Expression profile of cyclin D1 and cyclin D2 proteins
Expression of cyclin D1 and cyclin D2 proteins was highly variable among the samples, particularly in the case of cyclin D1. Expression values of cyclin D1 ranged from 0 to 15.05, while those of cyclin D2 varied from 0 to 1.18 (Figure 1A). Excluding non-expressed values, the non-parametric coefficients of variation for cyclin D1 and cyclin D2 were 96% and 70%, respectively.
Patients were divided into four groups based on their cyclin D1 and cyclin D2 protein expression: expression of cyclin D1 exclusively (54 out of 165, 33%); expression of cyclin D2 exclusively (30 out of 165, 18%); coexpression of both proteins (14 out of 165, 8%); no expression of either cyclin D protein (67 out of 165, 41%) (Figure 1B).
The group of MM patients expressing only cyclin D1 contained all the 23 cases with t(11;14) and 55% (21/38) of the cases with 11q13 gains. In other words, cyclin D1 expression was associated with t(11;14) or 11q13 gains in 82% (44/54) of the patients. We next dichotomized the expression of this group of patients by fitting a Gaussian mixture model that differentiated two groups, one with a high level of cyclin D1 expression (cyclin D1 > 0.057), and the other with a low level (cyclin D1 ≤ 0.057) (Figure 1C). Eighteen of the 23 patients (78%) with t(11;14) were classified in the group with high cyclin D1 expression, while only two of the 38 patients (5%) with 11q13 gains were included in that group (Figure 1D). IGH translocations other than t(11;14) were rarely found in the group of patients expressing only cyclin D1. In fact, t(4;14) was detected in only two cases that also featured 11q13 gain, which were in turn classified into the low cyclin D1 expression group.
In the group of patients who exclusively expressed cyclin D2, none had t(11;14) as expected, although 11q13 gain was present in three of the 30 patients (10%). The distribution of the other cytogenetic abnormalities in this group was as follows: t(4;14) and t(14;16) were each present in 13% (4/30) of the cases; 1q gains, and 1p and 17p deletions, were found in 80% (24/30), 27% (8/30) and 20% (6/30) of cases, respectively. t(14;16), 1q gains and 1p deletions were significantly enriched in the group of patients expressing only cyclin D2 compared with the other MM patients (13% vs. 0%, p < 0.001; 80% vs. 40%, p < 0.001; 27% vs. 10%, p = 0.03, respectively). FISH studies yielded normal results in only two of the 30 patients expressing solely cyclin D2. In the same way as for cyclin D1, patients with cyclin D2 expression were dichotomized into two groups, one with high cyclin D2 expression (cyclin D2 > 0.058) and the other with low expression (cyclin D2 ≤ 0.058) (Supplementary Figure 1A). Cytogenetic abnormalities were uniformly distributed throughout the two groups (Supplementary Figure 1B).
In the group of patients coexpressing both cyclins D a preference for the expression of one of them was observed in 10 of the 14 patients (71%) (Figure 1E). Based on the level of expression of each cyclin D, most cases (71%) expressed low levels of cyclin D1 and D2. Two cases each exhibited high levels of expression of cyclin D1 and of cyclin D2. None of the patients belonging to this group showed t(11;14), while 11q13 gain was detected in three patients who expressed low levels of both cyclins D.
Finally, the largest group of patients analyzed (41%) expressed neither of the cyclins D. The distribution of cytogenetic abnormalities analyzed by FISH within this group is summarized and compared with the other three groups of cyclin D expression in Supplementary table 2. Interestingly, more than half of the patients with t(4;14) did not express cyclin D2, whereas all the four samples with t(14;16) did express it.
Expression profiles of CCND1 and CCND2 mRNA
Quantifying cyclin D1 and D2 proteins showed a high proportion of MM patients without expression of any of the cyclins D. To gain more insight into this unexpected finding, we evaluated the expression of CCND1 and CCND2 at the mRNA level using qRT-PCR in 110 of the 165 samples for which RNA was available.
Expression of CCND1 and CCND2 mRNA was quantified in 16 normal plasma cells (NPC) to establish the baseline expression level for both mRNAs in the cohort. CCND1 and CCND2 mRNAs were considered to be overexpressed when their expression in MM samples was above the upper 95th percentile expression level in NPC (ΔCt = -5.99 for CCND1 and ΔCt = -3.51 for CCND2) (Figures 2A and 2B). According to these criteria, exclusive overexpression of CCND1 or CCND2 was detected in 53% (58 of 110) and 21% (23 of 110) of patients, respectively. Overall, 6% (7 of 110) of the samples simultaneously expressed CCND1 and CCND2 at the mRNA level.
The Spearman's rank-order correlation between mRNA and protein expression levels was stronger for cyclin D1 than for cyclin D2 (rho = 0.7 vs. rho = 0.53; p < 0.001) (Supplementary Figure 2A and 2B).
Almost all the samples that exclusively expressed cyclin D1 protein overexpressed CCND1 mRNA (40 of 41 samples for which protein and mRNA material was available) (Figure 2C). The highest levels of CCND1 mRNA were observed in MM patients with t(11;14).
However, when we compared the expression of cyclin D2 at the protein and mRNA levels in the samples for which both molecules were available, we found that 71% (15 out of 21) of the patients exclusively expressing cyclin D2 protein also overexpressed CCND2 mRNA (Figure 4C). Finally, the 41% of patients who did not express either cyclin D1 or cyclin D2 protein expressed mRNAs at levels lower than those observed in NPC, whereas 59% of those patients expressed the mRNA of at least one cyclin D.
Prognostic effect of cyclin D protein expression
The survival analysis considered only the patients who exclusively expressed cyclin D1 or D2, and compared them with patients who did not express the corresponding cyclin D. Expression of cyclin D1 protein was significantly associated with longer overall survival (OS) (HR [95% CI] = 0.44 [0.22-0.91], p = 0.022) (Figure 3A). Conversely, expression of cyclin D2 was significantly associated with shorter OS (HR [95% CI] = 2.14 [1.13-4.05], p = 0.017) (Figure 3B). No statistically significant differences were found in the time to progression (TTP) among the patients classified by their cyclin D1 or D2 expression status (Figures 3C and 3D). A positive effect of CCND1 mRNA overexpression on OS was also observed (Supplementary Figure 3).
Given the significant association between cyclin D2 protein expression and 1q gains, we investigated how this relationship was related to survival. We found that the prognosis of patients with 1q gains was not affected by cyclin D2 protein levels, while cases expressing cyclin D2 exhibited short survival only if they also had 1q gains (Figure 4).
A subsequent survival analysis considering the groups of high and low expression of both cyclins D revealed no significant differences in OS between the two groups (Figure 5A). However, TTP was significantly shorter among patients with high levels of cyclin D1 (HR [95% CI] = 2.43 [1.06-5.55], p = 0.03), indicating a less favorable prognosis for patients with t(11;14) than for those with 11q gains (Figure 5B). Partitioning the patients into the high and low level cyclin D2 groups revealed no differential association with survival.