Fig 1. Schema of the analysis pipeline. Left panel (orange) illustrates the DNA processing flow. DNA is extracted from CD138⁺ tumor cells from bone marrow and CD3⁺ or granulocytes (GRN) from peripheral blood as a control. Whole-exome and/or targeted panel sequencing (seq) is performed, and the obtained reads are mapped to the reference genome and analyzed for the identification of somatic mutations and copy number alterations, which are then prioritized on the basis of their actionability. Right panel (blue) illustrates the RNA processing flow. RNA is extracted from CD138⁺ tumor cells, and RNA seq is performed. The obtained reads are mapped to the reference genome and summarized at the gene level. Gene expression analysis is then performed to calculate outlier genes, pathway activation, and drug repurposing through inverse match with drug-induced gene expression profiles. DNA- and RNA-based drug recommendations are then summarized in reports. CIViC, Clinical Interpretations of Variants in Cancer (https://civicdb.org).

Fig 2. Summary of DNA findings. (A) Distribution of mutational burden (ie, number of total mutations per megabase [Mb] detected; n = 41 patients with whole-exome sequencing [WES] data available). (B) Pie chart illustrates the percentage of nonintronic mutations found per category. Bold print indicates categories of potentially pathogenic mutations. (C) Top 30 mutated genes (n = 55 patients with mutation data available from WES and/or targeted sequencing). (*) Gene carried actionable mutations (according to Clinical Interpretations of Variants in Cancer [https://civicdb.org]). (D) Genes with actionable copy number alterations (according to CIViC) found in the 41 patients with WES data available. The colors indicate that the corresponding gene had mostly gain of copies (red) or loss of copies (blue). IGR, intergenic region; lincRNA, long intergenic noncoding RNA.

Fig 3. Summary of RNA findings. (A) Actionable outlier genes (according to Clinical Interpretations of Variants in Cancer [https://civicdb.org]). The colors indicate that the corresponding gene was mostly overexpressed (red) or underexpressed (blue). (B) Pathway activation calculated by gene set variation analysis for the 60 patients with RNA sequencing data available. (C) Distribution of activated pathways (ie, positive score) for the 60 patients with RNA sequencing data available. (D) Drugs recommended by RNA-based drug repurposing on the basis of inverse matching of patient gene expression profiles with drug-induced profiles from L1000. FGFR3, fibroblast growth factor receptor 3; HDAC, histone deacetylase; IL-6, interleukin-6; MAPK, mitogen-activated protein kinase; mTOR, mammalian target of rapamycin; P13K AKT, phosphoinositide 3-kinase; XBP1s, Xbox binding protein 1 spliced.

Fig 4. Depth of response and timeline of treatments. (A) Chart shows depth of response as the percentage change from baseline (start of therapy), for patients who received our recommended drugs. Response was determined after the International Myeloma Working Group criteria. Patients ISMMS42 and ISMMS09 received two different drugs at different times; thus, their responses are shown as two separate bars. Colors indicate the source data used to generate the recommendation, either DNA or RNA or both (see legend). The arrow indicates that the patient had an ongoing response at the end of the study. (B) Each bar indicates the time that each patient was receiving our recommended treatment. Each color represents a different drug, coded as follows: trametinib (light blue), venetoclax (dark gold), panobinostat (dark blue), dabrafenib (salmon), and etoposide (dark red). Multicolored portions of the bars indicate time receiving a drug combination on the basis of our recommendation. Triangle indicates discontinuation of treatment, where reasons are color coded as follows: disease progression (gold), physician’s choice (green), adverse event (pink), and remission achieved (white). Red diamond indicates death of the patient. (+) Additional drugs were used in the specific time frame (see Table 2 for details). Gray bars indicate time receiving a different treatment outside of our recommendation. Red arrow indicates patient had an ongoing response at the end of the study.

Fig 5. Schematic of the trial with results, limitations, and proposed solutions. We recruited 64 patients (pts) with relapsed and/or refractory multiple myeloma (MM) treated at the Mount Sinai hospital. We obtained RNA sequencing (RNAseq) for 94%, whole-exome sequencing (WES) for 64%, and targeted DNA panel data for 53% of the pts. Sequencing (Seq) data were analyzed by our pipeline (see Fig. 1), and drug recommendation reports were produced. Treatment was implemented in 40% of the pts, and 81% of these were evaluable (see Table 2). According to IMWG criteria, 76% of the pts had a clinical response (MR and above, where MR = minimal response and corresponds to a 25% reduction of disease marker), whereas 24% of the pts had stable disease (SD) or worse. Problems to address to improve recommendations include assessment of clonal heterogeneity, analysis of bone marrow microenvironment, and extension of reference data to include MM-specific drug profiles from in vitro and in vivo models. Treatment was not implemented in 60% of the pts, because either no drugs were identified, because insurance denied the proposed drugs, or because of rapid progression of the patient before the results of sequencing were available. CyTOF, mass cytometry; scRNA, single-cell RNA.