Upfront autologous haematopoietic stem-cell transplantation versus carfilzomib–cyclophosphamide– dexamethasone consolidation with carfilzomib maintenance in patients with newly diagnosed multiple myeloma in England and Wales (CARDAMON): a randomised, phase 2, non-inferiority trial

Summary Background Standard-of-care treatment for patients with newly diagnosed multiple myeloma is bortezomib-based induction followed by high-dose melphalan and autologous haematopoietic stem-cell transplantation (HSCT) and lenalidomide maintenance. We aimed to evaluate whether an immunomodulatory-free carfilzomib-based induction, consolidation


Introduction
The role of upfront autologous haematopoietic stem-cell transplantation (HSCT) for newly diagnosed transplantation-eligible patients with multiple myeloma has been examined in phase 3 clinical trials published between 2017 and 2022.The IFM 2009 study, 1 EMN02/HO95, 2 and DETERMINATION trials 3 showed an improvement in progression-free survival for individuals receiving upfront autologous HSCT compared with ongoing treatment.All studies used a bortezomib-containing induction regimen, then lenalidomide maintenance after HSCT.These studies incorporated lenalidomide maintenance, which has the greatest benefit in standard-risk disease, but there is evidence that extended proteasome inhibitor treatment after HSCT might benefit patients with high risk (eg, patients with multiple myeloma with 17p deletion). 4xtended therapy with bortezomib, however, is limited by neurotoxicity. 5 Although ixazomib is an oral proteasome inhibitor, results as a maintenance therapy have not shown a consistent benefit. 6,7arfilzomib is commonly used to treat relapsed multiple myeloma.However, there are increasing data for its use as a first-line therapy, [8][9][10] resulting in high response rates and high rates of minimal residual disease (MRD) negativity.The FORTE study 8 has reported responses to carfilzomib triplets in trans plan tation-eligible patients with multiple myeloma, including MRD-negativity rates.Data are accumu lating regarding the efficacy of carfilzomib in high-risk disease multiple myeloma, 11,12 in which extended use of proteasome inhibitor regimens might be advantageous.Alternatives to lenalidomide as a main tenance therapy are also being studied, most notably CD38 antibodies, 13 and carfilzomib, lenalidomide, and dexamethasone (KRd) in an MRD-risk-adapted way, 14 but the use of proteasome inhibitor-based main tenance remains to be established after autologous HSCT.
All randomised trials investigating the benefit of autologous HSCT as an upfront therapy have been designed to support the statistical superiority of transplantation over a non-transplantation consolidation approach.Despite a clear progression-free survival advantage of autologous HSCT in these studies, [1][2][3][4] an overall survival advantage has not yet emerged, which prompts the question of whether patients are harmed by not receiving upfront autologous HSCT.This is a pertinent question because, despite advances in supportive care, autologous HSCT remains associated with considerable morbidity, recovery time, and a small mortality rate.The lack of a substantiated longterm overall survival benefit of upfront autologous HSCT is also emphasised by increasingly effective first-line and salvage regimens, which allow the debate between upfront autologous HSCT and deferred HSCT to continue.
The CARDAMON trial was designed to investigate any detrimental effect of allocating patients to a non-transplantation group using a carfilzomib-based induction, consolidation, and maintenance approach.The primary aims were to establish the efficacy of the triple regimen of carfilzomib, cyclophosphamide, and dexamethasone (KCd) as induction in transplantation-eligible patients with newly diagnosed multiple myeloma and to evaluate the non-inferiority of KCd consolidation compared with upfront autologous HSCT in patients with at least a partial response to induction.

Study design and participants
CARDAMON is a randomised, open-label, phase 2 trial in 19 hospitals in England and Wales, UK (appendix p 17).The protocol was approved by the UK Medicines and Healthcare Products Regulatory Agency (London, UK) and the London-City and East Research Ethics Committee (London, UK; date of favourable ethical opinion Feb 23, 2015), and was managed by the Cancer Research UK and University College London Cancer Trials Centre (London, UK).The protocol can be found in the appendix (pp 18-161).
Eligible patients were aged 18 years or older; had symptomatic multiple myeloma eligible for high-dose therapy and HSCT; had an ECOG performance status of 0-2 (>2 was permitted if resulting from complications related to myeloma); had measurable disease according to standard criteria; 15 had a life expectancy of 3 months or more; and had adequate neutrophil count, platelet count, haemoglobin, and creatinine clearance.Exclusion criteria included previous systemic treatment for myeloma (except local palliative radiotherapy or corticosteroids for a maximum of 4 days) and substantial cardiac comorbidity (appendix pp 18-161).Sex was reported on case report forms by trial investigators during patient registration.The options for this data field were male and female.Data on gender were not collected.All patients provided written informed consent in accordance with the World Medical Association Declaration of Helsinki.

Randomisation and masking
Patient randomisation was done after completion of four cycles of KCd induction and peripheral blood stem cell (PBSC) harvesting for patients that had at least a partial response.Patients were randomly assig ned (1:1) to either high-dose melphalan and autologous HSCT followed by carfilzomib maintenance or to KCd consolidation followed by carfilzomib maintenance.

Procedures
As induction therapy, patients received four 28-day cycles of carfilzomib (56 mg/m² administered intravenously on days 1, 2, 8, 9, 15, and 16 of every cycle; 20 mg/m² administered intravenously on days 1 and 2 of the first cycle), cyclophosphamide (500 mg administered orally See Online for appendix on days 1, 8, and 15 of all cycles) and dexamethasone (40 mg administered orally on days 1, 8, 15, and 22 of all cycles) before PBSC harvesting and randomisation.
For the consolidation phase, responders received highdose melphalan according to local protocols and autologous HSCT (HSCT group), or four further cycles of KCd (KCd group), at the same dose, route, and schedule of administration as the induction protocol, including the step-up dosing in the first cycle of consolidation.This step-up dosing was introduced halfway through the trial to reduce adverse events.
All randomly assigned patients received carfilzomib maintenance (56 mg/m² on days 1, 8, and 15 of a 28-day cycle for up to 18 cycles, with 20 mg/m² on the first day of the first cycle).Each dose of carfilzomib was administered with dexamethasone 10 mg pre-medication (intravenously or orally).Treatment delays of up to 4 weeks were permitted.During the COVID-19 pandemic, after March 25, 2020, delays were permitted up to 14 weeks and when more than 12 months of maintenance had been completed, treatment could stop at the discretion of the investigator. 16After cases of thrombotic microangiopathy were reported, the protocol was amended to include a step-up carfilzomib dose at the reduced amount of 20 mg/m² for any patient resuming treatment after a break of more than 4 weeks, before returning to the protocol dose of 56 mg/m² (or the last dose amount administered after any reduction).
Prespecified carfilzomib dose modifications were permitted for grade 4 thrombocytopenia (or grade 3 with active bleeding), grade 4 neutropenia, and grade 3-4 carfilzomib-related, non-haematological toxic effects (appendix pp 18-161).Unless explicitly withdrawing consent, patients who withdrew from trial treatment continued to be followed up for collection of outcome data.
Disease response assessments were recorded for each cycle of treatment according to modified International Myeloma Working Group (IMWG) criteria 17 and were centrally reviewed.Bone marrow examination to assess complete response was mandated after PBSC harvesting at the end of KCd consolidation, 100 days after autologous HSCT, 6 months after maintenance start, and at the end of maintenance.MRD was assessed in all patients centrally, regardless of serological response, by flow cytometry (threshold of 10 -⁵; one tumour cell in 100 000 bone marrow cells) after PBSC harvesting, after autologous HSCT or KCd consolidation, and at 6 months of maintenance.Adverse events, recorded from initiation of trial treatment until 30 days after final trial treatment administration, were graded according to the Common Terminology Criteria for Adverse Events (CTCAE), version 4.03.

Outcomes
This phase 2 study had two stages with no halt to recruitment between stages and separate primary endpoints for each stage.The primary outcome for the induction stage was the proportion of patients with at least a very good partial response after induction and the difference in 2-year progression-free survival (defined as the time between randomisation and first progression or death from any cause) after the randomisation phase.
The secondary outcomes were disease response and MRD rates at each timepoint, time until second progression or death, overall survival (measured from both registration and randomisation until death from any cause), toxicity, and quality of life.Follow-up for time until second progression or death and overall survival data are immature and the quality-of-life assessment is ongoing.These data will be reported elsewhere.
An important prespecified exploratory analysis assessed the relationship between fluorescence in-situ hybridisation (FISH) abnormalities and clinical outcomes.An adverse FISH result was defined as the presence of one or more of the cytogenetic abnormalities: t(4;14), t(14;16), and t(14;20) and del(17p) in 50% or more of bone marrow plasma cells.Post-hoc analyses examined progressionfree survival in subgroups defined by MRD status at different timepoints.

Statistical analysis
The sample size for the induction stage was calculated with an A'Hern single-stage phase 2 design, 18 con sidering the first primary endpoint (ie, induction response).We aimed to rule out a major response (eg, at least very good partial response) rate of less than 30%, with an anticipated rate of more than 50% with KCd induction.Using a one-sided 5% α value and 90% power, we required a minimum of 22 (42%) of 53 patients to respond to treatment.For the phase after randomisation we aimed to show that, in combination with KCd induction and carfilzomib maintenance, KCd consolidation was non-inferior to autologous HSCT in terms of progression-free survival.Assuming a 2-year progression-free survival rate of 85% in the HSCT group, 19 we estimated that 210 ran domly assigned patients (43 events of disease progression or death) would achieve 80% power with a one-sided 15% α value and non-inferiority margin for a difference in 2-year progression-free survival of 10%.Assuming at least a 75% partial response rate to induction, we aimed to recruit 280 patients.The 10% non-inferiority margin is justified by the fact that transplantation will remain a salvage option for most patients with disease progression in the KCd consolidation group, and by evidence that a 10% (or smaller) difference in progression-free survival is expected to translate to a much smaller, if any, difference in overall survival. 2  randomised groups.Patients were censored at the date last seen if they did not have an event, or at the date of new, off-trial treatment if it happened before progression (for the progression-free survival endpoint only).When examining associations between MRD and progressionfree survival, we used landmark analyses from the date of MRD assessment.p-values for comparisons between treatment groups were calculated with Cox regression analyses for survival outcomes and the χ² test for response rates.The difference in 2-year progression-free survival between the two treatment groups and 70% CI (corresponding to our choice of a one-sided 15% α value) was calculated by applying the hazard ratio (HR) to the 2-year rate in the KCd consolidation group to get the corresponding rate in the autologous HSCT group, then subtracting this from the 2-year rate in the KCd consolidation group.A separate, post hoc, per-protocol analysis was done, excluding patients who did not start randomised treatment.If the proportional hazards assumption was violated, the difference in restricted mean survival time (RMST) 20 was reported instead of the HR.The timepoint used for calculating the difference in RMST was taken as the largest observed time to event of either treatment group.A sensitivity analysis excluding patients whose maintenance was delayed or stopped because of the COVID-19 pandemic was performed on the primary progression-free survival outcome.Safety data were assessed in all patients who received at least one dose of any study drug.All analyses were performed with Stata version 16.1.This study is registered with ClinicalTrials.gov,number NCT02315716 and all patients are in follow-up.

Role of the funding source
The funders of the study had no role in the study design, data collection, data analysis, data interpretation, writing of the report, or the decision to submit the paper for publication.

Results
Between  All adverse events of grade 1-2 that occurred in at least 10% of patients, and all adverse events of grade 3 or worse, that occurred during the trial, except those related to transplantation (standard of care), have been summarised (table 3).The most common grade 3 or worse adverse events during induction treatment were lymphocytopenia (occurring in 72 [26%] of 278 patients), infections (53 [19%]), hypertension (33 [12%]), neutropenia (32 [12%]), and anaemia (29 [10%]).11 (4%) of 278 patients had grade 3 or worse cardiac disorders during induction, with five (2%) of 278 patients having myocardial ischaemia (myocardial infarction or acute coronary syndrome).The proportion of patients with grade 3 or worse hypertension during KCd consolidation (12 [11%] of 109) was similar to that seen during induction, whereas the proportion of patients who had grade 3 or worse lymphocytopenia (15 [14%]), neutropenia (one [1%]), anaemia (two [2%]), or infection (15 [14%]) were reduced.During maintenance treatment, there was a higher incidence of grade 3 or worse events in patients who had HSCT (65 [66%] of 99) than in patients who received KCd consolidation (44 [45%] of 97; p=0•0042).This difference was because of a higher proportion of lung infections and cytopenia in the HSCT group than in the consolidation group.There were five treatmentemergent deaths during the induction phase.Four patients died while receiving induction treatment (three from infection and one from a cardiac event).The fifth patient withdrew due to sepsis before dying 4 months later of renal failure.One patient in the HSCT group died from an infection 114 days after transplantation and another patient from the HSCT group died during maintenance due to an oesophageal malignancy.
Of 259 patients with complete FISH data, 52 (20%) had high genetic risk.Response rates to KCd induction were similar irrespective of genetic risk (at least a very good partial response occurred in 31 [60%] of 52 patients with high risk vs 120 [58%] of 207 patients with standard risk).MRD negativity rates were also similar, occurring in 12 (23%) of 52 patients with high risk versus 48 (23%) of  3

Discussion
Studies published in the past 6 years [1][2][3]8 have all shown a superior progression-free survival for auto logous HSCT compared with consolidation in newly diag nosed, transplantation-eligible patients with multiple myeloma. Hwever, only one study has shown an improve ment in overall survival, 21 and with increasingly effective induction regimens, any true long-term benefit to patients remains unclear.Rather than showing superiority, our study sought to assess if ongoing treatment with KCd was non-inferior to KCd induction followed by autologous HSCT in newly diagnosed, transplantationeligible patients with multiple myeloma, with both groups receiving carfilzomib maintenance.Our results show that KCd consolidation did not meet the criteria for non-inferiority when compared with upfront transplantation.However, the non-inferiority margin was only exceeded by a small amount (confidence limit -11•1%, prespecified margin -10%), although this study was a phase 2 trial with 15% significance level.Therefore, there are likely to be subgroups of patients for whom deferred trans plantation might be an option, which should be explored in future prospective trials.
Our progression-free survival outcomes are similar to those in the IFM 2009 study that also used restricted duration maintenance, albeit with lenalidomide.We observed a median progression-free survival from randomisation of 33•8 months in the KCd consolidation group versus 42•4 months in the HSCT group (HR 1•35), whereas the IFM 2009 trial reported a median progression-free survival from registration of 35•0 months for lenali domide-bortezomib-dexametha sone (VRd) con sol i dation versus 47•3 months for HSCT (HR 1•43). 22tudies with lenalidomide maintenance until progression 3 reported longer pro gression-free survival overall than trials with fixed-duration lenalidomide maintenance, although the relative difference between treatment groups is similar across trials (appendix p 16).
Supporting the efficacy of KCd in patients with high risk, the very good partial response rate and MRD-negative rate after KCd induction were similar between patients with standard risk and patients with high risk, and these measures of response were also similar after randomised treatment.Although very good partial response rates were similar between randomisation groups, MRD-negative rates were higher in the HSCT group for both patients with high risk and patients with standard risk than in the KCd consolidation group, with little difference between patients with high risk and patients with standard risk.Despite this finding, patients with high-risk genetics had inferior progression-free survival compared with patients with standard risk.Reasons for this finding could be the restricted duration of maintenance carfilzomib (18 months) and the absence of consolidation after autologous HSCT, which could be beneficial in high-risk multiple myeloma.In the FORTE study, 8 patients with high risk benefitted from the addition of carfilzomib to lenalidomide, thus, a proteasome inhibitor and immunomodulatory drug approach to maintenance is likely to be required for optimal outcomes in this group.The interim results of the ATLAS study showed a median progression-free survival of 59•0 months for KRd followed by lenalidomide maintenance after autologous HSCT, with more benefit for patients with high risk than with lenalidomide maintenance alone. 14he importance of MRD as a prognostic marker has led to speculations on its use in treatment decisions, with several current trials addressing this important question in the context of first-line treatments. 23The IFM 2009 study showed that MRD was the strongest driver of progression-free survival, 24 regardless of randomisation group.However, as MRD testing was not done before autologous HSCT or consolidation, the IFM 2009 study could not address the relative benefit of HSCT in MRD-negative patients after induction.In our current study, there was no clear benefit of HSCT in patients who had an MRD-negative response after induction, whereas for individuals who are MRD positive, autologous HSCT could be beneficial.We also showed that there was no difference in progression-free survival between the autologous HSCT and consolidation groups for patients with an MRD-negative response after autologous HSCT or consolidation.Acknowledging that patient numbers were small and analyses were not prespecified, our results suggest that patients who are MRD positive after induction could benefit more from autologous HSCT than from consolidation chemotherapy.Conversely, individuals who are MRD negative should be investigated on a deferred autologous HSCT pathway in a larger prospective randomised trial.Ongoing trials, such as MIDAS (NCT04934475) and MASTER-2 (NCT05231629), could answer these questions. 25An interim analysis from the ATLAS study also showed that MRD could be used to de-escalate the intensity of maintenance. 15verall, this carfilzomib-based treatment pathway was feasible, although discontinuations because of adverse events were noted.This study used a higher dose of carfilzomib (56 mg/m²) than other carfilzomib triplets used in twice per week dosing.Discontinuations related to adverse events during KCd induction were more common in our CARDAMON study than in the FORTE study (7% vs 3%). 8However, 5 (25%) of the 20 patients who withdrew from our study were older than 65 years and would have been excluded from FORTE.Overall, 78% of patients proceeded from induction to randomisation in CARDAMON, a rate that is similar to that in the KCd group (84%) in FORTE, especially given the age limit of 65 years.Although the discontinuation rates at randomisation from the DETERMINATION trial (5%; NCT01208662) are lower than in our study, this finding is partly because of design; randomisation occurred after one induction cycle whereas in CARDAMON randomisation occurred after four induction cycles.However, the adverse event profile of carfilzomib requires careful proactive management, particularly in individuals older than 65 years, in whom a low carfilzomib dose might be better tolerated than a high carfilzomib dose.Fewer discon tinuations were observed after the implementation of further safety measures than before.For example, eight cases of thrombotic microangiopathy were noted, 26 with no further cases after proactive manage ment of hypertension and introduction of step-up dosing at the start of maintenance.The main adverse event during carfilzomib maintenance was hypertension, suggesting that cumulative drug doses might have had an effect.More patients in the autologous HSCT group discontinued maintenance than in the KCd consolidation group, potentially because of the higher incidence of infections and cytopenias.Withdrawals from carfilzomib maintenance might also have been influenced by the COVID-19 pandemic, as guidance was issued to allow discontinuations in patients who had completed at least 12 months of maintenance.The quality-of-life data are currently under analysis and might provide further insights into the tolerability of this regimen.
KCd induction resulted in high response rates (more than the prespecified target) and high MRD-negativity rates without the use of an immunomodulatory drug, results that are similar to the KCd plus autologous HSCT group in FORTE (very good partial response rates of 57•7% in CARDAMON vs 53% in FORTE; 77•1% after autologous HSCT in CARDAMON vs 77% in FORTE). 8he MRD-negativity rates at the start of maintenance are also similar (47•7% in CARDAMON vs 43% in FORTE).Although the KRd regimen was shown to be superior to KCd, 8 cyclophosphamide might be useful in countries where the combination with lenalidomide could be costly or for patients who are intolerant to lenalidomide.There were also fewer mobilisation failures with KCd than with KRd in the FORTE study, and we did not see any mobilisation failures in CARDAMON.Continuous proteasome-inhibitor-based treatment could be effective in patients who are unable to tolerate long-term immuno modulatory drugs or who have high-risk multiple myeloma in which lenalidomide-based treatments are less effective than for patients with standardrisk disease.Carfilzomib has been previously shown to be superior to bortezomib for patients with disease relapse, 27 making it a preferred option in this setting.However, this superiority has not been shown when used as a first-line treatment, 28,29 and further comparative studies are required.However, the requirement for clinic visits once or twice per week for carfilzomib maintenance, depending on the schedule, needs to be balanced against the convenience of oral lenalidomide or subcutaneous daratumumab adminis tered once per month.The FORTE study has reported superior outcomes of combined main tenance with carfilzomib and lenalidomide, and ATLAS showed more of a progression-free survival benefit for KRd than for lenalidomide.
The addition of CD38 antibodies to standard-of-care triplets has been shown to improve responses and MRD negativity rates that were maintained after autologous HSCT. 30,31Daratumumab plus bortezomib-thalidomidedexametha sone has already become the standard of care in many countries and daratumumab plus bortezomiblenalidomide-dexamethasone could also become a standard approach, with tolerability advantages of lenalidomide.The addition of a CD38 antibody to a carfilzomib-based triplet has encouraging preliminary data. 23,32The benefits of upfront autologous HSCT should continue to be examined in individuals who have deep, sustained responses to induction therapy.
There are some limitations to our study.Due to the non-inferiority design of CARDAMON, we cannot comment on the superiority of autologous HSCT in the context of this study.Furthermore, to keep the trial at a feasible size, it was not powered to assess non-inferiority in subgroup analyses and there might be additional subpopulations (eg, based on individual cytogenetic factors) that deserve further evaluation.Transplantationrelated events were not reported as autologous HSCT is standard of care and the toxicity profile is already well documented.However, by not doing so we are unable to make a direct comparison of toxicity between treatment groups during the random ised phase of treatment.The reported rate of patients who received a deferred transplant after disease progression might be an underestimation of the true value, as these data are collected on long-term follow-up forms.The COVID-19 pandemic led to patient with drawals during maintenance (five withdrawals in the KCd consolidation group and ten in the autologous HSCT group), alongside treatment delays and interruptions, which could have resulted in worse clinical outcomes overall for the trial cohort than if COVID-19 had not affected treatment.
The CARDAMON trial provides preliminary evidence that MRD assessment after induction could be used to guide treatment choice.Future trials that stratify by a composite of genetic risk and depth of response will be able to accurately identify patients likely to benefit from autologous HSCT, leading to a personalised treatment approach.
In conclusion, the CARDAMON trial showed that carfilzomib-based induction, consolidation, and maintenance did not meet the criteria for non-inferiority when compared with upfront transplantation.Research should continue to explore the benefit of upfront autologous HSCT in specific patient populations (eg, patients with standard risk genetics vs high risk genetics and patients who are MRD negative or positive after induction therapy).Although autologous HSCT continues to be regarded as a first-line, standard-of-care treatment for multiple myeloma, its role will continue to be assessed, especially with highly efficacious quadruplet regimens and for patients who have MRD negativity early on.Current risk-stratified trials should also aim to identify patients who benefit from upfront autologous HSCT, and the optimal post-autologous HSCT therapy needed to maintain disease response.

Figure 2 :
Figure 2: Progression-free survival (A) In all registered patients (n=281).(B) In all randomly assigned patients (n=218), per treatment group.HSCT=haematopoietic stem-cell transplantation.HR=hazard ratio.KCd=carfilzomib-cyclophosphamide-dexamethasone.Number at risk (number censored)Autologous HSCT KCd consolidation Randomisation was done centrally by a computer program set up by the data service at the Cancer Research UK and University College London Cancer Trials Centre with a minimisation approach stratified by hospital, depth of response to induction, International Staging System stage, and genetic risk.Patients were approached and assessed for eligibility for the trial by investigators at participating hospital sites (appendix p 17). Patient eligibility was confirmed by Cancer Research UK and University College London Cancer Trials Centre before study entry.Investigators are responsible for site-based responsibilities as delegated by Principal Investigators.Cancer Research UK and University College London Cancer Trials Centre are responsible for central trial activities as delegated by the Chief Investigator.Patients, people giving the interventions, those assessing outcomes, and those analysing the data were not masked to group assignment.Individuals who did not have at least a partial response to induction therapy were treated with salvage therapy outside of the trial.

table 2 )
Withdrawals due to adverse events or patient or clinician decision were more common in the HSCT group than in the KCd group (six [6%] of 97 patients in the KCd group vs 16 [16%] of 99 patients in the HSCT group, p=0•027).15patients(five in the KCd group and ten in the HSCT group) discontinued maintenance due to the COVID-19 pandemic, and 41 patients (23 in the KCd group and 18 in the HSCT group) had their treatment interrupted or delayed because of it.The first primary endpoint was met, with 43 (81%) of the first 53 patients having at least a very good partial response after induction (revised to 37 [70%] of the first 53 patients after central review).The proportion of patients who had at least a very good partial response after induction therapy in the entire intention-to-treat .Randomised groups were balanced in terms of depth of response and MRD negativity rate after induction (table1).Serological response rates increased after randomised treatment but were similar across both treatment groups (at least a very good partial response occurred in 85 [78%] of 109 patients in the KCd group vs 84 [77%] of 109 in the HSCT group, p=0•87; table2).After randomised treatment, more patients in the HSCT group had MRD negativity than patients in the KCd group (33 [30%] of 109 vs 52 [48%] of 109, p=0•0083).This difference in MRD negativity was still apparent at the 6-month maintenance timepoint (34 [31%] of 109 vs 50 [46%] of 109], p=0•026; appendix p 6).Exclusion of patients who were negative for MRD with less than a very good partial response did not alter these conclusions (appendix p 6).For the second primary endpoint, 150 events of disease progression or death have been reported, of which 117 (64 in the KCd group and 53 in the HSCT group) occurred after randomisation.Median progression-free survival for all 281 patients from registration was 38•1 months (95% CI 33•1 to 44•8; figure2A).Median progression-free survival for all 218 patients from randomisation was 42•4 months (32•6-59•5) for the HSCT group and 33•8 months (28•3-40•6) for the KCd group (HR 1•35, 70% CI 1•11 to 1•64, p=0•11; figure2B).The 2-year progression-free survival was 75% (95% CI 65-82) in the HSCT group and 68% (95% CI 58-76) in the KCd group, resulting in a calculated difference of -7•2% (70% CI -11•1 to -2•8).In the 208 patients who started randomised treatment in the per-protocol population, this difference was -8•5%, 70% CI -12•3 to -4•1 (2-year progression-free survival 77%, 95% CI 68-84 in the HSCT group vs 2-year progression-free survival 68%, 95% CI 58-76 KCD group; HR 1•44, 70% CI 1•18-1•75, p=0•058).The 2-year progression free survival Data are n (%), unless otherwise specified.Some patients had more than one genetic risk factor.HSCT=haematopoietic stem-cell transplantation.del=deletion.ECOG=Eastern Cooperative Oncology Group.KCd=carfilzomibcyclophosphamide-dexamethasone.t=translocation.MRD=minimal residual disease.*Demographics case report form introduced Sept 21, 2018, to record patient ethnicity.†Excluded from denominator.