Introduction

Since the introduction of proteasome inhibitors (PIs), immunomodulatory (IMIDs) drugs, and other novel agents, a continuous improvement of survival has been shown in multiple myeloma (MM), predominantly in younger patients [1, 2]. Novel agent-based induction, followed by high-dose melphalan and autologous stem cell transplantation (ASCT) is the standard of care for newly diagnosed, transplant-eligible, MM patients based on randomized trials showing improved progression-free survival (PFS) and overall survival (OS) [3,4,5,6,7].

MM is a heterogeneous disease with variability in response to treatment and survival due to the interaction between host factors and those intrinsic to disease biology [8]. In this setting, there is a continued need for and interest in devising reliable prognostic tools, not only in order to provide the accurate prognostic information possible to patients but also for adopting risk-adapted strategies to improve their survival and quality of life.

Several risk-stratification models have been developed. In 2005, the International Staging System (ISS) emerged, based solely on the serum albumin and β2-microglobulin concentration. Although the ISS has been widely validated, its prognostic value in the era of novel agents and transplant settings needs to be readdressed [9,10,11]. Evaluation of cytogenetic abnormalities (CA) and levels of lactate dehydrogenase (LDH) are important as these reflect disease biology, and both markers have independent prognostic value [12, 13]. Several studies have shown that high-risk CA, such as the presence of chromosome 14 translocations or 17p abnormalities, are the most important prognostic biomarkers in MM patients [14,15,16]. Other studies proposed a risk stratification model based on the combination of both host and disease factors: ISS and CA [17, 18] or LDH [19]. Finally, a new Revised-International Staging System (R-ISS) was developed by the International Myeloma Working Group (IMWG) that integrates the aforementioned ISS, LDH, and high-risk CA, resulting in three groups with different outcomes [20].

However, there is limited data on the applicability of this prognostic model in the setting of novel induction therapy followed by ASCT. Thus, the aim of this study is to determine whether the new R-ISS is a valid risk model for predicting survival in a cohort of unselected transplant MM patients, treated at Mayo Clinic in Arizona and the University Hospital of Salamanca in Spain. We show that R-ISS model allows a more accurate stratification of the transplant patients, identifying a high-intermediate risk group based on the presence of high-risk CA in the R-ISS II group.

Methods

We retrospectively studied a cohort of newly diagnosed MM patients who underwent ASCT and were followed up at Mayo Clinic in Arizona and University Hospital of Salamanca in Spain from January 2005 to December 2014. Both Institutional Review Boards approved the study and it was performed in accordance with the Declaration of Helsinki. Informed consent was obtained from all individual participants included in the study. To be eligible, we required a complete dataset including ability to determine ISS, LDH, and CA. Since these institutions are tertiary referral centers many patients were excluded as fluorescence in situ hybridization (FISH) was either not done, done without selection of plasma cells by cell separation or cIg technique, or LDH or β2-microglobulin data were lacking. Patients were often seen after induction therapy had been initiated by the referring physicians and obtaining testing for FISH at that point is not informative. All patients were required to have PI- or IMIDs-based therapies as induction treatment.

Baseline data were collected by searching at medical records database of Mayo Clinic and University Hospital of Salamanca and included data required for the assignment of the R-ISS described by Palumbo et at. [20]. Serum LDH level was classified as normal or high according to the normal range given by the local laboratory. High LDH was defined as higher than the upper limit of normal range and normal LDH was defined as a serum level lower than the upper limit of normal. Cytogenetic evaluation was performed by FISH as previously published [12, 14]. Response to treatment was evaluated according to International uniform response criteria for MM 2006 [21].

The primary and secondary endpoints were OS and PFS from ASCT, respectively. OS was considered the time from date of transplantation to death. PFS was defined as the time from date of transplantation to relapse, progression or death, regardless of cause. Patients without a recorded progression or death date were censored for PFS or OS at their last follow-up. Survival curves were plotted by Kaplan-Meier method, with differences assessed with the log-rank test. Effects of potential risk factors of progression were analyzed in a Cox proportional hazards model. The chi-square, t student and Mann-Whitney U tests were used to estimate the statistically significant differences among baseline characteristics. P values were considered at the conventional 5% significance level. All statistical analyses were performed using IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp.

Results

Patient characteristics

In all, 134 patients had complete data and subsequently were included in the study. A total of 92 out of 509 patients who consecutively underwent ASCT at Mayo Clinic and 42 out of 107 from the University Hospital of Salamanca. The baseline characteristics are summarized in Table 1. There were 70 (52%) men and 64 (48%) women. The median age at diagnosis was 62 years (range, 29–78 years), and 41 (45%) patients were older than 65 years at the time of ASCT. Fifteen (11%) patients presented with renal impairment (RI) at the time of diagnosis.

Table 1 Baseline characteristics of 134 myeloma patients who underwent autologous stem cell transplantation (2004–2014)
Full size table

According to ISS, 62 (46%) patients had stage I, 38 (28%) stage II, and the remaining 34 (25%) stage III disease at diagnosis. In addition, there were 39 patients (29%) with high-risk CA: 18 patients (14%) with del17p; 17 patients (13%) with t(4;14); and 7 (5%) with t(14;16). Seventeen patients (13%) had high LDH levels. Consequently, patients were re-staged at diagnosis according to the R-ISS, resulting 44 patients (33%) with stage I, 75 (56%) with stage II, and 15 (11%) with stage III. Thus, 18 patients previously categorized as having low risk (ISS I) and 15 patients as high risk (ISS III) were reclassified as intermediate risk (R-ISS II), according to the new revised staging system (Table 2).

Table 2 Clinical features according to R-ISS stage at diagnosis
Full size table

ASCT features and treatment response

Median time from diagnosis to ASCT was 6.7 months (range 4–69 months, Table 3). All patients received induction therapy before ASCT; 104 (78%) patients received PI-based therapy, with CyBorD (cyclophosphamide, bortezomib, and dexamethasone) as the preferred choice. One hundred and eleven (83%) patients responded to first line of induction and 23 (17%) were initially refractory and received more than one line of induction treatment (Table 4). In the majority of cases, high-doses of melphalan were used as conditioning, followed by infusion of autologous stem cells, with the median CD34+ cells-dose infused over 2 × 106/kg. Median times to engraftment were 11 (range 5–19) and 17 (range 7–34) days for neutrophils and platelets, respectively. Only one case of graft failure was reported.

Table 3 Characteristics of Autologous stem cell transplantation, Mayo Clinic (Arizona) and University Hospital of Salamanca (2004–2014)
Full size table
Table 4 Induction and maintenance treatment in myeloma patients who underwent autologous transplantation at Mayo Clinic in Arizona and University Hospital of Salamanca (2004–2014)
Full size table

Most patients underwent ASCT in at least partial response (PR), and there was an improvement in response category after ASCT, with an increase of complete response (CR) rate from 43 (32%) patients to 72 (54%) patients in CR before and after ASCT, respectively.

Survival analysis

The median follow-up was 59.6 months (range, 7.3–135.4 months). Forty-one (31%) patients died and 93 (69%) patients progressed or died at any time after ASCT. The median OS was 110.9 months (95% CI 86.6–135.2 months) from ASCT and the median PFS was 34.5 months (95% CI 29.7–42.1 months) from ASCT.

In our series, there were no statistically significant differences in terms of OS or PFS among groups according to ISS classification. However, patients classified as having high-risk CA showed a statistically significant shorter median OS than those with standard-risk CA 57.4 months vs. median OS not reached (NR) (P = 0.001), respectively. Patients with high LDH levels had a significantly inferior OS, 36.1 vs. NR (P < 0.001), compared with patients with normal LDH levels at diagnosis.

As expected, patients with R-ISS III had a significantly shorter median OS from ASCT compared to patients with R-ISS II or R-ISS I (37.2 vs. 110.9 months vs. not reached, respectively, P = 0.0001) (Fig. 1). The probability for 5-year OS was 28, 68, and 86% per R-ISS III, II, and I, respectively. Although there were more cases with renal impairment and anemia at diagnosis in the R-ISS III group, no statistically significant differences in baseline characteristics were identified among groups to explain the differences in OS observed, as shown in Table 2. Patients who belonged to R-ISS III group had also a significantly shorter median PFS than the R-ISS II and R-ISS I groups: 19.2 vs. 36.1 vs. 35.2 months, respectively (P = 0.05) (Fig. 2).

Fig. 1
figure 1

Overall survival (OS) by R-ISS risk group. OS was defined as time from date of transplantation until date of death. There were statistically significant differences in OS with median OS not reached, 110.9 and 37.2 months in R-ISS risk groups I, II, and III, respectively. The probability of OS at 5 years was 28, 68, and 86% for R-ISS III, II, and I, respectively

Full size image
Fig. 2
figure 2

Progression free survival (PFS) by R-ISS risk group. PFS was defined as time from date of transplantation until date of relapse, progression or death. Shorter PFS in R-ISS risk group III versus risk groups I or II was observed—median PFS 19.2 vs. 35.2 vs. 36.1 months, respectively

Full size image

In order to explore whether the presence of high-risk CA had an impact on the survival of the R-ISS II group, we re-classified R-ISS II into two subgroups according to the presence or not of high-risk CA and we compared both of them with either R-ISS I and III. As a result, we had four groups with significantly different OS: 44 (33%) patients with R-ISS I and a median OS NR; 50 (37%) patients with R-ISS IIa (not R-ISS stage I or III and absence of high-risk CA) with a median OS of 110.9 months; 25 (19%) patients with R-ISS IIb (not R-ISS stage I or III but presence of high-risk CA) with a median OS of 69.8 months; and 15 (11%) patients with R-ISS III and a median OS of 37.2 months, P < 0.001, (Fig. 3). Although there were no statistically significant differences among groups, there was a trend towards shorter PFS in the R-ISS IIb group compare with PFS in either R-ISS I or IIa: 32.9 vs. 35.2 vs. 48.2 months, respectively (P = 0.06).

Fig. 3
figure 3

Overall survival (OS) by R-ISS, incorporating new subclassification of R-ISS group II by presence or absence of high risk chromosomal abnormalities. R-ISS group II was subclassified into groups IIa (defined as not group I or III and lack of high risk chromosomal abnormalities (t(4;14), t(14;16) or del17p) and groups IIb (defined as not R-ISS group I or III and presence of high risk chromosomal abnormalities). Median OS was not reached, 110.9 months, 69.8 months and 37.2 months in R-ISS groups I, IIa, IIb, and III, respectively

Full size image

Additional risk factors for OS were identified in the univariate analysis such as renal impairment at diagnosis (median OS 37.2 months; P < 0.001) and ≥ 2 lines of induction treatment (median OS 60.9 months; P = 0.005) (Table 5). All patients were included in the multivariate analysis and R-ISS was selected as an important independent predictor for OS. An increase HR of 6.9, 95% CI 2.4–20.3; P = 0.001 was observed between R-ISS III and R-ISS I, and a HR of 1.5, 95% CI 0.6–3.6; NS, between R-ISS II and R-ISS I. Another discriminating factor for OS was ≥ 2 lines of induction before ASCT (HR = 3.8, 95% CI 1.7–8.5; P = 0.001).

Table 5 Univariate and multivariate analysis of covariates affecting PFS and OS in transplant myeloma patients at Mayo Clinic in Arizona and University Hospital of Salamanca (2004–2014)
Full size table

In addition, independent-risk factors for shorter PFS were identified in the multivariate analysis such as R-ISS III (HR = 2.9, 95% CI 1.3–6.6; P = 0.008) and having received more than 1 line of induction before ASCT (HR = 2.8, 95% CI 1.5–5.1; P = 0.001). The achievement of CR on day 100 after ASCT and having received maintenance treatment were independently associated with longer PFS, as shown in the multivariate analysis in Table 5.

Discussion

We showed in this study that R-ISS is a reproducible and applicable method to robustly predict survival in MM patients who underwent ASCT. This is one of the first evaluations and validations of the R-ISS in transplant MM patients.

We present a representative series of exclusively transplant patients who received novel agents with outcomes comparable to those recently reported in prospective and randomized controlled studies, with 80% of patients achieving very good partial response or better after ASCT, median PFS of 45 months and 4 year-OS of 81% [3, 4, 6, 7].

According to the results recently reported by several studies which evaluate the R-ISS in myeloma patients [20, 22,23,24,25,26], the distribution of patients is quite similar, particularly stage III (10–17% in these studies), with a slightly higher proportion of R-ISS I cases in our series (33% vs. 18–28%). This could be explained by the fact that the aforementioned studies included elderly population as well, with potentially more cases of hypoalbuminemia and chronic renal disease [27]. Regarding survival, although our R-ISS stage III patients displayed a shorter OS than patients with stage III in the study of Palumbo et al. [20], it is important to point out that the original study was developed in a selected cohort of patients enrolled in clinical trials, and our results are closer to those reported by the Greek and the British group in unselected MM patients cohorts [23, 25].

Although only 134 patients had complete data, they were precisely those who were consecutively diagnosed and treated at Mayo Clinic in Arizona and University Hospital of Salamanca, whereas the remaining patients were referred to these centers after induction treatment for the transplant procedure and diagnosis data were not completely available. Thus, our series may be considered a representative sample of newly diagnosed transplant patients.

An immediate consequence of re-staging myeloma patients with the R-ISS model is that patients belonging to either ISS I with additional high-risk features (high LDH or high-risk CA) or ISS III with low-risk features (normal LDH and standard-risk CA) are now reclassified as R-ISS II, with accurate identification of low and high-risk patients according to their OS: median NR and 37 months, respectively. As showed in the above studies, there is a higher proportion of patients categorized as R-ISS II, including patients with high-risk CA belonging to either ISS II or ISS I. Several studies have shown that genetic events have a determinant role in prognosis in myeloma patients [12, 14, 19]. Based on this, we explored the impact of the presence of high-risk CA on R-ISS II survival. This group of patients was split into IIa and IIb groups, according to the absence or presence of high-risk CA, respectively. Interestingly, statistically significant differences in survival were found between IIa and IIb stages: median OS 111 and 70 months, respectively. These findings need to be validated in independent and prospective studies, and perhaps will allow improvement of the risk assessment of not only the low and high-risk transplant myeloma patients, but also the intermediate risk group, identifying low-intermediate and a high-intermediate risk subgroups according to the presence or not of del17p, t(4;14) or t(14;16).

In addition, we compared the prognostic value of R-ISS with other potential predictors of survival in our series of transplant patients and we showed that R-ISS is an independent prognostic marker. R-ISS model employs a feasible evaluation of both host factors and disease biology, encompassing ISS staging, LDH, and genetics. All of these factors should be assessed in the diagnosis work-up according to the IMWG recommendations [28, 29]. ISS was not selected as a predictor of survival in this cohort of patients exclusively treated with novel agents; this result is consistent with other studies [10, 11, 26] and could be explained by the fact that ISS model was developed before 2002 when only a minority of patients had received IMID- or PI-based therapies.

Interestingly, all of the patients were treated with a novel agent-based induction and most of them achieved partial response or better before ASCT; however, those patients who were refractory to the first line of induction and needed at least a second line had a 3-fold higher risk of progression or fourfold higher risk of death after ASCT. In fact, treatment response is a well-known surrogate marker of PFS and, most importantly, OS [30,31,32]. Indeed, response 100 days after ASCT was an independent predictor for PFS, with 60% reduction in risk of progression or death in those who achieved CR after ASCT, as shown in the multivariate analysis performed in our study.

In summary, R-ISS assessed at diagnosis was an independent predictor for OS after ASCT in our series, with median OS for the different R-ISS groups comparable to those reported by Palumbo et al. in their subgroup of younger patients [20]. Thus, this study lends further support for the R-ISS as a reliable prognostic tool for estimating OS in transplant-eligible MM patients with one contribution: the identification of R-ISS II subgroups according to the presence or not of high-risk CA. If this observation is validated in other independent studies, this approach might help to improve the risk-stratification of transplant patients and thus, accurately predict survival and define tailored treatment strategies.