Skip to main content

16-02-2016 | Hematologic cancers | Article

Risk Stratification in Multiple Myeloma

Current Hematologic Malignancy Reports

Authors: Melissa Gaik-Ming Ooi, Sanjay de Mel, Wee Joo Chng

Publisher: Springer US


There are many prognostic variables in multiple myeloma and the difficulty is in deciding which is truly significant. The widely used International Staging System (ISS) does not incorporate genetics, age, and other important variables in its risk stratification. Although it has its own limitations, the recently published Revised International Staging System (R-ISS) that was built upon the framework of ISS, is a more comprehensive and predictive tool for multiple myeloma patients and should be henceforth utilised. We will review the current prognostic variables and their significance in this paper.


Multiple Myeloma (MM) is a clonal plasma cell disorder characterised by the secretion of a monoclonal protein in the majority of patients. MM shows a variable clinical presentation ranging from the premalignant condition of monoclonal gammopathy of undetermined significance (MGUS) to smouldering MM, and symptomatic MM, extra-medullary myeloma or plasma cell leukaemia. The life expectancy of patients with this disease has improved with the advent of novel therapies. However, MM remains a highly heterogeneous disease with the overall survival (OS) ranging from a few months to decades among different cases. This variability results from the heterogeneity in the tumour and bone marrow microenvironment.
Prognostic factors in MM can be divided into tumour factors or host factors. Tumour biology factors relate to the MM chromosomal abnormalities, gene expression profile, and tumour burden. Lactate dehydrogenase (LDH), plasma cell proliferative rate, extramedullary disease and initial presentation as plasma cell leukaemia are additional measures of tumour burden. Host related factors include age, performance status, and comorbidities. Response to therapy is a factor related to both host and tumour biology.

Tumour Factors

Tumour Biology

Cytogenetic Abnormalities

Chromosomal changes are present in the plasma cells of virtually 100 % of MM patients [1], but only 20–30 % of patients have demonstrable abnormal karyotypes. This observation is related to the low proliferative index observed in most of the patients. Thus, abnormal cytogenetics can be considered a surrogate of high proliferative index and patients presenting with demonstrable karyotypic abnormalities will have a shorter OS. Due to the limitation of conventional cytogenetics, the International Myeloma Working Group (IMWG) recommendation is for baseline testing with either cytoplasmic immunoglobulin-enhanced fluorescent in situ hybridization (FISH) or FISH carried out on the nuclei from purified plasma cells. The probes used should be able to detect t(4;14)(p16;q32), t(14;16)(q32;q23), t(11; 14)(q13;q32), 17p13 deletions, chromosome 13 deletion, ploidy category, and chromosome 1 abnormalities.
In general, MM can be broadly categorised into two genetic categories: hyperdiploid MM and non-hyperdiploid MM.

Hyperdiploid MM

Hyperdiploid MM is diagnosed by FISH in about 50 % of the cases [24] and is generally associated with a better outcome. These patients’ clonal cells harbour multiple trisomies of the odd-numbered chromosomes, with the exception of chromosome 13. Hyperdiploidy has been observed in MGUS [5, 6] and is, therefore, considered to be an initiating pathogenetic event.
Rarely, hyperdiploidy is associated with high-risk cytogenetics such as IgH translocations, del(17p) and +1q abnormality. It is still unclear if this favourable prognostic impact is lost in such cases as recent studies give us conflicting results, with two studies demonstrating a negative impact on survival [7, 8], whereas Kumar et al. [9] found that the presence of even one chromosomal trisomy was able to ameliorate the adverse effect of the t(4;14), t(14;16), and t(14;20).

Non-hyperdiploid MM

Non-hyperdiploid MM is characterised by a relative high prevalence of IgH chromosome translocations (>85 %) [3, 4]. Most of these patients, with the exception of those with a t(11;14)(q13;q32) and t(6;14) translocation, have an aggressive disease course, characterised by rapid relapse and shorter survival [1018].

IgH Translocations

Patients with translocations t(4;14)(p16;q32) and t(14;16)(q32;q23),seen in 15 and 5 % of cases, respectively, are associated with high-risk MM [12, 14]. The t(4;14)(p16;q32) is associated with FGFR3 and MMSET dysregulation [19], while t(14;16)(q32;q23) is associated with increased c-maf expression [20]. These patients also have a higher frequency of additional genetic abnormalities that are associated with poor prognosis such as chromosome 13 and 14 deletion, chromosome 17 abnormalities, chromosome 1q amplication, and 1p deletion [24]. Another maf gene is b-maf, the expression of which is highly correlated with t(14;20)(q32;q12), a translocation occurring in about 1–2 % of all MM cases and strongly associated with poor prognosis [2123].

17p Deletion [del(17p)]

These patients present with more aggressive disease, extramedullary disease (EMD), and central nervous system involvement [14, 24]. During the disease course, any primary genetic subtype can acquire del(17p), thereby conferring a worse prognosis.

Other Cytogenetic Abnormalities

Initially, chromosome 13 deletion was thought to negatively affect both event-free(EFS) and overall survival (OS) but more recent studies have suggested that its prognostic value is actually related to other genetic changes, especially t(4;14) and del(17p) [10]. Hypodiploidy detected by karyotype has also been identified as an adverse prognostic factor [2529].

Gene Expression Profiling

Two large studies using gene expression profiling have been reported. These two reports describe two different sets of genes that identify poor-risk patient populations. One of the two models—the University of Arkansas for Medical Sciences (UAMS) 70-gene model—has identified genes that are involved in cell cycle regulation, angiogenesis, cell adhesion, cell migration, and proliferation. About 30 % of the informative genes are mapped to chromosome 1 [17]. In the other model—the Intergroupe Francophone du Myélome (IFM) 15-gene model—high-risk patients were enriched in genes controlling proliferation, whereas low-risk patients were enriched in hyperdiploid karyotypes [30].
Multivariate analysis comparing the GEP70-gene model with the 15-gene model revealed that the GEP70-gene model was significant in all datasets tested, but the 15-gene model was significant in bortezomib trials only [31]. It is interesting to note that the two models do not have a single common gene, reflecting either differences in techniques, the treatment used to define the patient population and/or redundancy in the genes and pathways that control growth, proliferation, and survival.

Other Significant Molecular Abnormalities

Amplification of 1q21 increases the risk of MM progression and incidence of the amplification is higher in relapsed than in newly diagnosed MM [32, 33]. Gains of 1q are associated with upregulation of genes involved in intracellular protein transport, including COPA and ARF1 and downregulation of three genes involved in protein translation (RPLP2, RPL21, and FAU) [34]. Cases with 1q gains also showed significantly modulated expression of genes involved in ER stress-induced responses. The unfolded protein response promotes cell survival by reducing the accumulation of aberrantly folded proteins through translation arrest, production of chaperone proteins, and increased degradation [35].
In the MRC Myeloma IX trial, DNA methylation profiling was carried out. A few locally regulated genes were identified—CD38, GPX3, NCAM1/CD56, PDK4, RASD1, RBP1, SPARC,and TGFBI [36]. Hypermethylation of these genes was associated with significantly shorter OS, independent of age, International Staging System score (ISS), and adverse cytogenetics [36]. Although not commonly seen in MM, BRAF V600E is associated with significantly shorter OS in patients and increased aggressiveness [3739].

Tumour Burden

Indicators of tumour burden such as LDH, plasma cell proliferative rate, extramedullary disease, and disease aggressiveness play an important role in MM prognosis. Assessment of disease burden includes imaging modalities and laboratory measurements.

Disease Presentation

Initial presentation of the disease also has prognostic significance. A high stage at diagnosis, plasma cell leukaemia and extramedullary disease (EMD) affect response to therapy and survival.
Extramedullary Disease
Extramedullary disease portends poor prognosis. For the patients enrolled into the Total Therapy protocols, a 5-year OS was 31 % for those with EMD documented by baseline PET/CT, compared with 59 % for those who did not have EMD [40]. Several cytogenetic abnormalities, especially those associated with proliferation and aggressive disease, such as t(14;16), t(14;20), and del(17p), has been associated with EMD [41, 42]. High-risk molecular subtypes such as BRAF V600E is up to four times more frequent in EMD compared to marrow disease [37]. As plasmacytomas are a manifestation of EMD, patients presenting with this condition generally have a shorter OS [4346].
Plasma Cell Leukaemia
Plasma cell leukaemia (PCL) represents a unique subset of patients with MM and can be associated with del(13q) and del(17p) [47] and MYC translocation [42]. These patients tend to do poorly compared to other MM patients as demonstrated in a large series, where the median OS was only 16 months despite treatment with novel agents [48].


The type of imaging modality is more important than ever with the updated IMWG criteria for diagnosis of MM. The new criteria consider any bone lesions seen on skeletal survey (SKS), CT, and PET/CT significant enough to make a diagnosis of MM [49••].
Whole Body MRI (WBMRI)
In several studies including one of the largest using MRI, additional skeletal abnormalities were detected by MRI compared to SKS [50, 51]. This is because MRI has high sensitivity for the early detection of marrow infiltration by myeloma cells. IMWG regards more than one focal lesion of a diameter >5 mm as diagnostic of symptomatic MM [52].
To be of prognostic significance, seven or more lesions are needed before MRI can be an independent prognostic factor of survival [51]. The authors reported that the resolution of focal lesions correlated with superior OS [51]. The number of focal lesions on WBMRI significantly correlates with and predicts OS in patients in the context of ASCT [53]. WBMRI has been shown to predict risk of relapse after ASCT if residual disease is detected [53, 54].
Studies have shown that patients with a diffuse pattern of infiltration had features of more advanced disease and poorer OS [5557]. It is thought that the diffuse pattern was associated with the extent of angiogenesis and increased microvessel density in the BM. Moulopoulos et al. demonstrated that there was an improvement of survival in patients who had a diffuse pattern of bone marrow involvement on MRI if they were treated with novel agents [58]. In contrast to these findings, a study from the University of Arkansas involving MM patients treated with a tandem ASCT-based protocol found that heterogeneity of diffuse bone marrow signal did not emerge as an independent adverse feature for OS on multivariate analysis.
PET/CT can detect MM osteolytic lesion with a sensitivity of 80–90 % and specificity 80–100 %. PET/CT can detect more lesions than MRI, with the exception of spinal disease. FDG PET/CT can predict response to ASCT, in that a negative FDG PET/CT after autologous SCT can be considered predictive for non-relapse and, conversely, a short time-to-relapse can be anticipated in patients in whom significantly increased metabolic activity is still persistent after therapy [5961]. The presence of more than three FDG-avid focal lesions is an independent parameter associated with decreased OS and EFS [62]. Many have reported that the presence of extramedullary lesions and the increased value of SUVmax predicted poor prognosis with shorter OS [40, 61, 63].

Laboratory Measurement

Lactate Dehydrogenase
Serum lactate dehydrogenase (LDH) was not included in the ISS staging system because it was not found to be an independent prognostic factor on multivariate analysis [64]. This finding was challenged by other studies which have found that LDH does provide a convenient and dependable prognostic indicator in MM, and a high LDH is associated with worse outcomes [65, 66].
Measurement of Plasma Cell Proliferation
Plasma cell labelling index (PCLI) has a prognostic value, but this technique is expensive and can only be performed in a few laboratories. Cell morphology has also been considered a prognostic variable in MM but the drawback of this method is the considerable interobserver variability. A test looking at the fractal dimension on chromatin of May-Grunwald–Giemsa-stained myeloma cells has shown prognostic significance, where a higher fractal dimension is associated with poorer survival [67]. This test would need to be validated but may be a simple, inexpensive alternative.
Free Light Chain
Measurement of serum free light chain (sFLC) has long been established as part of the diagnostic workup of plasma cell dyscrasias. The quantity of the free light chain measured and the kappa to lambda ratio (rFLC) has prognostic significance from MGUS to active MM. One of the new diagnostic criteria for MM is a serum involved/uninvolved free light chain ratio of 100 or greater, provided the absolute level of the involved free light chain is at least 100 mg/L [49••]. Solitary plasmacytoma also demonstrated that an abnormal rFLC and a persistent serum M protein level of ≥0.5 g/dL was an additional risk factor for progression to MM. Using these two variables, a risk-stratification model was constructed with low-risk, intermediate-risk and high-risk groups having a 5-year progression rate of 13, 26 and 62 %, respectively (P < 0.001) [68, 69]. Even in active MM, several studies have shown that baseline sFLC is prognostic for survival in patients with newly diagnosed active myeloma [6873]. Van Rhee et al. have also demonstrated that among 301 patients enrolled to receive Total Therapy III, those with the highest levels of FLC—greater than 750 mg/l, had the poorest outcomes [71].
There is a concern that the widely used ISS staging system may not prognosticate accurately for ethnic Chinese patients, as there is less accuracy for determination of disease stages in Chinese patients with MM especially between stages II and III [74]. A recent study confirms that both sFLC and rFLC has prognostic significance in MM for Chinese patients as seen in European patients [68, 69, 7173, 75]. The authors of the study suggested combining rFLC to the current ISS, using rFLC (<0.04 or >25), serum β2M (>3.5 mg/L) and serum albumin (<35 g/L) as three adverse risk factors. Patients in the low risk group (no factor, the low-intermediate risk group (one factor), the high-intermediate risk group (two factors) and the high-risk group (three factors) had a median OS of NR, NR, 24 months and 13 months,respectively (p < 0.05) [75].
Heavy Light Chain Specific Immunoglobulin Ratios
Traditional screening tests for monoclonal gammopathies are serum protein electrophoresis (SPE) with scanning densitometry and/or immunofixation electrophoresis (IFE) together with serum free light chain immunoassays. While SPE is a simple, low cost test, it is not very sensitive and quantification of proteins at low concentrations (1–3 g/L) is inaccurate. Improved sensitivity is achieved with IFE, but IFE is a non-quantitative assay. A new method was recently developed and validated for the separate quantification of the kappa- and lambda-bounded amounts of circulating IgG and IgA [7678]. This was achieved by developing antibodies targeting unique junctional epitopes between the heavy and light chains of each Ig molecule. Studies have shown that involved Ig (iHLC) is related to parameters of disease activity, severity, and tumour burden, including anaemia, sFLCR, and extensive bone marrow infiltration [7780]. Koulieris et al. also concluded that the ratio of involved/uninvolved Ig (HLCR) correlated with time to treatment and, most importantly, with outcome; thus, patients with ‘high’ HLCR had a significantly shorter OS [79]. The impact of HLCR on PFS and OS has since been confirmed by others [77, 78, 81, 82]. Ludwig et al. concluded that patients who are deemed having achieved CR by conventional testing (SPEP, NEPH and IFE) may still have residual disease by HLC testing. Vice versa, conversion of a normal to an abnormal HLCR indicated evolving relapse, weeks or months before relapse became evident by conventional methods [78].

Host Factors

Prognosis in MM also depends on host factors (age, performance status, frailty, comorbidities and response to therapy) which will be discussed below.


Increasing age has been purported to be a poor prognostic factor in MM. In a recent study of 682 newly diagnosed patients, 23 % were 80 years or older [83]. These patients had worse performance status, advanced ISS, and a median OS of 22 months, with 14 % dying within 2 months of initial therapy [83]. In a meta-analysis of 1435 patients 65 years or older enrolled in four European phase III trials that included thalidomide and/or bortezomib, the risk for death was increased in patients 75 years or older (HR, 1.44; 95 % CI, 1.2–1.72; P < 0.001) in multivariate analysis [84]. The inferior outcome seen may be that older patients have more comorbidities, their MM is more aggressive biologically, or they tolerate therapies more poorly. The observation by San Miguel et al. that the initial performance status had a highly significant relationship with survival can be generalised to reflect age as it tends to be associated with performance status [85]. However, even in the ‘younger’ myeloma patients (60–65 years of age), there is evidence that age affects the outcome with more ISS II and III in the ‘older’ group (60–65 years) vs. the ‘younger’ group (<60 years) [86]. There was no difference in high-risk cytogenetics between the two groups but the ‘older’ group had a higher beta 2 microglobulin (β2m) which in turn, accounted for higher ISS [86]. It is currently unclear why the β2m levels are increased with age.

Frailty Score

In cancer patients, frailty and Comprehensive Geriatric Assessment (CGA) are being incorporated to guide treatment decisions [87] Frailty is a state of cumulative decline in many physiological systems, resulting in a diminished resistance to stressors, such as cancer and its treatment [8890]. It must not be confused with age and performance status. Frailty profiles were associated with a shorter OS, regardless of staging and treatment administered. The higher risk of death for disease progression was related with lower dose intensity as a consequence of higher rate of drug discontinuation and/or dose reduction and a higher cumulative incidence of nonhematologic toxicities and drug discontinuation [84].
By combining the frailty score with the established ISS, the 3-year OS rate was 55 % in the frail-ISS 3 group, and 94 % in the fit-ISS 1 group. The combination of these two independent parameters significantly improved the prognostic value of each and is an important strategy for predicting outcome. Although not evidence based, treatment can be tailored to frailty scores, with fit patients receiving intensive therapies including stem cell transplant, intermediate-fitness patients treated with doublet treatments or less intense triplets and frail patients receiving gentler, reduced-dose therapy or even a palliative/supportive treatment.

Response to Treatment

There are data to show that depth of response is associated with better outcome. Most studies showed that achieving good response is an important surrogate for improved OS [9193]. In fact, in the Spanish [92] and Italian [94] trials, the achievement of complete remission (CR) has a significant impact on survival when compared with very good partial response only. The Arkansas group and our own data have demonstrated that it is not only the achievement of CR but the ability to sustain the CR that is important for overall survival [9597]. Loss of CR within a short time of achieving it (early relapse) is associated with a worse outcome compared to those who never achieved a CR. IMWG and EBMT harmonised their response criteria and a new category of stringent CR (sCR) was established as a way to measure the quality of response, although there was prognostic significance. sCR is defined as having normal rFLC, absence of clonal cells in the bone marrow, negative IFE on the serum and urine, disappearance of any soft tissue plasmacytomas, and <5 % plasma cells in the bone marrow [98]. Despite its wide use as a clinical endpoint, only one study [99] has reported a benefit of sCR over CR,whereas other studies suggested that the rFLC do not provide additional prognostication [100102]. The issue is further complicated by oligoclonal bands that develop post ASCT that causes a false positive sFLC result [103].
A more robust measure of quality remission is the achievement of immunophenotypic remission [104, 105] and molecular CR [106, 107] with multiple studies demonstrating association with longer survival. The prognostic implications of achieving molecular CR have been discussed elsewhere in this journal. A detection of disease by MRI [53] or PET/CT [61] after treatment is associated with a high risk of relapse. This suggests that response to treatment as well as depth of response is an important prognostic factor. The best treatment to bring a patient into deep remission as well as the best measure of response is still unclear. It is conceivable that a combination of measures would be needed to define the deepest response.

Current Prognostication Systems

There are many known prognostic factors in MM and several different risk stratification systems have been described in MM. Most of these risk stratification systems utilised prognostic factors that were described before the advent of novel agents and it is questionable if they still hold true as novel agents are able to overcome the high-risk features defined by one or more of these prognostic factors. They also do not help in making therapeutic decisions.

Durie-Salmon Staging System and International Staging System

The Durie-Salmon (DS) staging system was developed 30 years ago to provide a practical way to measure MM tumour burden [108]. It measures calcium, renal function, haemoglobin, and the number of lytic lesions. This system is predictive of clinical outcome after standard-dose chemotherapy, but one of the limitations of the DS staging system is the interobserver variability in the number of lytic lesions seen on a skeletal survey. Moreover, the DS staging system predates magnetic resonance imaging (MRI) and positron emission tomography-computed tomography (PET/CT), which are much more sensitive in detecting osseous disease. In 2003, the IMWG introduced the newer Durie-Salmon PLUS (DS+) staging system, which takes into account MRI and PET/CT findings [109, 110]. Comparing the two systems, the DS staging system is based on radiographic findings and clinical parameters, whereas the DS+ staging system relies mainly on modern imaging findings and does not include haemoglobin and immunoglobin levels and may be at a disadvantage in terms of clinical correlation.
In an effort to develop a more clinically relevant, objective and feasible staging system for MM, the International Staging System (ISS) was proposed. A number of statistically significant parameters were analysed and serum β2M and serum albumin were the most consistent, broadly applicable prognostic factors correlated with survival duration. Other parameters (low platelets, high serum creatinine, age, low haemoglobin, calcium, lactate dehydrogenase, and bone marrow plasma cell percentage) had a significant hazard ratio but they identified small patient subsets compared to β2M and serum albumin [64]. Cytogenetic abnormalities that can be identified by FISH have prognostic implications but practical application of this technique has been hampered by lack of standardisation, costs, and restricted availability (at the time of publication of the ISS staging system). Furthermore, to enable the staging system to be widely adopted, the consensus was to use parameters that are routinely used and easily available.
The ISS segregates patients with MM into stages I, II, and III which have an associated median OS of 62, 44 and 29 months, respectively [64]. The ISS is based on the measurement of serum albumin and β2M levels. However, cutoff levels remained a matter of controversy because renal failure could elevate β2M levels even in patients with low tumour burden. Therefore, this system cannot consistently provide a good estimate of tumour burden.

Development of a Better Stratification System

Combining iFISH with other prognostic factors has been examined by several groups. Previously published results by the Intergroupe Francophone du Myélome [10, 11], one of the German groups, [111] and Mayo Clinic mSMART model [112] showed that using iFISH produced a good separation of the patients, based on outcome. Avet-Loiseau et al. combined ISS with cytogenetic abnormalities to improve stratification of MM patients [11]. There was a good differentiation of patients with the 4-year PFS 39, 20, and 11 % and OS 71, 45, and 33 % for ISS-iFISH groups I, II and III, respectively. Although iFISH would be offered widely, cost may be a limiting factor for widespread use.


As mentioned previously, LDH is a relevant biomarker in MM. Recent data from three separate myeloma trials has shown that patients with t(4;14) and/or del(17p) in addition to ISS stage III and/or high levels of LDH are at high risk of progression-related death despite modern treatment strategies [113••]. These findings demonstrate that the prognostic accuracy of the ISS can be enhanced through the incorporation of both biochemical and genetic parameters. The new stratification system proposed the revised international staging system (R-ISS): R-ISS I, (ISS stage I, no high-risk CA [del(17p) and/or t(4;14) and/or t(14;16)], and normal LDH); R-ISS III, (ISS stage III and high-risk CA or high LDH level; and R-ISS II, including all the other possible combinations (Table 1). At a median follow-up of 46 months, the 5-year OS rate was 82 % in the R-ISS I, 62 % in the R-ISS II, and 40 % in the R-ISS III groups; the 5-year PFS rates were 55, 36, and 24 %, respectively. The limitations of this new stratification system are the exclusion of chromosome 1 abnormalities as a prognostic parameter, no interlaboratory standardisation of FISH analysis, and heterogeneous cutoff levels for LDH and cost.
Table 1
Revised International Staging System (R-ISS) [99]
High-risk CA [del(17p) and/or t(4;14) and/or t(14;16)]
LDH level
aif either positive


There are many variables that affect risk stratification. Table 2 summarises the main factors.
Table 2
Summary of main prognostic variables
Tumour factors
Cytogenetic abnormalities
•Stratifies patients into high risk or standard risk
Tumour burden
•Extramedullary disease and plasma cell leukaemia presentation confers worse prognosis
•>1 lesion of a diameter >5 mm is diagnostic of MM
•Diffuse pattern of infiltration had features of more advanced disease and poorer OS
•Good for detecting EMD
•EMD and increased SUVmax associated with poorer OS
•Newly incorporated into the revised ISS as a independent prognostic factor
Host factors
Frailty score
•Frailty profiles are associated with poorer OS
•Can help tailor treatment
Response to treatment
•Depth of response is associated with better outcome
•Immunophenotypic and molecular CR are more robust in predicting OS
It is important to predict outcome accurately by using simple and easily applicable tools. This has implications in patient counselling and risk-adapted treatment. Furthermore, development of a uniform risk stratification system allows better comparison of patient groups across different trials. The current ISS staging system is outdated as the Revised-ISS can prognosticate MM patients more accurately.
The challenge would be to incorporate other important prognostic factors into the current risk stratification systems. It is clearly demonstrated that GEP is important in risk stratification of MM but the lack of universal availability impedes its inclusion in the current staging system. Treatment response and further modification of the stratification systems based on response attained will also be important to evaluate. Eventually, establishing a staging system that is able to prognosticate accurately and enable a risk-adapted treatment strategy that is modifiable based on response achieved would be an important goal to advance our current management approaches in MM.

Compliance with Ethical Standards

Conflict of Interest

The authors declare that they have no competing interests.

Human and Animal Rights and Informed Consent

This article does not contain any studies with human or animal subjects performed by any of the authors.

Related topics

Image Credits