Blood, the hidden side of myeloma?

In this issue of Blood, Kogure et al explore the prognostic prediction of circulating tumor DNA (ctDNA) in patients with relapsed/refractory myeloma treated with the same regimen.¹ They show that TP53 and KRAS mutations predicted for a poor prognosis, and proposed a scoring system based on the number of detected mutations.

Currently, there is a cooperative effort to identify patients with high-risk multiple myeloma (MM) to provide the best risk-adapted therapy. Cytogenetics evaluation assessed by fluorescence in situ hybridization and performed on sorted bone marrow (BM) plasma cells is one of the most robust tools to assess risk in myeloma. The abnormalities associated to shorter survival include 17p deletion, 1p32 deletion, 1q gain, and transactions t(4;14) and t(14;16). With the advent of next-generation sequencing (NGS), it is now possible to detect all these abnormalities using a single targeted panel. In addition, recurrent mutations can also be detected, with TP53 being the only widely accepted mutation associated with shorter survival in patients with myeloma.² Despite these efforts, ≈15% of patients without any identified high-risk factor will still relapse early.³ Furthermore, most studies have focused on newly diagnosed patients and not on those at relapse.⁴ 

At the same time, the use of BM samples has been accused of all sort of evils: too invasive of course, but also of not always representative of the disease, because of hemodilution, patchy distribution of MM cells, and spatial heterogeneity. The discovery of the latter, by multiregion sequencing,⁵ let us understand why some patients have a poor outcome despite the lack of high-risk cytogenetic factors in their BM sample. The emerging field of liquid biopsies could theoretically overcome the limitations of BM sampling. Indeed, myeloma can be detected in blood by cell-free ctDNA. Some studies have showed that most of the mutations detected in BM can be found in blood, with some ctDNA-specific mutations, suggesting that ctDNA sequencing recapitulates the spatial heterogeneity detected in the marrow.^6,7 

Kogure et al have explored targeted sequencing of ctDNA in a large cohort of 261 patients with relapsed/refractory myeloma treated with ixazomib, lenalidomide, and dexamethasone. They compared the results with those from targeted sequencing of 161 paired DNA samples from sorted BM plasma cells. Although the results obtained in the BM were expected, those obtained in the blood were less so: TP53 was the most common mutation (21.6% vs 12.9% in the BM), followed by KRAS (10.0% vs 21.5% in the BM). Indeed, among mutations detected only by ctDNA, TP53 was the highest (59.2% of cases), and in approximately a quarter of cases, there were multiple TP53 mutations. All KRAS and NRAS mutations detected in the BM were also detected in blood. Even if some cases of clonal hematopoiesis-related TP53 mutations cannot be excluded, this observation reflects the existence of multiple TP53 mutated subclones in MM, which were not detected in the BM. Importantly, most of these patients did not have extramedullary disease, despite the fact that >40% of patients received at least 3 prior regimens.

When only BM results were examined in multivariate analysis, the presence of del(17p), TP53, and KRAS mutations was significantly associated with shorter progression-free survival (PFS). Indeed, although KRAS mutations have no prognostic impact on newly diagnosed myeloma,⁸ they specifically negatively impact relapsed cases with KRAS mutations detectable in ctDNA (half the cases). Importantly, TP53 mutations only detectable in ctDNA were also associated with poor prognosis. In addition, the number of detected mutations in ctDNA and plasma DNA concentration were also correlated with PFS. The authors had the opportunity to analyze paired samples before and after treatment; TP53 and KRAS mutations were the most frequent emerging mutations in ctDNA. Finally, the authors proposed a prognostic scoring system able to successfully separate patients with relapsed/refractory disease into different risk subgroups, based solely on the number of mutations detected in ctDNA, plasma DNA concentration, and the number of lines of therapy.

This study brings a strong argument in favor of the value of ctDNA as a prognostic tool, detecting some aggressive subclones undetectable in the BM. However, if NGS was performed not at the bulk but at the single-cell level, the question arises whether these subclones would be identified.⁹ In addition, the panel used here was not designed to capture copy number alterations, which have the strongest clinical relevance in MM. The poor prognostic impact of KRAS mutation only detected in ctDNA should be further explored: does it remain true at first relapse only? Does it only reflect tumor aggressiveness of the disease, or is there a real role for KRAS mutations in disseminating clonal heterogeneity? The fact that KRAS mutations are also associated with a shorter time to progression in smoldering myeloma may be in favor of the second option.¹⁰ 

The future of myeloma will undoubtedly lie in liquid biopsy. Nevertheless, as long as myeloma diagnosis requires BM analysis, the latter will remain the gold standard, and liquid biopsy a complementary tool.

Conflict-of-interest disclosure: The author declares no competing financial interests.