient. To begin to address this, we have adopted machine learning approaches to predict individual responses to specific anti-myeloma therapies. We have previously shown that it is possible to select rationally between bortezomib- and lenalidomide-based therapy using a gene expression signature. Gene expression models are difficult to implement clinically, so more recently we presented a five gene mutational signature to identify carfilzomib-specific responses. Although easier to translate, such molecular signatures can still be hindered clinically by slow processing and technical failures. Because flow cytometry is well established in the diagnostic laboratory and is sensitive enough to detect minimal residual disease, we set out to develop an immunophenotypic signature of carfilzomib-specific responsiveness in primary myeloma. Methods We originally developed our mutational signature on whole exome sequencing of diagnostic samples from the CARDAMON clinical trial. For the current study, we performed RNA sequencing on 80 of these. Patients were divided into carfilzomib-responsive and non-responsive groups according to their mutational signature. Genes were filtered for those encoding cell surface proteins. A second round of filtering selected only those genes whose RNA expression correlated well with protein expression as determined by plasma membrane fractionation and mass spectrometry. Genes differentially expressed between the two groups were then detected using DeSEQ2. Logistic regression on these candidate genes was performed to assess whether they could be used to correctly classify patients for carfilzomib responsiveness and hence form a predictive flow cytometric panel. Results 76 patients from the CARDAMON trial had RNAseq that passed quality control. 128 genes encoded cell surface proteins, were reliable predictors of cell surface protein expression (R2 > 0.8), and were differentially expressed between carfilzomib-responsive and non-responsive patients. No single cell surface protein alone could correctly classify patients. However, using multiple logistic regression, six genes whose expression could be detected by flow cytometry (CLEC7A, STRA6, TMPRSS11E, CD27, GPR176, TK1, LPXN) together predicted carfilzomib responsiveness with an AUC of 0.77. Independent validation of these genes by RNA-Seq in the CoMMpass dataset showed that they could predict carfilzomib responsiveness with an AUC of 0.803 (Figure 1A). Indeed, survival for those patients who were predicted to be carfilzomib-responsive by our potential flow panel was significantly higher than those predicted to be resistant (median 34 months vs 15months ; p= 0.0032), figure 1B. Conclusions By combining transcriptomic and proteomic approaches, we have generated a panel of six cell surface proteins that are amenable to flow cytometry and can predict carfilzomib-responsiveness. Prospective validation of these markers is underway. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal