Original Study Prognostic Value of Serum Paraprotein Response Kinetics in Patients With Newly Diagnosed Multiple Myeloma Luis-Esteban Tamariz-Amador,1 Paula Rodríguez-Otero,1 Ana Jiménez-Ubieto,2 Laura Rosiñol,3 Albert Oriol,4 Rafael Ríos,5 Anna Sureda,6 Maria Jesus Blanchard,7 Miguel Teodoro Hernández,8 Valentin Cabañas Perianes,9 Isidro Jarque,10 Juan Bargay,11 Mercedes Gironella,12 Felipe De Arriba,13 Luis Palomera,14 Yolanda Gonzalez-Montes,15 Josep M. Martí,16 Isabel Krsnik,17 José María Arguiñano,18 María Esther González,19 Luis Felipe Casado,20 Ana Pilar González-Rodriguez,21 Lucía López-Anglada,3 Noemi Puig,22 Maria Teresa Cedena,2 Bruno Paiva,1 Maria-Victoria Mateos,22 Jesús San-Miguel,1 Juan-José Lahuerta,2 Joan Bladé,3 Iñaki F. Trocóniz23 Abstract Response kinetics is not well-established as a prognostic marker in multiple myeloma (MM). We developed a mathematical model to assess the prognostic value of serum monoclonal component (MC) response kinetics during 6 induction cycles in 373 newly diagnosed MM patients. The model calculated a “resistance” parame- ter that reflects the stagnation in the response after an initial descent, dividing the patients into two kinetics categories with significantly different progression-free survival (PFS). Introduction: Response kinetics is a well-established prognostic marker in acute lymphoblastic leukemia. The situation is not clear in multiple myeloma (MM) despite having a biomarker for response monitoring (monoclonal component [MC]). Materials and Methods: We developed a mathematical model to assess the prognostic value of serum MC response kinetics during 6 induction cycles, in 373 NDMM transplanted patients treated in the GEM2012Menos65 clinical trial. The model calculated a “resistance” parameter that reflects the stagnation in the response after an initial descent. Results: Two patient subgroups were defined based on low and high resistance, that respectively captured sensitive and refractory kinetics, with progression-free survival (PFS) at 5 years of 72% and 59% (HR 0.64, 95% CI 0.44- 0.93; P = .02). Resistance significantly correlated with depth of response measured after consolidation (80.9% CR and 68.4% minimal residual disease negativity in patients with sensitive vs. 31% and 20% in those with refractory kinetics). Furthermore, it modulated the impact of reaching CR after consolidation; thus, within CR patients those with refractory kinetics had significantly shorter PFS than those with sensitive kinetics (median 54 months vs. NR; P = .02). Minimal residual disease negativity abrogated this effect. Our study also questions the benefit of rapid responders compared to late responders (5-year PFS 59.7% vs. 76.5%, respectively [P < .002]). Of note, 85% of patients considered as late responders were classified as having sensitive kinetics. Conclusion: This semi-mechanistic modeling of M-component kinetics could be of great value to identify patients at risk of early treatment failure, who may benefit from early rescue intervention strategies. Clinical Lymphoma, Myeloma and Leukemia, Vol. 22, No. 9, e844–e852 © 2022 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Keywords: Early treatment failure, Newly diagnosed myeloma, Time to best response, Prognostic marker, Depht of response L.E.T.A. and P.R.O. contributed equally to this work. 4 Institut Català d’Oncologia i Institut Josep Carreras, Badalona, Spain 5 1 Clínica Universidad de Navarra, CCUN, Centro de Investigación Médica Aplicada Hospital Universitario Puerta de Hierro, Majadahonda, Madrid, Spain 6 (CIMA), IDISNA, CIBERONC, Pamplona, Spain Institut Català d’Oncologia – Hospital Duran i Reynals, IDIBELL, Universitat de 2 Hospital 12 de Octubre, Madrid, Spain Barcelona, Barcelona, Spain 7 3 Hospital Clínic de Barcelona, IDIBAPS, Barcelona, Spain Hospital Ramón y Cajal, Madrid, Spain 2152-2650/$ - see front matter © 2022 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) e844 Clinical Lymphoma, Myeloma and Leukemia September 2022 https://doi.org/10.1016/j.clml.2022.04.024 Luis-Esteban Tamariz-Amador et al Introduction Mathematical approaches, such as semi-mechanistic models The clinical landscape in multiple myeloma (MM) has signif- integrating relevant treatment and tumor related properties (ie, icantly improved over the last 20 years1 because of novel drugs pharmacodynamics, proliferation, and resistance development) and better tools for diagnosis and disease monitoring.2 Response have been used to predict clinical outcomes based on the time to therapy is a key element to evaluate treatment efficacy, which is course of circulating biomarkers. This has been reported in the critically associated with patients’ survival. Standard response crite- setting of small cell lung cancer, nonfunctioning gastro-entero- ria rely on the evaluation of plasma cells in bone marrow3-8 and pancreatic neuroendocrine tumors and breast cancer, among many monoclonal component (MC) in serum and urine.9 other types of tumors.17-20 However, to our knowledge, these Response kinetics is a well-established prognostic marker in acute semi-mechanistic models have not been explored in MM using lymphoblastic leukemia,10 , 11 while information about its value in serial assessment of the MC. Thus, we aimed to assess the MM is rather limited.12-16 prognostic value of serum MC response kinetics using popula- Two studies have analyzed response kinetics in newly diagnosed tion semi-mechanistic pharmacodynamic models, evaluated in MM12 , 13 and a third one, at relapse.14 Yan et al.12 demonstrated NDMM patients treated in the GEM2012Menos65 clinical that early responders (patients that reach their best response within trial. the first 3 months) had significantly worse survival compared with late responders. Additionally, they defined 4 distinct response kinet- Materials and methods ics patterns with different outcomes. Patients with gradual and Study design sustained remission ("U-valley" pattern) showed prolonged survival, Four hundred and fifty-eight newly diagnosed MM patients were whereas poor outcomes were observed in patients with rapid and included in the GEM2012Menos65 clinical trial conducted by the transient responses ("roller coaster" pattern). Spanish Myeloma Group.21 Patients with Bence Jones (n = 69) MM The Mayo Clinic group analyzed 840 patients treated between and non-secretory disease (n = 6) and incomplete data regarding 2004 and 2015.13 They found that patients harboring high-risk serum MC during the 6 induction cycles (n = 10) were excluded features, such as high tumor burden, high-risk cytogenetic abnor- from the analysis. Accordingly, 373 patients were evaluated in this malities or ISS 3 were more likely to achieve an early response (in post-hoc study. the first 2 cycles). However, the achievement of an early response Trial design and primary analysis were recently published.21 In (in the first 2 months) was not associated with a prolonged survival, brief, patients were uniformly treated with 6 induction cycles with suggesting that response kinetics has limited impact on long-term bortezomib + lenalidomide + dexamethasone (VRD), followed by outcomes in the era of novel therapies. autologous stem cell transplant (ASCT) conditioned with either In the context of RRMM patients, Garderet et al.14 showed Melphalan 200mg/m2 or Busulfan + melphalan and posttransplant that patients achieving an early (0-4 months) VGPR or better consolidation with 2 additional cycles of VRD followed by mainte- response had significantly shorter duration of response and PFS nance with lenalidomide +/- ixazomib. when compared to those achieving a late response (> VGPR after The following characteristics, documented at diagnosis, were 4 months). They suggested that patients with indolent disease and analyzed: MC concentration, hemoglobin, creatinine, calcium, lower tumor proliferation would slowly respond to therapy, with albumin, beta-2-microglobulin, lactate dehydrogenase categorized favorable long-term outcomes. as high and normal, the International Staging System (ISS), and cytogenetics. High-risk cytogenetics was defined by the presence of at least one of the following abnormalities detected by iFISH: 8 Hospital Universitario de Canarias, Santa Cruz de Tenerife, Spain t(4;14), t(14;16) and/or del(17p). 9 Hospital Virgen de la Arrixaca, Murcial, Spain For the purpose of this study, we analyzed the sequential measure- 10 Hospital La Fe, Valencia, Spain 11 Hospital Son Llatzer, Palma de Mallorca, Spain. ments of MC each cycle, during the 6 induction cycles and 12 Hospital Vall d’Hebron, Barcelona, Spain minimal residual disease status after consolidation. We also deter- 13 Hospital Universitario Morales Meseguer, IMIB-Arrixaca, Universidad de Murcia, Murcia, Spain mined in which cycle each patient achieved their best response 14 Hospital Clínico Lozano Blesa, Zaragoza, Spain and further classified them as early responders if they reached 15 Hospital Josep Trueta, Girona, Spain 16 Hospital Mutua de Terrassa, Terrassa, Spain their best response within the first 4 cycles or late responders 17 18 Hospital Puerta del Hierro, Madrid, Spain if it was after 4 cycles. Finally, we determined which patients Hospital Universitario de Navarra, Pamplona, Spain 19 Hospital de Cabueñes, Gijón, Spain displayed rises in the MC that did not qualify as progressive 20 Hospital Virgen de la Salud, Toledo, Spain disease (PD) at any time during the induction. Kinetics of response 21 Hospital Universitario Central de Asturias, Oviedo, Spain 22 Hospital Universitario de Salamanca, Instituto de Investigación Biomédica de were correlated with both the response and the minimal resid- Salamanca, Salamanca, Spain. ual disease (MRD) obtained after consolidation, defined accord- 23 Facultad de Farmacia y Nutrición, Universidad de Navarra, Pamplona, Spain ing to the International Myeloma Working Group (IMWG) 2016 Submitted: Jan 14, 2022; Revised: Apr 26, 2022; Accepted: Apr 29, 2022; Epub: 5 May 2022 criteria.9 Address for correspondence: Luis-Esteban Tamariz-Amador, MD, Clínica Universidad de Navarra, Av. Pio XII 36, 31008 Pamplona, Spain. Population semi-mechanistic pharmacodynamic E-mail contact: ltamariz@unav.es, paurodriguez@unav.es modelling of the M-component Address for correspondence: Paula Rodríguez-Otero, MD, PhD, Clínica Universidad de In the clinical setting, we expect an MC decline after each treat- Navarra, Av. Pio XII 36, 31008 Pamplona, Spain. E-mail contact: ltamariz@unav.es, paurodriguez@unav.es ment cycle, more quickly in some patients than others. Nonethe- Clinical Lymphoma, Myeloma and Leukemia September 2022 e845 Prognostic value of serum paraprotein response kinetics less, occasionally there is a stagnation in this decrease (stabilization) Model evaluation. The selected model was evaluated using or even, the MC slightly increases without qualifying as PD, or simulation-based model diagnostics as visual predictive checks. One shows overt progression in the following cycles. We estimated that thousand simulations, with the same design characteristics as the these different dynamics of MC reduction are the result of several original one, were done. For each simulated study and time of factors. First, the expected positive treatment’s effect on the tumor measurement, the 5th, 50th and 95th percentiles of the simulated leading to MC reduction. Second, factors that generate resistance to values were computed. Then the 90% prediction intervals of the therapy that may account for progressively diminished efficacy of the aforementioned percentiles were calculated and plotted together treatment, which translates into slower MC reduction and stagna- with the corresponding percentiles obtained from raw data. tion in the response. This resistance may be due both to intrinsic tumor’s characteristics and to external factors. Finally, the tumor’s Statistical analysis proliferative capacity, which makes the MC steadily increase over Figure 1 shows some features of the semi-mechanistic model time. Dynamic interaction between all these factors could explain including its schematic representation, model performance at the the different MC kinetics. Thus, we designed a mathematical model individual level, and the impact on the M-component kinetics of to calculate these parameters for each patient. Serum concentrations different degrees of proliferation and resistance. of the MC that were measured at the time of diagnosis and after We defined two groups for the resistance and two for the prolif- each induction cycle were plotted individually for each patient and eration parameters, depending on their magnitude through ROC analyzed through the nonlinear mixed effects modeling approach analysis in relation to PFS: low (< 0.001 for resistance and < implemented in the software NONMEM 7.4 using the first order 0.0004 for proliferation) and high (≥ 0.01 for resistance and ≥ conditional estimation with the laplacian option. 0.0004 for proliferation). The Deffect parameter was not considered Model building: in order to characterize the MC reduction curve for analysis since its values were assumed as a constant variable in and calculate the proposed parameters for each patient we developed the model due to lack of values of drug concentration. the following ordinary differential equation: We performed Kaplan Meier survival analysis for PFS using the log-rank test to determine significant differences. Later, patients that progressed during the induction cycles were excluded for the analy- dM = λ − DE f f ect × e −γ ×t × M sis. Univariate analysis was performed using independent-samples dt Kruskall-Wallis test, chi-squared or Fisher’s exact test, as appropri- ate, for the independent variables previously mentioned, only with Where dM/dt stands for the rate of change of the MC in serum. the categorized resistance variable. All the statistical analysis was Treatment’s effect is represented by the DEffect parameter. Since done with the SPSS software (IBM Corp. Released 2019. IBM SPSS values of drug concentration were not available for correlation with Statistics, Version 26.0. Armonk, NY: IBM Corp). the MC it was assumed as a constant variable in the model. The tumor’s resistance is represented in the formula by the gamma (γ ) Results parameter, which can predict the moment when the slope of the M- We established two categories for both the resistance and prolif- component reduction decreases to a half. Such a moment accounts eration parameters using the cut-off described in Materials and for an eventual stagnation of patients’ response measured by the Methods: high (≥ 0.01 for resistance and ≥ 0.0004 for prolifera- MC. Finally, the tumor’s proliferative capacity is represented by the tion) and low (< 0.001 for resistance and < 0.0004 for prolifera- lambda (λ) parameter, a zero-order rate constant, that estimates the tion). Initially, the combination of these parameters, which included MC daily increase if the treatment variable (DEffect ) had no effect. all patients necessary to develop the model, allowed to establish The initial condition of the system was set up at the value of the 4 kinetics categories. Nonetheless, after excluding the standard MC-component at diagnosis. progressions (according to IMWG criteria) detected during the 6 Data were logarithmically transformed for the analysis. Levels of induction cycles (38/373; 10.2%) only the resistance parameter MC below the quantification limit (0.009 g/dL) were treated as retained significant association with PFS, so we proceeded using censored information. Inter-individual variability (IIV) associated only this parameter, that categorized patients into two categories: with the parameter of the model (see below) was described with low resistance that reflects “sensitive kinetics” and high resistance an exponential model preventing negative values for the individ- that reflects “refractory kinetics” and we will use these two terms ual parameters. An additive error model in the logarithmic scale along the study. accounted for the residual variability. Accordingly, the sensitive kinetics group included 215/335 Model selection. Selection between competing models was based (64.2%) patients while within the refractory kinetics category there on the (i) visual inspection of the goodness of fit plot, (ii) were 120/335 (35.8%) patients. Although neither group reached the minimum value of the objective function approximately equal to PFS median, patients with sensitive kinetics had significantly longer -2 × Log(likelihood) (-2LL), where a drop of 3.84 points in -2LL PFS than those with refractory kinetics (72% and 59% at 5-years), between 2 nested models differing in one parameter is considered which translated into a 36% reduction in risk of progression for significant at the 0.05 level, and (iii) precision of model parameters the sensitive kinetics group (HR 0.64, 95% CI 0.44-0.93; P = .02) computed as the ratio between the standard error and the estimate of (Figure 2). the parameter (RSE). Model candidates with parameters presenting Univariate analysis (Table 1) did not show an association between RSEs greater than 50% were rejected. the kinetics patterns and baseline disease characteristics, except for e846 Clinical Lymphoma, Myeloma and Leukemia September 2022 Luis-Esteban Tamariz-Amador et al Figure 1 A, Schematic representation of the semi-mechanistic model used to describe the time profiles of the M-component. B, Measured values versus predictions. Solid line is the line of unity. C, M-component profiles of 4 patients chosen at random (measured, points; predicted, lines). D, Simulated profiles of M-component representing patients with same value at diagnosis, same sensitivity to treatment effects, and different magnitude of proliferation and resistance. Low, medium, and high, refer to the 2.5th, 50th, and 97.5th percentiles of the respective parameter distributions. an unexpected higher proportion of high-risk cytogenetics cases Next, we sought to determine whether the kinetic profiles had any within the sensitive kinetics group (22.8% vs. 13.3%). Interest- subsequent impact on PFS for patients that reached CR or negative ingly, patients with high-risk cytogenetics and sensitive kinetics MRD status after consolidation. Within patients who reached CR (n = 49) did not have a significantly different PFS from patients those who displayed sensitive kinetics (174/210, 82.8%) had signif- with standard cytogenetics and refractory kinetics (n = 62) (median icantly better PFS than those with a refractory kinetics (36/210, PFS 58 months [95% CI 47-69] vs. NR; P = .53). 17.2%) (median NR vs. 54 months [95% CI 45-62]; P = .02) We found a significant correlation between kinetic profiles and (Figure 4), with a 45% reduction in risk of progression or death depth of response measured after consolidation (Figure 3). Among (HR 0.55; 95% CI 0.32-0.95; P = .03). By contrast the kinetics patients with sensitive kinetics (n = 215), 174 (80.9%) reached CR, profile did not significantly modify the outcome for patients that 35 (16.3%) VGPR, 1 (0.5%) PR, 1 (0.5%) SD, and 4 (1.9%) PD; reached negative MRD status after consolidation (P = .46). and 147 patients (68.4%) reached negative MRD status. Within the We then analyzed the correlation of the kinetic profiles with the refractory kinetics group (n = 120), only 36 patients reached CR time to best response. We defined early responders as those who (31.6%), 60 VGPR (50.9%), 19 patients PR (14.9%), 1 (0.8%) SD achieved their best response within the first 4 cycles and late respon- and 4 (2.6%) PD. Within this cohort, only 25 patients (20.8%) ders as those who achieved it during the last two. Accordingly, reached negative MRD status. 185/335 (55.4%) patients were classified as early responders and 150/335 (44.6%) as late responders. There was a higher propor- Clinical Lymphoma, Myeloma and Leukemia September 2022 e847 Prognostic value of serum paraprotein response kinetics Figure 2 Survival plot for PFS comparing the two response kinetics profiles. PFS = progression-free survival. Table 1 Baseline Characteristics for the 2 Response Kinetics Patterns Sensitive Kinetics (n = 215) Refratory Kinetics (n = 120) Valor P Hemoglobin 10.9 (11.9) 10.9 (8.8) .82 Creatinine 0.85 (1.58) 0.89 (1.46) .38 Albumin 3.71 (3.6) 3.7 (3.15) .94 B2M 3.4 (14.5) 3.4 (17.4) .86 LDH Normal 180 (83.7%) 101 (84.2%) .85 High 25 (11.6%) 15 (12.5%) ISS I 87 (40.5%) 52 (43.3%) .36 II 86 (40%) 38 (31.7%) III 39 (18.1%) 29 (24.2%) Cytogenetics Standard 84 (39.1%) 62 (51.7%) .02 High-risk 49 (22.8%) 16 (13.3%) Missing data 81 (37.7%) 43 (35.8%) Abbreviations: ISS = International Staging System; LDH = lactate dehydrogenase. tion of late responders within the sensitive kinetics group (127/215 ing for disease progression as per IMWG at any time during the [59.1%] vs. 22/120 [18.3%], P < .001) as compared to the refrac- induction. Accordingly, 100/335 (29.9%) presented a rebound in tory kinetics group, whereas there was a higher proportion of early the MC, while 235/335 (70.1%) did not. Patients that presented a responders (98/120, 81.7%) among patients with refractory kinet- rebound at any time during the induction had a significantly shorter ics. Of interest, patients with sensitive kinetics who were late respon- PFS as compared to patients without MC rebound, with a median ders had significantly longer PFS than those with sensitive kinetics PFS of 61 months versus NR, respectively (P = .002) (Figure 4). who were early responders (P = .008), with PFS rates at 5 years of Interestingly patients with sensitive kinetics included a much lower 81.1% versus 64.8%, respectively. No significant differences were proportion of rebounds (19/215, 8.8%) as compared to the refrac- observed between early and late responders among patients with tory kinetics category (81/120, 67.5%) (P < .001). refractory kinetics (data not shown). Finally, we analyzed the correlation between the kinetic profiles and patients that displayed an increase in the MC not qualify- e848 Clinical Lymphoma, Myeloma and Leukemia September 2022 Luis-Esteban Tamariz-Amador et al Figure 3 A) Serum MC component response after consolidation among the response kinetics profiles. B) MRD status after consolidation among the response kinetics profiles C) Survival plot for PFS with the response kinetics profiles among patients that reached CR after consolidation. D) Survival plot for PFS with the integration of the response kinetics profiles and MRD status after consolidation. MC = monoclonal component; MRD = minimal residual disease; PFS = progression-free survival; CR = complete response. Discussion in patients with MM who received isatuximab as a single agent Depth of response to induction therapy is one of the most or in combination.25 Yet there are some differences. For example, relevant prognostic factors in MM.7 , 9 However, the relevance of we described disease proliferation with a linear model rather than response kinetics has not been carefully evaluated, despite the exponentially as Koiwai et al.,25 since according to the data shown fact that MM has a unique biomarker (the MC) for follow-up. by these authors, individual profiles presented an initial increase in Moreover, intriguingly some data suggests that early responders may M-protein values, a phenomenon that did not occur in our original have a worse outcome as compared to slow responders,13 and that data. early response with early relapse may be associated with an adverse Using this resistance parameter, we have defined two different outcome.12 patterns of response kinetics associated with different outcomes in Clinicians usually recognize the stagnation in the decrease of terms of PFS. Thus, patients corresponding to the sensitive kinetics the MC, after a rapid reduction during initial cycles, as an group, showed a significantly longer PFS as compared to the refrac- adverse kinetic pattern, since this may reflect treatment resistance.22 tory kinetics group (72% and 59% at 5-years) with a 36% reduction Moreover, minimal rebound in the M-component (below the 25% in risk of progression for the sensitive kinetics group. cut-off of increase defined as disease progression by IMWG crite- There are frequent debates about the relevance of an early ria) may also alert of an impending disease progression.23 Using a response in MM.12 Our study based on a prolonged induction, semi-mechanist model, we have analyzed the kinetics of response with 6 cycles before transplant, represents a unique opportunity to and defined an independent parameter, “resistance,” for each patient. evaluate this question. Interestingly, overall, 128 (85%) of patients The structure of the model is similar to that previously devel- considered as late responders were classified as having sensitive kinet- oped by Claret et al. in 200924 during the longitudinal analysis of ics in our model and the median PFS was not reached in this cohort, solid tumors, and recently applied to the time course of M-protein being clearly superior to that of early responders with sensitive kinet- ics (P = .008). These findings reflects that the stagnation in the Clinical Lymphoma, Myeloma and Leukemia September 2022 e849 Prognostic value of serum paraprotein response kinetics Figure 4 Survival plot for PFS comparing patients that presented an increment in the MC not qualifying as PD at any point during the induction with those who did not present it. MC = monoclonal component; PD = progressive disease; PFS = progression-free survival. response, identified by our resistance parameter, should probably be cohorts, especially in transplant-ineligible patients and in patients considered as a negative prognostic factor and may favor the design that have received more novel therapies such as monoclonal antibod- of early rescue interventions in curative oriented clinical trials. ies (antiCD38 or bispecific antibodies) or CAR-T cell therapies. The proportion of patients that reached either CR or negative In conclusion, this semi-mechanistic model based on the M- MRD status after consolidation was significantly higher among component kinetics has the capacity to identify two different groups those with sensitive kinetics (80.9% vs. 30% and 68.4% vs. 20.8%, of patients with different response kinetics and different survival. respectively). Furthermore, the resistance parameter identifies that Our goal would be to implement the model in the clinic with patients who reach a CR and have sensitive kinetics have signif- the help of a simple calculator. It might be particularly attrac- icantly better PFS than those that reach CR but have refractory tive to determine the utility of the model in non-transplant treat- kinetics. However, achievement of MRD negative status is able to ment approaches that are usually based on a prolonged/continuous abrogate the negative prognostic impact of refractory kinetics. This number of cycles. This tool, based on the response kinetics during would further support the value of MRD negativity as a highly the first initial 6 cycle, could be of great value to identify patients relevant clinical endpoint.26 at a risk of early treatment failure, before fulfilling IMWG criteria Cytogenetics is one of the most relevant prognostic parameters of disease progression. In this setting, an early detection/prediction at baseline.27 Intriguingly in our series, the frequency of adverse of treatment failure would avoid unnecessary physical and finan- cytogenetics was not higher among the high resistant kinetics cial toxicity. Moreover, this may offer a window of opportunity to category, albeit there was a high proportion of missing data. Never- overcome the dismal prognosis of early progressions, through early theless, of note, the resistance parameter modulates the prognostic rescue intervention strategies incorporating new immunotherapeu- impact of cytogenetics, turning patients who have standard cytoge- tic approaches, in a setting of low tumor burden and potentially less netics but refractory kinetics into a subgroup with adverse prognosis. compromised patient condition. Even though this study was based on data from a large random- ized trial (GEM2012menos65) of patients uniformly treated, with Clinical practice points external audit, and long-term follow-up, a potential limitation is What is already know about this subject? that we only used this one series of transplant-eligible patients, with only measurable serum MC that received 6 cycles of induc- • Although information about the relevance of response kinetics as tion treatment for the development and evaluation of the model. a prognostic marker in MM is limited, some studies have analyzed Additionally, MC measurements were not made in a central labora- its use and found that early responders (best response within the tory. Further studies are warranted to validate this model in other first 3 months) had significantly worse survival compared with e850 Clinical Lymphoma, Myeloma and Leukemia September 2022 Luis-Esteban Tamariz-Amador et al late responders, and defined several response kinetics patterns, Squibb (Inst), MSD (Inst), Novartis (Inst), Sanofi (Inst), Janssen such as patients with gradual and sustained remission ("U-valley" (Inst), Roche (Inst), AbbVie (Inst). pattern) that showed prolonged survival, whereas poor outcomes were observed in patients with rapid and transient responders References ("roller coaster" pattern). 1. Fonseca R, Abouzaid S, Bonafede M, et al. Trends in overall survival and costs of multiple myeloma, 2000-2014. Leukemia. 2017;31:1915–1921. doi:10.1038/leu. • Mathematical approaches have been used to predict clinical 2016.380. outcomes based on the time course of circulating biomarkers for 2. Pinto V, Bergantim R, Caires HR, Seca H, Guimarães JE, Vasconcelos MH. Multi- ple myeloma: available therapies and causes of drug resistance. Cancers. 2020;12:1– some solid tumors with success. 32. doi:10.3390/cancers12020407. 3. Paiva B, Vidriales MB, Mateo G, et al. The persistence of immunophenotypically What are the new findings? normal residual bone marrow plasma cells at diagnosis identifies a good prognos- tic subgroup of symptomatic multiple myeloma patients. Blood. 2009;114:4369– • Our mathematical model calculated a “resistance” parameter that 4372. doi:10.1182/blood- 2009- 05- 221689. 4. Paiva B, Vidriales MB, Pérez JJ, et al. Multiparameter flow cytometry quantifi- reflects the stagnation in the response after an initial descent and, cation of bone marrow plasma cells at diagnosis provides more prognostic accordingly, defined 2 patient subgroups based on low and high information than morphological assessment in myeloma patients. Haematologica. 2009;94:1599–1602. doi:10.3324/haematol.2009.009100. resistance that respectively captured sensitive and refractory kinet- 5. Paiva B, Cedena MT, Puig N, et al. Minimal residual disease monitoring and ics, with significantly different progression-free survival (PFS) at immune profiling in multiple myeloma in elderly patients. Blood. 2016;127:3165– 3174. doi:10.1182/blood- 2016- 03- 705319. 5 years. 6. Paiva B, Merino J, Miguel JFS. Utility of flow cytometry studies in the management • There was significant correlation between resistance and depth of of patients with multiple myeloma. Curr Opin Oncol. 2016;28:511–517. doi:10. 1097/CCO.0000000000000331. response measured after consolidation, with a greater proportion 7. Paiva B, van Dongen JJM, Orfao A. New criteria for response assessment: role of CR and negative MRD among patients with sensitive kinetics. of minimal residual disease in bmultiple myeloma. Blood. 2015;125:3059–3068. doi:10.1182/blood- 2014- 11- 568907. • The resistance parameter modulated the impact of reaching CR 8. Medina A, Puig N, Flores-Montero J, et al. Comparison of next-generation after consolidation. Refractory kinetics patients with CR had sequencing (NGS) and next-generation flow (NGF) for minimal residual disease (MRD) assessment in multiple myeloma. Blood Cancer J. 2020;10. doi:10.1038/ significantly shorter PFS than those with sensitive kinetics and s41408- 020- 00377- 0. CR. 9. Kumar S, Paiva B, Anderson KC, et al. International Myeloma Working Group consensus criteria for response and minimal residual disease assessment in multi- ple myeloma. Lancet Oncol. 2016;17:e328–e346. doi:10.1016/S1470-2045(16) How might it impact on clinical practice in the foreseeable future? 30206-6. 10. Wei AH, Ribera JM, Larson RA, et al. Biomarkers associated with blinatu- • These results might allow to implement the model in the clinic momab outcomes in acute lymphoblastic leukemia. Leukemia. 2021. doi:10.1038/ with the help of a simple calculator. s41375- 020- 01089- x. 11. Bassan R, Hoelzer D. Modern therapy of acute lymphoblastic leukemia. J Clin • This tool could be of great value to identify patients at a risk of Oncol. 2011;29:532–543. doi:10.1200/JCO.2010.30.1382. early treatment failure, before fulfilling IMWG criteria of disease 12. Yan Y, Mao X, Liu J, et al. The impact of response kinetics for multiple myeloma in the era of novel agents. Blood Advances. 2019;3:2895–2904. doi:10.1182/ progression. An early detection/prediction of treatment failure bloodadvances.2019000432. would avoid unnecessary physical and financial toxicity. 13. Tandon N, Sidana S, Vincent Rajkumar S, et al. Outcomes with early response to first-line treatment in patients with newly diagnosed multiple myeloma. Blood Advances. 2019;3:744–750. doi:10.1182/bloodadvances.2018022806. 14. Garderet L, Laubach JP, Stoppa AM, et al. Association between response Authorship kinetics and outcomes in relapsed/refractory multiple myeloma: analysis Contribution: P.R-O, L-E.T-A, B.P and J.S-M conceived the from TOURMALINE-MM1. Leukemia. 2018;32:2032–2036. doi:10.1038/ s41375- 018- 0091- 3. manuscript. J-I.T. developed the semi-mechanistic model. L-E.T- 15. Xu L, Liu J, Chen M, et al. Early monoclonal protein decline pattern is an A performed statistical analysis. P.R-O, L-E.T-A, J-I.T and J.S-M independent prognostic factor in patients with multiple myeloma. Ann Hematol. 2020;99:581–589. doi:10.1007/s00277- 020- 03915- 9. wrote the manuscript. L.R and J.B designed the clinical trial. All 16. Gassiot S, González Y, Morgades M, et al. Response to first cycle is the major authors contributed in the enrollment and treatment of the patients predictor of long-term response to lenalidomide and dexamethasone therapy in relapsed and refractory multiple myeloma: can we spare patients the toxicity and included in the clinical trial, reviewed the manuscript and provided costs of additional agents? Clin Lymphoma Myeloma Leuk. 2019;19:585–592 e1. final approval. doi:10.1016/j.clml.2019.05.020. 17. Buil-Bruna N, Sahota T, López-Picazo JM, et al. Early prediction of disease progression in small cell lung cancer: Toward model-based personalized Acknowledgment medicine in oncology. Cancer Res. 2015;75:2416–2425. doi:10.1158/0008-5472. CAN- 14- 2584. To all the patients, families, and healthcare professionals that 18. Buil-Bruna N, Dehez M, Manon A, Nguyen TXQ, Trocóniz IF. Establish- have contributed to the development and execution of the ing the quantitative relationship between lanreotide autogel®, chromogranin A, and progression-free survival in patients with nonfunctioning gastroenteropan- GEM2012menos65 clinical trial. creatic neuroendocrine tumors. AAPS Journal. 2016;18:703–712. doi:10.1208/ s12248- 016- 9884- 3. 19. Solans BP, López-Díaz de Cerio A, Elizalde A, et al. Assessing the impact of Disclosure the addition of dendritic cell vaccination to neoadjuvant chemotherapy in breast L-E.T-A has nothing to declare. P.R-O declares honoraria for cancer patients: a model-based characterization approach. Br J Clin Pharmacol. 2019;85:1670–1683. doi:10.1111/bcp.13947. lectures from and membership on advisory boards with Celgene- 20. Almufti R, Wilbaux M, Oza A, et al. A critical review of the analytical BMS, Janssen, Amgen, GSK, Kite Pharma, Oncopeptides, Sanofi, approaches for circulating tumor biomarker kinetics during treatment. Ann Oncol. 2014;25:41–56. doi:10.1093/annonc/mdt382. Abbvie and Takeda. Consultancy for Celgene, Janssen and GSK. 21. Rosiñol L, Oriol A, Rios R, et al. Bortezomib, lenalidomide, and dexamethasone A.J-U. declares no conflict of interest. R.R. declares honoraria as induction therapy prior to autologous transplant in multiple myeloma. Blood. 2019;134:1337–1345. doi:10.1182/blood.2019000241. from Becton-Dickinson, Sanofi and The Binding Site. FdA declares 22. Rodriguez-Otero P, Paiva B, San-Miguel JF. Roadmap to cure multiple myeloma. honoraria from Janssen, Celgene, Amgen, GlaxoSmithKline and Cancer Treat Rev. 2021;100. doi:10.1016/j.ctrv.2021.102284. 23. Lahuerta JJ, Paiva B, de Ubieto AJ, et al. Early detection of treatment failure and Takeda J.S-M has received honoraria from consulting or Advisory early rescue intervention in multiple myeloma: time for new approaches. Blood Role: Amgen (Inst), Celgene (Inst), Takeda (Inst), Bristol-Myers Advances. 2021;5:1340–1343. doi:10.1182/bloodadvances.2020003996. Clinical Lymphoma, Myeloma and Leukemia September 2022 e851 Prognostic value of serum paraprotein response kinetics 24. Claret L, Girard P, Hoff PM, et al. Model-based prediction of phase III overall in patients with multiple myeloma. CPT Pharmacometrics Syst Pharmacol. survival in colorectal cancer on the basis of phase II tumor dynamics. J Clin Oncol. 2021;10:928–940. doi:10.1002/psp4.12666. 2009;27. doi:10.1200/JCO.2008.21.0807. 26. Paiva B, Puig N, Cedena MT, et al. Measurable residual disease by next-generation 25. Koiwai K, El-Cheikh R, Thai HT, et al. PK/PD modeling analysis for dosing flow cytometry in multiple myeloma. J Clin Oncol. 2019;38:784–792. doi:10. regimen selection of isatuximab as single agent and in combination therapy 1200/JCO.19. 27. Caro J, al Hadidi S, Usmani S, Yee AJ, Raje N, Davies FE. How to treat high-risk myeloma at diagnosis and relapse. American Society of Clinical Oncology Educational Book. 2021:291–309. doi:10.1200/edbk_320105. e852 Clinical Lymphoma, Myeloma and Leukemia September 2022