Multiple myeloma immunoglobulin λ translocations portend poor prognosis

Multiple myeloma is a malignancy of antibody-secreting plasma cells. Most patients benefit from current therapies, however, 20% of patients relapse or die within two years and are deemed ‘high-risk’. To better understand and identify high-risk myeloma, we analyzed the translocation landscape of 826 newly-diagnosed patients by whole genome sequencing as part of the CoMMpass study. Translocations at the IgL locus were present in 10% of myeloma patients, and corresponded with poor prognosis. Importantly, 70% of IgL translocations co-occurred with hyperdiploid disease, a marker of standard risk, which is routinely diagnosed clinically whereas IgL-translocations are not. Thus, it is likely that the majority of IgL-translocated myeloma is being misclassified. The IgL enhancer is among the strongest in myeloma cells, indicating it can robustly drive oncogene expression when translocated. Consistent with this, IgL-translocated patients failed to benefit from immunomodulatory imide drugs (IMiDs), which target the lymphocyte-specific transcription factor Ikaros. These data implicate the IgL enhancer as resistant to IMiD-inhibition, and when translocated, as a driver of poor prognosis.


37
Multiple myeloma is the second most common hematological cancer, which affects terminally 38 differentiated antibody secreting B cells, known as plasma cells, and results in hypercalcemia, anemia, 39 renal failure, and lytic bone lesions. Over the past decade, there have been significant Improvements in 40 survival due to current therapies, which include autologous stem cell transplant 1 , proteasome inhibitors 41 2 , the immunomodulatory imide drugs (IMiDs), thalidomide 3 , lenalidomide 4,5 , and pomalidomide, and 42 more recently monoclonal antibodies 6,7 . Despite these advances, approximately 20% of patients relapse 43 or die within two years of diagnosis 8,9 . These patients are referred to as 'high-risk'. Understanding how 44 to identify and treat these patients as well as the mechanisms underlying the biology of high-risk myeloma 45 is critical for improving outcomes. and/or treatment promote the outgrowth of sub-clones harboring specific mutations that confer a survival 58 advantage 15 . Fortunately, modern combination therapies are mostly effective despite disease 59 heterogeneity, with the majority of patients responding to frontline treatments that target plasma cell 60 biology rather than specific genetic lesions 16 . 61 IgK [t(8;2); N=16], FOXO3 (N=8), and FAM46C (N=6). Unlike IgH, which was translocated primarily to a 136 few specific loci, a larger proportion of MYC translocations (N=73; 39.2%) occurred throughout the 137 genome. Also, unlike IgH translocations, MYC translocation breakpoints were clustered across two broad 138 regions, one centered on MYC and the other 600 kb to the telomeric side of MYC (Fig. 3b). MYC 139 translocations occurred near regions enriched for H3K27ac, a histone modification present at active 140 enhancers, in the myeloma cell line MM.1S (Fig. 3b). Analysis of copy-number alterations (CNA) across 141 the MYC locus identified focal amplifications that commonly occurred near the most frequently 142 translocated regions, which were more often observed among myelomas that contained a MYC 143 translocation (Fig. 3b). Indeed, analysis of MYC CNV data derived from exome sequencing in 754 144 samples, confirmed that MYC translocations corresponded with copy number gains in the region, and 145 showed that among the 118 (15.6%) myelomas containing MYC amplification, 50% also had a MYC 146 translocation (Fig. 3c). This is in contrast to non-MYC amplified myeloma which only contained MYC 147 translocations in 17.8% of cases (Fig. 3c, see inset table). Both genomic alterations resulted in increased 148 MYC expression, and there was no difference in expression between MYC amplified myelomas and those 149 with different types of MYC translocation (Fig. 3d). Nor was there a difference in expression between 150 MYC translocations that occurred proximal or distal to MYC (data not shown). Interestingly, despite 151 similar levels of MYC expression, distinct MYC-translocation partners exhibited differences in outcome. 152 Indeed, among MYC translocations, MYC-IgL t(8;22) translocations had the worst PFS and OS (Fig. 3e). 153 These data indicate that aberrant MYC expression resulting from MYC amplification or translocation is a 154 common feature of myeloma, but the MYC-IgL translocated subset is unique among MYC alterations in 155 that it portends a poor prognosis. 156

Translocation of IgL is an independent marker of poor prognosis 157
The above findings prompted us to examine IgL translocations in more detail. IgL was translocated in 158 9.8% (N=81/826) of newly diagnosed myeloma with 40.7% of IgL translocations being juxtaposed to MYC 159 and the remaining were scattered throughout the genome, but included recurrent translocations proximal 160 to MAP3K14, CD40, MAFB, TXNDC5, CCND1, CCND2, and CCND3 (Fig. 4a). As suggested by the 161 MYC analysis, t(IgL) patients experienced a worse PFS and OS as compared to non-t(IgL) myeloma ( Fig. 4b). This was further confirmed through permutation analysis and was not attributable to differences 163 in patient population as t(IgL) patients were of similar age, disease stage, sex, race, had similar levels of 164 serum M-protein and β2-Microglobulin, and were treated with similar therapies as compared to patients 165 without t(IgL) (Fig. S2a-g). Multivariate survival analysis was performed to identify any potentially 166 confounding factors, and t(IgL) remained a significant marker of poor outcome when considering other 167 known prognostic variables including age and stage (ISS) (Fig. S2h). Additionally, survival of patient's 168 with t(IgL) was compared to patients with other immunoglobulin translocations (Fig. S3a), which indicated 169 that patient's with t(IgL) had a worse outcome than those with t(IgH) or t(IgK). Finally, patients with IgL-170 MYC translocations and IgL-non-MYC translocation were compared to MYC-non-IgL translocation and 171 non-IgL-non-MYC translocations (Fig. S3b), which showed that patients with IgL translocations had a 172 worse prognosis, regardless of the partner. Taken together, these data identify t(IgL) as an independent 173 marker of poor prognosis regardless of the transposed loci. 174 To determine whether there might be other molecular features that contribute to the poor 175 outcomes of t(IgL) patients, the mutational repertoire of t(IgL) myeloma was interrogated using high-depth 176 exome sequencing on 814 of the 826 specimens characterized by long-insert whole genome sequencing. 177 This analysis indicated that there was no difference in the total and nonsynonymous mutational burden 178 among t(IgL) myeloma compared to those with IgK and IgH translocations or no immunoglobulin 179 translocation ( Fig. S4a,b). The frequency of specific mutations types also indicated that t(IgL) myeloma 180 had a similar mutational spectrum as other myelomas where all myelomas had a higher rate of transitions 181 (C>T and A>G) than transversions (Fig. S4c). Analysis of specific mutations indicated that KRAS and 182 NRAS were the most frequently mutated genes, as previously reported 14 (Fig. S4d). Although t(IgL) 183 myeloma had a slightly higher rate of mutations in RYR1, CSMD1, USH2A, and FLG than other 184 myelomas, none of these were statistically significant, and in general, t(IgL) myeloma contained a similar 185 frequency of mutations in the most commonly mutated genes as compared to all other myelomas. Most 186 mutations had a marginal impact on prognosis as assessed by a univariate analysis (data not shown). 187 Furthermore, a bivariate analysis considering each mutation in combination with t(IgL), identified t(IgL) 188 as prognostic of poor outcome independent of any other common mutation (Fig. S4e, right).

IgL-translocated myeloma is not defined by gene expression subtype 190
To gain insight into the pathogenesis of IgL translocation, the relationship between t(IgL) and gene 191 expression subtypes were determined using consensus clustering 23 on 654 samples for which whole 192 genome sequencing and RNA-seq data were available. Similar to previous reports 24 this identified 7 193 gene expression subtypes, where samples within a given subtype were highly correlated with each other 194 but not with samples from other subtypes (Fig. 5a). Both up-and down-regulated genes in each cluster 195 were determined ( Fig. S5a; Table S2), and annotated using gene set enrichment analysis (GSEA; Table  196 S3) 25  subtype harbored t(4;14), the HY gene expression subtype corresponded with genetic hyperdiploidy, the 204 PR subtype contained amp(1q), the CD subtype had t(11;14), and MF had t(14;16) ( Fig. S5c, Fig. 5a -205 see annotation above). Notably, IgL translocations were found in every expression subtype with a modest 206 but non-significant enrichment in the HY and PR gene expression subtypes and depletion in the CD 207 subtype (Fig. 5b). 208 To identify putative molecular mechanisms contributing to t(IgL) pathogenesis, genes differentially 209 expressed in t(IgL) were determined, revealing 330 upregulated and 186 downregulated genes in t(IgL) 210 myeloma as compared to non-t(IgL) myeloma (Table S4; FDR <0.01). Expression of these genes 211 modestly aggregated t(IgL) myelomas using hierarchical clustering (Fig. 5c), but clearly grouped many 212 non-t(IgL) with t(IgL) myelomas, suggesting that t(IgL) myeloma is not clearly defined by a baseline gene 213 expression signature. GSEA identified several gene sets and pathways corresponding with t(IgL) 214 differentially expressed genes including overexpression of ribosomal genes RPS2 and RPL4 that are 215 involved in protein translation, components of energy metabolism including oxidative phosphorylation 216 and respiratory electron transport chain, such as the NADH ubiquinone oxidoreductase complex 217 component NDUFAF4, and genes upregulated downstream of MYC, including MYC itself, and the 218 eukaryotic translation elongation factor EIF3J (Fig. S6a, top; Table S5). These differentially expressed 219 genes, which while statistically significant, showed only modest differences between t(IgL) from non-220 t(IgL) myeloma (Fig. S6b, top). Likewise, the genes most downregulated in t(IgL) myeloma which 221 included those normally repressed during B cell to plasma cell differentiation as well as genes involved 222 in cytokine and chemokine signaling (Fig. S6a, bottom), showed only subtle differences between t(IgL) 223 and non-t(IgL) myeloma (Fig. S6b, bottom). These data indicate that t(IgL) occurs across all gene 224 expression subtypes of myeloma and indicate that translocation of IgL does not drive a unique gene 225 expression program. 226

IgL translocations co-occur with hyperdiploid disease 227
As t(IgL) was not associated with any specific mutations and had only modest correlations with gene 228 expression, we investigated structural variants that might be associated with t(IgL). The frequency of 229 specific loci translocated directly to IgL as well as those that co-occur with t(IgL) were compared to 230 myeloma with IgH and IgK translocations (Fig. S7a). This indicated that IgH was more frequently 231 translocated to CCND1, WHSC1, and MAF, whereas IgK and IgL were more frequently translocated to 232 MYC. Approximately 40% of both IgK and IgL translocations were to MYC and another 20% of both t(IgK) 233 and t(IgL) myelomas contained a MYC translocation but to a different locus (Fig. S7b). t(IgL) myeloma 234 contained very few unique translocation partners, in that most loci transposed to IgL were also transposed 235 to IgK or IgH. The most common translocations unique to IgL were MAP3K14 and 3q26.2, which only 236 accounted for 7.4% and 4.9% of t(IgL) myeloma, respectively. Thus, these data suggest that the unique 237 pathologic effects of IgL translocation are directly related to the IgL locus and not necessarily to a gene 238 dysregulated by IgL transposition. 239 Other structural variants were also interrogated including deletions, duplications, and inversions, 240 and showed that t(IgL) myeloma had a slightly larger number of duplications than t(IgH) myeloma (Fig.  241 S7c). Thus, common CNAs were annotated for 754 newly diagnosed myelomas for which exome 242 sequencing and whole genome sequencing data were available. A heatmap of CNAs identified a 243 bifurcation in samples, where approximately half (N=382; 50.7%) showed a characteristic hyperdiploid 244 pattern, exhibiting aneuploidy of odd numbered chromosomes including 3, 5, 7, 9, 11, 15, 17, and 19 245 ( Fig. 6a). Hyperdiploid myelomas were mostly mutually exclusive with IgH translocations t(11;14), t(4;14), 246 and t(14;16) as previously reported 27 . t(IgL) was found in both groups but was more common in 247 hyperdiploid disease (Fig. 6a). Indeed, del(17p), del(1p), amp(1q), del(13q) were found at similar 248 frequencies in t(IgL) as in all other myelomas, but hyperdiploid disease was found in 70.4% of t(IgL) 249 myeloma, which was a significant overrepresentation (Fig. 6b). Multivariate survival analysis of t(IgL) and 250 the aforementioned CNAs indicated that t(IgL) was prognostic of poor PFS and OS even when accounting 251 for all the other common CNAs, some of which have themselves been independently associated with a 252 poor prognosis 28 (Fig. 6c). Notably, patients with hyperdiploid disease generally experience better 253 outcomes 28 , but this was not the case for patients with t(IgL) and hyperdiploid disease, who had 254 significantly worse outcomes, as compared to other patients with hyperdiploid disease or other myeloma 255 (Fig. 6d). These data identify IgL translocation as a poor prognostic prevalent in 14.9% of hyperdiploid 256 myeloma that would otherwise be classified as standard-risk. Thus, a large portion of t(IgL) myeloma is 257 likely being misclassified. 258 The above multivariate analysis also indicated amp(1q) as an independent marker of poor 259 prognosis. Indeed, amp(1q) was associated with worse PFS and OS as compared to non-amp(1q) 260 myeloma (Fig. 6f), similar to previous reports 28 . Interestingly, the poor prognosis associated with 261 amp(1q) was further exacerbated by the concurrent presence of t(IgL), which resulted in a median PFS 262 and OS of 1.37 and 2.23 years, respectively (Fig. 6f). This suggests that the pathogenic effects of t(IgL) 263 and amp(1q) have additive effects on myeloma outcomes. 264 The IgL locus is a super-enhancer 265 The above data indicate that t(IgL) is an independent marker of poor prognosis regardless of translocation 266 partner or other known molecular features, which suggests a factor intrinsic to the IgL locus may be 267 mediating this pathology. One possibility is that the IgL enhancer is particularly potent in driving oncogene 268 expression. Thus, we analyzed the chromatin structure of the IgL locus using ChIP-seq data in the 269 myeloma cell line MM.1S 29 . These data indicated that the IgL locus contained several regions with the 270 activating histone modification histone 3 lysine 27 acetylation (H3K27ac), and these coincided with 271 MED1, BRD4, and MYC occupancy (Fig. S8a). Although a similar 200 kb region at the 3' end of the IgH 272 locus also contained these marks, the IgL enhancer appeared to be more densely clustered than that of 273 the IgH (Fig. S8b). Analysis of H3K27ac-or MED1-defined 'super enhancers' indicated that multiple 274 enhancers at the IgL locus were among the biggest in the genome (Fig. S8c). Indeed IgL super-275 enhancers were larger than those of the IgH locus. These data indicate that the IgL locus contains 276 several super-enhancers that when translocated likely serve as potent inducers of juxtaposed 277 oncogenes. 278

t(IgL) patients do not benefit from treatment with IMiDs 279
Given that t(IgL) patients experienced worse outcomes than other patients we hypothesized that this may 280 be due to certain therapies being ineffective against t(IgL) myeloma. Treatment of myeloma in the 281 (33.5%), and cyclophosphamide (30.1%) (Fig. S2). We tested the outcome of t(IgL) patients in the 284 context of these agents and this indicated that t(IgL) patients showed a similar PFS and OS regardless 285 of IMiD treatment (Fig. 7a, compare blue and red lines). This is in stark contrast to non-t(IgL) patients 286 who derive clear benefit from treatment with IMiDs (Fig. 7a, compare gray and green lines). Indeed, a 287 significant survival benefit was provided by IMiDs for non-t(IgL) patients whereas no IMiD survival benefit 288 was realized for patients with t(IgL) who had a PFS and OS similar to patients who did not receive an 289 IMiD (Fig. 7a, see table right). In contrast, patients with IgH translocations who received IMiDs did 290 significantly better than t(IgH) patients that did not (Fig. 7a). Importantly, bootstrap sampling of t(IgH) 291 patients based on the number of t(IgL) patients indicated that t(IgH) patients who received IMiDs had a 292 better progression-free survival than t(IgL) patients (PFS P=0.045; OS P=0.163). Thus, the lack of IMiD 293 benefit observed in t(IgL) patients is not simply due to the smaller number of t(IgL) patients. Together, 294 these data indicate that IMiDs are less affective against t(IgL) myeloma than other myelomas. 295

Discussion 296
Here we present the first comprehensive catalogue of translocations in newly diagnosed multiple 297 myeloma using whole genome sequencing on 826 tumor specimens with matched germline controls. 298 Importantly, these results are placed in the context of clinical covariates and outcome, which allows for 299 the identification of high-risk genetic alterations. This analysis identified 19 translocation hotspots present 300 in ≥2% of the population and 8 recurrent translocations. As expected, the most common translocated 301 region was IgH, which was primarily juxtaposed to known myeloma oncogenes CCND1, WHSC1, MYC, 302 and MAF. Sequencing analysis provided high resolution of breakpoints and indicated that IgH Several studies have reported IgL translocations 12,33-36 , but to our knowledge this is the first study 327 in such a large cohort (N=826) of newly diagnosed myeloma and to report a frequency approaching 10%. 328 This high frequency of IgL translocations was somewhat surprising given that 65% of myelomas express 329 IgK, which was only translocated in 4.3% of patients. However, studies in murine B cells, 95% of which 330 express IgK, found equivalent numbers of IgL and IgK translocations after acute activation with minimal 331 time for oncogenic selection 37 , suggesting that IgL is prone to translocation regardless of expression. 332 Interestingly, the RNA-seq data here indicated that t(IgL) myeloma had the expected ratio of light chain 333 expression with two-thirds of t(IgL) myeloma expressing IgK. This somewhat surprising result is 334 consistent with the IgL enhancer being active even in IgK-expressing myeloma, which is supported by 335 H3K27ac ChIP-seq data in IgK-expressing KMS11 and IgL-expressing MM.1S (Fig. S9a,b). However, 336 100% of IgK translocated myeloma expressed IgK (Fig. S9c). This observation can be explained by the 337 fact that at least 75% of IgL-expressing B cells somatically delete the IgK constant region including the 338 IgK 3' enhancer 38 , thus minimizing the potential for translocation and/or subsequent oncogenic 339 propagation. Indeed, copy number data derived from the whole genome sequencing here clearly 340 indicated that the majority of IgL-expressing myelomas had somatically deleted the IgK 3' enhancer ( Fig.  341

S9d). Conversely, the IgL enhancer, which is one of the strongest enhancers in B cells, plasma cells, 342
and myeloma, appears to be active regardless of whether or not a productive IgL is expressed as 343 H3K27ac is found at IgL in IgK-expressing myeloma cell lines. Indeed, 'super-enhancers' are particularly 344 prone to AID-induced genomic instability and translocation 39 and the IgL enhancer has been reported to 345 be the strongest super-enhancer in myeloma 40 . Subsequently, the IgL enhancer represents an 346 unchecked and potent inducer of gene expression, that when transposed near an oncogene, robustly 347 drives oncogenesis.
Although IgL translocations did not coincide with other small insertions/deletions, mutations, or 349 patient clinical characteristics, t(IgL) was more common in hyperdiploid disease. The poor prognosis of 350 t(IgL) is independent of hyperdiploidy and other major copy number alterations, which is exemplified by 351 the additive risk when concurrent with amp(1q). This indicates that t(IgL) and amp(1q) function in 352 independent pathways. Consequently, patients who harbor both of these high-risk genetic alterations 353 experience an exceedingly poor outcome. Identification of t(IgL) myeloma will require a rapid and reliable 354 diagnostic that can be used for appropriate risk-stratification. This is particularly important as t(IgL) is 355 rarely identified clinically, whereas FISH is routinely performed to identify hyperdiploidy, which 356 corresponds with better prognosis 28 . Resultantly, t(IgL) represents an unrecognized high-risk marker 357 prevalent in a subset of patients that are considered to be standard risk. Resultantly, the majority of t(IgL) 358 are currently being misclassified as standard risk. 359 One indication of how t(IgL) may contribute to outcome is provided by the interaction of t(IgL) and 360 treatment, which indicated that t(IgL) patients did not benefit from IMiDs as much as other translocated 361 myelomas. The therapeutic efficacy of IMiDs may be mediated, in part, by their ability to inhibit expression 362 from IKZF1 regulated enhancers, including those enhancers juxtaposed to oncogenes by translocation. 363 While translocation of the immunoglobulin loci is a common mode of oncogene overexpression in 364 myeloma, the IgL locus is unique in that, regardless of IgL expression, the IgL enhancers appear to be 365 very active. This may render the IgL locus particularly insensitive to IMiD-based inhibition and potential 366 explain why t(IgL) patients do not benefit from IMiDs. However, it will be important to further understand 367 if and how IKZF1 regulates the IgL enhancers in the context of translocation. Recent work has shown the 368 IgL locus to be one of the largest super-enhancers as measured by MED1 or BRD4 occupancy in MM.1S 369 cells 40 , suggesting several distinct transcriptional regulators may be mediating IgL-driven oncogene 370 expression in t(IgL) myeloma. Thus, it will be important to understand the efficacy of emerging 371 transcriptional therapeutics such as BET inhibitors 41 or degraders 42 that abolish BRD4 as well as other 372 small molecule inhibitors that target the transcriptional machinery that drive oncogene expression. The 373 data herein provide motivation for determining the efficacy of such transcriptional regulators in the context 374 of t(IgL) myeloma, but also a rationale for the better understanding of cis-regulatory factors affecting transcription at translocated loci. This daunting task not only requires an understanding of the 376 combinatorial effects of the trans-acting molecular machinery, but also a cartography of the chromatin 377 structure, epigenetic mechanisms known to influence plasma cell fate 43,44 , and enhancer function in the 378 context of translocation breakpoints and micro-environmental ques. Such results will ultimately need to 379 be placed in the context of multicellular organisms and bone marrow micro-environmental ques to help 380 identify drug targets with high specificity for the transcriptional program and genetic architecture of 381 myeloma. 382

CoMMpass data 385
Use of CoMMpass data (interim analysis 11) was approved by the data access use committee and 386 downloaded from dbGaP (phs000748.v6.p4). Summarized data was provided by TGen. 387

Whole genome sequencing 388
Long-insert whole genome library preparation and sequencing was performed by the Translational (Bio-Rad) prior to library amplification. Library size was assessed on a Bioanalyzer (Agilent) and 395 quantified using the Qubit dsDNA HS assay (Invitrogen) and on a TapeStation (Agilent). Paired-end 396 sequencing was performed using a HiSeq2000 or HiSeq2500 (Illumina) with either v3 or v4 chemistry 397 and 86 bp reads. Raw sequencing data was extracted from BCL files using BCL2FASTQ v2.17.1 to 398 extract FASTQ files. 399

Determination of structural variants 410
Structural variants were determined using DELLY (v0.7.6) 22 with the following filter module options: 411 altFrac = 0.1, ratioGeno = 0.75, coverage = 5, controlContamination = 0, minSize = 500, maxSize = 412 500000000. DELLY BCF files were converted to VCF files using BCFTOOLS and VCF files were parsed, 413 formatted, and annotated using custom R scripts available upon request. DELLY translocations were 414 subject to further quality control that included homology searches of 1 kb on either side of the 415 translocation break point between the two transposed regions where translocations with homology of 416 80% or more in any 100 bp window were removed as false positives. Translocations were compared to 417 the ENCODE 50 20bp mappability tracks and those translocations that had an average mappability of less 418 than 20% across 1 kb on either side of the translocation breakpoint were also removed. Finally, 419 translocations were visually inspected and compared to the mapped reads in BAM files resulting in 420 elimination of translocations in certain regions with sequencing anomalies (Table S6). 421

Identification of translocation hotspots 422
Commonly translocated regions were identified using a 1 Mb window incremented 500 kb across each 423 chromosome. Contiguous 1 Mb regions translocated at a frequency of 1% or more were stitched together. 424 immunoglobulin genes were excluded from analysis due to somatic hypermutation. Synonymous and 474 non-synonymous mutations were determined using the 'locateVariants' and 'predictCoding' functions of 475 the 'VariantAnnotation' (v1.24.5) package 60 . Mutational differences between t(IgL) and non-t(IgL) 476 myelomas were assessed using Fisher's exact test with an FDR correction for all mutations present at 477 frequency of ≥4% of the population. 478

Copy number alteration analysis 479
Copy number alterations (CNAs) were determined separately for exome sequencing and whole genome 480 long-insert sequencing using the TGen tool tCoNut (https://github.com/tgen/tCoNuT). Exome-sequencing 481 derived CNAs were used to define large cytogenetic abnormalities including hyperdiploidy, del(1p), 482 amp(1q), del(13q), and del(17p). CNA gains and losses were defined as log2 CNA ratio of myeloma to normal of ≥0.2 and ≤0.2, respectively. The following regions were used to define common myeloma CNAs 484 based on the average CNA segmentation call for the region. 485

Survival analysis 491
Survival analysis was conducted using in R using the 'survival' (v2.41-3) package. Differences in 492 progression-free and overall survival were determined using a cox proportional hazards regression fit to 493 either a continuous (e.g. the number of deletions, duplications, inversions, or translocations) or discrete 494 (e.g. t(IgL) versus other) variable. P-values were calculated using a Wald's test. When more than two 495 discrete variables existed a P-value of differences between all groups was first calculated followed by 496 pairwise comparisons and FDR correction. Hazard ratios associated with translocations, mutations, and 497 other clinical variates were also calculated using a cox proportional hazards regression and 95% 498 confidence intervals are shown. Bivariate analysis was performed for the most common mutations in 499 combination with t(IgL) and multivariate analysis was conducted with clinically relevant parameters and 500 t(IgL) as well as common CNAs and t(IgL). Bootstrapping of outcome was performed using 1,000 501 permutations and comparing the PFS and OS hazard ratios as compared to the actual hazard ratio. The 502 comparison of IMiD survival benefit in t(IgL) versus t(IgH) sampled t(IgH) patients according to the 503 number of t(IgL) patients. 504

ChIP-seq analysis 505
H3K27ac ChIP-seq data generated as part of the ENCODE project 50 for the myeloma cell lines KMS11 506 and MM.1S or by Lin et al. 29 for MM.1S were downloaded from the short read archive and mapped to 507 the same GRCh37 genome used above for CoMMpass data using bowtie 2 (v2.2.6) 61 . Mapped SAM 508 files were converted to BAM files and putative PCR duplicates were marked using SAMtools (v1.7) 48 . 509 H3K27ac and IKZF1 enriched regions were determined using MACS2 (v2.1.0.20151222) 62 using 510 default parameters and a q-value of 0.01. Fragment size was estimated using the R package 'chipseq' 511 (v1.28.0) and reads were extended to the estimated fragment size for visualization using the R package 512 'rtracklayer' (v1.38.3) 63 . Super-enhancer analysis was performed using custom R code in a manner 513 analogous to that done previously 40 . Briefly, this involved stitching together IKZF1-enriched regions 514 that were within 15 kb of each other for enriched regions that did not overlap a 2.5 kb proximal to a 515 GRCh37.74 defined promoter. Regions were then ranked by IKZF1 occupancy measured as reads per 516 million (RPM) and regions that were past the inflection point were considered super-enhancers. 517