Late versus early response and depth of response are associated with improved outcomes in patients with newly diagnosed multiple myeloma enrolled in the TOURMALINE‐MM2 trial

Abstract Deeper responses are associated with longer survival in multiple myeloma (MM); however, limited data exist on the impact of response kinetics on outcomes. We investigated progression‐free survival (PFS) and duration of response (DOR) by response depth and in early (best confirmed response 0–4 months; n = 424) versus late responders (best confirmed response >4 months; n = 281). Newly diagnosed patients enrolled in TOURMALINE‐MM2 receiving ixazomib‐lenalidomide‐dexamethasone (IRd) (n = 351) or placebo‐Rd (n = 354) were evaluated post hoc. Deeper responses were associated with longer PFS (complete response [CR] not reached [NR], very good partial response [VGPR] 37.2 months, partial response [PR] 16.4 months) and DOR (CR NR, VGPR 42.6 months, PR 15.4 months). Among patients with a PFS (n = 511) or DOR (n = 484) of ≥6 months who achieved ≥PR, median PFS was prolonged among late versus early responders receiving IRd (59.7 vs. 17.9 months) or placebo‐Rd (56.6 vs. 12.4 months), as was median DOR (IRd, NR vs. 20.9 months; placebo‐Rd, 58.2 vs. 11.7 months). While the treatment paradigm for newly diagnosed MM is treatment to progression, our findings suggest slowness of response to a proteasome inhibitor‐immunomodulatory drug‐steroid combination is not a negative predictor of outcome.


INTRODUCTION
The therapeutic landscape for multiple myeloma (MM) has improved with the introduction of novel agents, including proteasome inhibitors (PIs) and immunomodulatory drugs (IMiDs), which have led to an improvement in overall survival (OS) and higher response rates [1][2][3][4].
There is now interest in quality and depth of response to treatment and how these parameters impact outcomes [5,6].
Similarly, a separate analysis from another study in patients with newly diagnosed MM (NDMM) receiving IMiD-or PI-based induction showed that early responders who achieved a best objective response within 3 months had inferior PFS and OS compared with late responders with a response after 3 months [14].
Building on these observations, we have conducted an analysis to further investigate the relationship between depth and timing of response on clinical outcomes in MM.We have used PFS and duration of response (DOR) data from the randomised, double-blind, phase 3 TOURMALINE-MM2 trial (NCT01850524) of IRd versus placebo-Rd in transplant-ineligible patients with NDMM [17].The primary results from TOURMALINE-MM2 showed a clinically meaningful PFS benefit with IRd versus placebo-Rd, although the difference between the two treatments was not statistically significant (median PFS, 35.

Patients
Full methods for TOURMALINE-MM2 have been reported previously [17].Adult patients with a new, confirmed diagnosis of symptomatic MM, according to International Myeloma Working Group (IMWG) 2011 criteria [18], and who were eligible for treatment with Rd but ineligible for autologous stem cell transplantation (ASCT), were enrolled.The trial was conducted in accordance with the International Conference on Harmonization Good Clinical Practice guideline and appropriate regulatory requirements.Local ethics committees or institutional review boards approved the protocol.All patients provided written informed consent.
In addition, all patients received oral lenalidomide 25 mg on days

Assessments
In TOURMALINE-MM2, progression and response assessments were based on central laboratory results and IMWG 2011 criteria [17,18].
Response assessments were performed every cycle until PD, or every 4 weeks in patients who discontinued treatment prior to PD [17].Cytogenetic abnormalities were assessed by a central laboratory using bone marrow aspirate samples taken at screening [17].

Statistical analysis
For this analysis, PFS (time from randomisation to first documentation of PD or death of any cause) and DOR (time from first documentation of best confirmed response to first documentation of PD) were analysed post hoc in subgroups defined by depth of response and by time-tobest confirmed response."Early" and "late" responders were defined by time-to-best confirmed response of 0-4 months and >4 months, respectively.Patients in either subgroup could have recorded an initial response prior to their best response.
As in the main study [17]

Best confirmed response in the intent-to-treat population
In total, 705 patients in the intent-to-treat (ITT) population underwent an assessment of IRC-assessed best confirmed response: 20% had a CR or stringent CR (sCR), 35% achieved a VGPR, 26% had a PR, 10% had stable disease (SD), and 3% had PD.Best-confirmed responses, overall and by treatment arm, are shown in Table 1.

PFS and DOR by depth of best confirmed response
In a pooled analysis of patients from both treatment arms, achieving a deeper response was associated with improved PFS and a longer DOR.
Analyses of PFS and DOR in late and early responders receiving IRd versus placebo-Rd are summarised in Table S2.

Multivariable analysis
Multivariable analyses by Cox proportional hazards modelling were conducted for IRd-treated patients who achieved ≥PR to identify associations between PFS and the timing of response after controlling for baseline covariates.covariates (HR 0.31; 95% CI: 0.21-0.46;p < 0.001; Figure S5).From the predictive model obtained from the multivariable analysis, achieving a late response remained independently associated with longer PFS (HR 0.31; 95% CI: 0.21-0.44;p < 0.001; Figure S6).

DISCUSSION
This post hoc analysis of patients with NDMM in TOURMALINE-MM2 demonstrated that achieving a greater depth of response was associated with prolonged PFS and DOR.The PFS benefit with IRd versus placebo-Rd on the ITT analysis appeared to be driven by higher rates of deep responses, with 63% versus 48% of patients achieving ≥VGPR.
Moreover, the analysis of the effect of time to response on outcomes showed that PFS and DOR were also longer in patients achieving a late versus early best confirmed response of ≥PR or ≥VGPR, with a higher proportion of late responders achieving a deeper response (≥VGPR vs. ≥PR) than early responders.
Consistent with results from a similar analysis in RRMM in the TOURMALINE-MM1 trial [15], our findings indicate that transplantineligible patients with NDMM achieving a late ≥VGPR may have longer PFS and DOR than those achieving an earlier ≥VGPR.Since survival of late responders was guaranteed to be >4 months, PFS estimates may be biased towards late responders.Another potential bias is that achievement of a deeper response may typically take longer, To explore potential biological differences between early and late responders, baseline characteristics were compared in an exploratory analysis.Interestingly, we observed differences in ISS stage, CrCl, cytogenetic risk, and LDH levels; patients who demonstrated high-risk features according to these variables [19] were more likely to be in the early response group.While the comparison of baseline variables was only exploratory, thus limiting any conclusions we can make, our results do support the hypothesis proposed by Garderet et al. [15] that an early response is associated with higher proliferative activity.
Patients with indolent disease may be slower to respond to therapy but an early response, characterised by higher plasma cell proliferation rates, may lead to an initial increased sensitivity to treatment and, consequently higher rates of loss of response and poorer long-term outcomes [15,16].Due to significant differences observed in several baseline characteristics, we performed a multivariate analysis to eval-  [20].
The relationship between MRD and response kinetics was not assessed in the current study, thus further analyses are needed to assess whether patients who respond later are more likely to achieve MRD negativity.Accordingly, our results demonstrate that slowness of response to treatment with a PI-IMiD-steroid combination is not a negative predictor of outcome for patients with NDMM.The potential of treatment adaptation according to response depth and duration has been acknowledged and may lead to a more tailored approach to therapeutic decision-making [21].
The exploratory nature of this analysis is a limitation of the study as our results are inconclusive and require further confirmation.However, our conclusions do support previously reported findings [15] and contribute to the understanding of response kinetics in MM.Analyses of studies involving other classes of MM agents are needed to determine whether response kinetics impact outcomes in a similar manner for other therapies.Encouragingly, recent data on daratumumab-based triplets in RRMM also report better outcomes (PFS and DOR) in late versus early responders [22], supporting the notion that patients without an early response can benefit from continued antimyeloma treatment.Due to the exploratory nature and complexity of the analyses, which are based on treatment until progression and analysis of best response achieved, it is difficult to apply the results clinically, and to understand the relative contributions of kinetics and depth of response.
In conclusion, our results confirm that a greater depth of response was associated with both prolonged PFS and DOR in transplantineligible patients with NDMM.Importantly, we evaluated the relationship between rapidity of response and outcomes and observed that PFS and DOR were longer in patients achieving a late (≥4 months) versus early (<4 months) best confirmed response, and that a higher proportion of late responders achieved a deeper response (CR vs. VGPR or PR) than early responders.Moreover, we demonstrated superior outcomes for late versus early responders in both treatment arms.
Our findings thus indicate that a slow response to therapy may be a powerful prognostic factor and highlight that slowness of response to treatment with a PI-IMiD-steroid combination is not a negative predictor of outcome in NDMM.

F I G U R E 3
PFS lasting ≥6 months by treatment arm in early versus late responders achieving (A) ≥PR or (B) ≥VGPR.CI, confidence interval; HR, hazard ratio; IRd, ixazomib-lenalidomide-dexamethasone; PFS, progression-free survival; PR, partial response; Rd, lenalidomide-dexamethasone; VGPR, very good partial response.and thus late responders would be enriched in deeper responses.To control for these potential biases, we conducted sensitivity analyses in patients with a PFS or DOR of ≥6 months who achieved ≥PR or ≥VGPR; the results confirmed the association of a late response with improved outcomes.Furthermore, in an analysis by treatment based on early versus late responders, our results demonstrate superior PFS for late responders in both arms, although it should also be noted that, in early responders, PFS was longer in patients who received IRd com-pared with those treated with placebo-Rd (≥VGPR median PFS of 28.5 months vs. 21.6).

Table S1 . 3.4 PFS and DOR in early and late responders
Patient demographics and baseline characteristics were generally similar between early versus late responders, although there was an imbalance in ISS stage (stage III, 21% vs. 12%), CrCl (CrCl ≤60 mL/minute, 46% vs. 37%), cytogenetics (expanded high-risk cytogenetic abnormalities, 44% vs. 33%; presence of amp(1q21), 34% vs. 24%), and LDH levels (high LDH, 14% vs. 5%) (Table2).Baseline characteristics by treatment and time to best response are shown in and 64.1 versus 17.2 months with placebo-Rd (HR 0.27; 95% CI: 0.19-0.39;p<0.0001), respectively, in patients achieving ≥PR (FigureS4A).Among patients with ≥VGPR, median DOR for late versus early responders was NR versus 42.8 months with IRd (HR 0.40; 95% CI: 0.25-0.64;p < 0.0001) and 64.2 versus 25.1 months with placebo-Rd (HR 0.31; 95% CI: 0.18-0.51;p < 0.0001), respectively (Figure Baseline demographics and characteristics in early versus late responders.Values were calculated to identify possible covariates to include in the Cox proportional hazards model; however, they should be interpreted with caution as the analysis is post hoc and does not adjust for multiplicity. Randomisation stratification variables and baseline covariates characterising a difference between early and late responders were included in the adjusted full Cox proportional hazards model.Using early response as reference, a late response was significantly associated with longer PFS when adjusted for baseline TA B L E 2 a Determined by Pearson's chi-squared test, Wilcoxon rank sum test, or Fisher's exact test depending on the variable.pb Asian Indian, Chinese, Filipino, Japanese, Korean, Vietnamese.c American Indian or Alaskan Native, Native Hawaiian or other Pacific Islander, Other.d Includes t(4 ;14), t(14 ;16), del(17p), amp(1q21).e Six early responders and five late responders carried the t(14;16) cytogenetic abnormality.f Totals do not sum due to 88 and 32 missing early responder and later responder patients, respectively.
was achieved in patients who had obtained a response in ≥4 months versus <4 months.By assessing late responders for what drives this superior PFS, a higher proportion had deeper responses versus early responders.There is a clear association between deeper responses and improved outcomes, as demonstrated by the prolonged PFS in patients with measurable residual disease (MRD) negativity in a pooled analysis of two previous studies investigating ixazomib maintenance uate the impact of early versus late response on PFS after adjusting for potentially confounding factors.Overall, timing of response was an independent prognostic factor for PFS on multivariable analyses of IRd-treated patients who achieved ≥PR, with late response associated with superior PFS.Alongside depth of response, our findings indicate that a slow, gradual response may also serve as a powerful prognostic factor given that a longer PFS