|Home | About | Journals | Submit | Contact Us | Français|
Analyzed gene array data: YZ, MAC, JH, JDS
Enrolled and treated patients according to protocol directives: BN, EA, FvR, BB
Collected and analyzed data: JH, JC
The clinical significance of cytogenetic abnormalities (CA) present in randomly sampled (RS) or focal lesion (FL) bone marrow sites was examined in 419 untreated myeloma patients. Among 290 patients with gene expression profiling (GEP) data generated from RS sites, GEP-defined high-risk was present in 52% of the RS+/FL+ group but in only 9% of the remainder (p<0.001). The RS+/FL+ constellation (18%) was an independent predictor of poor survival, also after adjusting for GEP-derived risk and TP53 status (HR=2.42, p=0.004). The prevalence of high-risk myeloma in the RS+/FL+ group may reflect a dissemination-prone condition not shared by the other 3 groups.
The presence of osteolytic lesions distinguishes multiple myeloma (MM) from its precursor condition, monoclonal gammopathy of undetermined significance (MGUS), and is the cause of major morbidity (Kyle, et al 2007). Long recognized as a prognostic variable in the original clinical staging system named after Durie and Salmon (Durie and Salmon 1975), bone destruction has since been recognized not only as a consequence of MM, but also as an active participant in its propagation and component of failure to therapeutic interventions (Yaccoby, et al 2002). We and others have demonstrated that magnetic resonance imaging (MRI) is useful for the early detection of intramedullary focal lesion (FL) growth long before the development of X-ray-detectable bone destruction (Walker, et al 2007). In these investigations, we observed, in confirmation of the Durie-Salmon staging system, that the number of MRI-defined FL adversely affected survival, second only to the presence of metaphase cytogenetic abnormalities (CA). The hypothesis that the sites of FL growth in MM harbored particularly aggressive disease had motivated our efforts over the years to obtain, from consenting subjects, fine needle aspirates (FNA) from MRI-FL under CT guidance, specifically for the purpose of CA analysis.
The details of Total Therapy (TT) protocols, approved by the UAMS Institutional Review Board, and outcome results have been reported previously (Barlogie, et al 1997; Barlogie, et al 2006; Barlogie, et al 2007). All participating patients had signed a written informed consent acknowledging the investigative nature of these studies and awareness of alternative treatment options, in keeping with institutional, federal and Helsinki Declaration guidelines. Among 1379 patients successively enrolled in and treated on TT protocols (TT1, n=231; TT2, n=668; TT3, n=480), cytogenetic information prior to transplant was available on RS in 1357; additional information on FL sites was available for 419 of the 1357 patients.
For the purpose of cytogenetic examination, an effort was made to examine at least 20 metaphases, applying Giemsa banding techniques. The presence of CA required the detection of at least 2 abnormal metaphases in cases of hyperdiploidy and translocations, whereas at least 3 metaphases with clonal abnormalities were required in cases of whole and partial chromosome deletions. Gene expression profiling (GEP) data from RS were available from 290 patients prior to therapy. GEP analysis was performed after CD138 purification; both laboratory processing and data analyses have been previously reported regarding a molecular classification (Zhan, et al 2006), 70-gene risk stratification (Shaughnessy, et al 2007) and TP53 deletion (Xiong, et al 2008).
All patients underwent repeated laboratory and imaging examinations as specified in the protocols and in keeping with good medical practice. Definitions of clinical endpoints of response (complete response, CR; near-complete response, n-CR), response duration, event-free survival (EFS) and overall survival (OS) have been previously reported and were similar to recently agreed upon conventions (Durie, et al 2006).
The Kaplan-Meier method was used to estimate EFS and OS. EFS was defined from date of registration to the occurrence of death from any cause, disease progression, or relapse, or was censored at the date of last contact. OS was defined from date of registration to the date of death from any cause or censored at the date of last contact. Duration of complete response was defined from the earliest date of complete response to the date of death from any cause, disease progression or relapse, or censored at the date of last contact. The Cox regression method was used to examine multivariate prognostic factor models for OS and EFS (Cox 1972).
The overall incidence of CA in RS (RS+) among 1357 patients with CA data was 37%, similar to the 34% detected in RS of 419 patients with concurrent RS and FL examinations. Among the latter, CA was absent from both RS and FL (RS−/FL−) in 48%, present in both sites (RS+/FL+) in 18%, only in RS (RS+/FL−) in 16%, and only in FL (RS−/FL+) in 18%. Seven-year estimates of OS and EFS were lowest at 28% and 23%, respectively, among the 75 patients designated as RS+/FL+ (Figure 1a, ,b).b). The remainder had comparable EFS, whereas OS was inferior in the 67 subjects with RS+/FL− in comparison with the 277 with RS−/FL− (n=200) or RS−/FL+ (n=77). In the RS+/FL+ category, discordance in CA (unique structural or numerical aberrations in only one of the two sites) was observed in 48 patients whose survival outcomes were similar to those of 27 patients with CA concordance in both sites (p=0.83, data not shown). In case of CA discordance, the presence of greater CA complexity, in RS or FL, likewise did not influence outcome.
We next examined the relative distributions of baseline prognostic variables among the 4 groups, relative to findings in all 1357 patients. There were no significant differences in the proportions of patients with certain age, C-reactive protein (CRP) and creatinine cut-off values of recognized prognostic importance. The 75 patients with RS+/FL+ exhibited higher frequencies of elevated levels of lactate dehydrogenase (LDH, >= 190U/L) (p<0.0001), whereas the 159 with RS+ status regardless of FL+ status had higher frequencies of elevated B2M and lower hemoglobin compared to those with RS-with or without FL+ (p=0.007 and p=0.002, respectively) (data not shown).
Examination of GEP data, derived from RS sites in 290 patients, revealed that the proportion of patients with high GEP-defined risk scores (>0.66) decreased from a high of 52% among the 54 subjects with RS+/FL+ to 27% in the 49 patients with RS+/FL−(p=0.009) to 6% and 4% (p<0.001) among the 50 patients with RS−/FL+ and the 130 patients with RS−/FL− status (p =0.65). Relevant to molecular subgroup distribution (Zhan, et al 2006), no significant differences were observed in RS/FL constellations among MF (MAF/MAFB), MS (MMSET/FGFR3), CD-1 (CCND1 without CD20 expression) and LB (low bone disease) subtypes; however, RS+/FL+ was uniquely over-represented in the PR (Proliferation) group (p<0.0001) (data not shown).
Univariate analysis confirmed the prognostic implications of many well recognized standard and also GEP-derived parameters such as high-risk designation, PR subgroup and TP53 deletion (Table 1). Multivariate analyses, exclusive of GEP data, revealed that RS+/FL+ status imparted inferior OS (HR = 3.52; p < 0.001) and EFS (HR = 2.36; p = <0.001) (see Table 1). This significance was upheld in the context of GEP data available in 284 patients for OS (HR=2.43, p=0.004). Among the 236 patients lacking RS+/FL+, high-risk status segregated 22 patients with markedly inferior clinical outcomes resembling those of the 54 patients with RS+/FL+ whose poor OS and EFS were not affected by GEP risk designation (Figure 1c, ,d).d). GEP-derived TP53 deletion status affected both endpoints adversely (OS: HR=2.02, p=0.013; EFS 1.84, p=0.017), whilst GEP-defined high risk and low hemoglobin level were significantly associated with EFS (see Table 1).
The novel finding of our work relates to the discovery that the adverse prognostic consequences of CA requires the detection of CA at both RS and FL sites, a circumstance linked in turn to a predominance of high-risk and PR GEP designations in RS sites. The unexpected favorable outcomes of patients with CA present solely in FL may be explained by better adhesion of DKK1-secreting FL-resident MM cells, resulting in impaired proliferation and dissemination potentials (Tian, et al 2003). By way of a cell adhesion-mediated drug resistance mechanism, such FL-MM cells respond less favorably and more slowly to systemic therapy, explaining the striking time delay of 1 to 2 years observed in FL resolution compared to MM protein response (Walker, et al 2007). Persistence of FL in complete remission is consistent with a tumor dormancy state from which late relapses can ensue. Given the difference in MM biology in FL and RS sites, presumably with a more diffuse rather than focal growth pattern in the latter, work is in progress to investigate the intra- and inter-tumor heterogeneity in molecular signatures of both MM cells and adjacent stroma at RS and FL sites.
Supported: Program Project grant (CA 55819) from the National Cancer Institute, Bethesda, MD.
Disclosures: Research support (JDS, EA, FvR, BB) from Celgene, Millennium and Novartis corporations
Honoraria for speaking engagements (JDS, EA, FvR, BB)
Royalties from Novartis (JDS, BB)