Abstract 195

GEP analysis is a robust method to distinguish low- and high-risk multiple myeloma (MM), pertaining to 85% and 15% of newly diagnosed patients, respectively (Shaughnessy et al., Blood, 2007; 109:2276–84). As developed in TT2 and validated in TT3A and TT3B, we are now examining, similar to previous work in high-risk MM, whether we can define outliers among low-risk MM, i.e., patients not living up to the low-risk prediction model. Toward this end, we scrutinized early relapses in TT3A and TT3B within three years of protocol entry. Using logistic regression analysis, we identified baseline parameters including GEP, en route for distinguishing this high-risk subset among low-risk MM. Also examined was whether a new model could be built within low-risk disease that allowed for the identification of a high-risk subset. Our database was interrogated for patients known to have GEP-defined low-risk in the GEP-70 model. Table 1 summarizes the 3-year events among GEP-70 low-risk subjects per protocol. An optimal cut-point at +0.146 distinguished, among the combined TT2 and TT3 patients, inferior progression-free survival (PFS) and overall survival (OS) (Figure 1a). Next, we examined outcomes among all TT2 and TT3 patients with GEP data, including those with traditionally-defined high-risk (>=0.66). Here, we were able to distinguish three subgroups with distinctly different PFS and OS (Figure 1b). Utilizing logistic regression analysis, limited to traditionally-defined GEP-70 low-risk MM (=<0.66), three-year progression events during the this period were adversely dominated by the following: GEP-70 scores >0.146 (HR=2.61, p=0.0005), the presence of cytogenetic abnormalities (CA) (HR=1.93, p=0.018), B2M >5.5mg/L (HR=1.95, p=0.04) and LDH >190U/L (HR=1.93, 0.02). These are all reported in Table 2. In conclusion, we have identified, within GEP-70 low-risk patients, a new cut-point. This allows a better categorization of patients having truly low risk disease. Also, above which a prognosis intermediate to the traditional high-risk prognostic group (>=0.66) could be identified. GEP >0.146 dominated a multivariate logistic regression model. Further efforts will be presented on unique genes characterizing this intermediate risk group in relationship to low and high-risk subsets.

Table 1.

Three-year Events Among GEP-70 Subjects Per Protocol

ProtocolTotal with GEPGEP-70 low-riskGEP-70 low-risk, event within first 3 years
TT2 - thalidomide 176 156 55 
TT2 + thalidomide 175 149 36 
TT3A 275 235 39 
TT3B 166 129 23 
ProtocolTotal with GEPGEP-70 low-riskGEP-70 low-risk, event within first 3 years
TT2 - thalidomide 176 156 55 
TT2 + thalidomide 175 149 36 
TT3A 275 235 39 
TT3B 166 129 23 

Table 2.

Logistic Regression for 3-year Event Factors, TT2+3 GEP-70 Low-Risk (<0.66)

Event in first three years on protocol
VariableNWith FactorWithout FactorOR (95% CI)P - value
Multivariate B2M > 5.5 mg/L 666 24/69 (35%) 59/328 (18%) 1.95 (1.03, 3.69) 0.0401 
 LDH >= 190 U/L 668 33/99 (33%) 50/298 (17%) 1.93 (1.10, 3.40) 0.0229 
 Cytogenetic abnormalities 665 35/116 (30%) 48/281 (17%) 1.93 (1.12, 3.32) 0.0182 
 GEP-70 score > 0.146 669 37/104 (36%) 46/293 (16%) 2.61 (1.52, 4.47) 0.0005 
Event in first three years on protocol
VariableNWith FactorWithout FactorOR (95% CI)P - value
Multivariate B2M > 5.5 mg/L 666 24/69 (35%) 59/328 (18%) 1.95 (1.03, 3.69) 0.0401 
 LDH >= 190 U/L 668 33/99 (33%) 50/298 (17%) 1.93 (1.10, 3.40) 0.0229 
 Cytogenetic abnormalities 665 35/116 (30%) 48/281 (17%) 1.93 (1.12, 3.32) 0.0182 
 GEP-70 score > 0.146 669 37/104 (36%) 46/293 (16%) 2.61 (1.52, 4.47) 0.0005 

OR - Odds Ratio, 95% CI - 95% Confidence Interval, P - value from Wald Chi - Square Test in Logistic Regression.

NS2 - Multivariate results not statistically significant at 0.05 level. Univariate p - values reported regardless of significance.

Multivariate model uses stepwise selection with entry level 0.1 and variable remains if meets the 0.05 level.

A multivariate p - value greater than 0.05 indicates variable forced into model with significant variables chosen using stepwise selection.

Disclosures:

No relevant conflicts of interest to declare.

This icon denotes a clinically relevant abstract

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution