The Changing Landscape of Maintenance Therapy in Newly Diagnosed Multiple Myeloma: A Systematic Review With Network Meta‐Analysis of the European Myeloma Network (EMN)

ABSTRACT

We performed a comprehensive review with meta‐ and network meta‐analyzes of maintenance‐therapy studies. Lenalidomide, proteasome inhibitors, and CD38 antibodies improved progression‐free survival (PFS). However, overall survival (OS) benefit appeared only with lenalidomide in transplant‐eligible (TE) patients, while CD38‐directed therapy showed a trend toward improved OS not seen with proteasome inhibitors. The network meta‐analysis ranked regimens against observation: daratumumab–lenalidomide (DR) yielded the greatest PFS benefit, followed by carfilzomib–lenalidomide–dexamethasone (KRd), KR in TE, and DRd in transplant‐ineligible patients. Significant OS prolongation in TE patients occurred with KR, DR, and lenalidomide alone.

Our meta‐analysis showed significant improvement of PFS with lenalidomide, proteasome inhibitors, and CD38‐based therapies. A significant OS benefit was noted only with lenalidomide in transplant‐eligible (TE) patients, while CD38‐directed therapy showed a trend toward improved OS. The network meta‐analysis showed the greatest reduction in the risk of progression with Dara‐R, followed by KRd, Dara‐Rd, KR, and Isa‐Rd with risk reductions from 78% to 66%.

1. Introduction

Maintenance therapy has become a central pillar of frontline management for newly diagnosed multiple myeloma (NDMM), aiming to deepen responses, prolong remission, delay disease progression, and improve overall survival (OS). Unlike induction, autologous stem‐cell transplantation (ASCT), and consolidation, maintenance approaches focus on sustaining therapeutic benefit with minimal toxicity, thereby balancing efficacy and quality of life [1, 2, 3].

Several studies have established lenalidomide maintenance following ASCT as the standard of care, demonstrating significant improvements in both progression‐free survival (PFS) and OS [4, 5, 6, 7]. In patients not eligible for transplantation (TI), continuous treatment—often with a reduced induction regimen—has likewise shown remarkable benefits in terms of PFS and OS [8, 9].

More recently, novel immune and cellular therapies with unprecedented efficacy have been introduced. Some have already been introduced in clinical practice, and several are now being investigated as emerging options for maintenance therapy [10]. This development motivated us to review the current evidence on maintenance therapy and to conduct both a meta‐analysis and a network meta‐analysis, enabling a robust comparison of established and emerging approaches and providing support for treatment selection in clinical practice.

2. Methods

2.1. Literature Review

Literature for this analysis was retrieved by searching PubMed, EMBASE, and the Cochrane Library up to November 2025. Search terms included maintenance therapy and multiple myeloma (Table S1). Abstracts presented at ASH 2025 were also screened. Articles were excluded if irrelevant to the research question, lacking a control group, using outdated treatment regimens, involving CAR‐T therapy, missing PFS data, or lacking Kaplan–Meier curves for studies that continued induction therapy without randomization. These curves were essential for deriving individual patient data for the landmark analysis after the estimated end of induction. Table 1 summarizes the included publications and their characteristics. Of note, we included trials without randomization at the maintenance time point and those where ongoing therapy was intended for TI patients defining the start of “maintenance” as the time at which there was a step down from more intensive to less intensive combinations.

TABLE 1.

List of included studies.

Study names	Maintenance treatment		Number of patients in maintenance phase	Transplant‐eligibility	Median age (range)	Treatment duration
	Control	Experimental
ATLAS	R	KRd	180	Eligible	59 (IQR 49–63)	KRd: 8–36 cycles (28 days each) R: until PD or intolerance
AURIGA	R	DR	200	Eligible	62 (35–78)	Max. 36 cycles
CALGB 100104	Obs	R	460	Eligible	R: 59 (29–71) Obs: 58 (40–71)	Until PD or intolerance
CASSIOPEIA	Obs	D	886	Eligible	D: 59 (53–63) Obs: 59 (53–63)	Induction: 4 cycles (28 days each) Consolidation: 2 cycles (28 days each) Maintenance: max. 2 years
CEPHEUS	Rd	DRd	354	Ineligible	DRd: 70 (42–79) Rd: 70 (31–80)	Induction: 8 cycles (21 days each) Maintenance: until PD (28‐day cycles)
EMN01	R	Rp	402	Ineligible	RP: 73 (65–87) R: 73 (50–89)	Induction: 9 cycles (28 days each) Maintenance: until PD
FORTE	R	KR	356	Eligible	KR: 56 (52–62) R: 57 (51–62)	Induction: 4 cycles/8 cycles (28 days each) Consolidation: 4 cycles (28 days each)/none Maintenance: KR up to 2 years, R until Pd or intolerance
GEM2014MAIN	Rd	IRd	332	Eligible	58 (32–67)	24 cycles in MRD‐neg patients, 36 additional cycles in MRD‐pos patients (4 weeks each)
GMMG‐HD6	Rd	ERd	559	Eligible	RVd/R: 59 (IQR 52–64) RVd/E‐R: 60 (IQR 53–63) E‐RVd/R: 59 (IQR 52–64) E‐RVd/E‐R: 59 (IQR 52–65)	Induction: 4 cycles (21 days each) Consolidation: 2 cycles (21 days each) Maintenance: 26 cycles (28 days each)
GRIFFIN	R	DR	207	Eligible	DR: 59 (IQR 51–65.5) R: 61 (IQR 54–66)	Induction: 4 cycles (21 days each) Consolidation: 2 cycles (21 days each) Maintenance: 7–32 cycles (28 days each)
HOVON‐126	Obs	I	78	Ineligible	I: 72 (66–80) Obs: 73 (67–82)	Induction: 9 cycles (28 days each) Maintenance: until PD
IFM 2005–02	Obs	R	614	Eligible	R: 57.5 (22.7–68.3) Obs: 58.1 (32.3–67.0)	Consolidation: 2 cycles (28 days each) Maintenance: until PD or intolerance
IMROZ	Rd	IsaRd	446	Ineligible	IsaRd: 72 (60–80) Rd: 72 (55–80)	Induction: 4 cycles (6 weeks each) Maintenance: until PD or intolerance
JCOG1911/B‐DASH	D	DV	224	Ineligible	DV: 73 (64–86) D: 74 (65–85)	Up to 18 cycles (3 weeks each)
MM‐015	Obs	R	182	Ineligible	R: 71 (65–87) Obs: 71 (65–86)	Induction: 9 cycles (28 days each) Maintenance: until PD or intolerance
Myeloma XI	Obs	R	833	Both	CRda: 75 (60–92) a CTda: 74 (54–89) a	Induction: min 6 cycles (21 or 28 days each) until maximum response or intolerance Maintenance: until PD or intolerance
NCT04497961	R	D	89	Both	64 (NA‐NA)	Max. 36 cycles (28 days each)
Palumbo 2014	Obs	R	251	Eligible	R: 57 (IQR 50–61) Obs: 57 (IQR 50–61)	Induction: 4 cycles (28 days each) Consolidation: 6 cycles (28 days each) or 2 cycles (4 months each) Maintenance: until PD or intolerance
PERSEUS	R	DR	709	Eligible	DR: 61 (32–70) R: 59 (31–70)	Induction: 4 cycles (28 days each) Consolidation: 2 cycles (28 days each) Maintenance: until PD or intolerance
RV‐MM‐EMN441	R	RP	223	Eligible	RP: 57 (53–61) R: 56 (51–61)	Induction: 4 cycles (28 days each) Consolidation: 6 cycles or 2 cycles (28 days each) Maintenance: until PD or intolerance
RV‐MM‐PI‐0752	Rd	R	89	Ineligible	R: 75 (IQR 73–77) Rd: 76 (IQR 74–79)	Induction: 9 cycles Maintenance: until PD or intolerance
SeaLAND (ALLG MM23)	R	SR	149	Eligible	SR: 62 (39–75) R: 62 (34–76)	Until PD
TOURMALINE‐MM3	Obs	I	656	Eligible	I: 58 (52–63) Obs: 60 (54–64)	Max. 26 cycles (28 days each)
TOURMALINE‐MM4	Obs	I	706	Ineligible	I: 72 (42–89) Obs: 73 (52–90)	Max. 26 cycles (28 days each)

Abbreviations: C, Cyclophosphamide; D, Daratumumab; d, Dexamethasone; E, Elotuzumab; I, Ixazomib; Isa, Isatuximab; K, Carfilzomib; M, Melphalan; NA, not available; Obs, Observation; P, Prednisone; PD, progressive disease; R, Lenalidomide; S, Selinexor; T, Thalidomide; TE, transplant eligible; TIE, transplant ineligible; V, Bortezomib.

^a Patients were randomized to CRda and CTda before half of each group received maintenance. Age was only reported for the groups of the first randomization.

The quality of the selected studies was assessed by two reviewers (HL and SB) using the Cochrane Risk of Bias tool (RoB2) [11]. The results are shown in a heat map (Figure S1). To increase transparency, a Preferred Reporting Items for Systematic Reviews and Meta‐Analyzes (PRISMA) checklist was completed (Supplementary: PRISMA 2020 Checklist) (https://www.prisma‐statement.org/prisma‐2020‐checklist).

2.2. Statistical Methods

Retrieved outcomes included hazard ratios (HR) for PFS and OS with their standard errors (SE) and/or 95% confidence intervals (CI). When these outcomes were not directly reported, we digitized the published Kaplan–Meier curves using a Shiny application [12]. For studies with continued induction therapy, we used the reported treatment design to calculate the planned start of the maintenance phase as the time point corresponding to the last dose administered during the intensive induction/consolidation regimen for each individual study. This enabled us to perform a landmark analysis of the maintenance part using reconstructed data. A comparison of digitized vs. original outcome measures is provided in Table S2. Sensitivity analyzes assessed the impact of the chosen landmark by shifting the time point in both directions and comparing the resulting HRs (Table S3).

The meta‐ and network meta‐analyzes were conducted using the meta [13] and netmeta [14] packages in R. We used comparison‐adjusted funnel plots and Egger's tests to assess publication bias. The consistency of the network was tested using I [15] while Q‐statistics were checked for within‐design heterogeneity.

Pre‐planned subgroup analyzes included cytogenetic risk, transplant‐eligibility, treatment duration, digitized or originally retrieved data, and risk of bias.

To obtain a connected network, we assumed that placebo treatment was equivalent to observation and that there were no differences in efficacy due to dosages or schemes.

All statistical analyzes and figures were generated using RStudio (version 4.3.3; R Foundation for Statistical Computing, Vienna, Austria). The present systematic review and meta‐and network meta‐analysis was registered at the PROSPERO data base (ID CRD420251271990).

2.3. Role of the Funding Source

The study was supported by the Austrian Forum Against Cancer.

3. Results

3.1. Study Selection and Characteristics

The literature search identified 885 unique records, of which 24 were included in this analysis (Figure 1). Reasons for exclusion are listed in Table S4. Evaluation of risk of bias is shown in Figure S1, comparison‐adjusted funnel plot and Egger's test are shown in Figure S2.

FIGURE 1.

PRISMA flow chart of study selection. *Unsuitable article types: corrections, reviews, systematic reviews and meta‐analyses, real‐world studies, cost‐effectiveness/economic studies, congress summaries, non‐phase III or randomized phase II clinical trials (phase I or non‐randomized II, community‐based studies), smoldering myeloma, other cancers. ** No publications available because studies are too recent or results remain unpublished. [Color figure can be viewed at wileyonlinelibrary.com]

3.2. Lenalidomide

Lenalidomide was initially evaluated as post‐transplant maintenance therapy and three pivotal randomized trials comparing lenalidomide maintenance with observation after ASCT established the basis for this strategy [6, 16, 17]. A subsequent meta‐analysis demonstrated significant improvements in PFS and OS [5]. These findings were later confirmed by a network meta‐analysis [18] and an update of the CALGB study and partly by the DETERMINATION trial with continued lenalidomide use [19, 20].

The Myeloma XI trial evaluated lenalidomide maintenance in both transplant‐eligible (TE) and transplant‐ineligible (TI) patients [4]. PFS and OS were significantly increased in the TE cohort; in TI patients, a significant prolongation of the PFS but not of OS was observed. PFS benefits were also noted in patients with high‐risk cytogenetics, particularly in those with t(4;14) and 1q21 gain, whereas effects in t(14;16) and del(17p) did not reach significance. Patients with ≥ 2 high‐risk abnormalities had poor outcomes, with a PFS of 24 months and an OS of 47.7 months, underscoring the need for intensified maintenance in this group [9].

Lenalidomide maintenance deepened the quality of response in about 30%–40% of patients, which is in part reflected by the transformation from MRD‐positive to MRD‐negative status, resulting in superior survival compared with those remaining MRD‐positive [21]. Landmark analyzes suggested persistent benefit for PFS of lenalidomide maintenance at 2, 3, and 4 years in TE patients, although the effect after that time point was not statistically significant at the reported follow up [8]. Retrospective data from other studies support treatment durations beyond 3 years [22]. MRD‐positive patients derived the greatest benefit, with sustained reductions in risk at all landmark timepoints [8]. In MRD‐negative patients, benefit was evident at 2 and 3 years but not statistically significant beyond 4 years at the reported follow up time, although it should be recognized that in this analysis the comparator group had never received lenalidomide maintenance rather than stopping at the specified time point, and MRD assessment was at 3 and 9 months after ASCT rather than later. Taken together, these data suggest that discontinuation of maintenance treatment could be considered after at least 2–3 years of sustained MRD‐negativity in standard‐risk patients, whereas continuation appears necessary in MRD‐positive patients [8]. In individuals with high‐risk cytogenetic features, continuation of maintenance should be considered irrespective of MRD status. Outcomes remain particularly poor in patients with two or more high‐risk abnormalities, underscoring the need for improved strategies [23].

A recent study integrating annual PET/CT with MRD assessment evaluated a structured discontinuation strategy [24]. Patients who remained PET/CT‐ and MRD‐negative for 3 years stopped maintenance and were followed for three further years. Twenty‐three percent reconverted to MRD‐positive status; half with high‐risk disease and all restarted lenalidomide. Only 7.6% experienced clinical relapse after a median follow‐up of 3 years. These findings support discontinuation in standard‐risk patients with sustained deep remission, including negative PET/CT findings, provided that MRD status is being investigated over time.

As the depth of MRD‐negativity reported across studies varies between 10⁻⁶ and 10⁻⁵, uniform reporting would substantially improve comparability between trials. Harmonizing the presentation of MRD results therefore seems advisable, particularly in light of recent data indicating that sustained MRD < 10⁻⁵ provides the most robust model for estimating the risk of MRD resurgence or progression in the context of fixed‐duration therapy [25].

The median duration of lenalidomide maintenance in the Myeloma XI trial was 36 cycles; half of the patients who stopped maintenance did so due to progressive disease [23]. Lenalidomide exerts long‐term immunomodulatory effects on the bone‐marrow microenvironment [26]. Improvements in PFS2 after relapse have been reported, accompanied by increases in naïve CD8⁺ and memory T‐cell subsets and reductions in terminal effector T‐cells, resulting in a net immunostimulatory effect [27].

Exposure to lenalidomide is associated with an increased risk of secondary primary malignancies (SPM), most notably hematologic cancers and specially in multiple myeloma [28]. This phenomenon may, at least in part, reflect the higher incidence of clonal hematopoiesis observed in myeloma [29]. Earlier studies such as the meta‐analysis of the impact of lenalidomide maintenance therapy after ASCT reported an incidence of 5.3% of hematological SPM in the lenalidomide group compared to 0.8% in the control arm [5]. The Myeloma XI trial group reported a 3‐year cumulative incidence of 5.3% in the lenalidomide patients versus 3.1% in the observation group (HR, 1.85; 95% CI, 1.18–2.90), but this was not further investigated in the TE versus TI cohorts [4]. The SPM risk appears to increase with cumulative lenalidomide exposure, particularly when used at both induction and maintenance. A small, randomized trial demonstrated a dose‐related effect on both efficacy and general adverse events [30], but it did not reveal an association between lenalidomide dose and SPM incidence.

Rates of solid SPMs do not appear increased [31]. Cytopenias and infections occur in 20%–30% of patients. Rare events include herpes zoster and hepatitis B reactivation, requiring close monitoring in anti‐HBc–positive individuals. Rash is usually mild but may require temporary interruption and very rarely permanent discontinuation. Fatigue and diarrhea are common; the latter may respond to bile‐acid sequestrants [32]. Muscle cramps and thromboembolic events can occur, and thromboprophylaxis with aspirin or a direct oral anticoagulant is recommended [33].

Our meta‐analysis of five studies including 2340 patients showed no heterogeneity for PFS (I ² = 0.0%) and a HR of 0.53 (95% CI 0.49–0.57), corresponding to a doubling of PFS with lenalidomide maintenance after ASCT (Figure 2A). OS analyzes showed moderate heterogeneity (I ² = 46.5%) (Figure 2B). The largest OS benefit was observed in CALGB 100104 [17], with additional benefit in the TE cohort of Myeloma XI [4]. In this study, no OS benefit was observed in TI patients [8].

FIGURE 2.

Meta‐analysis of lenalidomide versus observation for (A) PFS and (B) OS.

Individual trials further support these findings. In the GIMEMA RV‐MM‐PI‐2095 study, which compared lenalidomide maintenance with placebo for 2 years in TE patients after induction and consolidation with either ASCT or MPR, showed a significantly longer PFS with lenalidomide maintenance and a trend toward improved OS [16]. In the MM‐015 trial, lenalidomide maintenance significantly prolonged PFS in TI patients receiving lenalidomide maintenance therapy after induction with melphalan‐prednisone‐lenalidomide compared to patients randomized to induction therapy with either melphalan‐prednisone‐lenalidomide or melphalan‐prednisone only [34]. In IFM 2005‐02, lenalidomide maintenance vs. placebo improved PFS but not OS after 2 years of therapy [6]. In Myeloma XI, lenalidomide maintenance doubled PFS in TE patients across standard‐risk, high‐risk, MRD‐negative, and MRD‐positive subgroups, with only modest OS improvement; no OS benefit was observed in TI patients. In TE patients previously exposed to immunomodulatory drugs during induction, the PFS benefit was marginally attenuated [8].

Overall, lenalidomide maintenance after ASCT is well established and remains the gold standard of maintenance therapy and is the only approved agent for this indication. MRD‐guided and risk‐adapted strategies are emerging to refine treatment duration and patient selection.

3.3. Proteasome Inhibitors

The use of proteasome inhibitors (PI) for maintenance therapy is more common in patients with cytogenetic high‐risk [9]. Initial support for PIs in maintenance emerged from the Phase 3 HOVON‐65/GMMG‐HD4 trial. Patients received either 3 cycles of vincristine‐adriamycin‐dexamethasone (VAd) induction followed by 1–2 ASCT and thalidomide maintenance or 3 cycles of bortezomib‐adriamycin‐dexamethasone induction followed by 1–2 ASCT and bortezomib maintenance each for 2 years. The median PFS was 34 months in the bortezomib‐adriamycin‐dexamethasone arm and 28 months in the VAd arm, respectively (HR 0.76, 95% CI 0.65–0.89; p < 0.001). However, long term OS at 96 months was similar, 45% in patients receiving thalidomide and 48% (p = 0.24) in those treated with bortezomib maintenance therapy [35, 36].

Carfilzomib has been evaluated in only a few maintenance studies. In the FORTE trial [37], TE patients were randomized to 4 cycles of carfilzomib‐lenalidomide‐dexamethasone (KRd) followed by ASCT followed by 4 cycles of KRd, or to 12 cycles of KRd, or to 4 cycles of carfilzomib‐cyclophosphamide‐dexamethasone (KCd) followed by ASCT and 4 cycles of KCd. Patients were then assigned to maintenance with carfilzomib plus lenalidomide or lenalidomide alone. Primary endpoints were depth of response after induction and PFS during lenalidomide or KR maintenance.

A total of 356 patients entered the maintenance phase. After a median follow‐up of 37.3 months, the 3‐year PFS was 75% (95% CI 68%–82%) with KR versus 65% (95% CI 58%–72%) with lenalidomide alone (HR 0.64, 95% CI 0.44–0.94; p = 0.023). High‐risk patients with at least a t(4;14), t(14;16), or del(17p) aberration also appeared to benefit, though not significantly, probably due to small numbers and missing data. In patients with ≥ 2 HRCAs the 3‐year PFS rate was 67% with KR compared to 42% with lenalidomide maintenance therapy (HR 0.53, 95% CI 0.24–1.15) [38]. The three‐year OS was similar between arms (94% vs. 90%). The most common grade 3–4 adverse events (AE) were neutropenia (13% vs. 10% vs. 11% across KRd plus ASCT, KRdx12, and KCd plus ASCT), dermatologic toxicity (6% vs. 8% vs. 1%), and hepatic toxicity (8% vs. 7% vs. 0%).

The ATLAS study compared KRd with single agent lenalidomide in 180 patients after ASCT [39]. The median age was 59 years. After a median follow‐up of 33.8 months, the median PFS was 59.1 months with the triple‐combination versus 41.4 months with lenalidomide alone (HR 0.51, 95% CI 0.31–0.80, p = 0.012). Serious adverse events (SAE) occurred in 30% of patients on KRd and in 22% of those receiving lenalidomide alone.

The TOURMALINE‐MM3 randomized 656 patients after single ASCT within 12 months of diagnosis in a 3:1 ratio to oral ixazomib or placebo on days 1, 8, and 15 in 28‐day cycles for 2 years [40]. The starting dose of 3 mg was increased to 4 mg from cycle 5, if tolerated. After a median follow‐up of 31 months, ixazomib reduced the risk of progression or death by 28% (median PFS 26.5 months [95% CI 23.7–33.8] vs. 21.3 months [95% CI 18.0–24.7]; HR 0.72, 95% CI 0.58–0.89; p = 0.0023). However, the absolute PFS gain was only 5.3 months, substantially less than with lenalidomide in comparable settings [40]. Moreover, no OS benefit was seen after 64 months follow‐up [41]. SAEs occurred in 27% in the ixazomib and in 20% in the placebo group. Rates of SPMs were similar (3%) in both groups.

The TOURMALINE‐MM4 trial randomized 706 TI patients 2:1 to receive ixazomib (n = 425) or placebo (n = 281) [42]. Ixazomib reduced the risk of progression or death by 34.1% (median PFS 17.4 vs. 9.4 months; HR 0.66, 95% CI 0.54–0.80; p < 0.001). The greatest benefit was seen in patients achieving complete or very good partial response after induction (median PFS 25.6 vs. 12.9 months; HR 0.59; p < 0.001). Grade ≥ 3 treatment‐emergent adverse events (TEAE) occurred in 36.6% versus 23.2% of patients. Treatment discontinuation because of TEAEs was noted in 12.9% versus 8.0%. Gastrointestinal events were more common with ixazomib, including nausea (26.8% vs. 8.0%), vomiting (24.2% vs. 4.3%), and diarrhea (23.2% vs. 12.3%). Rates of new primary malignancies (5.2% vs. 6.2%), and on‐study mortality (2.6% vs. 2.2%) were similar.

The HOVON‐126 trial randomized 143 patients after 9 cycles of induction with ixazomib, thalidomide, and dexamethasone to either ixazomib or observation [43]. A total of 78 patients were randomized to ixazomib maintenance or placebo. Median PFS was 9.5 months in ixazomib‐treated patients and 8.4 months for those receiving placebo (HR 0.80, CI 0.45–1.34, p = 0.39). Median OS has not been reached in the primary publication and was significantly improved after long‐term follow‐up, but these data are limited by low patient numbers [44]. Outcome was independent of cytogenetic risk. Age > 75 and frailty were associated with lower 2‐year OS (73% vs. 90% for ≤ 75 years, p = 0.002; 74% vs. 89%–90% for unfit/fit, p = 0.08). Ixazomib maintenance was well tolerated, with comparable PFS in older, frail, and fit patients.

In our meta‐analysis maintenance with PI compared with lenalidomide or observation showed a PFS benefit (HR 0.67, 95% CI 0.59–0.77) but no improvement of OS (HR 1.01, 95% CI 0.86–1.20) (Figure 3).

FIGURE 3.

Meta‐analysis of PIs versus other treatments for (A) PFS and (B) OS.

3.4. CD38 Antibodies

The CASSIOPEIA study evaluated daratumumab maintenance versus observation in patients previously randomized to either daratumumab‐bortezomib‐thalidomide‐dexamethasone (Dara‐VTd) or VTd induction therapy before ASCT [45]. Daratumumab maintenance significantly prolonged PFS (median not reached vs. 45.8 months; HR 0.49, 95% CI 0.40–0.59, p < 0.0001). A benefit was seen even in patients who had already received Dara‐VTd during induction (PFS not reached vs. 72.1 months; HR 0.76, 95% CI 0.58–1.00; p = 0.048), though the effect was more pronounced in daratumumab‐naïve patients (PFS median not reached vs. 32.7 months; HR 0.34, 95% CI 0.26–0.44; p < 0.0001).

MRD‐negativity rates were highest in patients receiving daratumumab during both induction and maintenance (65.1%), followed by those treated only during induction (58.1%), only during maintenance (53.5%), and those without daratumumab exposure (36.3%).

The PERSEUS [46] and GRIFFIN [47] trials compared daratumumab plus lenalidomide continuous therapy stopping after 2 years if MRD‐negative, with single‐agent lenalidomide continuous treatment after induction and consolidation with either daratumumab‐bortezomib‐lenalidomide‐dexamethasone (Dara‐VRd) or VRd. Of note, our analysis is based on reconstructed PFS curves with landmark analysis at the end of induction therapy. Both studies showed significantly prolonged PFS when daratumumab was incorporated into induction, consolidation, and maintenance therapy.

In PERSEUS, the estimated PFS rate at 48 months was 84.3% with continuous daratumumab‐lenalidomide versus 67.7% with continuous lenalidomide only (HR 0.42, 95% CI 0.30–0.59; p < 0.001) [48]. Complete response or better was achieved in 87.9% in the Dara‐VRd arm versus 70.1% in the VRd group (p < 0.001), and MRD‐negativity was obtained in 75.2% versus 47.5% of the respective groups (p < 0.001). Death occurred in 34 versus 44 patients, respectively. The benefit was consistent across nearly all risk groups, including patients with cytogenetic abnormalities. Only in the subgroup of patients with ≥ 2 high‐risk aberrations [49] and those aged 65 years or older, no clear advantage was observed (HR 0.97, 95% CI 0.52–1.81). However, a pooled analysis of elderly patients (≥ 65 years) from PERSEUS and GRIFFIN showed a trend toward improved PFS with Dara‐VRd (HR 0.56, 95% CI 0.30–1.01) followed by Dara‐R maintenance [48]. Daratumumab‐based therapy also improved rates of complete response or better (82.8% vs. 67.0%; OR 2.37, 95% CI 1.28–4.39; p = 0.0046) and MRD‐negativity (10⁻⁵; 66.4% vs. 41.7%; OR 2.75, 95% CI 1.61–4.71; p = 0.0002) in comparison to VRd alone.

In the GRIFFIN trial, with a design similar to that of the PERSEUS study, Dara‐VRd improved the rate of stringent complete response (67% vs. 48%; OR 2.18, 95% CI 1.22–3.89; p = 0.0079), and the 4‐year PFS (87.2% vs. 70%) compared to VRd (70.0%, HR 0.45; 95% CI 0.21–0.95; p = 0.032). Median OS was not reached for either arm (HR 0.90, 95% CI 0.31–2.56; p = 0.84). Cytopenias were more frequent with daratumumab‐based therapy, whereas SAEs were slightly more common in patients with VRd (52% vs. 46%).

As daratumumab was administered continuously throughout treatment in both the PERSEUS and GRIFFIN study, precluding conclusions on the added value of daratumumab maintenance, we performed a landmark analysis using individual patient data digitized from the Kaplan–Meier curves after consolidation to assess the impact of daratumumab‐lenalidomide (Dara‐R) versus lenalidomide maintenance therapy. This showed a benefit for Dara‐R in both studies: PERSEUS, HR 0.39 (95% CI 0.27–0.57, p < 0.001) and GRIFFIN, HR 0.47 (95% CI 0.21–1.05, p = 0.067) (Figure 4A).

FIGURE 4.

Meta‐analysis of CD‐38 monoclonal antibodies versus other treatments for (A) PFS and (B) OS.

The CEPHEUS study compared 8 cycles of Dara‐VRd or VRd followed by Dara‐Rd or Rd until progression in TI patients [50]. At a median follow‐up of 58.7 months, the MRD‐negativity rate at 10⁻⁵ (primary endpoint) was 60.9% with Dara‐VRd versus 39.4% with VRd (OR 2.37, 95% CI 1.58–3.55; p < 0.0001). Sustained MRD‐negativity (≥ 12 months) was significantly higher with Dara‐VRd versus VRd (48.7% vs. 26.3%; p < 0.0001). Risk of progression or death was 43% lower for Dara‐VRd versus VRd (HR 0.57, 95% CI 0.41–0.79; p = 0.0005). AEs were consistent with the known safety profiles for daratumumab and VRd.

The AURIGA trial evaluated Dara‐R vs. lenalidomide maintenance in 200 patients with NDMM, who achieved a very good partial response or better, remained MRD‐positive at 10⁻⁵, and were anti‐CD38‐naïve after ASCT [51]. Maintenance was given for up to 36 cycles. The 12‐month MRD‐negativity rate at 10⁻⁵ (primary endpoint) was significantly higher with Dara‐R (50.5% vs. 18.8%; OR 4.51, 95% CI 2.37–8.57; p < 0.0001). Sustained ≥ 12 months MRD‐negativity rate (29.3% vs. 7.9%) was also improved in the Dara‐R arm, indicating greater depth and durability of response with Dara‐R [52]. At a median follow‐up of 32.3 months, Dara‐R continued to show superior outcomes, including a higher overall MRD‐negative conversion (60.6% vs. 27.7%; OR 4.12, 95% CI 2.26–7.52; p < 0.0001) and a higher rate of complete response or better (75.8% vs. 61.4%; OR 2.00, 95% CI 1.08–3.69; p < 0.0255). PFS favored Dara‐R over lenalidomide (HR 0.53, 95% CI 0.29–0.97), with estimated 30‐month PFS rates of 82.7% versus 66.4%. A post hoc analysis [27], using multiple high‐risk definitions including the most recent model published by International Myeloma Society (IMS) [53], confirmed higher MRD conversion rates in high‐risk patients receiving Dara‐R.

The IMROZ trial randomized 446 TI patients to either 4 cycles isatuximab‐bortezomib‐lenalidomide‐dexamethasone (Isa‐VRd) induction therapy followed by continued treatment with Isa‐Rd or to the same regimens but without isatuximab [54]. At a median follow‐up of 59.7 months, the estimated PFS at 60 months was 63.2% in the Isa‐VRd, as compared to 45.2% in the VRd arm (HR 0.60, 95% CI 0.41–0.88; p < 0.001). The percentage of patients with a complete response or better was significantly higher in the Isa‐VRd group than in the VRd arm (74.7% vs. 64.1%, p = 0.01), as was the percentage of patients with MRD‐negative status and with a complete response (55.5% vs. 40.9%, p = 0.003). No new safety signals were observed with the Isa‐VRd. The incidence of SAEs during treatment and the incidence of AEs leading to discontinuation were similar in both groups.

An investigator‐initiated, single‐center trial (NCT04497961) comparing daratumumab (1800 mg weekly in Cycles 1–2, every 2 weeks in Cycles 3–6, and every 4 weeks in Cycles 7–36) with lenalidomide maintenance (10 mg on Days 1–21 of a 28‐day cycle) for 36 cycles was presented at ASH 2025 [55]. Eighty‐nine patients achieving very good partial response or better after induction, with or without ASCT, were enrolled. The primary endpoint was health‐related quality of life, assessed using the EORTC QLQ‐C30 instrument. Secondary endpoints included response rates, MRD, PFS, and toxicity. Quality of life scores improved gradually from a baseline score of 72, by 0.18 points per month, without significant difference between arms. Rates of complete response or better (82.2% vs. 68.2%) and MRD‐negativity (57.5% vs. 55.26%) were numerically higher in the daratumumab arm, although the trial was not powered for efficacy comparisons. PFS at 24 months from the start of maintenance was similar between daratumumab and lenalidomide (78.9% vs. 79.5%). The authors concluded that single‐agent daratumumab maintenance is an attractive alternative to lenalidomide maintenance.

Our meta‐analysis compared CD38‐based maintenance treatments with other treatments, with the former showing longer PFS (HR 0.50, 95% CI 0.44–0.58) and a tendency for longer OS (0.80, 95% CI 0.63–1.02), though OS data is still immature (Figure 4).

3.5. Other Agents

The GMMG‐HD6 trial investigated the effect of adding elotuzumab to bortezomib‐lenalidomide‐dexamethasone during induction and to lenalidomide during maintenance therapy in 555 TE patients with NDMM [56]. Patients were randomized into four groups: Rd induction, followed by lenalidomide maintenance, VRd induction followed by elotuzumab‐lenalidomide (E‐R) maintenance, E‐Rd induction followed by lenalidomide maintenance and E‐VRd induction followed by E‐R maintenance. Patients were transplanted after four induction cycles and received two consolidation cycles with the same regimen that was administered before. Maintenance was administered for 2 years. Results showed no difference in PFS and OS after 49 months follow‐up.

The Australian–New Zealand Myeloma Study Group evaluated whether combining selinexor with lenalidomide as maintenance therapy after ASCT would improve outcomes [57]. The study planned to enroll 290 patients but was stopped early after a futility analysis of 149 patients showed no PFS difference between the two arms. Thirty‐month PFS rates were 67% (95% CI 49%–79%) with lenalidomide and 71% (95% CI 57%–81%) with selinexor‐lenalidomide (SR). No PFS benefit was seen in patients with del(17p) (HR 1.06, 95% CI 0.54–2.09; p = 0.87) or any HRCA (HR 0.68, 95% CI 0.15–3.03; p = 0.61). Lenalidomide mean relative dose intensity (RDI) was lower in the SR arm (68%) compared with lenalidomide alone (81%; p = 0.003), while the mean RDI of selinexor was 55%. Grade ≥ 3 AEs were more frequent in the SR arm compared with lenalidomide alone (85% vs. 45%; p < 0.001). Although complete response rates were higher with SR, this came at the cost of increased toxicity, including more infections, cytopenias, and gastrointestinal AEs. Given available alternatives, selinexor is unlikely to be pursued further in maintenance therapy.

3.6. Duration of Maintenance Therapy

The optimal duration of maintenance remains under active investigation and depends on a variety of factors, particularly depth of response, sustained MRD‐negativity, cytogenetic risk, previous induction therapy, drug tolerance, and availability of drugs. Current evidence supports at least 2 years of maintenance in patients with sustained deep response [58]. Data with lenalidomide maintenance show a continued benefit up to 3 years, particularly in patients who remain MRD‐positive. In those patients, treatment may be prolonged, but the benefit is less well documented. In patients with ultra‐high‐risk features (including those with two or more HRCAs) long PFS was reported in studies like the GMMG‐CONCEPT [59], OPTIMUM/MUKnine [60], and IFM 2018–04 [61] trials which employed extensive intensified therapy [59].

The incremental benefit of continuing maintenance therapy beyond 2 years appears to diminish in patients with sustained MRD‐negativity, suggesting that treatment duration should be individualized according to MRD status and biological risk [62]. Prospective MRD‐guided trials are expected to further refine the balance between long‐term disease control and treatment burden.

3.7. Network Meta‐Analysis

The network of maintenance treatments included 14 treatment regimens, each compared with observation. The network was largely open looped (Figure S3), with low heterogeneity within and between studies as documented by Q‐statistics (Table S5). I ² values for both PFS and OS were 0% and 12.5%, respectively. Subgroup analyzes (Figures S4–S8) further support the robustness of the network.

3.7.1. Progression‐Free Survival

Our network meta‐analysis compared each maintenance regimen with observation in terms of PFS and OS using a random effects model (Figure 5). The results suggested a ranking of maintenance protocols; however, considerable overlap in confidence intervals and P‐scores (Table S6) indicated that the hierarchy is not clearly defined, hence the analysis should be considered explanatory rather than definitive. All pairwise comparisons are shown in Table S7. In TE patients, Dara‐R reduced risk of progression by nearly 80% compared to observation (HR 0.24, 95% CI 0.18–0.32). This is followed by two carfilzomib‐based regimens, namely KRd (HR 0.27, 95% CI 0.16–0.46) and KR (HR 0.34, 95% CI 0.23–0.50). In TI patients, Dara‐Rd (HR 0.29, 95% CI 0.16–0.52) and Isa‐Rd (HR 0.36, 95% CI 0.21–0.64) were ranked third and fifth and were found to reduce risk of progression by 71% and 64%, respectively.

FIGURE 5.

Network meta‐analysis of all included treatment regimens for (A) PFS and (B) OS.

Further improvements were seen with lenalidomide‐prednisone: HR 0.45 (95% CI 0.37–0.55), daratumumab monotherapy: HR 0.49 (95% CI 0.40–0.59), and lenalidomide alone: HR 0.53 (95% CI 0.49–0.58), all in TI patients. Rd, HR 0.58 (95% CI 0.37–0.92), and ixazomib monotherapy, HR 0.69 (95% CI 0.60–0.80), provided modest benefit both in TE and TI patients. E‐Rd improved PFS in TE patients (HR 0.56, 95% CI 0.33–0.96) [56]. Regimens such as ixazomib‐Rd [63], SR [57] and daratumumab‐bortezomib [64] showed wider confidence intervals and did not reach statistical significance.

3.7.2. Overall Survival

Our analysis identified two regimens with a significant OS benefit: KR, HR 0.45 (95% CI 0.21–0.97), and Dara‐R, HR 0.60 (95% CI 0.36–1.00). Both regimens emerge as the most promising maintenance treatments for improving OS in TE patients. Lenalidomide monotherapy, HR 0.84 (95% CI 0.73–0.96), also showed a clear survival benefit in both TE and TI patients. In TE patients, KRd provided a modest but consistent survival benefit, HR 0.69 (95% CI 0.28–1.72). Presently available data for Isa‐Rd, HR 0.98 (95% CI 0.46–2.11), showed no clear survival advantage, and results for lenalidomide‐prednisone, HR 1.06 (95% CI 0.84–1.33), ixazomib monotherapy, HR 1.06 (95% CI 0.75–1.52), and Rd, HR 1.18 (95% CI 0.63–2.22), showed no significant effect on survival. For E‐Rd a HR 1.71 (95% CI 0.75–3.90) was noted with the point estimate suggesting worse survival.

3.8. Ongoing Studies

Ongoing studies using novel maintenance approaches are summarized in Table 2. Several trials evaluate isatuximab either as a single‐agent maintenance or in combination regimens. In the German HD7 trial (NCT03617731), patients were first randomized to three 42‐day cycles of VRd with or without isatuximab followed by ASCT. Results of this phase have already been published [65]. In the ongoing maintenance part, patients are randomized again to receive lenalidomide alone or lenalidomide plus isatuximab therapy for 3 years.

TABLE 2.

Ongoing studies including a maintenance phase with novel treatments in NDMM.

Study names	Trial register numbers	TE/TI	Treatment
			Experimental arm	Control arm
Monoclonal antibodies
BENEFIT	NCT04751877	TI	Isa‐VRd ×12 → Isa‐VR ×6 → Ida‐R cont	Isa‐Rd → Isa‐R cont
FiTNEss (UK‐MRA Myeloma XIV)	NCT03720041	TI	Standard or frailty adjusted dosing → IR	Standard or frailty adjusted dosing → R
ECOG‐ACRIN EAA181 (EQUATE)	NCT04566328	TI	DRd → Rand VRd → DR	DRd → Rand DRd → DR
GMMG‐HD7	NCT03617731	TE	Isa‐RVd → ASCT → Rand Isa‐R or R 36mos	RVd → Rand Isa‐R or R 36mos
SWOG S2209	NCT05561387	TI	VRd lite → R	DRd → Rand to R or DR
IsKia	NCT04483739	TE	Isa‐KRd → ASCT → Isa‐KRd ×4 → Isa‐KRd ×12	KRd → ASCT → KRd ×4 → KRd ×12
Bispecific antibodies
EMN30/MajesTEC‐4	NCT05243797	TE	Tec‐R or Tec	R
IFM 2024‐06	NCT06931704	TE	ASCT → Elra‐R 24mos	ASCT → DR 24mos
IFM 2025‐01 (ElLen)	NCT06918002	TE	ASCT → Elra‐R or Elra	ASCT → SOC
MagnetisMM‐7	NCT05317416	TE	Elra	R
CAR‐T cells
CARTITUDE‐5	NCT04923893	TI	VRd ×6 + VRd ×2 → Cilta‐cel	VRd ×6 + VRd ×2 → Rd until PD/Tox
EMAGINE/CARTITUDE‐6	NCT05257083	TE	DVRd ×6 → Cilta‐cel → R for 2yrs	DVRd ×4 → ASCT → DVRd ×2 → R for 2yrs
KarMMa‐9	NCT06045806	TE	ASCT → Ide‐cel‐R	ASCT → R
Cereblon E3 ligase modulating drugs
EMN26	NCT04564703	TE	ASCT → Iber in 3 cohorts of Iber with starting dose 1.3, 1.0, or 0.75 mg
GEM2021menos65	NCT05558319	TE	VRd → ASCT → extended VRd OR Isa‐VIber‐d → ASCT → Isa‐Iber	Isa‐VRd → ASCT → Isa‐R
GMMG‐HD9/DSMM XVIII	NCT06216158	TE	Isa‐Iber 36mos	Iber 36mos
MRD‐guided maintenance
Remnant	NCT04513639	TE	MRDneg patients → Rand to second line TX at MRD reappearance	MRDneg patients → or second line TX when meeting IMWG progression criteria

Abbreviations: Belamaf, Belantamab Mafodotin; C, Cyclophosphamide; Cilta‐cel, Ciltacabtagen Autoleucel; Cont, continued; D, Daratumumab; d, Dexamethasone; Elra, Elranatamab; Iber, Iberdomide; Ide‐cel, Idecabtagen‐Vicleucel; Isa, Isatuximab; K, Carfilzomib; PD, progressive disease; R, Lenalidomide; Rand, Randomization; Tox, unbearable toxicity; TX, therapy; V, Bortezomib.

A key question in maintenance therapy is the optimal treatment duration, particularly in regard to the MRD status. The DRAMMATIC trial (NCT04071457) randomizes patients to Dara‐R or lenalidomide maintenance after induction therapy and ASCT [66]. After 2 years, MRD‐negative patients are re‐randomized to continue or stop maintenance, while MRD‐positive patients remain on therapy. Treatment continues for up to 7 years from maintenance randomization, with follow‐up extending to 15 years.

Another pending question is how to manage relapses after loss of MRD‐negativity. The REMNANT study (NCT04071457) randomizes patients to start second line treatment either at MRD reappearance or at IMWG defined progression [67]. Most ongoing studies are testing the addition of CD38 or of bispecific antibodies to maintenance regimens. Early EMN/MajesTEC‐4 data show MRD‐negativity rates of 100% with teclistamab or teclistamab‐lenalidomide after ASCT [10]. Teclistamab combined with daratumumab also achieved promising results in RRMM [68]. BCMA‐directed CAR‐T cell therapy is being evaluated in the CARTITUDE‐5 trial (NCT04923893) [69] which randomizes TI patients after 6 cycles of VRd to either two additional cycles of VRd followed by Rd maintenance until progression/intolerance or to a single infusion with ciltacabtagene autoleucel [70]. The cereblon E3 ligase modulating drug iberdomide is investigated in various dose levels, both in combination with isatuximab‐bortezomib, or as monotherapy compared with isatuximab‐iberdomide in the maintenance setting [69].

The GMMG‐HD9 study randomizes patients who received induction therapy and ASCT in the GMMG‐HD8/DSMMXIX trial (NCT06216158) to receive 39 cycles of iberdomide maintenance therapy or the same number of cycles with iberdomide plus isatuximab [71]. In both arms, patients will receive 20 mg dexamethasone in cycle 1 on the same days as the isatuximab administration in Arm B. The end of study will occur after 36 months of the maintenance therapy. The primary objective is MRD‐negativity after 24 months of therapy. The GMMG‐HD9/DSMM XVIII (NCT06216158) is the subsequent maintenance therapy following the GMMG‐HD8/DSMM XIX trial, which compared 3 cycles of VRd with either subcutaneous or intravenous isatuximab followed by ASCT. The maintenance part of the study randomizes patients to either iberdomide or iberdomide plus isatuximab maintenance therapy for 2 years. The EMN26 study [69] was a dose‐finding study for iberdomide as maintenance therapy after ASCT. The established recommended dose of 75 mg is now compared to lenalidomide in 1.216 patients in the EXCALIBER maintenance study (NCT05827016), where treatment will be continued until progression and/or intolerance. The MAGNETISMM‐7 study (NCT05317416) compares maintenance therapy with the bispecific antibody elranatamab with lenalidomide in patients who are MRD‐positive after ASCT.

4. Discussion

Our meta‐analysis confirmed that lenalidomide significantly prolongs PFS in both TE and TI patients (Figure 2). A survival benefit, however, was observed only in CALGB 1001048 [5] and in the TE cohort of the MYELOMA XI trial [4]. PIs also improved PFS (Figure 3) but have not shown an OS effect. CD38 antibodies result in a similar pattern (Figure 4) with clear PFS benefits and a trend toward longer OS, though data remain immature. When results from AURIGA, PERSEUS, IMROZ, and CEPHEUS were pooled in a random‐effects model, the combined estimate suggested a 21% reduction in mortality risk (HR 0.79, 95% CI 0.59–1.05) but the confidence interval still includes 1. Finally, comparisons between lenalidomide and various combinations must be interpreted cautiously, as lenalidomide was tested against observation, whereas Dara‐R and KR were evaluated against lenalidomide itself, inherently favoring the single agent comparator.

Our network meta‐analysis compared each regimen with observation, thereby providing a ranking of the most effective maintenance regimens (Figure 5). The greatest reduction in the risk of progressive disease—nearly 80%—was observed with Dara‐R maintenance therapy, followed by KRd, Dara‐Rd, KR, and Isa‐Rd, which reduced the risk of progression by 73%, 71%, 66%, and 64%, respectively. Other regimens, including lenalidomide‐prednisone, single‐agent daratumumab, lenalidomide, ixazomib, E‐Rd, and Rd also prolonged PFS, though with more modest risk reductions of 31%–55%.

A significant OS benefit was observed with KR, Dara‐R, and lenalidomide alone, reducing the risk for mortality by 55%, 40%, and 16%, respectively. These benefits were confined to TE patients. Data for Isa‐Rd in this group are still pending.

Subgroup analyzes showed similar patterns (Figures S4–S8). However, many treatment combinations were present in only one subgroup, limiting the strength of these comparisons. The best treatment combinations in TE and TI patients were Dara‐R and Dara‐Rd, respectively.

Data on patients with high‐risk cytogenetics remain limited because of small sample sizes and the scarcity of studies reporting detailed subgroup analyzes. In this population, Dara‐R and KR improved progression‐free survival by 74% and 66%, respectively. Among standard‐risk patients, a significant prolongation of progression‐free survival was observed with six regimens (Dara‐R, KR, lenalidomide, daratumumab, lenalidomide and ixazomib), with the greatest benefit again seen with Dara‐R and KR. These two regimens also emerged as preferred options for high‐risk patients as well, although the magnitude of benefit is lower.

Independently from maintenance duration (continued vs. limited duration), KRd demonstrated superior efficacy, which was matched by Dara‐R in studies with limited treatment duration. The optimal duration of maintenance therapy remains one of the most clinically relevant open questions. Based on currently available data, it is unclear whether continuing maintenance until progression—an approach recommended by many experts—is superior to a fixed, limited treatment duration. In the subgroup analysis comparing continued vs. limited maintenance, only three regimens (Dara‐R, lenalidomide, and ixazomib) appeared in both subnetworks, and their hazard ratios were similar, with broadly overlapping confidence intervals. Hence, the statistical data are not mature enough to clarify this issue. Several groups are currently investigating the potential benefits of an individualized strategy in which the duration of maintenance therapy is tailored to a patient's MRD status, particularly the achievement of sustained MRD‐negativity. Currently, MRD‐negativity at a threshold of 10⁻⁵ for at least 12 months, and ideally for 24 months, is the best validated and clinically meaningful predictive parameter for outcome of patients with multiple myeloma [25]. Standard‐risk patients have a low risk of MRD resurgence [62], making treatment discontinuation a reasonable option. In contrast, patients who remain MRD‐positive and those with ultra‐high risk are generally advised to continue therapy until disease progression or treatment intolerance. Recently, data from Costa et al. suggested that, even in high‐risk patients, treatment discontinuation may be considered, provided they have been in sustained MRD‐negativity at 10⁻⁵ for at least two years [25] as the risk of MRD resurgence at 5 years was lower than 10% in their patient cohort.

To test the robustness of the network meta‐analysis we defined subgroups for digitized versus published data and low vs. high risk of bias, which were broadly consistent with the main analysis, supporting its validity.

Our study has several important limitations. Some trials reported their data without clearly distinguishing between the induction or consolidation phases and the maintenance period. To perform a landmark analysis from the start of maintenance, we therefore had to make assumptions about the initiation of maintenance therapy and rely on digitized data for the corresponding calculations. Nevertheless, the quality of the digitization and landmark process has been validated and demonstrated high accuracy with low sensitivity to methodological variation (Tables S2, S3) [12]. Network meta‐analysis relies on the assumption of transitivity and is inherently prone to bias arising from differences in patient populations, dosing schedules, and maintenance duration. As a result, it cannot fully account for heterogeneity across trials, even though the Q‐statistics, I ², and subgroup analyzes showed minimal evidence of heterogeneity and indicated strong robustness of the network. In addition, survival data are often immature or incompletely reported, and information on post‐progression therapies is frequently lacking. These factors should be considered when interpreting the results. Despite these limitations, the methodology allows indirect comparisons of treatments that have not been evaluated head‐to‐head and offers a structured approach to ranking therapeutic options to support clinical decision‐making.

Several ongoing studies are evaluating novel treatments—including bispecific antibodies, CelMoDs, CAR‐T cells, and combinations of these agents with established therapies—for use in maintenance, raising hope for deeper and more durable responses and, ultimately, the elimination of malignant clones and/or reinforcement of the bone marrow microenvironment to enhance immune surveillance. These innovations, together with others likely to emerge [72], may finally tip the balance of treatment success in multiple myeloma from incurable to curable in the years ahead.

Author Contributions

Heinz Ludwig and Sarah Bernhard designed the study and discussed areas of greatest importance with all authors. Sarah Bernhard conducted all statistical analyzes. All authors had access to and verified the data, participated in data interpretation, contributed to drafting and revising the manuscript, and approved the final version prior to submission.

Funding

The study was supported by the Austrian Forum Against Cancer.

Ethics Statement

This study is a systematic review with meta‐ and network meta‐analyzes based exclusively on data from previously published studies. As no new data were collected and no individual patient‐level data were used, ethics committee approval was not required.

Consent

This study did not involve the collection or use of individual patient data. Therefore, patient consent was not required.

Conflicts of Interest

Heinz Ludwig: consulting/speakers bureau/research funding: AMGEN, Takeda, Celgene‐BMS, Janssen, Pfizer, Sanofi; Evangelos Terpos: advisory board/honoraria/research funding: Amgen, Antengene, AstraZeneca, EUSA Pharma, BMS, Forus, GSK, J&J, Menarini/Stemline, Novartis, Pfizer, Sanofi, Swixx, Takeda; Francesca Gay: Honoraria/consultancy fees: AbbVie, Amgen, AstraZeneca, BMS, Celgene, Gilead, GSK, J&J, Kite, Oncopeptides, Pfizer, Roche, Sanofi, Takeda; Niels W.C.J. van de Donk: research support/advisory board: Janssen Pharmaceuticals, Amgen, Celgene, Novartis, Cellectis, BMS, Sanofi, Takeda, Roche, Bayer, Adaptive, Galapagos, Kite Pharma, Merck, Pfizer, AbbVie, Servier; Sarah Bernhard: None; Monika Engelhardt: Advisory support: Amgen, GSK, Janssen, Stemline, Oncopeptides, Pfizer, Sanofi, Takeda, BMS; Gordon Cook: Honoraria/grants/funds: Takeda, Amgen, BMS, J&J, Roche, Menarini, Pfizer; Fredrik Schjesvold: Honoraria/consultancy fees/research funding: AbbVie, Amgen, BMS, Caedo Therapeutics, Galapagos, GSK, J&J, Menarini, Oncopeptides, Pfizer, Regeneron, Sanofi, Takeda, XNK Therapeutics, Circio; Hermann Einsele: Honoraria/research funding/advisory board: Amgen, BMS, J&J, Novartis, Sanofi, Takeda; Graham Jackson: None; Charlotte Pawlyn: Honoraria/advisory board: AbbVie, Amgen, BMS/Celgene, Cellcentric, GSK, J&J, Menarini Stemline, Opna Bio, Pfizer, Sanofi, iTEOS therapeutics; María‐Victoria Mateos: Advisory board/speakers bureau: Amgen, BMS/Celgene, GlaxoSmithKline, Janssen Cilag, Kite Pharma, Oncopeptides, Pfizer, Roche, Sanofi, Stemline; Sonja Zweegman: Research funding/advisory board: J&J, BMS, Sanofi, Oncopeptides, Amgen; Martin Schreder: None; Meral Beksaç: Honoraria/advisory board/research funding: Sanofi, Celgene, Amgen, Janssen, Novartis, BMS, AbbVie, GlaxoSmithKline, Mundipharma; Christoph Driessen: Consulting/advisory role/speaker bureau: BMS, J&J, Amgen, Novartis, GSK; Enrique M. Ocio: Consultancy/honoraria: AbbVie, Amgen, Astra Zeneca, BMS, GSK, Janssen, Menarini, Oncopeptides, Pfizer, Regeneron, Sanofi, Takeda; Rakesh Popat: Honoraria/consultancy/research funding: J&J, BMS, AbbVie, GSK, Pfizer, Roche; salary supported by the NIHR UCLH Biomedical Research Centre; Leo Rasche: Honoraria/research support: Janssen, BMS, Sanofi, Pfizer, Amgen, GSK, Roche; Cyrille Touzeau: Advisory board/honoraria/research funding: Amgen, BMS, GSK, J&J, Pfizer, Sanofi, Takeda, AbbVie; Pellegrino Musto: Honoraria/lectures/advisory board: AbbVie, Alexion, Amgen, Astellas, Astra‐Zeneca, Bei‐Gene, BMS, Gilead, GlaxoSmithKline, Grifols, Incyte, J&J, Jazz, Novartis, Pfizer, Roche, Sanofi, Sobi, Takeda; Annemiek Broijl: Advisory role/speakers bureau: BMS, J&J, Amgen, Sanofi; Meletios Dimopoulos: Advisory board/satellite symposia: Amgen, Sanofi, Regeneron, Menarini, Takeda, GSK, BMS, Janssen, BeiGene, Swixx, AstraZeneca; Roman Hajek: Honoraria/consultancy fees/research funding: AbbVie, Amgen, BMS, Celgene, GSK, J&J, Novartis, Oncopeptides, PharmaMar, Sanofi, Takeda; Mario Boccadoro: Honoraria/research funding/advisory board: AbbVie, Amgen, BMS, Celgene, GSK, J&J, Mundipharma, Novartis, Sanofi; Pieter Sonneveld: Research funding/advisory board/patents/royalties: Amgen, BMS, Celgene, J&J, Karyopharm, Oncopeptides, Pfizer; President of the EMN.

Supporting information

Table S1: Detailed search strategies used in different databases.

Table S2: Quality assessment of the digitization process comparing (A) PFS and (B) OS outcome measures of the digitized data and the original data.

Table S3: Sensitivity analysis for the landmarked studies regarding (A) PFS and (B) OS.

Table S4: List of excluded studies.

Table S5: Q‐statistics for (A) total heterogeneity in PFS, (B) design‐specific heterogeneity in PFS, (C) total heterogeneity in OS, (D) design‐specific heterogeneity in OS.

Table S6: Ranking of treatments with their corresponding P‐score for (A) PFS and (B) OS.

Table S7: League tables showing the pairwise comparison between all treatment regimens for (A) PFS and (B) OS.

Figure S1: Heat plot showing the risk of bias in the included studies according to the Cochrane Risk of Bias 2.0 tool (RoB2).

Figure S2: Comparison‐adjusted funnel plot.

Figure S3: Network plot of the network meta‐analysis. The red line indicates, that the comparison could not be considered in the analysis due to lacking connections to other treatments.

Figure S4: Subgroup analysis of PFS according to transplant eligibility. (A) Forest plot and (B) network plot of treatments in TE patients; (C) Forest plot and (D) network plot of treatments in TI patients. The red line indicates, that the comparison could not be considered in the analysis due to lacking connections to other treatments.

Figure S5: Subgroup analysis of PFS according to cytogenetic risk. (A) Forest plot and (B) network plot of treatments in patients with high‐risk cytogenetics; (C) Forest plot and (D) network plot of treatments in patients with standard cytogenetics. The red line indicates, that the comparison could not be considered in the analysis due to lacking connections to other treatments.

Figure S6: Subgroup analysis of PFS according to treatment duration. (A) Forest plot and (B) network plot of treatments in studies with continuous treatment; (C) Forest plot and (D) network plot of treatments in studies with a predetermined treatment duration. The red line indicates, that the comparison could not be considered in the analysis due to lacking connections to other treatments.

Figure S7: Subgroup analysis of PFS according to data retrieval. (A) Forest plot and (B) network plot of treatments in studies where maintenance data was published; (C) network plot of treatments in studies that were digitized and landmarked. The red line indicates, that the comparison could not be considered in the analysis due to lacking connections to other treatments. A network meta‐analysis was not feasible in the digitized subgroup due to the low number and lack of connection between treatment regimens in this group.

Figure S8: Subgroup analysis of PFS according to risk of bias according to RoB2. (A) Forest plot and (B) network plot of treatments in studies with high or moderate risk of bias; (C) Forest plot and (D) network plot of treatments in studies with low risk of bias. The red line indicates, that the comparison could not be considered in the analysis due to lacking connections to other treatments.

Data Availability Statement

Data were obtained as stated in publications and by digitizing survival curves. The dataset can be obtained from the corresponding author on reasonable request. In case of reasonable request, please contact Heinz Ludwig at heinz.ludwig@extern.gesundheitsverbund.at