Response and biological subtype of myeloma are independent prognostic factors and combine to define outcome after high-dose therapy.

Response to therapy has been linked to the outcome of multiple myeloma (MM) patients treated with autologous stem cell transplantation (ASCT) (Alvares et al, 2005; Cavo et al, 2007), however, an association between response and improved survival has not been seen in all studies. Myeloma is clearly a heterogeneous disease, with different biological risk groups having different clinical outcomes (Shaughnessy et al, 2007; Decaux et al, 2008), but whether the achievement of a high quality response is differentially important within a biological subgroup is not well understood. We have previously defined a fluorescent in situ hybridization (FISH)-based prognostic model, identifying three risk groups characterized by the presence of no (standard risk), 1 (high risk) or >1 (ultra-high risk) adverse FISH lesion [t(4;14), t(14;16), t(14;20), del(17)(p13) and +(1)(q21)] (Boyd et al, 2012). We have explored the relationship between response within different biological subtypes of MM and survival in patients treated in the intensive arm of the UK Medical Research council (MRC) Myeloma IX trial. The MRC Myeloma IX trial (International Standard Randomised Controlled Trial Number 68454111) enrolled 1960 patients with newly diagnosed MM, of which 1111 were treated intensively with the full design and primary results of the trial having previously been reported (Morgan et al, 2011). In summary, the trial compared standard therapy with that containing thalidomide. More primary end-points included progression-free survival (PFS), overall survival (OS) and response. Median follow-up was 5·9 years. Response was assessed post-induction and at 100 days (d100) post-ASCT, according to the International Uniform Response Criteria (Durie et al, 2006). Patients with incomplete data or experiencing early death (n = 114), and patients lacking a confirmatory bone marrow result (n = 20 post-induction, n = 29 post-ASCT) were excluded. PFS and OS were defined as the time from initial randomization to documented progression or death, as appropriate. FISH was performed on purified bone marrow plasma cells (CD138 magnetic microbeads; Miltenyi Biotec, Bisley, UK) to detect the presence of an IGH@ translocation, the common IGH@ translocation partners (WHSC1 at 4p16, CCND3 at 6p21, CCND1 at 11q13, MAF at 16q23 and MAFB at 20q12), hyperdiploid status, deletion of 1p32, 13q14, 16q23, 17p13, 22q11 and gain of 1q21 as previously described (Chiecchio et al, 2006). Results with a complete data set for adverse immunoglobulin heavy chain gene (IGH) translocations, +(1)(q21) and del(17)(p13) were available for 511 cases. Statistical analysis was performed using SPSS version 19.0 (SPSS Inc., Chicago, IL, USA). The analysis was performed per-protocol and survival curves were plotted using the Kaplan–Meier method with differences between curves tested using the log-rank test. Differences in response rates were assessed by the chi-squared or Fisher exact test. Multivariate analysis of variables associated with survival used a backwards elimination Cox proportional hazards model, with factors entered if they were significant at P < 0·05 in univariate analysis and retained if they were significant at P < 0·05 in the Cox model. High quality response rates, either pre or post-transplant, were not influenced by the presence of high risk genetic lesions (Boyd et al, 2012) or International Scoring system (ISS) stage, with the exception of del(17p) (associated with a lower rate of complete response post-ASCT, 25% vs. 50·1%, P = 0·020). Post-induction 39·4% of patients (393/977) achieved ≥ very good partial response (VGPR). We found that ≥VGPR post-induction was associated with a significantly improved PFS and OS (P < 0·001 and P = 0·007 respectively, median PFS 32·8 vs. 24 months, median OS 81·3 vs. 66·4 months). At this time point there was no significant difference between patients achieving CR or VGPR (P = 0·082 and P = 0·739 for PFS and OS respectively). At d100 post-ASCT, 47·2% of patients achieved a CR (326/691). The achievement of a CR predicted for an improved PFS (median PFS 38·5 vs. 30·4 months, P = 0·0006), with a similar OS (median OS 85·3 vs. 88 months, P = 0·877). Failure to achieve a ≥VGPR post-induction and the presence of a high or ultra high risk FISH were variables independently associated with a shorter PFS; the presence of a high or ultra high risk FISH was also independently associated with an impaired OS. Achievement of ≥VGPR post-induction was associated with a significantly longer PFS in both low- and high-risk ISS groups as well as in patients with standard- and high-risk FISH; in contrast, in the ultra high risk group, response did not correlate with PFS. At the d100 post-ASCT the achievement of a CR was associated with an improved PFS in the ISS I and in the standard risk FISH groups, however, in the ISS III and the high risk FISH groups, response was not associated with significant differences in outcome (Table 1). These data show that response to therapy is an important endpoint in ASCT-treated patients, and that reaching ≥VGPR pre-transplant has the strongest association with PFS and OS. Interestingly, response rates were similar in both standard- and high-risk FISH subgroups. In addition, response and biological subgroup defined by FISH were noted to be independent prognostic factors. We also show that the achievement of CR in the ultra-high risk FISH patients does not impact on survival. Maximizing response is an important endpoint, however it is clear that there is variability in the significance of response dependent upon the biology of the disease. Our data support a model in which patient outcomes are independently influenced by response and biological risk features. We show that response is independent of risk status, and that in the high risk group an impaired survival is not due to low response rate, but rather to earlier relapse. These data can be put together into a model whereby better responses associate with better outcomes, with response being important within each biological risk group. Standard and high biological risk groups have similar rates of response, but the biological difference, defined by their genetic abnormalities, leads to earlier relapse in the latter group (Fig 1) and impaired OS. The exceptions to this model are patients with a monoclonal gamopathy of undetermined significance-like clone, where the M protein can remain stable for many years, because of the indolent biology of the malignant cells (Pineda-Roman et al, 2007). The ability to distinguish indolent from more aggressive clones, and tailor therapy accordingly, will be an important investigation for the future clinical care of myeloma patients. G.J.M. and G.H.J were chief investigators of the MRC Myeloma IX trial; A.B. K.D.B. and G.J.M designed research, analysed data and wrote the paper; G.H.J. and F.E.D. designed research and wrote the paper; M.C. wrote the paper; M.F.K., C.P. and P.W. collected and analysed data; F.M.R. designed research and performed research; W.M.G. and R.O. analysed data. The authors declare no competing financial interests.