Introduction

Multiple myeloma is a clinically and biologically heterogeneous disease with highly variable patient survival ranging from <1 to >20 years. This variable clinical outcome is determined largely by the underlying tumor genetic makeup. Conventional cytogenetic and fluorescence in-situ hybridization (FISH) studies and, more recently, next-generation sequencing methods are commonly used to identify high-risk myeloma. Among the well-recognized adverse factors, del(17p13)/TP53, t(4;14)FGFR3-IGH and t(14;16)IGH-MAF are associated with shorter overall survival and are independent predictors of outcome. The prognostic value of 1q21/CKS1B gains or amplification (gain/amp) is controversial. In the Revised International Staging System proposed by the International Myeloma Working Group in 2015, which incorporates chromosomal abnormalities detected by FISH and serum lactate dehydrogenase levels into the preexisting International Staging System (ISS), only t(4;14), t(14;16), and del(17p13)/TP53, but not 1q21/CKS1B gain/amp, are listed as high-risk abnormalities [1, 2]. However, in the Mayo Clinic model—Mayo Stratification of Myeloma and Risk-Adapted Therapy—CKS1B gain/amp is listed as one of the high-risk aberrations [3].

Abnormalities of chromosome 1, mainly gain/amp of 1q21 or deletion of 1p21, are observed in ~40–45% of patients with multiple myeloma, resulting in disruption of the cyclin-dependent kinase regulatory subunit 1B (CKS1B) or cyclin-dependent kinase inhibitor 2C (CDKN2C) genes, respectively [4]. CKS1B gain/amp has been described as an adverse factor in the literature, observed more commonly in patients with an abnormal karyotype or with resistance to bortezomib [4,5,6,7,8,9,10,11,12,13]. Studies of patients treated with autologous stem cell transplantation (SCT) showed that CKS1B gain/amp was associated with deletion of 13q and had a negative impact on survival [12, 14]. Another study showed that patients with concurrent CKS1B amplification, defined as ≥4 copies, and ISS-III diseases, or biallelic TP53 inactivation, had a poor outcome, with a median overall survival of 20.4 months [15]. However, clinically we have also observed CKS1B gain/amp in patients with smoldering myeloma. Fonseca et al. reported that CKS1B gain/amp is often associated with other high-risk factors such as t(4;14) (p < 0.001), as well as a high proliferation signature as measured by plasma cell labeling index. These factors superseded CKS1B gain/amp in multivariate survival analysis [16]. That study was based on FISH data and did not incorporate karyotype. These previous findings led us to wonder if the prognostic implications of CKS1B gain/amp itself are independent of other coexisting adverse events, such as TP53 deletion or background karyotype.

In the current study, we examined the prognostic impact of CKS1B gain/amp in the broader context of karyotype and the status of TP53, FGFR3-IGH, MAF-IGH, MYEOV/CCND1-IGH, and RB1 analyzed concurrently by FISH. Our aim was to better understand the clinical impact of CKS1B gain/amp, alone or in combination with other abnormalities.

Materials and methods

We retrospectively reviewed myeloma patients whose specimens were assessed by conventional cytogenetics and tested for CKS1B using interphase FISH at our hospital between June 1, 2011 and June 30, 2015. Clinical and laboratory data were obtained by reviewing the electronic medical record. Any patients who were clinically diagnosed with smoldering myeloma or who had <10% abnormal CKS1B gain/amp signals were excluded. For patients who had a preceding diagnosis of smoldering myeloma, overall survival was calculated from the time of symptomatic myeloma when a therapy was required. We also excluded patients whose death resulted from causes other than myeloma, such as solid tumors or cardiovascular diseases, and those with therapy-related myeloid neoplasms.

For comparison, we included two control groups. Group 1 comprised myeloma patients with a complex karyotype (CK) without CKS1B gain/amp and group 2 comprised patients with standard-risk myeloma without CKS1B gain/amp. These patients were identified during the same period as those in the CKS1B group. A subset of these cases overlapped with those reported in a previous study of TP53-deleted myeloma [17, 18]. Our study was approved by the Institutional Review Board of The University of Texas MD Anderson Cancer Center.

Cytogenetics and FISH

Conventional karyotyping was performed on cultured (unstimulated for 24 or 48 h) bone marrow aspirate samples as part of the routine clinical workup and following standard laboratory procedures. At least 20 metaphases were fully analyzed whenever possible for the identification of clonal cytogenetic aberrations. The karyotypic results were reported according to the International System for Human Cytogenetic Nomenclature 2009 and 2013. In this system, a clone is defined as chromosomal changes in two or more metaphases and a CK is defined as ≥3 non-related chromosome abnormalities. A simple karyotype (SK) is defined as <3 abnormalities. A panel of four probe sets, including TP53/CEN17, MYEOV/CCND1-IGH/t(11;14), RB1 (Vysis-Abbott Molecular, Downers Grove, IL, USA), and CKS1B (CytoCell, Cambridge, UK), was performed on cultured (unstimulated for 24 or 48 h) bone marrow samples or enriched samples using a magnetic cell sorting procedure (MACS; Miltenyi Biotec, Bergisch Gladbach, Germany) as described when plasma cells were <20% of total cellularity in the differential count of bone marrow aspirate smears [19]. FGFR3-IGH/t(4;14) and IGH-MAF/t(14;16) were performed as reflex tests if an abnormal signal was observed in MYEOV/CCND1-IGH/t(11;14), but no fusion signals were observed. Three signal copies of CKS1B were interpreted as a gain of CKS1B, and four or more signal copies were interpreted as an amplification of CKS1B in our study. Details of the TP53, CKS1B, and FGFR3-IGH/t(4;14), MAF-IGH/t(14;16) FISH assays have been previously described [17]. Clone size was estimated by the number or percentage of positive cells in 200 interphase cells analyzed. The clone size of CKS1B gain/amp and TP53 deletion were compared with those of deletion of RB1 and/or MYEOV/CCND1-IGH and against the plasma cell percentage reported in differential counts of bone marrow aspirate smears to estimate whether the CKS1B aberration was more likely a major clone or a subclone. For the purpose of our study, patients with gain/amp of CKS1B only were considered to have a “single hit,” (SH), whereas patients with additional TP53 deletion were considered to have a “double hit” (DH). “Triple hit” (TH) was defined as DH cases with an additional t(4;14) or t(14;16). In 13 patients, FISH was performed on CD138-enriched bone marrow samples. These patients were excluded from the analysis of correlation between clone size of CKS1B and survival.

Statistical analysis

The follow-up interval and overall survival were calculated from the time of initial diagnosis until time of last follow-up or death. Statistical analyses were performed using the GraphPad Prism 6 software (GraphPad Software, La Jolla, CA, USA) and R version 3.6.1. Kaplan–Meier curves for overall survival were plotted and the log-rank test was applied, with p < 0.05 considered statistically significant. Multivariate analysis was performed using SPSS version 9.3 (SPSS Institute, Chicago, IL, USA) and R version 3.6.1 for Cox proportional hazards regression. The parameters included age, gender, tumor load as estimated from the bone marrow biopsy, ISS stage, clone size of CKS1B gain/amp, clone size of TP53 deletion, karyotype, and SCT. Different thresholds were used to test the impact of tumor load, using bone marrow plasma cell percentage as a surrogate, as well as clone size of CKS1B and clone size of TP53 deletion on overall survival in the multivariate analysis. TP53 was also analyzed as a continuous variable. Fisher’s exact  and Mann–Whitney U tests were also performed where indicated.

Results

Clinicopathologic findings

The study group included 132 patients, comprising 72 men and 60 women with a median age of 65 years (range 39–88 years; Table 1). The vast majority (n = 106, 80.3%) were newly diagnosed, and the remaining patients had either been previously treated (n = 9, 6.8%) or had relapsed/refractory disease (n = 17, 12.9%). A normal karyotype (NK) was found in 51 patients (39.5%), SK in 7 patients (5.4%), and CK in 71 patients (55.0%). In three patients, the karyotype was unknown. The patients with a CK had higher tumor loads, as estimated from the percentage of plasma cells in the bone marrow, than those of the patients with a NK or SK (p < 0.0001; Table 2).

Table 1 Clinicopathological features of the CKS1B study patients (n = 132).
Full size table
Table 2 Comparison of CKS1B+ patients with a complex (CK) vs. normal or simple (NK/SK) karyotypea.
Full size table

Twenty-five patients had CKS1B gain/amp and TP53 deletion (DH) and five patients had additional t(4;14) FGFR3-IGH (n = 3) or t(14;16), IGH-MAF (n = 2; TH). DH/TH was highly associated with CK: DH/TH occurred in 30.6% of patients with a CK compared with 13.7% in patients with a NK/SK, p = 0.02; Table 3. The clone size of TP53 deletion ranged from 8.5 to 86% (median 27%). The clone size was larger (>20% allelic burden) in patients with a CK (18/21, 85.7%) than in patients with a NK/SK (1/8, 12.5%; p = 0.0005). The one patient who had a SK with 27% TP53 deletion also had a TH. None of the patients in the NK/SK subgroup had TP53 deletion >50%. CKS1B amplification was identified in 44 patients, whereas CKS1B gain was identified in 92 patients, with the clone size ranging from 10 to 88% (median 26%). Comparison of the two groups showed that TP53 deletion was enriched in the CKS1B amp group (p = 0.04), but other clinical features, including frequency of CK, were similar between the two groups (Table 4).

Table 3 Comparison double (DH) or triple hit (TH) vs. single-hit (SH) cases.
Full size table
Table 4 Comparison between CKS1B gain vs. amplification cases.
Full size table

Twelve patients had FGFR3-IGH fusion (n = 10) or IGH-MAF (n = 2); nine had a CK and three had a NK/SK. Five of the 12 patients also had TP53 deletion (TH). MYEOV/CCND1-IGH was identified in 20 patients (15.2%) and RB1 deletion was identified in 89 patients (67.4%). Based on the percentage of plasma cells in the bone marrow aspirate and core biopsy samples, the clone size of CKS1B+ was smaller than that of RB1 or MYEOV/CCND1-IGH in ten patients. The clone size of CKS1B+ was larger than that of RB1 deletion or MYEOV/CCND1-IGH in eight patients and similar to tumor load in the remaining patients. In 13 patients, CKS1B gain/amp was the sole aberration detected by the FISH panel.

The clinical and pathologic features of the two control groups without CKS1B gain/amp (group 1, CK; group 2, standard risk) are summarized in Table 5. TP53 deletion was identified in 6/15 patients in control group 1, including one with concurrent t(4;14). Other clinical features were similar to those of the CKS1B+/CK study group. Control group 2 included 30 patients without high-risk FISH findings. The clinical features were similar to those of the CKS1B+ single-hit NK/SK group.

Table 5 Clinicopathological features of the control group 1—complex karyotype without CKS1B+ and control group 2—standard risk without CKS1B+.
Full size table

Survival

The median follow-up interval was 48.7 months (range 2–104 months). Univariate analysis of the whole study group showed that CK was associated with poorer overall survival compared with NK/SK (Fig. 1a). The median overall survival of patients with a CK was 38.6 months, and that of patients with a NK/SK has not been reached. Similarly, DH/TH was associated with poorer overall survival compared with SH (Fig. 1b). Multivariate analysis confirmed only CK (p = 0.0009), TP53 deletion (p = 0.0031), and advanced stage (p = 0.0336) as independent risk factors. TP53 deletion was a continuous variable with a higher allelic burden correlated with a greater risk of death (Table 6). Thus, among patients with a NK/SK, TP53 deletion was associated with a worse outcome compared with no deletion (Supplemental Fig. 1A). Among patients with a CK, a larger TP53 clone size was associated with higher mortality rates (Supplemental Fig. 1B). Linear regression showed some correlation between the percentage of tumor cells in the aspirate smears and core biopsy specimen and the signals of TP53 deletion after excluding patients with CD138-enriched bone marrow samples (r = 0.68, Supplemental Fig. 2). Thus, the high TP53 deletion allelic burden could be related to the high tumor burden seen in these patients. There was no survival difference between patients with a SK and those with a NK (p = 0.79).

Fig. 1: Survival curves and comparison of study subgroups.
figure 1

a Kaplan–Meier curves comparing overall survival in patients with CKS1B gain or amplification with a normal or simple karyotype (NK/SK) and those with a complex karyotype (CK). Patients with a CK had significantly worse overall survival (p < 0.0001). b CKS1B single hit compared with CKS1B and TP53 deletion (double/triple hit [DH/TH]). Patients with a DH/TH had significantly worse overall survival (p = 0.0001) in the entire study cohort. c Patients with CKS1B amplification had significantly worse overall survival (p = 0.02) than those with CKS1B gain for the entire study cohort. d Patients with CKS1B amplification or gain showed no overall survival difference after patients with a double or triple hit or patients with a complex karyotype were excluded (p = 0.2). e Survival comparison of four study subgroups: CKS1B gain/amp with a normal or simple karyotype (NK/SK; n = 58), standard-risk myeloma without CKS1B gain/amp; (n = 30), complex karyotype (CK) with CKS1B gain/amp (n = 71), and CK without CKS1B gain/amp (n = 15). Overall survival of patients with CKS1B NK/SK and that of standard-risk patients were similar (curve 1 compared with curve 2, p = 0.75). Overall survival of patients with a CK was similarly poor regardless of CKS1B status (curves 3 and 4, p = 0.18). However, between the two major risk subgroups, those with a CK fared significantly worse (p < 0.0001). f–h Stem cell transplantation was associated with better overall survival in the entire cohort (f). The survival advantage was largely due to improved overall survival in the subset of patients with a complex karyotype or double or triple hit (p = 0.0031; g). Such a survival difference was not evident in the subset of patients with a single hit and normal or simple karyotype (h).

Full size image
Table 6 Multivariate Cox regression analysis of overall survival considering TP53 deletion clone size as a continuous variable.
Full size table

Patients with CKS1B amp had a worse outcome that those with CKS1B gain (p = 0.02; Fig. 1c) for the entire cohort. However, after excluding patients with a CK and patients with a DH/TH, the overall survival difference was not evident using a cutoff of ≥4 CKS1B gene copies (Fig. 1d). Raising the cutoff to five copies did not make a difference (data not shown). As mentioned above, the CKS1B amp group was enriched with DH/TH (p = 0.04), suggesting that the inferior outcome of the patients with CKS1B amp was attributable to DH/TH. Multivariate analysis did not confirm a survival difference between the gain and amp subsets (p = 0.109; Table 6). In addition, CKS1B allelic burden (using a cutoff of >50%) did not impact survival.

The survival of patients with CKS1B SH and a NK or SK was similar to that of patients with standard-risk myeloma (Fig. 1e, curve 1 compared with curve 2, p = 0.75), and the outcome was not affected by the presence of MYEOV/CCND1-IGH or RB1 deletion (data not shown). In the presence of a CK, patients with CKS1B gain/amp did as poorly as those without (Fig. 1e, curves 3 compared with curve 4, p = 18).

All patients were treated with proteasome inhibitors, in various combinations with immunomodulatory drugs and steroids. In addition, 99 patients (75%) also underwent SCT. SCT was associated with a superior outcome (p < 0.01; Fig. 1f) for the entire group, and the survival advantage was most apparent in patients with a CK or a DH/TH (p = 0.0031; Fig. 1g). In multivariate analysis, SCT maintained its prognostic value (p = 0.006, Mantel–Cox test; Table 6). To exclude the possibility that the better overall survival of patients who underwent SCT was not attributable to older patients being excluded from SCT, a Mann–Whitney U test was performed and showed that the better outcome was independent of age (p = 0.17).

We also tested whether SCT impacted overall survival equally between patients with a CK and those with a NK/SK. As shown in Fig. 1g, SCT did not impact overall survival among patients with a NK/SK (p = 0.8; Fig. 1h), although the sample size of the NK/SK group was limited: only 12 patients were treated without SCT compared with 44 patients who underwent SCT.

Discussion

The introduction of immunomodulatory drugs, proteasome inhibitors, monoclonal antibodies, and high-dose chemotherapy and autologous SCT have remarkably improved the survival of myeloma patients. However, outcomes of patients with high-risk myeloma remain dismal. Identification of patients with high-risk myeloma and stratification of therapy is a key to better patient outcomes. Conversely, treatment for patients without high-risk myeloma could be managed as standard risk to minimize toxicity and therapy-related myeloid neoplasms. A risk-stratified approach is key to personalized medicine.

Cyclin-dependent kinase regulatory subunit 1B protein, encoded by CKS1B, is a cofactor for ubiquitination and regulates degradation of the cell cycle inhibitor p27Kip1 [20]. Investigators have found a correlation between increased expression of CKS1B protein and +1q21. Earlier studies showed a possible role of CKS1B gain/amp in cell cycle disruption and activation of the MEK/ERK and JAK/STAT3 signaling pathways, resulting in drug resistance and disease progression [4,5,6,7, 9,10,11, 14, 21,22,23]. One study showed that 10 of 11 patients with anaplastic myeloma had CKS1B amplification; five also had hemizygous 17p (TP53) deletion, four had 13q14 deletions, and four had t(4:14) [5]. However, we have observed CKS1B gain/amp in indolent smoldering myeloma clinically, as have others [24], and there is controversy about CKS1B [16, 25]. We therefore decided to reevaluate the prognostic impact of CKS1B gain/amp and hypothesized that the survival impact of CKS1B gain/amp could be dependent on or modified by the overall genetic context.

In the current study, we showed that CKS1B gain/amp is highly associated with a CK and TP53 deletion and that these associations likely account for the inferior outcome of affected patients. In contrast, patients with isolated CKS1B gain/amp in a background of a NK or SK had better overall survival than those with a CK, with a median overall survival duration similar to that of patients with standard-risk myeloma treated similarly. Most previous studies of CKS1B did not dissect the background CK or TP53 data. In the current study, about 50% of patients with de novo myeloma carrying CKS1B gain/amp also had a CK, exceeding the 30–40% typically reported in patients with newly diagnosed myeloma overall [26]. Similarly, TP53 deletion, reported in 10% of patients newly diagnosed with myeloma overall, was observed in ~20% of our CKS1B study cohort. These findings suggest that the CKS1B aberrations tend to coexist with other high-risk features.

Furthermore, the impact of combined CKS1B gain/amp and TP53 deletion, i.e., DH or TH, was correlated with the allelic burden of TP53 deletion, which was shown to be a continuous variable, e.g. the higher the allelic burden of TP53 deletion, the greater the risk of death. With a hazard ratio (HR) of 1.019, this represents a 2% increase in risk per 1% increase in cells with TP53 deletion. Similarly, others have reported that the size of TP53 deletion clonal fraction is a predictor of outcome [27]. These findings were also consistent with our previously reported observations that the background genetic context of TP53 deletion has more influence than isolated TP53 deletion status [17, 18]. Our results also supported the notion that adverse genetic lesions tend to co-segregate and that the accumulation of adverse aberrations is associated with a worse outcome [28].

Neoplastic plasma cells with multiple aberrations could be composed of subclones with more therapeutic resistance. The literature regarding the survival impact of CKS1B gain compared with amplification is controversial [4, 29,30,31]. In the current study, patients with CKS1B amp carried more TP53 deletions than did those with CKS1B gain; these two subgroups showed no survival differences in multivariate analysis. One cannot rule out the possibility that other genes in the +1q region beyond CKS1B, such as MCL1, IL6R, and ILF2 [32], could contribute to the poor prognosis of these patients.

Finally, autologous SCT significantly improved the survival of patients with a CK or DH/TH in our study, supporting the positive role of SCT in management of high-risk myeloma. The impact of SCT among patients with a NK/SK without a DH was not statistically significant, although the sample size was small. This observation raises the issue of whether SCT should be reserved at relapse for this small subset of patients, despite CKS1B gain/amp, in view of recent advances in myeloma treatment such as immunotherapy and newer generation of proteasome inhibitors. Further studies in a larger cohort of patients will be needed.

We acknowledge that patients with TP53 deletion constituted a small fraction of the study cohort and that the TP53 deletion clone size may reflect the tumor burden. We also acknowledge that the results of the current study do not address the potential clinical impact of CKS1B gain/amp in smoldering myeloma; these patients were specifically excluded from the study. Others have suggested that CKS1B gain/amp in smoldering myeloma is likely to herald progression to active disease [33]. Our data suggested that CKS1B could be the major clone in some cases or a subclone in others, indicating that it could be either an early or late genetic event. We also acknowledge that other than comparing SCT, we did not further address whether specific chemotherapy regimens were applied in different patient subgroups, which could impact survival.

As molecular and genetic studies become more readily available, there is a tendency to shift to more advanced techniques in clinical practice to decode the complex molecular genetics in myeloma. We believe that standard techniques such as karyotyping remain the backbone, and an integrated approach is the ultimate solution to clinical questions. Conventional cytogenetics typically yields an abnormal result in only about 30% of patients, although abnormal results usually turn out to be CK, because myeloma cells do not divide in vitro in most cases. This yield has led to the practice of limiting the diagnostic workup of myeloma to FISH or microarray-based testing in some centers. In some risk models, aberrations identified by FISH are the key or the only elements in defining biology and genetic risk stratification. The role of chromosomal abnormalities detected by conventional karyotyping is neither emphasized nor discussed. Nonetheless, karyotyping reveals a fundamental biologic feature of high-risk of myeloma—the ability to grow in vitro independent of the bone marrow microenvironment or proliferation capacity, and this affords a global view of genetic composition. The data presented in the current study indicate that conventional chromosomal analysis remains a powerful tool that can be used to refine risk assessment and supplement FISH and next-generation sequencing data, particularly if high-resolution genomic analysis such as microarray-based testing is not available. We suggest that myeloma risk models should consistently include karyotype in the schema, as is the case in the workup of acute leukemia and myelodysplastic syndromes.