Investigation Of The Genetic Disparities Between African Americans and Causasians Utilizing Gene Expression Profiles (GEP) Of Purified Plasma Cells (PC) and Whole Bone Marrow Biopsies (BMBx)

MGUS and Multiple Myeloma (MM) are two to three times more common in African Americans (AA) compared to Caucasians (CAU). They also differ in many disease characteristics, most notably the better disease specific survival of AA compared to CAU. Recent improvements in outcomes have helped the CAU population, not benefitting the AA to the same degree. Although this disparity is well appreciated, little is know regarding the associated molecular differences.

Availability of large sample sizes of GEPs from CD138+ selected PC offered the opportunity to investigate whether race related differences in PC-GEP exist. The training set included 467 Caucasians and 43 African Americans, and the test set included 209 CAU and 13 AA. A model was developed based on the 27 most differentially expressed probesets (by q-value) in the training set (ROC-AUC, 0.999), and was validated in the test set (ROC-AUC, 0.979). These data were further corroborated in a validation set of asymptomatic MM (AMM) (63CAU/7AA, ROC-AUC, 0.941). Heatmaps of the 27-gene model for the training, test and validation sets are shown in Figure 1a -c. Performance analysis of this 27-gene model revealed a sensitivity of 69.2% and 90.9%, a specificity of 99.1% and 98.1% and a positive predictive value of 99.2% and 96.4% in the MM test set (table 1) and AMM validation set (table 2) respectively.

Figure 1a

View largeDownload slide

Heatmap for 27-gene model, training set

(Key: blue = African-American, red=Caucasian)

Figure 1a

View largeDownload slide

Heatmap for 27-gene model, training set

(Key: blue = African-American, red=Caucasian)

Close modal

Figure 1b

View largeDownload slide

Heatmap for 27-gene model, test set

(Key: blue = African-American, red=Caucasian)

Figure 1b

View largeDownload slide

Heatmap for 27-gene model, test set

(Key: blue = African-American, red=Caucasian)

Close modal

Figure 1c

View largeDownload slide

Heatmap, 27-gene model, AMM validation set

(Key: blue = African-American, red=Caucasian)

Validation set: 63 Caucasian, 7 African-American with AMM baseline GEP samples

Figure 1c

View largeDownload slide

Heatmap, 27-gene model, AMM validation set

(Key: blue = African-American, red=Caucasian)

Validation set: 63 Caucasian, 7 African-American with AMM baseline GEP samples

Close modal

Table 1

Summary measures of the 27-gene model, test set

Measure .	Value .
Sensitivity	9/13 (69.2%)
Specificity	205/209 (98.1%)
Positive Predictive Value	214/222 (96.4%)

Measure .	Value .
Sensitivity	9/13 (69.2%)
Specificity	205/209 (98.1%)
Positive Predictive Value	214/222 (96.4%)

View Large

Table 2

Summary measures of the 27-gene model, AMM validation set

Measure .	Value .
Sensitivity	5/7 (71.4%)
Specificity	62/63 (98.4%)
Positive Predictive Value	67/70 (95.7%)

Measure .	Value .
Sensitivity	5/7 (71.4%)
Specificity	62/63 (98.4%)
Positive Predictive Value	67/70 (95.7%)

View Large

The bone marrow microenvironment (ME) has a significant impact on the behavior of myeloma cells. We therefore also sought to investigate the differences in the bone marrow ME between AA and CAU. Using a set of 783 probesets that are unique to the bone marrow microenvironment, we analyzed GEPs of whole bone marrow biopsies in a training set of 250 CAU and 22 AA patients, identifying only 7 probesets with a false discovery rate <0.1. A score created based on these 7 genes had a sensitivity of only 66.7%, a specificity of 76.5% and a positive predictive value of 76.1%.

In summary we identified a list of genes, which are differentially expressed between AA and CAU and have developed a model that can accurately identify race based on the expression of 27 genes. With the data available we were not able to develop a suitable model that identified differences of the bone marrow ME between AA and CAU. Correlation of the gene expression differences of plasma cells with clinical characteristics and outcome will provide us with a better understanding of the pathophysiology of MM in AA and will hopefully be the basis of individualized therapy that also takes ethnicity into consideration.