The SKY92 prognostic marker is validated in eight multiple myeloma clinical datasets

Van Vliet Duin M., Kuiper R., Ubels J., Dumee B., Bosman L., Van Beers E., De Best L., Waage A., Zweegman S., Morgan G., Levy J., Auclair D., Fonseca R., Goldschmidt H., Van Vliet Duin M., Sonneveld P.

Research output: Contribution to journalArticleAcademicpeer-review


Background: Prognostic markers for Multiple Myeloma (MM) are recommended in most guidelines and consensus papers for clinical use and for risk stratified trials. A prognostic marker is considered suitable for clinical use when it identifies a reasonable fraction of all patients with sufficiently differential outcome (hazard ratio), and is reproducible in multiple independent clinical studies. SKY92 risk stratification was performed in eight independent clinical datasets for which gene expression profiling was available along with overall survival (OS). Three datasets also had progression free survival (PFS), which is an important measure to identify the patient's response to therapy and which is often a relatively early indication of the severity of the disease. Patients with a shorter PFS require care at a higher frequency, possibly a different treatment strategy and likely need higher intensity treatment than patients with longer progression free survival. It would therefore be useful to reliably identify patients that are expected to have a shorter OS and PFS in an early stage. The SKY92 signature has been reported to be a robust predictor for OS as well as PFS (Kuiper et al. 2012, Kuiper et al. 2015). Aims: To validate the prognostic value for PFS and OS of the SKY92 high risk marker in newly diagnosed and relapsed refractory Multiple Myeloma across a wide variety of clinical trial datasets. Methods: OS data and gene expression profiling (GEP) data, either publicly available or obtained by analysis of the CE IVD MMprofiler assay (that includes the SKY92 signature), of eight clinical datasets (Figure 1) were used to evaluate the proportion of cases, hazard ratios and p-values associated with the SKY92 risk marker. The datasets included both newly diagnosed MM (TT2, TT3, MRC-IX, MMGI, HOVON87) and relapsed refractory MM (APEX and TT6) and also elderly (>65 yrs) MM (HOVON87 and about half of MRC-IX). The Cox proportional hazards model was used to calculate Hazard Ratios (HR), and associated p-values. Results: Figure 1 shows Kaplan Meier curves for eight datasets comparing cases that are high risk for SKY92 versus those that are standard risk for SKY92. The proportion of high risk cases was very similar in all cohorts (range 15.4% - 20.7%). For OS the (high risk) / (standard risk) hazard ratios range from HR=2.2 (MRC-IX) to HR=10.3 (TT6) and are all significant, see Figure 1A, including the relapsed stage cohorts TT6 and APEX. For PFS, the hazard ratios range from HR=1.7 (APEX) to HR=2.2 (HOVON87) and are all significant, see Figure 1B. The hazard ratios for OS (Figure 1A) are larger than for PFS (Figure 1B), which seems to suggest that the SKY92 signature is prognostic beyond the first line of treatment. (Figure presented) Summary/Conclusions: SKY92 is a powerful and robust prognostic marker, not only for OS but also for PFS in younger, elderly, newly diagnosed, relapsed (refractory) MM patients across various treatments including transplant, thalidomide, lenalidomide and bortezomib. It can therefore be used reliably as a predictor for survival to optimize follow-up and treatment strategies in an early stage.
Original languageEnglish
Pages (from-to)83
Number of pages1
Publication statusPublished - 2016


  • Kaplan Meier method
  • aged
  • bortezomib
  • cancer epidemiology
  • clinical article
  • clinical trial
  • diagnosis
  • drug therapy
  • follow up
  • gene expression profiling
  • genetic marker
  • genetic susceptibility
  • hazard ratio
  • human
  • lenalidomide
  • multiple myeloma
  • overall survival
  • progression free survival
  • proportional hazards model
  • statistical significance
  • thalidomide
  • treatment response
  • tumor resistance

Cite this