General MM

The gene signature of MM patients can predict response to IMiD treatment

With an increased number of therapeutic options available for Multiple Myeloma (MM) patients, clinicians are faced with the difficult task of choosing which will be the most effective for a given patient. Immunomodulatory drugs (IMiDs) are routinely used for MM treatment and form the backbone of several regimens. In order to establish criteria for the use of IMiDs, in terms of identifying patients that will be either sensitive or resistant, a study was conducted to develop and identify a gene-expression profile (GEP) signature for patient response to IMiD therapy. This retrospective study, conducted by Manisha Bhutani from The Carolinas HealthCare System, Charlotte, NC, USA, and colleagues, and published in the September edition of The Lancet Haematology, used a cohort of patients with either newly diagnosed Multiple Myeloma (NDMM) or relapsed and refractory MM (RRMM), who had been treated in clinical trials with IMiD-containing regimens, and for whom GEP data was publically available.

Key Highlights:
  • A training cohort of patients from the thalidomide group of the Total Therapy (TT) 2 trial (thalidomide induction and consolidation with autologous stem cell transplant (ASCT), and maintenance) was used to establish the model
  • For validation, 4 independent MM datasets were used: TT3a trial (thalidomide in induction, consolidation after tandem ASCT, and thalidomide maintenance), TT3b trial (thalidomide in induction and consolidation with tandem ASCT, and lenalidomide maintenance), TT6 trial (thalidomide in induction, lenalidomide in consolidation with tandem ASCT, lenalidomide in maintenance), and vincristine, doxorubicin, and dexamethasone (VAD) group of the HOVON65/GMMG-HD4 trial (thalidomide maintenance)
  • Using a paired significance analysis of microarrays (with a p value cut-off of 0·05), 176 genes were identified with expression levels that changed significantly relative to baseline, 48 h after IMiD test dose
  • Out of these genes, 14 had p values <0·05 in univariate Cox regression analysis of progression-free survival (PFS) in the thalidomide group of the TT2 trial
  • These 14 genes were combined to create an IMiD-based risk score (IMiD-14): defined as the average log2-scale differential expression of the 4 prognosis-unfavourable genes (HR >1) and the 10 prognosis-favourable genes (HR <1)
  • Genes identified as having prognostic significance (Increased = I or Decreased = D): Prognosis-unfavourable genes:  XPO1 (I), DDR2 (I), TRAF3IP3 (I) and FAIM3 (D) ; Prognosis-favourable genes: IL5RA (D), TNFRSF7 (D), AMPD1 (D), ENO2 (D), ITGA6 (D), FLJ22531 (D), LAMA5 (D), PGRMC2 (D),  SLC39A14 (I), KIAA0247 (I)
  • An IMiD-14 score higher than –1·075 = resistant disease; lower than –1·075 = IMiD-sensitive disease
  • Training set (n= 175 pts): IMiD-14 high score = 83 pts; IMiD-14 low score = 92 pts

Data is given as IMiD-14 high group vs IMiD-14 low group:

  • PFS (3 year):  52% (95% CI 42–64) vs 85% (78–92); HR = 2·51 (95% CI 1·72–3·66; p<0·0001)
  • Validation cohorts:
    • TT3a = 115 pts vs 160 pts; 3 year PFS = 63% (95% CI 55–73) vs 87% (82–92); HR = 1·54 (1·11–2·15), p = 0·010
    • TT3b = 77 pts vs 89 pts; 62% (52–74) vs 80% (72–89); HR = 2·07 (1·28–3·34), p = 0·0024
    • TT6 = 20 pts vs 36 pts; 39% (22–68) vs 74% [61–90]; HR = 2·40 (1·09–5·30), p=0·026
    • VAD group of HOVON65/GMMG-HD4 = 65 pts vs 77 pts; 16% (9–28) vs 54% (44–67); HR = 2·29 (1·52–3·45), p<0·0001
  • Top 5 pathways most affected by IMiDs: nuclear factor of activated T cell (NFAT) pathway, phospholipase C signalling, interferon signalling, phosphoinositide 3-kinase (PI3K) signalling in B lymphocytes, and integrin signalling; significant enrichment for genes of the NFAT signalling pathway was also observed  
  • Gene signatures from the IMiD-14 and GEP80 models (a previously described bortezomib-response gene model) were combined and used to divide patients from the TT3a, TT3b, and TT6 trials (regimens that included both an IMiD and a proteasome) into 4 different risk subgroups: IMiD-14 low and GEP80 low, IMiD-14 high and GEP80 low, IMiD-14 low and GEP80 high, and IMiD-14 high and GEP80 high
  • Pts with the IMiD-14 high and GEP80 high signature had the worst outcomes
  • Chromosome 1q has 3 out of 4 prognosis-unfavourable genes in the IMiD-14 model, but prognostic capacity did not appear to be associated with increased 1q copy number
  • High-risk defined by the IMiD-14 model correlated with other high-risk indicators, such as ISS stage III
  • The IMiD-14 model had higher prognostic performance (across all datasets) when compared with the expression of IMiD-related biomarkers: cereblon, Ikaros, or Aiolos, which overall yielded lower AUCs (although some similarities between AUCs using certain datasets were found)

Using data GEP-data from patients treated with IMiD-based regimens, the IMiD-14 model was developed and extensively validated. This now provides a valuable tool for risk stratification in order to identify patients that will be best suited to IMiD-based regimens or to follow patients during treatment. A high IMiD-14 score will indicate poor PFS and OS, compared with patients with low IMiD-14 scores that have a more favourable prognosis. A question mark over the study is whether it is truly predictive of an IMiD-related response, as patients evaluated received other drugs in their treatments, such as dexamethasone, doxorubicin and high-dose melphalan. The next step will be a functional validation of these gene changes in order to better understand the factors leading to IMiD response versus resistance.

In a Comment by Sinto Sebastian, from the Division of Hematology and Oncology, Mayo Clinic, Scottsdale, Arizona, USA, and also published in The Lancet Haematology, the benefits of such a gene profiling signature were discussed. He proposed that future tests might combine cereblon, Ikaros and Aiolos immunohistochemistry with IMiD-14 gene expression to improve predictive accuracy, as well as future studies to dissect the biological significance of the observed changes.

  1. Bhutani M. et al. Investigation of a gene signature to predict response to immunomodulatory derivatives for patients with multiple myeloma: an exploratory, retrospective study using microarray datasets from prospective clinical trials. Lancet Haematol. 2017 Sep;4(9):e443-e451.  DOI: 10.1016/S2352-3026(17)30143-6
  2. Sinto Sebastian. IMiD-induced gene expression profiling predicts multiple myeloma prognosis. Lancet Haematol. DOI: