Response kinetics as a dynamic prognostic marker in patients with multiple myeloma

To the Editor

Multiple myeloma (MM) exhibits heterogeneous clinical outcomes despite advances in therapeutic agents. Prognostic assessment has traditionally relied on static baseline biomarkers at serologic, biologic, genomic, and imaging levels, which incompletely capture the dynamic nature of high-risk disease during treatment [1–4]. Dynamic indicators assessed during therapy, including depth of response, refractoriness to therapy, and early relapse, have provided additional prognostic value [5–7]. Response kinetics, defined as the timing of achieving therapeutic response, has been studied as a meaningful dynamic prognostic marker in various hematologic malignancies. In this study, we investigated the prognostic significance of response kinetics and its associated clinical and disease-related factors in patients with MM treated with carfilzomib- or ixazomib-based regimens.

This retrospective study reviewed the medical records of patients with relapsed or refractory multiple myeloma (RRMM) who received carfilzomib- or ixazomib-based regimens as second-line therapy between January 2015 and June 2021 across 17 medical centers in South Korea. Carfilzomib-based regimens included dexamethasone alone (Kd) or lenalidomide plus dexamethasone (KRd), whereas ixazomib-based therapy was administered as IRd. Patients achieving partial response (PR) or better were considered responders. Among responders, response kinetics were analyzed using two measures: time to first response (TTFR) and time to best response (TTBR). TTFR was defined as the interval from the initiation of second-line therapy to the first documentation of PR or better, and TTBR as the time to the deepest response achieved. Median values were used as cut-offs for TTFR and TTBR.

A total of 384 patients with RRMM received carfilzomib- or ixazomib-based therapy at first relapse. Of these, 80.7% received carfilzomib-based therapy (KRd, n = 268; Kd, n = 42), while 19.3% received ixazomib-based therapy (IRd, n = 74). Seventy-two patients (18.8%) had light chain MM, and high-risk cytogenetic abnormalities were present in 24.5%. Most patients (85.9%) had prior exposure to bortezomib, and 69 patients (18.0%) received maintenance therapy; 58.3% experienced early relapse within 2 years of frontline therapy. Thirty-five patients (9.1%) received lenalidomide-containing regimens as frontline or maintenance therapy, among whom 26 patients (6.8%) were lenalidomide-refractory. The overall response rate was 84.1% (n = 323); only responders were included in the analysis of response kinetics. Baseline characteristics of all patients are summarized in Supplementary Table S1. Among 323 responders, the median time to the first documented response was 1.9 months (range, 0.2–36.7), and the median time to best response was 3.0 months (range, 0.5–44.7). For TTFR, the median value of 1.9 months was rounded to 2 months as cut-off, and the TTBR cut-off was defined as 3 months. The median follow-up duration for responders was 29.4 months (range, 1.6–86.5). A landmark of 3 months, corresponding to the median TTBR, was applied. After excluding patients who died (n = 3) or had insufficient follow-up (n = 4) before the landmark, 316 patients were included in the final survival analysis (Figure S1). Among these patients, 168 (53.2%) achieved TTFR < 2 months and demonstrated significantly inferior outcomes compared to those with TTFR ≥ 2 months (2-year PFS: 54.0% vs. 74.5%, p < 0.001; 2-year OS: 69.4% vs. 80.0%, p = 0.023) (Figure S2A). Similarly, patients achieving TTBR < 3 months (n = 153 (48.4%)) exhibited inferior survival outcomes compared to those who achieved it after 3 months (2-year PFS: 51.8% vs. 73.9%, p < 0.001; 2-year OS: 65.8% vs. 82.1%, p < 0.001) (Figure S2B).

Among 168 patients achieving TTFR < 2 months, 67.9% (n = 114) also achieved TTBR < 3 months, whereas among 148 patients with TTFR ≥ 2 months, 73.6% (n = 109) achieved TTBR ≥ 3 months. Based on these patterns, patients were categorized as rapid responders (both TTFR and TTBR below median, n = 114) or slow responders (both above median, n = 109) (Figure S3). Slow responders had significantly better PFS and OS than rapid responders (2-year PFS: 76.9% vs. 45.7%, p < 0.001; 2-year OS: 81.7% vs. 62.6%, p < 0.001) (Fig. 1A, B). This trend persisted across subgroups stratified by response depth—CR (PFS, p = 0.001; OS, p = 0.103), VGPR (PFS, p = 0.113; OS, p = 0.099), and PR (PFS, p = 0.016; OS, p = 0.047) (Figure S4A). Comparable survival benefits were observed across cytogenetic abnormalities (Figure S4B) and in both carfilzomib- and ixazomib-based second-line therapy subgroups (Figure S4C). In a multivariable Cox proportional hazards model for PFS, slow responder status was significantly associated with superior outcomes at diagnosis (hazard ratio [HR] 0.646, p < 0.001). At second-line therapy initiation, plasmacytoma (HR 2.393, p = 0.036) and a best response of PR (HR 1.595, p = 0.027) were associated with inferior PFS, while slow responder status (HR 0.454, p < 0.001) remained favorable. For OS, early relapse (HR 1.883, p = 0.025) was an adverse prognostic factor, while slow responder status (HR 0.740, p = 0.031) was favorable at diagnosis. At second-line therapy initiation, a best response of PR (HR 1.816, p = 0.009) was an adverse factor, whereas slow responder status was consistently favorable (HR 0.628, p = 0.024) (Table 1 and Table S2).

Fig. 1. Response kinetics patterns and survival outcomes.

A Progression-free survival (PFS) according to response kinetics patterns. B Overall survival (OS) according to response kinetics patterns.

Table 1.

Multivariable Cox regression analysis for PFS and OS in rapid and slow responders (n = 223).

(A) At diagnosis
	PFS		OS
	HR (95% CI)	p value	HR (95% CI)	p value
ISS stage (III vs. I)	–	–	1.236 (0.871–1.753)	0.236
Cytogenetics abnormalities (High vs. Standard)	1.466 (0.884–2.433)	0.139	1.164 (0.656-2.065)	0.603
Maintenance therapy (Yes vs. No)	0.719 (0.378–1.367)	0.314	–	–
Lenalidomide-refractory (Yes vs. No)	–	–	1.699 (0.708–4.076)	0.235
Early relapse (<2 years vs. ≥2 years)	–	–	1.883 (1.084–3.271)	0.025
Response kinetics (Slow responders vs. Rapid responders)	0.646 (0.507–0.824)	<0.001	0.740 (0.563–0.972)	0.031

(B) At second-line therapy
	PFS		OS
	HR (95% CI)	p value	HR (95% CI)	p value
Second-line treatment (Ixazomib vs. Cafilzomib)	0.713 (0.292–1.736)	0.456	–	–
Albumin (≥3.5 mg/dL vs. <3.5 mg/dL)	0.472 (0.220–1.013)	0.054	0.480 (0.204–1.130)	0.093
LDH (High vs. Low)	–	–	1.116 (0.494–2.524)	0.791
Beta-2 microglobulin (≥3.5 mg/dL vs. <3.5 mg/dL)	1.798 (0.805–4.020)	0.153	1.261 (0.551–2.885)	0.583
Plasmacytoma (Yes vs. No)	2.393 (1.058–5.414)	0.036	1.846 (0.771–4.419)	0.169
Hb. (≥10.0 g/dL vs. <10.0 g/dL)	0.463 (0.202–1.063)	0.069	0.828 (0.356 –1.928)	0.662
Platelets (≥ 100,000/μL vs. < 100,000/μL)	1.235 (0.512–2.981)	0.638	0.728 (0.309–1.717)	0.469
ALC/WBC ratio (≥20% vs <20%)	–	–	0.403 (0.138–1.178)	0.097
Best response to second-line therapy (PR vs. CR)	1.595 (1.055–2.412)	0.027	1.816 (1.159–2.843)	0.009
Response kinetics (Slow responders vs. Rapid responders)	0.454 (0.304–0.678)	<0.001	0.628 (0.420–0.941)	0.024

PFS progression-free survival, OS overall survival, HR hazard ratio, CI confidence interval, LDH lactate dehydrogenase, Hb hemoglobin, ALC/WBC absolute lymphocyte count/white blood cell count, PR partial response, CR complete response, ISS International Staging System.

Bold values indicate statistical significance (p < 0.05).

Factors associated with response kinetics were also explored (Table S3). Among baseline clinical factors, light chain MM and high-risk cytogenetics were more frequently observed among rapid responders (p = 0.001 and p = 0.027). Among variables at second-line therapy initiation, high light chain differences (≥300) (p = 0.002), anemia (<10 g/dL) (p = 0.064), and thrombocytopenia (<100,000/μL) (p = 0.004) were more common in rapid responders, with a trend toward higher prevalence of plasmacytoma (p = 0.054). Although slow responders tended to achieve deeper response than rapid responders (CR: 41.3% vs. 29.8%), this difference was not statistically significant (p = 0.078). In multivariable logistic regression analysis, light chain MM (odds ratio [OR] 3.720, p = 0.020), high-risk cytogenetic abnormalities (OR 2.950, p = 0.011), thrombocytopenia (OR 3.500, p = 0.032), and a best response of CR (OR 0.580, p = 0.022) were independently associated with rapid responder status (Table S4).

In this multicenter retrospective study, patients who achieved responses rapidly experienced significantly inferior PFS and OS compared with those who achieved responses more slowly. These findings are consistent with previous studies. Rapid early reduction in serum free light chain was associated with inferior survival outcomes in the setting of intensive cytotoxic chemotherapy [2]. Similar findings have also been observed in studies based on novel agent regimens. In a large cohort of 1,099 patients treated with novel agents as initial therapy, gradual responders to best response demonstrated longer survival than those achieving a rapid response [8]. Another study of 626 patients with newly diagnosed MM found that a TTBR of ≤3 months was significantly associated with inferior PFS and OS regardless of response depth [9]. Likewise, the TOURMALINE-MM1 and -MM2 trials demonstrated inferior PFS among early responders (defined as achieving best response) in both IRd versus placebo-Rd arms [10, 11]. Not all studies, however, have reported similar findings. Achievement of ≥VGPR after 2 cycles was not associated with improved survival outcomes in newly diagnosed MM [12], and a pooled analysis of daratumumab-based trials showed no significant difference in PFS or OS between early and late responders achieving ≥VGPR [13]. Notably, these analyses generally defined early versus late response based on the time to achieving a specific depth of response, such as ≥VGPR, rather than incorporating the time to best response, which was considered in our study and other studies demonstrating prognostic significance of response kinetics. This methodological difference may explain the discrepancy.

A distinct aspect of our study was the joint assessment of both TTFR and TTBR to characterize overall response pattern – an approach that, to our knowledge, has not been previously employed. Using this approach, approximately 70% of patients exhibited consistent response kinetics patterns: patients with a rapid first response tended to also achieve best response quickly (rapid responders), whereas those with a slow response showed delayed achievement of best response (slow responders). Slow responders demonstrated improved survival outcomes regardless of response depth, high-risk cytogenetics, and treatment regimen, and response kinetics remained independently prognostic for both PFS and OS after adjustment for baseline variables and factors at second-line therapy initiation. Additionally, our analysis focused on patients with RRMM rather than newly diagnosed MM, suggesting that response kinetics may retain prognostic relevance across different lines of therapy, even in the context of prior treatment exposure and refractory disease biology.

Although the underlying biological mechanisms were beyond the scope of this study, high-risk MM is characterized by stepwise accumulation of genetic alterations in subclones through branching evolutionary processes that drive treatment resistance and disease progression [14]. In this context, rapid responders may initially present with more proliferative or treatment-sensitive disease while harboring aggressive and latent subclones that expand during therapy, leading to early progression. In contrast, slow responders may reflect less proliferative disease with more stable clonal dynamics, resulting in more durable disease control. This interpretation is supported by the observation that rapid responders more frequently exhibited light chain MM, adverse cytogenetic abnormalities, and features associated with higher tumor burden, including higher light chain differences, anemia, thrombocytopenia, and a trend toward higher prevalence of plasmacytoma.

This study has several limitations, including its retrospective design and restriction to proteasome inhibitor-based therapy, which may limit generalizability to immune-based approaches such as monoclonal antibodies, bispecific antibodies, or CAR-T cell therapy. In addition, minimal residual disease was not evaluated and warrants investigation in future studies.

In conclusion, response kinetics—particularly the pattern of response from first to best response—represents a dynamic prognostic indicator in MM and may enhance risk stratification beyond conventional prognostic markers.

Author contributions

HJC collected and interpreted the data and drafted the manuscript. JHM and KK contributed to the study design and data interpretation and revised the manuscript critically for important intellectual content. MWL, JHL, HSE, JJ, HYY, DHY, HWC, SSY, JMB, YRD, HJK, JHL, SNL, HJK, HSL, YK, HJL, and SML contributed to patient enrollment, data collection, and revision of the manuscript and provided final approval of the version to be submitted.

Data availability

The dataset is available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The study was approved by the Institutional Review Board of Kyungpook National University Hospital (IRB 2023-04-033) and by the ethics committees of all participating centers, and was conducted in accordance with the Declaration of Helsinki. Informed consent was waived by the Institutional Review Board given the retrospective study design and the utilization of de-identified patient data.

Supplementary information

The online version contains supplementary material available at 10.1038/s41375-026-03007-z.