Real-world efficacy and safety of the carfilzomib–pomalidomide–dexamethasone regimen in patients with relapsed/refractory multiple myeloma: A multicenter retrospective study Authors: Huishou Fan, XiangYan Feng, Rongxia Wei, Xin Zhang, Tianlan Li, Lu Liu, Hui Wang, Lingjie Sun, Xue Shi, Hongguo Zhao, Zhao Li, Jingying Cui, Junqing Xu, Junjie Ma, Xianqi Feng, Wei Wang Published: 2026 ABSTRACT BackgroundRelapsed and refractory multiple myeloma (RRMM) is generally associated with a poor prognosis.ObjectivesThis real-world study aims to evaluate the efficacy and safety of the carfilzomib–pomalidomide–dexamethasone (KPd) regimen in patients with RRMM.DesignA multicenter, retrospective study was conducted in four centers in China.MethodsRRMM patients who received at least 1 cycle of KPd across four centers were retrospectively included and stratified by prior lines of treatment, survival outcomes and safety profile were analyzed.ResultsA total of 82 patients were enrolled. Based on the lines of treatment (LOT) at KPd initiation, patients were stratified into second-line (2L, n=39), third-line (3L, n=26), and fourth-line or beyond (4L+, n=17) groups. Among 72 response-evaluable patients, the overall response rate (ORR) was 69.4%, with ORRs of 79.4%, 73.9%, and 40.0% in the 2L, 3L, and 4L+ groups, respectively. After a median follow-up of 10.8 months, the median progression-free survival (mPFS) for the entire cohort was 22.5 months, while the median overall survival (mOS) was not reached (NR). The mPFS was NR, 22.5, and 10.1 months (P=0.178), and the mOS was NR, NR, and 11.6 months (P=0.019) for patients receiving KPd in the 2L, 3L, and 4L+ settings, respectively. Multivariable analysis identified elevated lactate dehydrogenase (LDH) level (HR=3.489, 95% CI: 1.557–7.823) and LOT ≥4 (HR=2.791, 95% CI: 1.086–7.177) as independent predictors of inferior PFS (P<0.05), while LOT ≥4 (HR=3.917, 95% CI: 1.258–12.200) was also associated with worse OS (P=0.019). Grade ≥3 adverse events (AEs) occurred in 12.8%, 50.0%, and 47.1% of patients in the 2L, 3L, and 4L+ groups, respectively.ConclusionThe KPd regimen demonstrated favorable clinical efficacy and manageable toxicity in RRMM, particularly in those with early-line relapse. Introduction Multiple myeloma (MM) is a hematologic malignancy characterized by the clonal proliferation of plasma cells in the bone marrow and the secretion of monoclonal immunoglobulin (M protein), often leading to CRAB symptoms and posing a serious threat to patient health.1,2 With advancements in novel therapies—including proteasome inhibitors (PIs), immunomodulatory agents (IMiDs)—as well as the introduction of upfront autologous stem cell transplantation (ASCT), the response rates and prognosis of patients with newly diagnosed MM (NDMM) have improved significantly over the past decades.2–5 Even in patients not eligible for immediate ASCT, the triplet regimen of bortezomib, lenalidomide, and dexamethasone (VRd) achieves high response rates, with a median progression-free survival (mPFS) of 41 months and a median overall survival (mOS) not yet reached after long-term follow-up.4,6 Despite recent therapeutic advances, MM remains incurable, and most patients eventually relapse and become refractory to existing treatments. The prognosis of patients with relapsed/refractory MM (RRMM) is significantly worse, making treatment decisions increasingly challenging.7,8 According to the European Society for Medical Oncology (ESMO) and European Hematology Association (EHA) guidelines, currently approved treatment options for RRMM include bortezomib-based combination regimens, next-generation PIs, IMiDs, and anti-CD38 monoclonal antibodies. 9 Carfilzomib, a second-generation PI, binds irreversibly to the proteasome, providing more sustained inhibition than bortezomib. Its efficacy and safety—either as carfilzomib–dexamethasone (Kd) or carfilzomib–lenalidomide–dexamethasone (KRd) in patients with RRMM were evaluated not only in phase III trials of ENDEAVOR and ASPIRE, respectively,10,11 but also in real-world settings.12,13 Owing to its efficacy and manageable toxicity, carfilzomib has been further explored in combination with dexamethasone and pomalidomide (KPd), daratumumab (DKd), or isatuximab (Isa-Kd), as well as in the quadruplet regimen with daratumumab, pomalidomide, dexamethasone (DKPd).14–19 Although quadruplet therapies may offer deeper responses, their broader use in real-world practice is often limited by cost, toxicity, and the need to preserve options for future relapse. In contrast, triplet regimens such as KPd, may offer a more feasible and widely applicable treatment option for RRMM patients in routine clinical practice. Although more intensive regimens, such as DRd, D-VRd, and Isa-VRd have been approved internationally for NDMM, VRd remains the most commonly used upfront induction regimen in China, largely due to limited drug accessibility, lack of reimbursement, and economic burden. Patients who progress during VRd treatment are often considered to have developed dual resistance to bortezomib and lenalidomide, representing a population with a poor prognosis. Previous studies have demonstrated that the KPd regimen exhibits promising efficacy in patients with RRMM. In the SELECT trial, KPd was evaluated in 52 patients with RRMM who had received one to two prior lines of treatment and were refractory to lenalidomide. The regimen showed encouraging efficacy, with an overall response rate (ORR) of 58%, mPFS of 11.1 months, and mOS of 18.8 months. 20 Furthermore, in the EMN011/HOVON114 trial, 111 patients who relapsed or progressed after frontline VRd induction received 8 cycles of KPd as second-line treatment, yielding superior efficacy with an ORR of 92% and mPFS of 26 months. 21 Despite these encouraging results from clinical trials, real-world evidence on the efficacy, safety, and survival outcomes of the KPd regimen in RRMM remains limited, particularly in patients who have previously received different lines of treatment. To address this gap, we conducted a multicenter retrospective study to evaluate the efficacy and safety of KPd in RRMM patients who had received at least one prior lines of treatment. Additionally, we analyzed potential prognostic factors that may impact the survival outcomes of the RRMM population receiving the KPd regimen, aiming to provide insights for real-world clinical practice. Methods Study design and participants It used a prospectively maintained institutional database of patients with MM. Additional data elements were obtained or confirmed by review of patient medical records and follow-up phone calls. In this study, the off-label treatment with KPd was performed after obtaining informed consent from all patients before therapy. The reporting of this retrospective multicenter study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement. 22 Patients were eligible for inclusion in this study if they met the following criteria: (1) RRMM treated at the participating research centers between February 2023 and July 2024; (2) received at least one cycle of carfilzomib-based regimen treatment. Among 98 eligible patients who met the inclusion criteria, 16 were excluded due to receiving a non-KPd regimen, including 10 who received KRd, 2 DKd, 2 KAd, 2 KCd, and 1 Kd. This resulted in a final cohort of 82 patients who fulfilled all criteria and were included in the study. Treatment regimen and response evaluation KPd was administered to RRMM patients in 28-day cycles as follows: carfilzomib (20/27 mg/m2, days 1, 2, 8, 9, 15, 16) with pomalidomide (4mg days 1–21) and dexamethasone (20mg days 1, 2, 8, 9, 15, 16). As all enrolled patients received at least 1cycle of KPd therapy, response assessment was performed every two cycles according to the IWMG response criteria. 23 Cytogenetic risk stratification Purified CD138+ plasma cells were isolated at diagnosis, and fluorescence in situ hybridization (FISH) was performed to assess cytogenetic abnormalities, including del(13q), del(17p), t(11; 14), t(14; 16), t(4; 14), t(14; 20), and 1q21 abnormalities. 24 High-risk was defined by the presence of any of the following CAs: t(4; 14), t(14; 16), del(17p), or 1q21 gain/amplification (1q21+). Patients without these CAs were classified as standard risk (SR). A 10% cut-off was used for defining translocations, deletions, and gains. 25 The International Staging System (ISS) was also applied for risk stratification prior to initiating KPd treatment. 26 Statistical analysis PFS was measured from the initiation of treatment of KPd to disease progression, death, or the last follow-up, whichever occurred first. OS was defined as the time period from treatment initiation of KPd to any-cause death or the last follow-up. Both of PFS and OS were estimated using the Kaplan–Meier method, and log-rank tests were used to compare the survival curves. The risk ratio was calculated using a univariate Cox proportional hazard model. Variables with p < 0.1 in the univariate Cox analysis were included in the multivariate Cox analysis. A p < 0.05 was considered statistically significant for the multivariable analyses. Statistical analyses were performed using SPSS (v25.0, IBM, Chicago, IL) and R (v4.1.2, Foundation for Statistical Computing, Vienna, Austria) software. Results Characteristics of the study population A total of 82 RRMM patients were enrolled from four participating centers. At the initiation of KPd treatment, the median age was 63 years (range: 38–77), and 45.1% were male. ISS staging at treatment initiation showed 33 (40.2%) patients in stage I, 20 (24.4%) in stage II, and 24 (29.3%) in stage III. M-protein subtypes included IgG in 49 (59.8%) patients, IgA in 16 (19.5%), light chain in 14 (17.1%), and non-secretory in 3 (3.7%). Due to the lack of FISH testing at relapse, cytogenetic risk classification was based on data from initial diagnosis (Table S1). Among 39 patients with available cytogenetic data, 28 (71.8%) had HRCAs, and 11 (28.2%) had SR features; the distribution of CAs is detailed in Table S1. Based on the lines of treatment (LOT) at KPd initiation, 39 in second-line (2L), 26 in third-line (3L), and 17 in fourth-line or beyond (4L+). Baseline disease characteristics were generally comparable across lines of treatment groups, except for a significantly higher incidence of thrombocytopenia with increasing lines of treatment (P = 0.036). Additional details are provided in Table 1.Table 1.Patient characteristic at KPd initiationCharacteristic at KPd initiationOverall(n=82)2L(n=39)3L(n=26)4L+(n=17) P value Gender, n (%)0.053 male37(45.1)14(35.9)11(42.3)12(70.6)​Age, year0.746 Median (Range)63.0(38,77)62.0(38,77)64.5(45,77)64.0(54,77)​ <6546(56.1)24(61.5)13(50.0)9(52.9)0.628≥6536(43.9)15(38.5)13(50.0)8(47.1)​ISS stage, n (%)0.132 I33 (40.2)19 (48.7)10 (38.5)4 (23.5)​ II20 (24.4)11 (28.2)6 (23.1)3 (17.6)​ III24 (29.3)7 (17.9)8 (30.7)9 (52.9)​ Unknown5 (6.1)2 (5.1)2 (7.7)1 (5.9)​ECOG performance status, n (%)0.927 015(18.3)7(17.9)4(15.4)4(23.5)​ 141(50.0)20(51.3)13(50.0)8(47.1)​ ≥224(29.3)11(28.2)9(34.6)4(23.5)​Unknown2(2.4)1(2.6)01(5.9)​Type of M protein, n (%)0.364 IgA16(19.5)7(18.0)5(19.2)4(23.5)​IgG49(59.8)22(56.4)16(61.5)11(64.7)​ Nonsecretory3(3.7)02(7.7)1(5.9)​ Light-chain14(17.1)10(25.6)3(11.5)1(5.9)​Light-chain subtype, n (%)0.074 K35(42.7)16(41.0)8(30.7)11(64.7)​ λ44(53.6)23(59.0)16(61.5)5(29.4)​ Unknown3(3.7)02(7.7)1(5.8)​Thrombocytopenia a , n (%)0.036 Yes23(28.0)6(15.4)11(42.3)6(35.3)​ No58(70.7)33(84.6)14(53.8)11(64.7)​ Unknown1(1.2)01(3.8)0​Elevated LDH b , n (%)0.709 Yes16(19.5)9(23.1)5(19.2)2(11.8)​ No63(76.8)30(76.9)19(73.1)14(82.4)​ Unknown3(3.7)02(7.7)1(5.9)​Cytogenetic risk at diagnosis c , n (%)0.045 High risk28(34.1)13(33.3)12(46.2)3(17.6)​ Standard risk11(13.4)10(25.6)1(3.8)0​ Unknown43(52.4)16(41.0)13(50.0)14(82.4)​2L, second line; 3L, third line; 4L+, fourth or later lines; ECOG, Eastern Cooperative Oncology Group; ISS, International Staging System; KPd, carfilzomib, pomalidomide, dexamethasone;LDH, lactate dehydrogenase.aThrombocytopenia was defined as a platelet count of <100 × 109/L;bElevated LDH was defined as ≥247 U/L;cHigh-risk cytogenetics is defined as the presence of any of the following: t(4;14), t(14;16), deletion 17p, and 1q gain or amplification. Table 1.Patient characteristic at KPd initiationCharacteristic at KPd initiationOverall(n=82)2L(n=39)3L(n=26)4L+(n=17) P value Gender, n (%)0.053 male37(45.1)14(35.9)11(42.3)12(70.6)​Age, year0.746 Median (Range)63.0(38,77)62.0(38,77)64.5(45,77)64.0(54,77)​ <6546(56.1)24(61.5)13(50.0)9(52.9)0.628≥6536(43.9)15(38.5)13(50.0)8(47.1)​ISS stage, n (%)0.132 I33 (40.2)19 (48.7)10 (38.5)4 (23.5)​ II20 (24.4)11 (28.2)6 (23.1)3 (17.6)​ III24 (29.3)7 (17.9)8 (30.7)9 (52.9)​ Unknown5 (6.1)2 (5.1)2 (7.7)1 (5.9)​ECOG performance status, n (%)0.927 015(18.3)7(17.9)4(15.4)4(23.5)​ 141(50.0)20(51.3)13(50.0)8(47.1)​ ≥224(29.3)11(28.2)9(34.6)4(23.5)​Unknown2(2.4)1(2.6)01(5.9)​Type of M protein, n (%)0.364 IgA16(19.5)7(18.0)5(19.2)4(23.5)​IgG49(59.8)22(56.4)16(61.5)11(64.7)​ Nonsecretory3(3.7)02(7.7)1(5.9)​ Light-chain14(17.1)10(25.6)3(11.5)1(5.9)​Light-chain subtype, n (%)0.074 K35(42.7)16(41.0)8(30.7)11(64.7)​ λ44(53.6)23(59.0)16(61.5)5(29.4)​ Unknown3(3.7)02(7.7)1(5.8)​Thrombocytopenia a , n (%)0.036 Yes23(28.0)6(15.4)11(42.3)6(35.3)​ No58(70.7)33(84.6)14(53.8)11(64.7)​ Unknown1(1.2)01(3.8)0​Elevated LDH b , n (%)0.709 Yes16(19.5)9(23.1)5(19.2)2(11.8)​ No63(76.8)30(76.9)19(73.1)14(82.4)​ Unknown3(3.7)02(7.7)1(5.9)​Cytogenetic risk at diagnosis c , n (%)0.045 High risk28(34.1)13(33.3)12(46.2)3(17.6)​ Standard risk11(13.4)10(25.6)1(3.8)0​ Unknown43(52.4)16(41.0)13(50.0)14(82.4)​2L, second line; 3L, third line; 4L+, fourth or later lines; ECOG, Eastern Cooperative Oncology Group; ISS, International Staging System; KPd, carfilzomib, pomalidomide, dexamethasone;LDH, lactate dehydrogenase.aThrombocytopenia was defined as a platelet count of <100 × 109/L;bElevated LDH was defined as ≥247 U/L;cHigh-risk cytogenetics is defined as the presence of any of the following: t(4;14), t(14;16), deletion 17p, and 1q gain or amplification. Patient characteristic at KPd initiation 2L, second line; 3L, third line; 4L+, fourth or later lines; ECOG, Eastern Cooperative Oncology Group; ISS, International Staging System; KPd, carfilzomib, pomalidomide, dexamethasone;LDH, lactate dehydrogenase. aThrombocytopenia was defined as a platelet count of <100 × 109/L; bElevated LDH was defined as ≥247 U/L; cHigh-risk cytogenetics is defined as the presence of any of the following: t(4;14), t(14;16), deletion 17p, and 1q gain or amplification. Treatment information before and post initiation of KPd A total of 76 cases (92.6%) had received prior bortezomib-based regimens, and 67 of these 76 cases were refractory. 59 cases (72.0%) had received lenalidomide-based regimens, and 53 of these 59 cases were refractory. 50 cases (60.9%) demonstrated documented dual-refractoriness to both bortezomib and lenalidomide in prior therapies. In addition, 18 cases (21.9%) had received Daratumumab-based regimens, and 15 of these 18 cases were refractory. Only 18 (21.8%) of patients had upfront ASCT prior to receiving KRd therapy (Table 2). Both drug exposure and refractoriness to PIs, IMiDs, and daratumumab progressively increased with advancing LOT in RRMM patients, corresponding to escalating treatment challenges.Table 2.Treatment information before and post initiation of KPd​Overall(n=82)2L(n=39)3L(n=26)4L+(n=17)Prior therapy, n (%) Conventional chemotherapy42(51.2)11(28.2)18(69.2)13(76.5)PIs Bortezomib76(92.6)34(87.2)25(96.2)17(100.0) Ixazomib11(13.4)4(10.2)3(11.5)4(23.5)IMiDs Thalidomide17(20.7)6(15.4)5(19.2)6(35.3) Lenalidomide59(72.0)21(53.8)22(84.6)16(94.1)Monoclonal antibody Daratumumab18(21.9)6(15.8)7(26.9)5(29.4)Upfront ASCT18(21.9)12(30.7)5(19.2)1(5.9)Refractory to previous treatment Thalidomide13(15.8)4(10.3)4(15.4)5(29.4) Lenalidomide53(64.6)18(46.2)19(73.1)16(94.1) Bortezomib67(81.7)28(71.8)22(84.6)17(100) Ixazomib9(10.9)3(7.7)2(7.7)4(23.5) Bortezomib & Lenalidomide50(60.9)17(43.6)17(65.4)16(94.1) Daratumumab15(18.3)5(12.8)6(23.1)4(23.5)KPd treatment cycles Median (Range)3.0(1,14)4.0(1,14)3.0(1,13)2.0(1,8)ASCT, Autologous stem cell transplantation; KPd, carfilzomib, pomalidomide, dexamethasone; PIs, Proteasome inhibitors; IMiDs, Immunomodulatory drugs. Table 2.Treatment information before and post initiation of KPd​Overall(n=82)2L(n=39)3L(n=26)4L+(n=17)Prior therapy, n (%) Conventional chemotherapy42(51.2)11(28.2)18(69.2)13(76.5)PIs Bortezomib76(92.6)34(87.2)25(96.2)17(100.0) Ixazomib11(13.4)4(10.2)3(11.5)4(23.5)IMiDs Thalidomide17(20.7)6(15.4)5(19.2)6(35.3) Lenalidomide59(72.0)21(53.8)22(84.6)16(94.1)Monoclonal antibody Daratumumab18(21.9)6(15.8)7(26.9)5(29.4)Upfront ASCT18(21.9)12(30.7)5(19.2)1(5.9)Refractory to previous treatment Thalidomide13(15.8)4(10.3)4(15.4)5(29.4) Lenalidomide53(64.6)18(46.2)19(73.1)16(94.1) Bortezomib67(81.7)28(71.8)22(84.6)17(100) Ixazomib9(10.9)3(7.7)2(7.7)4(23.5) Bortezomib & Lenalidomide50(60.9)17(43.6)17(65.4)16(94.1) Daratumumab15(18.3)5(12.8)6(23.1)4(23.5)KPd treatment cycles Median (Range)3.0(1,14)4.0(1,14)3.0(1,13)2.0(1,8)ASCT, Autologous stem cell transplantation; KPd, carfilzomib, pomalidomide, dexamethasone; PIs, Proteasome inhibitors; IMiDs, Immunomodulatory drugs. Treatment information before and post initiation of KPd ASCT, Autologous stem cell transplantation; KPd, carfilzomib, pomalidomide, dexamethasone; PIs, Proteasome inhibitors; IMiDs, Immunomodulatory drugs. All enrolled patients received at least one full cycle of the KPd regimen, with a median of 3 treatment cycles (range, 1–14). As shown in Table 2, there was a progressive decline in both the median and maximum number of KPd cycles with increasing lines of treatment. Specifically, the median number of cycles was 4, 3, and 2 for patients receiving KPd in the 2L, 3L, and 4L+ settings, respectively, while the corresponding maximum number of cycles was 14, 13, and 8. Best response in the overall population and across lines of treatment According to the study’s efficacy evaluation design and IMWG criteria, 72 of the 82 patients had evaluable responses after KPd treatment. Among the 72 patients, 50 (69.4%) achieved an overall response. Responses included 2 (2.8%) stringent complete responses (sCR), 20 (27.8%) complete responses (CR), 9 (12.5%) very good partial responses (VGPR), and 19 (26.4%) partial responses (PR). The remaining patients had minimal response (MR, 13.9%), stable disease (SD, 6.9%), or progressive disease (PD, 9.7%). By LOT, the ORR was 79.4% (27/34) in the 2L group, 73.9% (17/23) in the 3L group, and 40% (6/15) in the 4L+ group. These results indicate a decline in clinical benefit with increasing LOT. Further details are provided in Table 3.Table 3.Best response during KPd treatmentEvaluable patientsOverall(n=72)2L(n=34)3L(n=23)4L+(n=15)ORR,n(%)50(69.4)27(79.4)17(73.9)6(40.0)sCR,n(%)2(2.8)2(5.9)00CR,n(%)20(27.8)9(26.5)9(39.1)2(13.3)VGPR,n(%)9(12.5)7(20.6)1(4.3)1(6.7)PR,n(%)19(26.4)9(26.5)7(30.4)3(20.0)MR,n(%)10(13.9)3(8.8)1(4.3)6(40.0)SD,n(%)5(6.9)3(8.8)1(4.3)1(6.7)PD,n(%)7(9.7)1(2.9)4(17.4)2(13.3)CR, complete response; KPd, carfilzomib, pomalidomide, dexamethasone; MR, minimal response; ORR, overall response rate; PD, progressive disease; PR, partial response; sCR, stringent complete response; SD, stable disease; VGPR, very good partial response. Table 3.Best response during KPd treatmentEvaluable patientsOverall(n=72)2L(n=34)3L(n=23)4L+(n=15)ORR,n(%)50(69.4)27(79.4)17(73.9)6(40.0)sCR,n(%)2(2.8)2(5.9)00CR,n(%)20(27.8)9(26.5)9(39.1)2(13.3)VGPR,n(%)9(12.5)7(20.6)1(4.3)1(6.7)PR,n(%)19(26.4)9(26.5)7(30.4)3(20.0)MR,n(%)10(13.9)3(8.8)1(4.3)6(40.0)SD,n(%)5(6.9)3(8.8)1(4.3)1(6.7)PD,n(%)7(9.7)1(2.9)4(17.4)2(13.3)CR, complete response; KPd, carfilzomib, pomalidomide, dexamethasone; MR, minimal response; ORR, overall response rate; PD, progressive disease; PR, partial response; sCR, stringent complete response; SD, stable disease; VGPR, very good partial response. Best response during KPd treatment CR, complete response; KPd, carfilzomib, pomalidomide, dexamethasone; MR, minimal response; ORR, overall response rate; PD, progressive disease; PR, partial response; sCR, stringent complete response; SD, stable disease; VGPR, very good partial response. PFS and OS analysis in the overall population and across lines of treatment Among the 82 patients, the median follow-up was 10.8 months. During this period, 31 patients experienced events of disease progression/death, and the median PFS (mPFS) for the entire cohort was 22.5 months (95% CI: 11.6, NR) (Figure 1(A)). By LOT, PFS events occurred in 10, 12, and 9 patients in the 2L, 3L, and 4L+ groups, with mPFS of NR (95% CI: NR, NR), 22.5 (95% CI: 7.6, NR), and 10.1 (95% CI: 5.6, NR) months, respectively (P = 0.176) (Figure 1(C)). Additional 6- and 12-month PFS rates are detailed in Table S2.Figure 1.The progression-free survival and overall survival. (A) Progression-free survival and (B) Overall survival among the entire cohort. (C) Progression-free survival and (D) Overall survival by patients by lines of treatment. The progression-free survival and overall survival. (A) Progression-free survival and (B) Overall survival among the entire cohort. (C) Progression-free survival and (D) Overall survival by patients by lines of treatment. Among the 20 patients who experienced death events, 12 were due to PD and 8 from other causes. Median OS (mOS) was not reached (NR) in the overall population (Figure 1(B)). In detail, among the patients who died, there were 5 in the 2L group, 7 in the 3L group and 8 in the 4L+ group (Table S3). The mOS was NR (95% CI: NR, NR), NR (95% CI: 11.2, NR) and 11.6(95% CI:9.2, NR) months in patients with 2L,3L and 4L+ groups (Figure 1(D)), respectively (P value =0.019). Prognostic predictors for PFS and OS The univariate analysis of PFS was conducted using COX regression, and the results showed that extramedullary lesions and elevated LDH and LOT ≥ 4 at initiation of KPd regimen, were poor risk factors for PFS (both P < 0.1) (Figure 2). A multivariable analysis of PFS consisting of the established survival risk factors by univariate analysis revealed that an elevated LDH level (HR=3.489, 95% CI, 1.557-7.823) and LOT ≥4 (HR=2.791, 95% CI, 1.086-7.177) represent independent negative prognostic factors (both adjusted P value <0.05). Despite the patients with extramedullary lesions also had a worse PFS than patients without extramedullary lesions (HR=2.161, 95% CI, 0.959-4.872), the differences were not statistically significant (adjusted P value=0.063) (Figure 2).Figure 2.Univariate and multivariate analyses of progression-free survival. Univariate and multivariate analyses of progression-free survival. For OS, thrombocytopenia, and LOT ≥ 4 demonstrated poor prognostic impacts in the univariate analysis (both P < 0.1) (Figure 3). When these established prognostic factors were included in a multivariable analysis, the results revealed that only the LOT ≥ 4 remained a significant independent factor for shorter OS when compared with LOT =2 (HR=3.917,95% CI, 1.258–12.200) with adjusted P =0.019. Patients with third‑line treatment at initiation of KPd regimen had worse but not significantly shorter OS when compared with LOT = 2 (HR=1.740,95% CI, 0.537–5.632) with adjusted P =0.356 (Figure 3).Figure 3.Univariate and multivariate analyses of overall survival. Univariate and multivariate analyses of overall survival. Safety of KPd in the overall population and across lines of treatment A total of 46 patients (56.1%) reported treatment-emergent adverse events (TEAEs) among 82 included patients receiving KPd (Table 4). Overall, the most frequently reported TEAEs were neutropenia (29.3%) and infection (29.3%), followed by anemia (26.8%), thrombocytopenia (25.6%), and heart failure (23.2%) (Table 4). TEAEs of grade ≥3 were observed in 26 patients (31.7%), with infection (14.6%) and neutropenia (12.2%) being the most common severe events. Only 8(9.8%) patients discontinued the KPd treatment due to AEs and no patients died from KPd treatment. In the 2L subgroup (n = 39), 19 (48.7%) experienced TEAEs, including 5 (12.8%) with grade ≥3 toxicity. Among 3L group (n = 26), 17 (65.4%) developed TEAEs, of whom 13 (50.0%) had events of grade ≥3. In the patients with LOT ≥4 (n = 17), 10 patients (58.8%) experienced TEAEs, with 8 (47.1%) classified as grade ≥3.Table 4.Adverse events during KPd treatmentAdverse eventsOverall(n=82)2L(n=39)3L(n=26)4L+(n=17)Any GradesGrades 1-2Grades ≥3Grades 1-2Grades ≥3Grades 1-2Grades ≥3Grades 1-2Grades ≥3Any adverse events46(56.1)20(24.4)26(31.7)14(35.9)5(12.8)4(15.4)13(50.0)2(11.8)8(47.1)Neutropenia24(29.3)14(17.1)10(12.2)4(10.3)1(2.6)6(23.1)5(19.2)4(23.5)4(23.5)Thrombocytopenia21(25.6)13(15.9)8(9.7)7(17.9)1(2.6)3(11.5)5(19.2)3(17.6)2(11.8)Anemia22(26.8)15(18.3)7(8.5)7(17.9)1(2.6)5(19.2)3(11.5)3(17.6)3(17.6)Heart failure19(23.2)15(18.3)4(4.9)4(10.3)1(2.6)7(26.9)2(7.7)4(23.5)1(5.9)Arrhythmia8(9.8)7(8.5)1(1.2)4(10.3)02(7.7)1(3.8)1(5.9)0Infection24(29.3)12(14.6)12(14.6)4(10.3)2(5.1)5(19.2)6(23.1)3(17.6)4(23.5)Thrombosis2(2.4)2(2.4)01(2.6)0001(5.9)02L, second line; 3L, third line; 4L+, fourth or later lines Table 4.Adverse events during KPd treatmentAdverse eventsOverall(n=82)2L(n=39)3L(n=26)4L+(n=17)Any GradesGrades 1-2Grades ≥3Grades 1-2Grades ≥3Grades 1-2Grades ≥3Grades 1-2Grades ≥3Any adverse events46(56.1)20(24.4)26(31.7)14(35.9)5(12.8)4(15.4)13(50.0)2(11.8)8(47.1)Neutropenia24(29.3)14(17.1)10(12.2)4(10.3)1(2.6)6(23.1)5(19.2)4(23.5)4(23.5)Thrombocytopenia21(25.6)13(15.9)8(9.7)7(17.9)1(2.6)3(11.5)5(19.2)3(17.6)2(11.8)Anemia22(26.8)15(18.3)7(8.5)7(17.9)1(2.6)5(19.2)3(11.5)3(17.6)3(17.6)Heart failure19(23.2)15(18.3)4(4.9)4(10.3)1(2.6)7(26.9)2(7.7)4(23.5)1(5.9)Arrhythmia8(9.8)7(8.5)1(1.2)4(10.3)02(7.7)1(3.8)1(5.9)0Infection24(29.3)12(14.6)12(14.6)4(10.3)2(5.1)5(19.2)6(23.1)3(17.6)4(23.5)Thrombosis2(2.4)2(2.4)01(2.6)0001(5.9)02L, second line; 3L, third line; 4L+, fourth or later lines Adverse events during KPd treatment 2L, second line; 3L, third line; 4L+, fourth or later lines Discussion In real-world clinical practice, treatment choices for patients with RRMM vary widely, ranging from first- and next-generation PIs or IMiDs, to anti-CD38 monoclonal antibodies, and combinations of these novel targeted therapies. Treatment selection is typically individualized based on prior LOT, disease characteristics, and economic considerations.7,27 Although carfilzomib and pomalidomide have previously been approved for the treatment of RRMM, identifying patients who can derive optimal clinical benefit from the KPd regimen remains an active area of investigation. In this multicenter study, we aimed to evaluate the efficacy, survival outcomes, prognostic factors, and safety profile of the KPd regimen in RRMM patients stratified by LOT. The study cohort comprised relatively young patients (median age: 63 years) with preserved performance status, yet exhibiting features suggestive of treatment resistance. Notably, 29.3% were classified as ISS stage III, and 29.3% had an ECOG performance status ≥2. Despite the high rates of refractoriness to prior therapies—64.6% to lenalidomide, 81.7% to bortezomib, 18.3% to daratumumab, and 60.9% to both bortezomib and lenalidomide—it is theoretically reasonable to anticipate that the KPd regimen could remain effective in this population, given that carfilzomib and pomalidomide are a second-generation PI and a third‑generation IMiD, respectively. Our study showed an overall objective response rate (ORR) of 69.4% with the KPd regimen. Notably, the ORR was highest in the 2L group (79.4%), compared to 73.9% in the 3L group and 40% in the 4L+ group. Moreover, ≥VGPR was achieved in 52.9% , 43.5% and 20% of patients in the 2L, 3L and 4L+ groups, respectively. These findings suggest that earlier use of KPd is associated with deeper responses and greater clinical benefit in patients with RRMM. This pattern aligns with findings from Terpos et al., 13 who reported a progressive decline in Kd efficacy across later lines of therapy (ORR: 79.2% vs. 71.6% vs. 63.1% in 2L, 3L, and ≥4L; ≥VGPR: 60.4% vs. 47.8% vs. 34.6%). Taken together, these findings underscore the importance of deploying highly effective regimens such as carfilzomib-based therapy at earlier stages of relapse. This is particularly relevant as not only is the VRd regimen now widely used as first-line induction therapy, but the combination of anti-CD38 antibodies with VRd is also becoming increasingly adopted in both transplant-eligible and non-transplant-eligible patients. Differently, when compared with the KPd regimen of the SELECT trial, our study achieved a higher ORR (69.4% vs. 58%), likely reflecting differences in prior treatment exposure. 20 Notably, fewer patients in our study were refractory to lenalidomide (64.6% vs. 100%) and daratumumab (18.3% vs. 75%), whereas substantially more were refractory to bortezomib (81.7% vs. 35%). Moreover, although the EMN011/HOVON114 trial reported an even higher ORR compared with our study with KPd in the second-line setting (92% vs. 79.4%), 21 this difference likely reflects the longer treatment exposure in that trial, where patients received a median of eight cycles versus four in our study. In addition, the enrolled population in the EMN011/HOVON114 trial had more favorable baseline features, including a lower proportion of patients with ISS stage III disease (12% vs 17.9%). Future studies with extended follow-up are needed to comprehensively evaluate the potential benefit of prolonged KPd therapy in patients with RRMM. Survival analysis further demonstrated that both PFS and OS were superior in patients receiving KPd as 2L therapy compared to those treated in the 3L and 4L+ settings, supporting the notion that earlier use of KPd confers greater survival benefit. We also assessed whether AE incidence increased with later lines of treatment. In our cohort, both overall and grade ≥3 AEs were significantly more frequent in the 3L group than in the 2L group. However, contrary to previous reports, 13 AE incidence did not significantly increase in the 4L+ group compared to the 3L group. This discrepancy may be due to early disease progression in heavily pretreated patients, which shortened the duration of KPd exposure and limited the time for AEs to manifest. Overall, KPd remains effective and generally well-tolerated in the 2L–3L setting, with no unexpected or severe toxicities observed. These findings further support the early use of KPd in RRMM to maximize therapeutic benefit. To explore potential prognostic factors associated with KPd efficacy, we incorporated age at enrollment, platelet count, LDH levels, presence of Extramedullary disease (EMD)and LOT, into COX regression analyses. Our results identified elevated LDH and LOT ≥4 as independent adverse prognostic factors for PFS, with LOT ≥4 also predicting inferior OS. These findings highlight that patients with high LDH may experience more aggressive disease progression despite KPd treatment.25,26,28 Moreover, the limited survival benefit in patients with LOT≥ 4 suggests that KPd alone may be suboptimal in this setting. In line with current evidence, alternative strategies such as combining KPd with anti-CD38 monoclonal antibodies, adopting more intensive therapies like CAR-T, or enrolling in clinical trials may be more appropriate for these heavily pretreated patients.14,18,27,29–31 EMD is generally considered an adverse prognostic factor in MM.32,33 However, in our study, the presence of EMD in RRMM patients did not significantly affect survival following KPd treatment. This may be partly due to our relatively small sample size, and it is also possible that the KPd regimen attenuates the negative prognostic impact of EMD. Future studies with larger cohorts are warranted to clarify whether EMD influences outcomes in RRMM patients receiving the KPd regimen. Our study has several limitations:(1) As this was a retrospective study, no formal sample size calculation was performed, and the relatively small sample size may limit the statistical power of the analyses. Expanding the cohort in future studies would allow more robust analysis of KPd efficacy and safety across the different LOT. (2) Due to the retrospective nature, a small proportion of RRMM patients included in our study had prior exposure to daratumumab before receiving KPd, which may limit the generalizability of our findings to the current clinical scenario of RRMM. (3) Due to the lack of comprehensive genetic data in many patients, genetic background was not included in our prognostic model to avoid excessive patient exclusion from the Cox regression analysis. We plan to emphasize genetic testing at relapse in future studies to build a more comprehensive prognostic model. (4) Finally, as a retrospective multi-center study limited to Shandong province, China, further prospective studies incorporating data from other regions are needed to enhance the generalizability and representativeness of our findings. With clinical advancements and the ongoing development of personalized treatments, there has been increasing attention among physicians on studies exploring different administration regimens, timing, and doses of the same therapeutic approach. For example, adopting a weekly dosing schedule for carfilzomib and making adjustments to the dexamethasone dosing may further deepen the therapeutic effect and reduce the treatment-related side effects.34–36 In this context, future studies investigating KPd-based combination therapies and modifications in dosing regimens remain important areas for further exploration. Such research will help optimize the identification of patient populations likely to benefit from the KPd regimen, enable early identification of adverse prognostic factors, and ultimately improve clinical precision and outcomes for the broader patients with RRMM. Conclusion In conclusion, our study demonstrated that the KPd regimen had favorable clinical efficacy and manageable toxicity in RRMM, particularly in those with early-line relapse. Moreover, we also identified several independent prognostic factors, with LDH levels and LOT ≥4 significantly associated with inferior PFS, and the latter also linked to worse OS. These findings support the use of KPd as an effective therapeutic option in RRMM and underscore the importance of early intervention with potent regimens before the development of advanced refractoriness. Supplemental material Supplemental material - Real-world efficacy and safety of the carfilzomib–pomalidomide–dexamethasone regimen in patients with relapsed/refractory multiple myeloma: A multicenter retrospective study Supplemental material for Real-world efficacy and safety of the carfilzomib–pomalidomide–dexamethasone regimen in patients with relapsed/refractory multiple myeloma: A multicenter retrospective study by Huishou Fan, XiangYan Feng, Rongxia Wei, Xin Zhang, Tianlan Li, Lu Liu, Hui Wang, Lingjie Sun, Xue Shi, Hongguo Zhao, Zhao Li, Jingying Cui, Junqing Xu, Junjie Ma, Xianqi Feng, Wei Wang in Therapeutic Advances in Hematology.