RNA-sequencing based first choice of treatment and determination of risk in multiple myeloma

Background Immunotherapeutic targets in multiple myeloma (MM) have variable expression height and are partly expressed in subfractions of patients only. With increasing numbers of available compounds, strategies for appropriate choice of targets (combinations) are warranted. Simultaneously, risk assessment is advisable as patient’s life expectancy varies between months and decades. Methods We first assess feasibility of RNA-sequencing in a multicenter trial (GMMG-MM5, n=604 patients). Next, we use a clinical routine cohort of untreated symptomatic myeloma patients undergoing autologous stem cell transplantation (n=535, median follow-up (FU) 64 months) to perform RNA-sequencing, gene expression profiling (GEP), and iFISH by ten-probe panel on CD138-purified malignant plasma cells. We subsequently compare target expression to plasma cell precursors, MGUS (n=59), asymptomatic (n=142) and relapsed (n=69) myeloma patients, myeloma cell lines (n=26), and between longitudinal samples (MM vs. relapsed MM). Data are validated using the independent MMRF CoMMpass-cohort (n=767, FU 31 months). Results RNA-sequencing is feasible in 90.8% of patients (GMMG-MM5). Actionable immune-oncological targets (n=19) can be divided in those expressed in all normal and >99% of MM-patients (CD38, SLAMF7, BCMA, GPRC5D, FCRH5, TACI, CD74, CD44, CD37, CD79B), those with expression loss in subfractions of MM-patients (BAFF-R [81.3%], CD19 [57.9%], CD20 [82.8%], CD22 [28.4%]), aberrantly expressed in MM (NY-ESO1/2 [12%], MUC1 [12.7%], CD30 [4.9%], mutated BRAF V600E/K [2.1%]), and resistance-conveying target-mutations e.g., against part but not all BCMA-directed treatments. Risk is assessable regarding proliferation, translated GEP- (UAMS70-, SKY92-, RS-score) and de novo (LfM-HRS) defined risk scores. LfM-HRS delineates three groups of 40%, 38%, and 22% of patients with 5-year and 12-year survival rates of 84% (49%), 67% (18%), and 32% (0%). R-ISS and RNA-sequencing identify partially overlapping patient populations, with R-ISS missing, e.g., 30% (22/72) of highly proliferative myeloma. Conclusion RNA-sequencing based assessment of risk and targets for first choice treatment is possible in clinical routine.

Presence or absence and height of target expression is an evident selection criterion for treatment choice, as shown, e.g., for CD38 (43) and GPRC5D (41), or a potential use of g-secretase inhibitors (e.g.crenigacestat) (44) in case of low BCMA-expression.Recent studies identified BCMA-mutations conveying resistance to only part of respective T-cell bispecific antibodies and CAR-T treatments, suggesting the possibility to switch within BCMAtargeting agents to a different compound (45).
In this manuscript, we first assess applicability of RNA-sequencing in the GMMG-MM5 multicenter phase III clinical trial setting (604 patients) based on our low-input RNA-sequencing protocol (53).Secondly, we use a clinical routine cohort of 535 patients investigated by RNA-sequencing, GEP, and multi-parameter iFISH.We assess presence and expression height of actionable immunological targets, mutated BRAF V600E/K, and potential resistance-conveying mutations of these genes.We aim at delineating in what percentage of patients an "educated first choice" is possible on the simulated background of all immune-oncological compounds approved or in clinical trials (clinicaltrials.gov"active" or "completed", Table 1) being available.We further compare expression of identified targets in MM to plasma cell precursor populations, early-stage plasma cell dyscrasias, as well as to relapsed patients.The latter to first delineate whether a target might be specifically suited for early (e.g., CD19, lost in later myeloma stage) or late (e.g., cancer testis antigens, gained) treatment.Third, we take a fresh look at risk assessment by transferring proliferation and microarray-based risk scores to RNA-sequencing and establish a de novo risk score by RNAsequencing (termed LfM-HRS).Findings are validated in the independent MMRF CoMMpass-cohort (n=767 patients).

Feasibility of RNA-sequencing based on the GMMG-MM5 trial
Patients (n=604) were included in the prospective GMMG 54) between July 2010 and November 2013 in 31 transplant centers and 75 associated sites throughout Germany.As per protocol, bone marrow aspirates at study inclusion, i.e., before treatment, were available for n=573 patients (94.9%), of whom we were able to successfully perform plasma cell purification followed by quality control using flow cytometry for n=559 patients (97.6%).The 31 lacking samples (5.1%) were due to patients declining the bone marrow aspiration (2.5%) or punctio sicca (2.5%).Median purity according to CD38/CD138 double staining was 87.9% with a median cell number of 1.2 × 10 6 cells (51).
iFISH using cytospins from CD138-purified plasma cells was performed centrally (Multiple Myeloma Research Laboratory and Department of Human Genetics, Heidelberg).Data could be obtained for 556/573 patients with available bone marrow aspirates (97%) and 556/559 patients with available CD138purified plasma cells, respectively (99.5%).The median proportion of malignant plasma cells determined per iFISH, i.e., the highest percentage of a chromosomal aberration, was 95% (51).
Samples for RNA-extraction followed by quality control were collected over two weeks and subjected to GEP by DNAmicroarrays.In total, n=458 transcriptome datasets are available, i.e., 81.9% of patients with available CD138-purified plasma cells.Of these, two patients were excluded from further analysis for not fulfilling the trial's inclusion criteria.Gene expression profiling could not be performed in 53 cases due to low RNA quality (9.5%) and further 48 cases (8.6%) in which not enough RNA was available (51).
Using our standardized RNA-sequencing protocol (53) (see below), RNA-sequencing was possible in all patients considered as with "too low" amount of RNA for GEP.In total, in 506/559 patients with available CD138-purified plasma cells (90.5%) and 506/604 patients of the intention to treat population (83.8%).
RNA-sequencing data of 52/142/535/69 patients with MGUS/ AMM/MM/MMR and 26 HMCLs were used.GEP-data of 534 patients with MM were used for score definition and validation (i.e., translation of LfM-HRS into a GEP-based score).

Independent validation of risk assessment and target identification
Independent validation of risk assessment and target identification was performed using the Multiple Myeloma Research Foundation (MMRF) CoMMpass trial (NCT01454297), i.e., n=767 previously untreated myeloma patients with RNAsequencing data available (release 13).

Analysis of gene expression 2.4.1 RNA-extraction
RNA was extracted using the Qiagen AllPrep DNA/RNA kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions.Quality control and quantification was performed using an Agilent 2100 bioanalyzer (Agilent, Frankfurt, Germany).

RNA-sequencing
RNA-sequencing was performed as published (53).In brief, full-length double-stranded cDNA was generated from 5 ng of total RNA and amplified using the SMARTer Ultra Low RNA Kit (Illumina, San Diego, CA, USA).Library preparation was performed from 10 ng of fragmented cDNA using the NEBNext Chip-Seq Library Prep protocol (New England BioLabs, Ipswich, MA, USA).Libraries were sequenced on an Illumina Hiseq2000 with 2x50-bp paired-end reads.RNA-sequencing expression data are deposited in the European nucleotide archive (PRJEB37100, PRJEB36223).
Overall (OS) and event-free (EFS) survival were investigated for symptomatic multiple myeloma undergoing high-dose therapy using Cox's proportional hazard model as published (61).
Survival curves were computed with nonparametric survival estimates for censored data using the Kaplan-Meier method (62).Difference between the curves were tested using the G-rho Log-rank test (63).Wilcoxon ranks sum test and Jonckheere-Terpstra test were used to investigate differences in gene expression between groups and to test for an ordered alternative hypothesis within independent samples (between participants) design, respectively.A Chi-squared test for trend in proportion (Cochran Armitage trend test) was used for comparison of presence of expression from MGUS to AMM to MM to MMR and AMM to MM to MMR, respectively.For comparison of expression in longitudinal samples (MM vs. MMR), a paired Wilcox-test was performed.
Effects were considered statistically significant if the P-value of corresponding statistical tests was below 5%.For comparisons of parameters (Figure 1) and regarding OS and EFS (Figures 2, 3; Supplementary Figures S3, S4, S5), adjustment for multiple testing was made using the Benjamini-Hochberg correction.Adjustment was applied separately for the two cohorts assessed.
RNA-sequencing allows detection of somatic variants and mutated transcripts; e.g., for BCMA, we found in 25% of patients coding nonsynonymous single nucleotide variants present (median 1, maximum 10) with a median allele frequency of 1 in patients harboring the aberration.If present, a resistance conveying mutation would thus have been detected but expectedly could not, as none of the patients has been treated with BCMA-targeting agents prior to analysis.
Immunological target expression decreases from MGUS to AMM to MM to MMR, except for the aberrantly expressed genes MUC1 and NY-ESO1.

Frequency of presence and height of expression
Frequency of presence of expression of the immune-oncological targets BAFFR, TACI, CD19, CD20, CD22 and CD37 significantly decreases, those for MUC1 and NY-ESO1 significantly increases from MGUS to AMM to MM to MMR as well as from AMM to MM to MMR.Presence of CD38, CS1, BCMA, GPRC5D, FCRH5, CD74, CD44, CD1B, CD30, CD70 and CD79B does not vary significantly in both comparisons, see Supplementary Table S3.
In a hypothetical scenario in which all treatment options as in Table 1 are available, based on the expression pattern (Figure 4), for all patients a recommendation could be made.

Risk determination by RNA-sequencing
We first de novo generated a RNA-sequencing based score, termed LfM-HRS, using a method previously applied for DNA-microarrays (50).The LfM-HRS delineates three groups of patients with median EFS of 17 vs.33 vs. 41 months (P<0.001) and OS of 33 vs. 83 vs. 143 months (P<0.001); Figure 2.For independent validation, we used the CoMMpass-cohort (Supplementary Figure S2) and translated the LfM-HRS into a DNA-microarray based score (Supplementary Figure S3).In both cases, it retained its prognostic significance.
GEP and RNA-sequencing based scores showed a concordance of 71.7% -92.5% regarding patients identified as high risk (Supplementary Table S7).

Proliferation of malignant plasma cells
Proliferation of malignant plasma cells as biological variable is one of the strongest prognostic factors in myeloma (47, 65-68).We de novo generated a RNA-sequencing based proliferation index (RPI).In comparison to normal bone marrow plasma cells or nonproliferating memory B-cells, malignant plasma cells showed a significant and stage-dependent increase from early disease MGUS vs. asymptomatic vs. symptomatic, therapy-requiring multiple myeloma (Figure 3A; Jonckheere-Terpstra test P=0.001).Myeloma   3).The RPI was validated on the CoMMpass-cohort (Supplementary Figure S4).

Aplicability of RNA-sequencing in multicenter trials and clinical routine
We have previously shown GEP using DNA-microarrays to be possible in the GMMG-MM5 multicenter trial within four weeks in 81.9% of patients in which plasma cell purification was possible, and 75.8% of the total trial population (51).Here we show that using our small amount RNA-sequencing protocol (53) RNA-sequencing was possible in 92.5% of patients and 83.7% of the intention-to-treat ).Depicted are 20 targets with available antibody-drug conjugates (ADC) or antibody-radionuclide-conjugates (ARC), CART, or T-cell bispecific antibodies (TCB) with active trials in multiple myeloma, B-cell malignancies, or solid oncology (in case of MUC1, NY-ESO1, for which expression in a subfraction of myeloma patients has been reported).For each target, exemplary trials are stated (for ease of depiction, non-comprehensive list).
population.Risk-scores summing over prognosis-associated genes as the LfM-HRS introduced here, those translated from GEP (UAMS70, SKY92, RS-score) and proliferation scores (RPI introduced here) delineated validated groups of patients with highly different EFS and OS.RNA-sequencing allows comparable although not statistically better prognostication (Brier-score) compared to standard of care R-ISS (46).The patient population identified as high risk differs to a certain degree between R-ISS vs. RNA-sequencing vs. individual chromosomal aberrations especially for highly proliferative myeloma patients and presence of more than three copies of 1q21, in agreement with previous reports (47 -49, 52, 57, 69-72).R-ISS thus only identifies part of the high-risk population.
Considering potential immune-oncological targets, RNAsequencing allows suggestions for all patients (IOnc-advisor).This first relates evidently to targets not expressed in all myeloma patients, as being either lost in a subfraction of malignant plasma cells, i.e., BAFF-R, CD19, CD20, and CD22, or gained, i.e., NY-ESO1/2, MUC1, and CD30.Secondly, it relates to high inter-patient variation of target expression.Considering targets expressed in all myeloma patients, 10.9 log-fold (FCRH5) differences were found.With reported relation of expression height and response for CD38 (43) and GPRC5D (41) and the general mechanisms of action of immune-oncological compounds, this likely applies to other targets as well.In case of two treatment options of comparable populationbased response rates (assume BCMA-and FCRH5-CART) and lack of other deciding factors, odds of success could thus be potentially increased choosing the higher expressed target.Especially in patients having either limited reserves to tolerate rather aggressive treatment (e.g., CART), or limited coverage by health insurances, and considering that fewer patients receive subsequent lines of treatment (73), exemplified with 1 st (95%), to 2 nd (61%), 3 rd 38%, 4 th (15%) and 5 th line (1%).Furthermore, use of g-secretase inhibitors (e.g., crenigacestat) (44) in case of low BCMAexpression could be suggested.On a population basis, RNAsequencing as presented here allows including myeloma patients with expressed rare targets in basket trials in other indications.The  Targets can be divided in those expressed in all normal bone marrow plasma cells (BMPC) and (almost all, >99%) malignant plasma cells from therapy-requiring multiple myeloma patients (MMC; n=10, i.e., CD38, SLAMF7 [CS1], BCMA, GPRC5D, FCRH5, TACI, CD74, CD44, CD37 and CD79B), those constitutively expressed in all normal plasma cells with expression lost in a subfraction of malignant plasma cells (n=4, i.e., BAFF-R [81.3%, CD19 [57.9%],CD20 [82.8%],CD22 [28.4%]), and targets aberrantly expressed in malignant plasma cells (i.e., not expressed in BMPC) (n=3, i.e., NY-ESO1/2 [12%], MUC1 [12.7%],CD30 [4.9%]).CD70 is expressed in a subfraction of BMPC (50%) with decreasing expression frequency in MMCs (24.5%).Some suggested targets are not expressed (CD1B) or at a very low level in normal and malignant plasma cells (CD25 [2.6%]).Given are median expression in normal and malignant plasma cells, %age of patients expressing the respective gene, and standard deviation (SDV) within the respective population, i.e., BMPC or MMC.Note different expression height, e.g., detectable but low CD20 median expression.For graphical depiction of expression, see Figure 1.For expression in different MGUS and myeloma stages (asymptomatic myeloma, therapy-requiring myeloma, and relapsed myeloma), see Supplementary Table S3.For validation of target expression in the independent CoMMpass-cohort, see Supplementary Table S4.
BRAF-V600E/K mutation, the best documented small molecule target in myeloma, was present in 2.1% of patients in our cohort, in agreement with previous reports (29).It exemplifies the identification of mutated transcripts by RNA-sequencing, easily extended to other targets once additional clinical evidence emerges.
A further emerging use of RNA-sequencing is assessment of antigen downregulation or loss under targeted treatment, as reported, e.g., for CD38 (43), GPRC5D (45,74), or BCMA (45,75).Both downregulation as well as coding mutations can be detected by RNA-sequencing as exemplified for BCMA.For GPRC5D, Mailankody et al. (74) showed for 6/6 patients progressing after CART (MCARH109) GPRC5D-downregulation (2/6) or loss of expression (4/6).Lee et al. (45) showed 4/6 patients progressing under GPRC5D-TCB to harbor biallelic mutations abrogating compound efficacy.Loss of BCMA-expression after anti-BCMA CART was initially reported as rare event (3/71; 4%) (75).Subsequent studies by Lee et al. (45) showed that in 8/16 investigated patients progressing under BCMA-directed treatment, biallelic deletions or mutations of the TNFRSF17 (BCMA) locus occurred: in two patients, MM relapse post T-cell bispecific antibody or CART-therapy was driven by BCMA-negative clones harboring focal biallelic deletions at the TNFRSF17 locus at relapse or by selective expansion of pre-existing subclones with biallelic TNFRSF17 loss.In further five relapsing patients, newly detected non-truncating, missense mutations, or in-frame deletions in the extracellular domain of BCMA negated the efficacies of anti-BCMA T-cell bispecific antibody therapies, despite detectable surface BCMA protein expression.Of specific interest, for four BCMA mutational events, distinct sensitivities toward different anti-BCMA-targeting therapies could be found: first, a p.Arg27Pro mutation conferred resistance against teclistamab and elrantamab, abrogating binding and activity.Here, BCMA Arg27 interacts with the complementarity-determining regions of the heavy chain of the anti-BCMA variable region of teclistamab.In contrast, binding and activity of alnuctamab or Ide-cel-analogous CART is maintained.Secondly, a p.Pro34del in-frame deletions conveyed resistance Targets expressed in previously untreated myeloma vary in expression in relapsed disease.Assessment in 63 patients.Each row depicts an individual patient assessed longitudinally at treatment initiation and relapse.Targets expressed in normal bone marrow plasma cells and multiple myeloma remain stable in longitudinal samples, especially if highly expressed (CD38, CS1, BCMA, FCRH5, CD74, CD79B).Genes for which expression is stagedependently lost (e.g., CD19, CD22, BAFF-R) show high dynamics with changes occurring in 22/63 (35%), 21/63 (33%), 18/63 (19%) with comparable probability of gain and loss of expression.The cancer testis antigen expression regarding MUC1 [3 losses (red color) vs. 10 gains (light green color)] and NY-ESO1/2 and NY-ESO1/2 (2 losses vs. 15 gains) is predominantly gained in relapsed disease.Color code: dark green color, presence of expression in previously untreated myeloma and relapsed myeloma; light green, expression gained in relapsed myeloma; redexpression lost in relapsed myeloma; greyno expression in both previously untreated myeloma and relapsed myeloma.See Supplementary Tables S5, S6 for numerical depiction.
against teclistamab and elrantamab but maintained alnuctamab binding and activity, and third, a p.Ser30del to teclistamab but retained sensitivity to elranatamab and alnuctamab (45).As relapse under one BCMA-targeting agent does therefore not necessarily implicate resistance against others, and RNA-sequencing can easily identify these mutations, it can be used to guide subsequent treatment lines with different BCMAtargeting agents.As downregulations and mutations discussed here occur primarily under selection pressure of the respective treatment and, in part, different aberrations occur in a subclonal manner (45), with the lack of this evolutionary pressure, and on the background of clonal heterogeneity of myeloma, other subclones can grow out after subsequent lines of treatment, conveying, e.g., again expression of CD38 or GPRC5D, which can, in the same way, be identified by RNA-sequencing.

Longitudinal samples
Longitudinal samples showed dynamic changes of expression between diagnosis and relapse in 88.9% of patients for any of the investigated antigens, with both losses and gains occurring.These changes refer less to those targets likewise expressed in normal bone marrow plasma cells, especially those highly expressed (CD38, CS1, BCMA, FCRH5, CD74, CD79B).Genes with stage-dependent loss of expression, like CD19, CD22, and BAFF-R, showed high dynamics with both gains and losses of expression occurring.The cancer testis antigens MUC1 and NY-ESO1/2 were predominantly gained in MMR.These findings are in line with a subclonal architecture in myeloma leading to clonal tides (76) and spatiotemporal evolution between diagnosis and relapse (77).Molecular assessment should thus be repeated in relapse, especially for targets expressed with high dynamics, or if specific treatment has been applied, as e.g., CD38 or BCMA-directed therapy, leading to a selection pressure regarding target downregulation or loss.

Implementation of RNA-sequencing in standard work-up of myeloma patients
Introduced in myeloma research in 2002 (78) and 2011 (79), GEP and NGS revolutionized our understanding of myeloma biology, pathogenesis, and risk (80, 81) but the standard myeloma-workup is still based on morphological bone marrow assessment and iFISH.Why is this the case?
Several reasons can be identified.In particular, a knowledge gap between routine clinical care and the field of molecular profiling: GEP (23) and NGS (80) can be perceived as slow, complex, expensive, and not broadly applicable techniques that return results hard to interpret and reproduce, and with little clinical value.Of these, "practical issues" can be easily disproven: GEP can be applied in clinical routine in academic [e.g., GEP-R (82), UAMS70-score (48), IFM-score (49)] and commercial settings [e.g., MyPRS ® , Signal Genetics ™ (83), MMprofiler ™ , SkylineDiagnostics (72)] in most patients (51) within four weeks (51).NGS-based techniques, e.g., for mutational profiling or sequencing based FISH, can be performed in academic (CoMMpass) (84) or routine private laboratory setting (85), even within 14 days in a tertiary hospital (86).RNA-sequencing can be used in academic (39,58,84) or private laboratory settings (87) in over 90% of patients in clinical trials or routine within four weeks.But for rare circumstances, myeloma treatment is not an emergency, and a time interval of four weeks can be covered with a short course of steroids while waiting for test results (80).GEP or NGS-based sequencing are not expensive: cost in the range of 1000 US$ are comparable to iFISH (depending on the number of probe sets used) and frequently less than ten-fold compared with monthly treatment costs.
Are results then "hard to interpret and reproduce" and of "little clinical value"?Clinical value is given by risk assessment for patient counselling and respective trial-inclusion and targets selection for individualized treatment in a context of multiple equalseeming options.
But first, there is no consensus as to whether, and how, use these techniques to re-define risk.For GEP, a variety of prognostic scores (48,49,82) identifies partially overlapping patient populations and depend, to an extent, on the applied methodology (e.g., Affymetrix single vs. double amplification protocol).RNA-sequencing can apply translated GEP-based-or de novo generated scores (88) and sequencing-based R-ISS can be used (84).Mutational signatures can be prognostic (89, 90).Even if numerically superior, at the end of the day, GEP or NGS-based risk assessment are to be perceived, at large, as not better than R-ISS, not standardized and thus not warranted in routine application.This might change with NGSbased re-defining of adverse risk factors like the t(4;14) translocation depending on the breakpoint within the NSD2 gene (91).Secondly, although "risk" is part of treatment decisions, e.g., in the Mayo clinic's mSMART-stratification (www.msmart.org)(92), suggesting e.g., bortezomib maintenance for patients harboring t(4;14), del17p, t(14;20), t(14;16) (or transplant eligible additionally 1q-gain and double/triple hit multiple myeloma), the GMMG suggesting bortezomib maintenance after HDT and ASCT for previously untreated myeloma patients harboring a del17p13 or t(4;14), the GMMG-CONCEPT-trial (iFISH-based), the UAMS total therapy program (GEP-based) (93) or the MUKine-OPTIMUM trial (GEP/iFISH-based) (72).But there is still hesitation to apply risk-based approaches, because although low risk-patients might be spared "unnecessarily more effective" treatment and costs saved, for no treatment it is currently shown to works better in high risk compared to low-risk patients, and thus would be applicable for all patients.
Immediate usefulness would be perceived if either response could be predicted, or targets selected in for personalized treatment.Despite factors associated with response have been identified (24,25), it has not been possible to predict response to non-targeted small molecules at a clinically applicable level (22,23).In contrast, targetable mutations can be identified as exemplified by the BRAF V600E/K mutation (vemurafenib) (29), present in 2.1% of patients in our cohort, in agreement with previous reports (29).Currently, several clinical trials, e.g., "A Study to Evaluate Myeloma-Developing Regimens Using Genomics (MyDRUG, NCT02884102)" for patients with ≥30% mutation of CDKN2C, FGFR3, KRAS, NRAS, BRAF V600E, IDH2, or translocation t (11;14), and "Targeted Therapy Directed by Genetic Testing in Treating Patients With Advanced Refractory Solid Tumors, Lymphomas, or Multiple Myeloma (The MATCH Screening Trial, NCT02465060)" address this question.Although in principle GEP would have been applicable for target selection, attempts failed due to lack of compounds usable for personalized treatment, and suggested compounds like inhibitors of aurora kinase (57,82) or IGF1R (82,94) never made it to approval in myeloma.The situation is very different now with compounds available for personalized treatment both targeting immune-oncological targets and mutations which can be identified by NGS-techniques as RNA-sequencing.In the latter case, comprising change of BCMAtargeting treatment to, e.g., different bispecific antibodies in case of specific mutations.Clinical usefulness is thus now a priori evident.
GEP/NGS-based approaches would be significantly fostered by the use of appropriated clinical trial designs, especially for regulatory and approval purposes (81, 95) especially when considering the number of compounds and combinations.Traditional designs like phase III randomization of NGS (RNAsequencing) guided vs. investigators choice will be very difficult to implement, as, based on lack of governmental funding, pharmaceutical companies would need to be willing to provide their compounds without necessarily aligning business interests, which is, based on experience in IIT-trial design, very unlikely.NGS-based approaches are, however, implemented as part of tumor boards (96) as institutionalized framework for decision making and consecutively enabling refunding of treatments outside their specific indication by health insurances."Educated first choice" of treatment in case of lack of other guidance as part of personalized risk benefit assessment as suggested here is a further complementary possibility.Either way, precision oncology represents an epochal revolution in patients' management, and therefore it is conceivable that it involves substantial changes (at both cultural and practical levels) in the way we operate in order to cure cancer, that surely will need a long time to be realized.The concerted effort of all stakeholders involved in the development of precision oncology (researchers, clinicians, regulatory agencies, governments) is now mandatory to ensure that in the future it will become a reality in routine clinical practice (81).

Conclusion
RNA-sequencing is applicable in 90% of patients (comparable to iFISH), is of overall equal predictive power as the current gold standard R-ISS, but identifies a further fraction of myeloma patients, e.g., with highly proliferative myeloma cells.It allows personalized target identification for immune-oncological drugs based on presence and height of expression.RNA-sequencing used for "educated first guess" could be considered as "IOnc drug and risk advisor" in analogy to other decision tools.

FIGURE 1
FIGURE 1 FIGURE 2 RNA-sequencing based determination of risk.(A) De novo generated RNA-sequencing-based scores for risk (LfM-HRS) delineates 3 groups with significantly different overall (OS) (A1) and event-free (EFS) (A2) survival.(B-D) "GEP"-scores translated into RNA-sequencing.The scores of the Universities of Heidelberg and Montpellier (RS-score) (B), the University of Arkansas Medical School (UAMS70) (C), and the Erasmus Medical Center (SKY92), (D) in each case delineate symptomatic myeloma patients with significantly different EFS and OS.(E).The current clinical gold standard (revised ISS-score) delineates three groups of 30%, 56% and 14% of 535 patients with significantly different OS (E1) and EFS (E2).Depicted are Kaplan Maier curves with log-rank based P-value and patients at risk.P-values were adjusted for multiple testing using Benjamini-Hochberg correction.For validation of RNA-sequencing based scores on the independent CoMMpass-cohort, see SupplementaryFigureS5.

FIGURE 6
FIGURE 6 Determination of risk.Comparison of patients identified by RNA-sequencing scores, proliferation, R-ISS, and cytogenetic risk factors in 535 consecutive previously untreated myeloma patients.Percentage of patients identified as high risk and presence of t(4;14), del17p, 1q21 (>3 copies) or t(14;16) is depicted at the top of the figure and plotted in dark red color.Light red color delineates medium risk or presence of three copies of 1q21, green color low risk and/or absence of the respective aberrations.Grey color depicts missing values.Percentage of patients identified as high risk calculated excluding missing values.

TABLE 1
List of potentially actionable targets.