Phylogenetic reconstruction reveals early divergence and late expansion of focal lesions in multiple myeloma

Abstract

Abstract Introduction: During the precursor stages of multiple myeloma (MM), clonal plasma cells typically infiltrate the bone marrow (BM) in a diffuse pattern. In contrast, symptomatic MM is frequently marked by the emergence of focal lesions (FLs), which form as discrete nodular aggregates that can penetrate cortical bone and spread into adjacent soft tissue. While epidemiological and genomic evidence (e.g., clock-like mutational signatures) suggests that MM may be initiated decades before diagnosis, the timing of events driving spatial divergence and the formation of FLs, which are hotspots of subclonal evolution, remains unclear. This gap in knowledge is clinically relevant, as (1) current trials in high-risk precursor disease assume limited subclonal evolution, and (2) FLs often appear unexpectedly in patients with otherwise stable serum and BM parameters. To address this, we performed multi-region whole-genome sequencing (WGS) and applied mutational timing approaches in newly diagnosed MM (NDMM) patients. Methods:We analyzed 27 NDMM patients with paired iliac crest and FL samples. A total of 72 WGS datasets were processed using in-house pipelines. Subclones were identified using MOBSTER, VIBER, and Clonevol. Mutational signatures and clock-like SBS1/5 mutations were quantified with MMSig. Linear mixed-effects (LME) models were applied to estimate patient-specific mutation rates and time evolutionary events. Single-cell RNA-seq (scRNAseq) was performed using the 10× Genomics platform. Results:Phylogenetic trees revealed a median of 3 (range: 2–7) subclones per patient. In 21 of 27 pairs, distinct dominant subclones were identified at the respective sites. FL subclones harbored, on average, 390 additional SBS1/5 mutations compared to major BM subclones, representing a 28% increase (p=0.002). This likely reflects recent clonal expansions in FLs, while dominant subclones in the BM may have persisted for years. According to this interpretation, MM cells in the BM continued to accumulate mutations over time, but new subclones did not outcompete the previously dominant subclone and therefore did not expand. Supporting this, longitudinal WGS from a patient with precursor disease showed that the subclonal architecture in the BM remained stable over several years, with new mutations confined to minor subclones. In contrast, an FL that emerged at MM diagnosis in this patient displayed a dominant, genomically advanced profile, indicating recent selection and expansion at that site. Timing the most recent common ancestor (MRCA) of BM and FL branches revealed a median lag of 14 years (range: 0–41), indicating that regional clonal divergence is initiated early in most patients and may remain confined to single sites for years and provide the basis for later FL formation. As this implies that additional events are required to trigger focal outgrowth, we analyzed 10 patients with high-resolution data from scRNAseq-informed and/or dissection-based FL WGS for more detailed timing. These analyses dated the emergence of FL-expanded subclones to within only weeks or months before diagnosis. Notably, this included a patient with three distinct RAS-mutant subclones and another with a newly emerged TP53-mutant subclone. Extending these findings, RAS mutations were found in recent subclones (n=10, including 7 in newly emerging FLs) as well as ancestral branches (15–20 years pre-diagnosis, n=8). Del(1p) and del(17p) were predominantly subclonal, whereas gain(1q) was part of the phylogenetic trunk in 10 of 13 evaluable patients, suggesting that this aberration may persist silently for decades and may not represent the key event driving progression in these patients. Conclusions: Our data indicate that spatial clonal divergence in MM begins more than a decade before diagnosis, yet the emergence of FLs reflects a final wave of subclonal selection shortly before progression. These findings challenge the assumption of less advanced subclones in precursor stages and offer a genomic explanation for the abrupt clinical appearance of FLs. Moreover, the long persistence of gain(1q) in precursor states highlights the need for a deeper understanding of its contribution to early pathogenesis and poor prognosis.