The second revision of the International Staging System (R2-ISS) stratifies progression-free and overall survival in multiple myeloma: Real world data results in an Australian and New Zealand Population.

Abstract

To evaluate the performance of the second revision of the International Staging System 2 (R2ISS)1 in a real-world cohort, we applied the scoring system to patients with newly-diagnosed multiple myeloma (NDMM) in the Myeloma and Related Diseases Registry (MRDR). Our analysis establishes the validity of this risk scoring system in a real-world setting, in addition to being able to further risk-stratify the Revised International Staging System (R-ISS) group II category. The R2ISS is a recently described update to the established R-ISS for NDMM patients. In addition to β2-microglobulin, albumin, lactate dehydrogenase (LDH), del17p, and t(4;14) by fluorescence-in-situ hybridisation (FISH) included in the R-ISS, the R2ISS additionally considers gain of chromosome 1q (1q+), and adds additional weight when more than one high-risk cytogenetic abnormality is present.2 The cytogenetic abnormality t(14;16) is no longer included in the R2ISS. The impetus for this revision was poor discrimination within the intermediate risk stage (stage II) by the R-ISS, which categorises more than 50% of patients within this one group despite heterogeneous disease outcomes.3 The R2ISS stratifies patients by assigning a score value (in square brackets) to the included prognostic variables: ISS II [1], ISS III [1.5], del(17p) [1], high LDH [1], t(4;14) [1], and 1q+ [0.5]. The sum of these scores determines the risk group: I low (0), II low/intermediate (0.5–1), III intermediate/high (1.5–2.5) and IV high.3-5 A limitation of the creation of this system was its derivation from a platform that exclusively enrolled clinical-trial participants. Such participants are highly selected based on trial eligibility criteria and receive myeloma therapy that by definition deviates from current standards of care. Thus, prior to widespread adoption, there is a need to validate this risk stratification system within a non-clinical-trial ‘real-world’ population. To assess its real-world value, we retrospectively applied R2ISS scoring to patients in the MRDR, that prospectively collects data on NDMM patients from 56 Australian and New Zealand centres.4 All NDMM patients at each participating site are invited to join the registry, and their data are included unless they opt off, which is rare.5 Progression-free survival (PFS) was defined as the time from diagnosis until disease progression or all-cause mortality. Overall survival (OS) was defined as the time from diagnosis until death from any cause. FISH for del17p, (4;14) and 1q+ was performed at the designated central reference laboratory for each state/country on CD138+ enriched cells. The cut-off for FISH positivity differs between sites, ranging from 10% to 15% for IgH translocations and from 10% to 20% for numerical aberrations. The Kaplan–Meier method was used to calculate curves for PFS and OS, with groups compared using the log-rank test. Of the 3483 patients diagnosed from January 2012 to February 2022, complete staging data (ISS, LDH and FISH) was evaluable for 1289 (37%) patients. Disease progression and survival data were available for 1272 (36.5%) and 1275 (36.7%) patients respectively. Median follow-up was 32 months. Patient demographic and clinical characteristics are described in Table 1. R2ISS score-stratified PFS is shown in Figure 1. Median PFS in R2ISS group I was 40.2months [95% confidence interval (95% CI) 34.5–51.5 months], in group II it was 42.3 months (95% CI 36.1–50.6 months), in group III it was 24.0 months (95% CI 21.6–27.4 months), and in group IV it was 21.0 months (95% CI 14.1–25.4 months). There was a significant difference in PFS between R2ISS risk groups I versus III [hazard ratio (HR) 2.0 (95% CI 1.5–2.6), p < 0.001] and groups I versus IV [HR 2.9 (95% CI 2.0–4.1), p < 0.001], but not group I versus II [HR 1.0 (95% CI 0.8–1.4), p = 0.761]. R2ISS score-stratified OS is shown in Figure 1B. Median OS was not reached (NR) by 84 months in groups I and II (95% CI 73.4–NR months and 90.8–NR months respectively), in group III it was 51.3 months (95% CI 45.4–67.0 months), and in group IV it was 48.5 months (95% CI 41.9-NR months). There was a significant difference in OS of R2ISS risk groups I versus III [HR 2.9 (95% CI 2.0–4.2), p < 0.001] and of groups I versus IV [HR 3.5 (95% CI 2.1–5.7), p <0.001], but not of groups I versus II [HR 1.1 (95% CI 0.8–1.7), p = 0.526]. We further assessed the redistribution of R-ISS groups following R2ISS recategorisation. Of the patient cohort, 873 (67.7%) were categorised as R-ISS stage II. When recategorised by R2ISS these patients were redistributed across all four R2ISS risk groups: 0.9% (8/873) in group I, 47.8% (417/873) in group II, 48.1% (420/873) in group III, and 3.2% (28/873) in group IV, underscoring the heterogeneity of the R-ISS's stage II classification. It is worth noting that the recategorisation of patients in R-ISS group II as R2-ISS group I was due to t(14;16) not contributing to R2-ISS scoring. We analysed the outcomes of the R-ISS II risk group alone according to the R2ISS scores. R2ISS risk group II was used as a baseline comparative population due to low numbers in R2ISS risk group I. We found statistically significant differences in the outcomes of R2ISS risk group II versus III [PFS HR 1.7 [95% CI 1.4–2.1], p < 0.001; OS HR 2.1 (95% CI 1.6–2.9), p < 0.001] and of R2ISS risk group II versus IV [PFS HR 2.3 [95% CI 1.4–3.7], p = 0.002; OS HR 2.3 (95% CI 1.1–4.6), p = 0.020] in both PFS and OS (Figure 1C,D). To our knowledge, our study is the first study externally validating the R2-ISS risk stratification system with no additional published studies using this system to date. Our study has a number of strengths, including its large sample size from 56 sites across two countries. In addition, the MRDR's linkage to national death registries ensures the reliability of our survival outcome data. While a limitation of our analysis relates to the limited proportion of our real-world patients with all required staging data points, we believe the evaluable population is representative due to the broad inclusion criteria for the MRDR and the proportional representation of ISS and R-ISS staging categories within this population being consistent with those found in previous studies.2, 6 R2ISS was evaluable in 37% of patients, comparable to the D'Agostino training and validation cohorts at 2226/7072 (31.4%) and 1214/3771 (32.2%) respectively. There was no difference in PFS (p=0.61) between the evaluable and non-evaluable groups. Median OS was fractionally longer in the evaluable group (78.9 months, 95% CI 72.4–87.8 months) than the non-evaluable group (70.1 months, 95% CI 66.2–76.7 months; p = 0.009) which appears to relate to a younger median age in the former group than the latter (68.9 vs. 66.5 years respectively; p < 0.001). FISH was the most common missing factor (71.7%). This illustrates a potential limitation of this measure as a prognostic indicator for retrospective studies outside of a trial environment, and serves as a pertinent reminder to revise national and international guidelines recommending complete staging (including cytogenetic analyses) be performed at diagnosis. The greater availability of complete staging data would allow risk stratification systems such as the R2-ISS to be fully evaluated in future populations. The R2ISS was able to stratify our cohort PFS and OS into four risk groups; however, the clear distinction achieved between R2ISS groups I versus II and III versus IV was not reflected in this real-world analysis. This observation may be explained by the difference in up-front treatment approaches between our cohort and that of D'Agostino et al. (1), where the majority received either immunodulatory drugs (IMiDs) alone (training group 23%; validation group 87%) or in combination with a proteasome inhibitor (PI; training group 67%; validation group 13%), whereas the majority of our patients received either a PI alone (72.5%) or in combination with an IMiD (19.8%). The shorter length of follow-up (32 months vs 75 months in the D'Agostino cohort1) may also have been insufficient to observe a difference in survival between groups I and II. The four-risk-group system that the R2ISS proposes aids the design of future interventional trials by facilitating population dichotomisation (i.e. pairing R2ISS group I with II and group III with IV; Figure 1E,F). Our analysis of OS according to the R2ISS, where similar outcomes were experienced by groups I and II and by groups III and IV, is supportive of this. We did not see a significant difference in either OS or PFS between patients in R2ISS groups I and II, but this may reflect a lack of power in our cohort to resolve the difference of the magnitude observed by D'Agostino et al. Our analysis also concurs with the prior demonstration that patients in R2ISS groups III and IV have a particularly poor outcome with PFS and OS markedly lower when compared to the lower-risk groups, therefore representing a subset that requires novel approaches. In summary, we show that the R2ISS score effectively differentiates the R-ISS stage II category, addressing a shortfall of the R-ISS system. However, in this large real-world myeloma population, the clear distinction between R2ISS groups I versus II and III versus IV was not reproduced. Modifications to the ISS/R-ISS systems should similarly be derived and/or assessed within real-world populations in addition to, or potentially in preference to, the highly selected populations that typically comprise clinical trials. Tan Joanne L. C. Tan wrote the paper. Cameron Wellard analysed the data. Elizabeth M. Moore, Peter Mollee, Rajeev Rajagopal, Hang Quach, Simon James Harrison, McDonald Emma-Jane, P. Joy Ho, H. Miles Prince, Bradley M. Augustson, Philip Campbell, Zoe K. McQuilten and Erica M. Wood contributed substantially in the writing and editing of the draft. Andrew Spencer is the leading supervising author and has contributed in the analysis and writing of the paper. The authors thank the patients, hospitals, clinicians and research staff at participating institutions for their support of the MRDR. The Myeloma and Related Diseases Registry has received funding from the following companies: Abbvie, Amgen, Antengene, Bristol-Myers Squibb, Celgene, Gilead, Janssen, Novartis, Sanofi, & Takeda. All authors have no conflicts of interest to declare.