Infections are a major cause of death in patients with multiple myeloma. A post hoc analysis of the phase 3 FIRST trial was conducted to characterize treatment-emergent (TE) infections and study risk factors for TE grade ≥ 3 infection. The number of TE infections/month was highest during the ﬁrst 4 months of treatment (deﬁned as early infection). Of 1613 treated patients, 340 (21.1%) experienced TE grade ≥ 3 infections in the ﬁrst 18 months and 56.2% of these patients experienced their ﬁrst grade ≥ 3 infection in the ﬁrst 4 months. Risk of early infection was similar regardless of treatment. Based on the analyses of data in 1378 patients through multivariate logistic regression, a predictive model of ﬁrst TE grade ≥3infection inthe ﬁrst 4 months retained Eastern Cooperative Oncology Group performance status and serum β -microglobulin, lactate dehydrogenase, and hemoglobin levels to deﬁne high- and low-risk groups showing signiﬁcantly different rates of infection (24.0% vs. 7.0%, respectively; P < 0.0001). The predictive model was validated with data from three clinical trials. This predictive model of early TE grade ≥ 3 infection may be applied in the clinical setting to guide infection monitoring and strategies for infection prevention. Presented at the 21st Congress of the European Hematology Introduction Association; June 9–12, 2016; Copenhagen, Denmark Patients with multiple myeloma (MM) are more susceptible Electronic supplementary material The online version of this article (https://doi.org/10.1038/s41375-018-0133-x) contains supplementary to infections due to advanced age, immunodeﬁciency material, which is available to authorized users. * Thierry Facon Hospital 12 de Octubre, Madrid, Spain email@example.com Centre Hospitalier Yves Le Foll, Saint-Brieuc, France Hospices Civils de Lyon, Lyon, France Gachon University Gil Hospital, Incheon, Korea CHU Bordeaux, Bordeaux, France Service des Maladies du Sang, Hôpital Claude Huriez, National and Kapodistrian University of Athens, Athens, Greece Lille, France 4 16 CHU de Nancy, Université de Lorraine, Nancy, France Cross Cancer Institute, Edmonton, AB, Canada 5 17 University of Nantes, Nantes, France Mayo Clinic Cancer Center, Rochester, MN, USA 6 18 Hopitaux de Toulouse, Toulouse, France Wilhelminen Hospital, Wilhelminen Cancer Research Institute, Vienna, Austria CHU Purpan/IUCT Oncopole, Toulouse, France Centre Hospitalier, Périgueux, France Hôpital Saint-Antoine, Paris, France AZ Sint-Jan AV Brugge, Brugge, Belgium INSERM U1153, University Hospital Saint-Louis, Paris, France Blood Disease Hospital, Chinese Academy of Medical Science Quinten, Paris, France and Peking Union Medical College, Tianjin, China Celgene International Sàrl, Boudry, Switzerland Seràgnoli Institute of Hematology, Bologna University School of Medicine, Bologna, Italy YGM Consult, Paris, France Universitair Ziekenhuis Antwerpen, Edegem, Belgium 1234567890();,: 1234567890();,: A predictive model for risk of early grade ≥ 3 infection in patients with multiple myeloma not eligible. . . 1405 caused by the underlying disease, comorbidities, and treat- supporting the pooling of data from these two arms for the ment toxicities . Infections are a major cause of death, investigation of infections in the ﬁrst 4 or 18 months. particularly early death, in patients with MM, highlighting Demographics, medical history, and baseline character- the need for preventive or early treatment measures [2–6]. istics were analyzed to identify risk factors of early TE A scoring system can help identify patients at risk for grade ≥ 3 infection. Of 1613 treated patients, this analysis infections during MM treatment, enabling implementation was conducted on 1378 patients (prognostic analysis of risk-adapted strategies to prevent early infections. To population), which excluded patients who progressed, died, identify infection risk factors, we used data from the pivotal, or discontinued treatment and had no TE grade ≥ 3 infec- phase 3 FIRST trial, which compared the efﬁcacy and safety tions in the ﬁrst 4 months. of lenalidomide plus low-dose dexamethasone (Rd) until External validation of the results was conducted in three disease progression (Rd continuous) vs. Rd for 18 cycles independent data sets: MM-003 (NCT01311687) , MM- (Rd18) or melphalan, prednisone, and thalidomide (MPT) 009 (NCT00056160)/MM-010 (NCT00424047) [10–12], in transplant-ineligible patients with newly diagnosed MM and MM-015 (NCT00405756) , with 237, 444, and 391 (NDMM) [7, 8]. treated patients, respectively. These trials are described in In this post hoc analysis, a detailed characterization of the Supplement (External Validation Trials). infections in the FIRST trial was conducted and prognostic The numbers of patients in the various study populations factors of early treatment-emergent (TE) grade ≥ 3 infec- in MM-020 and the validation sets are described in Sup- tions were identiﬁed. The results were used to develop a plemental Table 1. predictive model to assess the risk of this event in patients receiving standard nonintensive treatment. Analysis of the impact of ﬁrst TE grade ≥ 3 infection in the ﬁrst 4 months on overall survival Methods A time-dependent Cox model analysis was performed to assess the impact of ﬁrst TE grade ≥ 3 infection in the ﬁrst Study design 4 months on patient overall survival (OS) . A multi- variate analysis was conducted with all baseline prognostic The FIRST study (MM-020/IFM07-01; NCT00689936) has factors identiﬁed in the study with the Q-Finder algorithm been previously reported . The protocol was approved by as described in the Supplement to assess the signiﬁcance of the appropriate institutional review board or independent the occurrence of ﬁrst TE grade ≥ 3 infection in the ﬁrst ethics committee before study initiation. Brieﬂy, the mul- 4 months on OS, independent of the role of potential con- tinational, open-label, randomized, phase 3 trial compared founding factors. Results were expressed using the hazard the efﬁcacy and safety of Rd continuous vs. MPT or Rd18 ratio (HR) of death and its 95% CI. in transplant-ineligible patients with NDMM. Infection prophylaxis was not mandatory in the protocol. Development and validation of ﬁrst TE grade ≥ 3 infection in the ﬁrst 4 months risk model Patients and assessments Overall, 853 variables were included in an analysis to Of the 1623 patients in the intent-to-treat population, TE identify rules that can predict the occurrence of the ﬁrst TE infections were investigated in 1613 patients who grade ≥ 3 infection in the ﬁrst 4 months, using the Q-Finder received ≥ 1 treatment dose (safety population), including subgroup discovery algorithm. A rule is 1 or a combination 532, 540, and 541 in the Rd continuous, Rd18, and MPT of a few variable modalities deﬁning a group with a high or arms, respectively. TE infections were deﬁned as infections low proportion of early TE grade ≥ 3 infection. Rules were occurring or worsening on or after the ﬁrst dose of any selected based on their P-value computed with the study drug and up to 28 days after treatment discontinua- hypergeometric law. The statistical signiﬁcance cutoff for −5 tion. Infections were identiﬁed by the investigator, classiﬁed retaining rules was determined at P < 5.10 × 10 to adjust per Medical Dictionary for Regulatory Activities and gra- for multiple testing. Twenty-ﬁve rules meeting the ded per Common Terminology Criteria for Adverse Events statistical signiﬁcance threshold were retained for expert v3.0. Early infection was deﬁned as occurring during the review. Additional details regarding this algorithm are ﬁrst 4 months of treatment. For comparison of the risk of provided in the Supplement (Q-Finder). Upon clinical infections between treatment arms, data from the Rd con- experts’ request, the cutoff value from statistically tinuous and Rd18 arms were pooled (Rd pooled) and a χ signiﬁcant rules was rounded to make it easier to use, and test was used. Patients in the Rd18 and Rd continuous arms additional tests were performed on variables with clinical received the same treatment in the ﬁrst 18 months, thereby signiﬁcance. 1406 C. Dumontet et al. Statistically signiﬁcant rules were selected by expert assessment based on their clinical and/or biological rele- vance to be included in a stepwise Akaike information criterion multivariate logistic regression model followed by an iterative variable selection process to remove variables with P ≥ 0.1 . Patients with missing data on ≥ 1 input variable were excluded from the model (n = 9). The ﬁnal model included six variables. A scoring system was developed by allocating points to factors of low (−1or −2 points) or high risk (1 or 2 points) based on their coefﬁcient in the multivariate logistic model. The cumulative score classiﬁed patients into high (2 to 5 points) or low (−3to1 points) infection risk groups. The concordance index (C-index), relative risk (RR) and its 95% CI, and number needed to treat (NNT) were determined. Assuming that a prevention treatment can reduce the risk of early TE grade ≥ 3 infection in 50% of the patients of the high-risk group, NNT is the number of patients in the high-risk group who had to receive the prevention treatment to avoid the occurrence of 1 early TE grade ≥ 3 infection. Thus, a higher NNT denotes a smaller beneﬁt of the treatment. A χ test was used to compare the proportions of patients with ≥ 1 early TE grade ≥ 3 infection in the high- vs. low-risk groups. The model was tested on three independent validation data sets, and all metrics (C-index, RR, and NNT) were computed to evaluate the model. As a conﬁrmatory analysis (in the MM-020 and valida- tion sets), time to ﬁrst infection was estimated in the safety population using the Kaplan–Meier method in the high- and low-risk groups and the log-rank test to assess statistical signiﬁcance of the difference. In addition, a competing risk analysis with progression or death without infection and infection as competing events was performed to conﬁrm the difference in risk of ﬁrst TE grade ≥ 3 infection in the ﬁrst 4 months between high- and low-risk groups in the prognostic analysis population (Supplement: Competing Risk Model) . Results Characterization of infections Demographic and baseline characteristics of the safety population in MM-020 are presented globally and per treatment group in Supplemental Table 2. History of infections before enrollment was similar across treatments (Rd pooled: 27.2%; MPT: 28.5%). During the study, anti-infective drugs were prescribed to 78.5% and 67.1% of patients in the Rd pooled and MPT groups, respectively. Among the three treatment arms, 3125 infections of any grade occurred during the study; 3031 infections were TE (1.9 TE infection events per patient). Of Table 1 TE infection events by grade and treatment arm in the safety population of the FIRST trial (1613 patients, including 532, 540, and 541 in the Rd continuous, Rd18, and MPT arms, respectively) TE Grade 1 (mild) infections Grade 2 (moderate) Grade 3 (severe) infections Grade 4 (life-threatening) Grade 5 (death) infections Unknown grade infections infection infections infections events, n Rd cont Rd18 MPT Total Rd cont Rd18 MPT Total Rd cont Rd18 MPT Total Rd cont Rd18 MPT Total Rd cont Rd18 MPT Total Rd cont Rd18 MPT Total Events in 134 136 85 355 157 170 114 441 57 68 62 187 17 16 15 48 11 10 9 30 0 3 0 3 the ﬁrst 4 months Events in 339 356 190 885 440 422 307 1169 148 145 105 398 35 27 24 86 20 20 15 55 0 3 2 5 the ﬁrst 18 months Events 175 4 4 183 174 3 1 178 62 1 0 63 6 0 0 6 2 0 0 2 1 0 0 1 beyond 18 months Total 514 360 194 1068 614 425 308 1347 210 146 105 461 41 27 24 92 22 20 15 57 1 3 2 6 MPT melphalan, prednisone, and thalidomide, Rd cont lenalidomide plus low-dose dexamethasone until disease progression, Rd18 lenalidomide plus low-dose dexamethasone for 18 cycles, TE treatment emergent A total of 79 infections occurred before the ﬁrst treatment administration, and 15 infections occurred > 28 days after treatment discontinuation A predictive model for risk of early grade ≥ 3 infection in patients with multiple myeloma not. . . 1407 3031 TE infection events of any grade that occurred during the study in 1104 patients, 610 in 321 patients were grade ≥ 3 (representing 20.2% of 3025 TE infection events of known grade) (Table 1). During the ﬁrst 18 months of treatment, 1055 patients (65.4%) and 340 patients (21.1%) experienced TE infections of any grade and TE grade ≥ 3 infections, respectively. The risk of TE infection of any grade in the ﬁrst 18 months was 69.4% with Rd pooled and 57.5% with MPT (P < 0.0001). The risk of having ≥ 1 TE grade ≥ 3 infection during the ﬁrst 18 months was 22.6% (120 patients) with Rd continuous, 22.6% (122 patients) with Rd18, and 18.1% (98 patients) with MPT (Rd pooled vs. MPT, P = 0.04). The risk of having a TE infection of any grade and a TE grade ≥ 3 infection beyond 18 months of treatment was 31.8% (169 patients) and 9.2% (49 patients), respectively, with Rd continuous. The risk of a TE grade 5 infection during the ﬁrst 18 months was 3.6% (19 patients) with Rd continuous, 3.3% (18 patients) with Rd18, and 2.6% (14 patients) with MPT (Rd pooled vs. MPT, P = 0.35). After 18 months of treatment, the risk of a TE grade 5 infection was 0.4% (two patients) with Rd continuous. TE infections occurring during the ﬁrst 4 months of treatment Fig. 1 Treatment-emergent (TE) infections in the FIRST trial. a Number of TE infections by month in the ﬁrst 18 months of the FIRST The number of TE infections per month was highest during trial (1613 treated patients). The numbers above the bars indicate the total number of TE infections of all grades during the treatment month. the ﬁrst 4 months of treatment (Fig. 1a). A total of 1064 TE b Number of new patients with TE grade ≥ 3 infections by month in infections of any grade occurred during the ﬁrst 4 months, the ﬁrst 18 months of the FIRST trial (1613 treated patients) including 265 TE grade ≥ 3 infections (representing 25.0% of 1061 TE infections of known grade) (Table 1). The lungs and respiratory tract were involved in 48.7% of early TE occurred during the ﬁrst 4 months (28 patients grade ≥ 3 infections, whereas 22.6% of these infections were [1.7%]). localized to the blood, with patients exhibiting sepsis, bacteremia, and viremia (Supplemental Table 3). The Impact of ﬁrst TE grade ≥ 3 infection in the ﬁrst pathogen was identiﬁed in 25.3% of early TE grade ≥ 3 4 months on OS infections; bacterial infections were implicated in 79.1% of cases in which a pathogen was identiﬁed (Supplemental The risk of death associated with a ﬁrst TE grade ≥3infec- Table 4). Streptococcal, staphylococcal, and clostridia tion in the ﬁrst 4 months, as assessed in a time-dependent infections were the most commonly speciﬁed bacterial Cox regression analysis, was signiﬁcant (HR, 2.9 [95% CI, infections. No statistical differences were seen between Rd 2.4–3.6]; P < 0.0001). A stepwise multivariate time- pooled and MPT in the rates of staphylococcal and strep- dependent analysis for baseline risk factors was then per- tococcal infections (P = 0.25 and P = 0.15, respectively). formed to adjust for potential confounding factors. The Overall, 56.2% of patients with a TE grade ≥ 3 infection occurrence of a ﬁrst TE grade ≥ 3 infection in the ﬁrst in the ﬁrst 18 months experienced their ﬁrst infection in the 4 months remained signiﬁcant in the ﬁnal OS predictive ﬁrst 4 months, and there were < 20 new patients with TE model (HR, 9.1 [95% CI, 5.6-14.6]; P < 0.0001) (Supple- grade ≥ 3 infections per month after 4 months of treatment mental Table 5). (Fig. 1b). A total of 191 patients (11.8%) experienced ≥ 1 TE grade ≥ 3 infection during the ﬁrst 4 months of treatment Baseline factors associated with risk of ≥ 1 early TE (12.2% Rd pooled and 11.1% MPT, P = 0.51); 54 patients grade ≥ 3 infection (3.3%) experienced > 1 TE grade ≥ 3 infection (Table 2). Of the 57 TE grade ﬁve infections that occurred Demographic and baseline characteristics of the intent-to- during the study (53 patients [3.3%]), 30 (52.6%) treat and prognostic analysis populations in MM-020 and 1408 C. Dumontet et al. the validation sets are presented in Supplemental Table 6. A comprehensive analysis was performed on the prognostic analysis population in MM-020 to identify risk factors associated with high or low risk of ﬁrst TE grade ≥ 3 infection in the ﬁrst 4 months using the Q-Finder algorithm (Supplemental Table 7). The most signiﬁcant variables associated with a high or low risk of infection included Sβ2M levels or International Staging System stage, number of CRAB (hypercalcemia, renal failure, anemia, and bone lesions) diagnostic criteria , M-protein urine levels, creatinine or urea levels, red blood cell counts, hematocrit or hemoglobin levels, LDH levels, triiodothyronine (thyroid hormone; T3) levels, α-1 globulin levels, and eosinophil counts. Patients with low quality-of-life score at baseline also had a signiﬁcantly increased risk of early grade ≥3TE infection. An exploratory analysis of baseline immunopar- esis on the risk of early grade ≥ 3 TE infection is presented in the Supplement (Immunoparesis and the Risk of Infec- tion at 4 Months). First TE grade ≥ 3 infection in the ﬁrst 4 months scoring system Of the statistically signiﬁcant variables identiﬁed by the Q-Finder algorithm, clinical experts in MM selected variables with high clinical relevance to be proposed to the multivariate logistic regression model (Supplemental Table 8). The multivariate analysis, which included eight rules identiﬁed by the univariate analysis to be associated with high or low risk of early TE grade ≥ 3 infection (ECOG PS < 1, ECOG PS ≥ 2, Sβ2M ≥ 6 mg/L, Sβ2M ≤ 3 mg/L, LDH ≥ 200 U/L, hemoglobin ≤ 9 g/dL, hemoglobin ≥ 11 g/dL, and creatinine ≥ 1.2 mg/dL), showed that six rules based on ECOG PS and Sβ2M, LDH, and hemoglobin levels were independently associated with ﬁrst TE grade ≥ 3 infection in the ﬁrst 4 months (Table 3). From the resulting predictive model, a scoring system (Table 3) was used to create high (2 to 5 points) and low (−3 to 1 points) infection risk groups. The cutoff between these groups was selected based on the best sensitivity/ speciﬁcity ratio. These high- and low-risk groups were associated with signiﬁcantly different rates of early TE grade ≥ 3 infections (24.0% vs. 7.0%, respectively; P < 0.0001; C-index, 0.66; RR, 3.43 [95% CI, 2.57–4.59]; NNT, 8.3). Validation of the predictive model for risk of ﬁrst TE grade ≥ 3 infection in the ﬁrst 4 months When tested on three independent cohorts (MM-015, MM- 009/010, and MM-003), [9, 11–13] the model discriminated between high- and low-risk patients regarding the risk of Table 2 Rate of TE grade ≥ 3 infections by treatment arm in the FIRST trial (safety population) Patients with indicated number 0–4 months 0–18 months Beyond 18 months of TE grade ≥ 3 infections, n (%) Rd cont Rd18 MPT Total Rd cont Rd18 MPT Total Rd cont Rd18 MPT Total (n = 532) (n = 540) (n = 541) (N = 1 613) (n = 532) (n = 540) (n = 541) (N = 1 613) (n = 532) (n = 540) (n = 541) (N = 1 613) 0 469 (88.2) 472 (87.4) 481 (88.9) 1 422 412 (77.4) 418 (77.4) 443 (81.9) 1 273 483 (90.8) 539 (99.8) 541 (100) 1 563 (88.2) (78.9) (96.9) 1 46 (8.6) 48 (8.9) 43 (7.9) 137 (8.5) 72 (13.5) 72 (13.3) 64 (11.8) 208 (12.9) 35 (6.6) 1 (0.2) 0 36 (2.2) 2 13 (2.4) 16 (3.0) 8 (1.5) 37 (2.3) 28 (5.3) 35 (6.5) 22 (4.1) 85 (5.3) 10 (1.9) 0 0 10 (0.6) ≥ 3 4 (0.8) 4 (0.7) 9 (1.7) 17 (1.1) 20 (3.8) 15 (2.8) 12 (2.2) 47 (2.9) 4 (0.8) 0 0 4 (0.2) MPT melphalan, prednisone, and thalidomide, Rd cont lenalidomide plus low-dose dexamethasone until disease progression, Rd18 lenalidomide plus low-dose dexamethasone for 18 cycles, TE treatment emergent A predictive model for risk of early grade ≥ 3 infection in patients with multiple myeloma not. . . 1409 Table 3 Multivariate logistic regression model for ﬁrst TE grade ≥ 3 infection during the ﬁrst 4 months of treatment (1369 patients included) Variable Coefﬁcient Odds ratio P-value Points Infection risk Estimate SE Sβ2M ≤ 3 mg/L −0.812 0.353 0.44 0.021 −2 Low ECOG PS of 0 −0.403 0.216 0.67 0.062 −1 Low Hemoglobin ≤ 11 g/dL 0.366 0.207 1.44 0.077 1 High ECOG PS of ≥ 2 0.457 0.189 1.58 0.016 1 High LDH ≥ 200 U/L 0.552 0.186 1.74 0.003 1 High Sβ2M ≥ 6 mg/L 0.820 0.176 2.27 < 0.001 2 High ECOG PS Eastern Cooperative Oncology Group performance status, LDH lactate dehydrogenase, Sβ2M serum β -microglobulin, TE treatment emergent Coefﬁcient in the multivariate logistic model Table 4 TE grade ≥ 3 infections during the ﬁrst 4 months of high- and low-risk populations in various studies Trial Grade ≥ 3 infections, % P-value*low risk vs. high risk RR (95% CI) NNT Low risk (−3 to 1 points) High risk (2 to 5 points) a −19 MM-020 (N = 1 369) 7.0 24.0 8.19 × 10 3.43 (2.57–4.59) 8.3 −13 Rd pooled (n = 918) 7.4 24.9 2.7 × 10 3.37 (2.39–4.76) 8.0 −7 MPT (n = 451) 6.2 22.4 9.15 × 10 3.63 (2.11–6.24) 8.9 MM-015 (n = 384) 6.3 12.9 0.0552 2.05 (1.07–3.92) 15.5 a −4 MM-009/10 (n = 404) 17.1 35.7 7.69 × 10 2.09 (1.41–3.10) 5.6 a −6 MM-003 (n = 222) 30.3 63.3 2.21 × 10 2.09 (1.54–2.83) 3.2 MPT melphalan, prednisone, and thalidomide, NNT number needed to treat, Rd cont lenalidomide plus low-dose dexamethasone until disease progression, Rd18 lenalidomide plus low-dose dexamethasone for 18 cycles, Rd pooled Rd cont and Rd18 patients combined, RR relative risk, TE treatment emergent *P-value computed with χ test Patients with missing data for ≥ 1 of the variables selected by the multivariate logistic regression were excluded from the high-/low-risk deﬁnition developing early TE grade ≥ 3 infection (Table 4), with had a signiﬁcantly shorter time to ﬁrst TE grade ≥ 3 infec- comparable RRs between high- and low-risk groups in all tion in the ﬁrst 4 months compared with the low-risk group three test sets (MM-015: RR, 2.05 [P = 0.055]; MM-003: (MM-020: HR, 3.6 [P < 0.0001], C-index, 0.65; MM-003: RR, 2.09 [P < 0.0001]; MM-009/010: RR, 2.09 HR, 2.7 [P < 0.0001], C-index, 0.64; MM-009/010: HR, 1.9 [P = 0.0008]). This was despite very different populations [P = 0.006], C-index, 0.57; MM-015: HR, 2.05 [P = 0.03], at baseline and different rates of early TE grade ≥ 3 C-index, 0.59). infection (MM-015, 9.4%; MM-009-010, 20.3%; MM-003, To conﬁrm our predictive model, a competing risks 43.7%) compared with MM-020 (13.9%). Due to the analysis with progression or death without infection as difference in infection risks in those populations, the competing events with ﬁrst TE grade ≥ 3 infection in the NNT differed greatly in the various populations (MM-015, ﬁrst 4 months was performed using the MM-020 data set; 15.5; MM-009/010, 5.6; MM-003, 3.2) compared with this analysis included the same eight rules and iterative MM-020 (8.3). selection process used in the multivariate logistic analysis. The competing risk analysis in MM-020 conﬁrmed Conﬁrmatory analyses of the predictive model for the signiﬁcance of the six rules as in the logistic model risk of ﬁrst TE grade ≥ 3 infection in the ﬁrst (Supplemental Table 9). As such, the competing risk 4 months analysis provided an identical model to the one obtained through logistic regression analysis. The ﬁnal For illustration, a time to ﬁrst infection analysis was per- model remained signiﬁcant (P < 0.05) in both the MM-020 formed in both the MM-020 and the independent validation and the independent validation sets in a competing risks sets (Fig. 2). In all test sets, patients in the high-risk group analysis with progression or death without infection as 1410 C. Dumontet et al. Fig. 2 Time to ﬁrst grade ≥ 3 TE infection in the ﬁrst 4 months for high- and low-risk groups in the a MM-020 (n = 1602), b MM-015 (n = 452), c MM-009/10 (n = 643), d MM-003 (n = 425) populations. C-index concordance index, HR hazard ratio competing events with ﬁrst TE grade ≥ 3 infection in the infection. The risk of infection in the ﬁrst 18 months was ﬁrst 4 months. different across treatments: all TE infections (Rd pooled, 69.4%; MPT, 57.5% [P < .0001]) and TE grade ≥ 3 infec- tions (Rd pooled, 22.6%; MPT, 18.1% [P = .04]). This was Discussion noted despite the higher rate of grade 3/4 neutropenia with MPT (44.9%) vs. Rd pooled (27.1%) . Nearly 75% of all Because infections remain an important cause of morbidity grade ≥ 3 infections occurred in the absence of neutropenia and mortality in patients with MM , analyses of large (data not shown), suggesting that dexamethasone may have clinical trials can help identify risk factors associated with a contributing role. severe and life-threatening infections. The FIRST trial, This post hoc analysis showed that in the ﬁrst 4 months which demonstrated a signiﬁcant progression-free survival of treatment, (1) of patients who experienced a TE grade ≥ 3 and OS beneﬁt with Rd continuous vs. MPT, is among the infection, the majority had their ﬁrst infection during this largest phase 3 studies in MM and represents a typical time; (2) nearly one-half of all TE grade ≥ 3 infections transplant-ineligible NDMM population per its eligibility occurred, including the majority of infection-related deaths; criteria; therefore, the prognostic factors of infection iden- and (3) ﬁrst TE grade ≥ 3 infection was associated with an tiﬁed for these patients may be quite common in this increased risk of death, independent of prognostic factors population . The FIRST trial conﬁrmed that the risk of for OS. Our results are consistent with previous studies that infection in MM is high: 65.4% of patients presented with ≥ have shown that infections occur more often in the ﬁrst and 1 TE infection and 21.1% presented with ≥ 1 TE grade ≥ 3 second months of treatment [18, 19]. Infection risk may be A predictive model for risk of early grade ≥ 3 infection in patients with multiple myeloma not. . . 1411 highest during this period due to the immunosuppressive speciﬁed [21, 22], additional MM studies with data on nature of active MM and antimyeloma agents coupled with infections with speciﬁed causes are needed to determine the likelihood that the antimyeloma agents have not yet possible patterns of speciﬁc types of infections and appro- maximally reduced tumor load and repaired organ and tis- priate preventative therapies for patients at risk. Our model sue damage [2, 18, 20]. The risk of early TE grade ≥ 3 also requires further prospective interrogation for additional infection was similar with Rd vs. MPT, highlighting the role validation, particularly in proteasome inhibitor-based stu- of baseline patient-speciﬁc factors in determining infection dies. Furthermore, it would be of interest for additional risk during early treatment. studies to investigate risk factors for TE grade ≥ 3 infection Multivariate analysis identiﬁed ECOG PS and Sβ2M, after the ﬁrst 4 months of treatment as just over half of all LDH, and hemoglobin levels as prognostic factors for early TE grade ≥ 3 infections occurred after the ﬁrst 4 months in TE grade ≥ 3 infection. The signiﬁcance of these variables this study. was conﬁrmed by a competing risk analysis of ﬁrst TE In conclusion, a majority of patients in the FIRST trial grade ≥ 3 infection and death or progression without infec- reported ≥ 1 TE infection, conﬁrming that the risk of TE tion during the ﬁrst 4 months. Given that only 94 of the infection in patients with MM is high. In addition, our 3125 infections of any grade that occurred during the study analysis identiﬁed a set of baseline patient characteristics were non-TE infections, it is unlikely that including non-TE that were associated with risk of developing a TE grade ≥ 3 infections in the analysis would alter the results. A risk- infection in the initial 4 months of treatment. The high- and scoring system was used to separate patients in the FIRST low-risk groups deﬁned by our scoring system were asso- trial into high- and low-risk groups, which were associated ciated with signiﬁcantly different infection rates, irrespec- with signiﬁcantly different rates of early TE grade ≥ 3 tive of treatment. Clinicians may be able to apply this model infections (24.0% vs. 7.0%, respectively). The predictive to adjust their monitoring and treatment strategies for model differentiated high-risk from low-risk patients in infection prevention. The results of the predictive model three independent data cohorts, which included patients could be integrated into current infection management with relapsed/refractory MM (RRMM; MM-003 and MM- guidelines, including those from the International Myeloma 009/010) and NDMM (MM-015). As expected, the risk was Working Group  and European Myeloma Network . greater in the three RRMM studies that used dexamethasone Future NDMM studies could apply this model to evaluate (high-dose dexamethasone in MM-009/010 and the control which patients (all or those at high infection risk) arm of MM-003 and low-dose dexamethasone in the should receive prophylactic anti-infective drugs and pomalidomide arm of MM-003). Although still relevant, the what type would be most beneﬁcial to each patient model showed a lower absolute beneﬁt in MM-015, which subpopulation. had a lower incidence of early TE grade ≥ 3 infections and used prednisone instead of dexamethasone. In the low-risk groups, the risk was similar in the MPT arms of MM-020 Disclaimer and MM-015, which investigated MP and MP+lenalido- mide (6.2% and 6.3%, respectively). The risk was margin- The authors were fully responsible for all content and ally higher in the Rd arms of MM-020 (7.4%) and highest in editorial decisions for this manuscript. MM-009/010 and MM-003 (17.1% and 30.3 %, respec- Acknowledgements Writing assistance was provided by Kristina tively). Similarly, RRMM studies had a signiﬁcant risk of Hernandez, PhD, and Apurva Davé, PhD, MediTech Media, Ltd, early TE grade ≥ 3 infections in the high-risk groups (up to through funding by Celgene Corporation. Research support for this 63.3% in the MM-003 study). Even though these ﬁndings study was provided by Celgene Corporation. should be interpreted cautiously, the results suggest that Author contributions All authors have contributed to the concept and dexamethasone is a risk factor for early TE grade ≥ 3 design of the work, acquisition, analysis, or interpretation of data for infections, with studies with prednisone being associated the work, contributed to the drafting of the work, revised the manu- with a lower risk. script critically for important intellectual content, approved the ﬁnal These post hoc analysis ﬁndings are informative; how- version to be published, and agree to be accountable for all aspects of the work. ever, cautious interpretation is warranted. The use of anti- biotic prophylaxis was neither mandated in the study Compliance with ethical standards protocol nor standardized, which may limit interpretability. A pathogen could not be speciﬁed in a substantial propor- Conﬂict of interest Charles Dumontet has received honoraria from tion of infections reported limiting further elucidation on the Sanoﬁ and Janssen, has received fees for a consulting/advisory role types of interventions that may be useful in this setting. from Merck, and has received research funding from Roche. Cyrille Hulin has received honoraria from Celgene, Amgen, Bristol-Myers Although it is common in MM trials and in practice that a Squibb, Novartis, Janssen-Cilag, and Takeda. Meletios A. Dimopoulos substantial proportion of infections have no pathogen 1412 C. Dumontet et al. has received honoraria from Amgen, Celgene, Janssen, and Takeda 3. Hsu P, Lin TW, Gau JP, Yu YB, Hsiao LT, Tzeng CH, et al. Risk and has received fees for a consulting/advisory role from Amgen, of early mortality in patients with newly diagnosed multiple Celgene, Janssen, and Takeda. Angela Dispenzieri has received myeloma. Med (Baltim). 2015;94:e2305. research funding from Alnylam, Celgene, Pﬁzer, Prothena, and 4. Blimark C, Holmberg E, Mellqvist UH, Landgren O, Bjorkholm Takeda. Heinz Ludwig has received speakers’ bureau fees from Cel- M, Hultcrantz M, et al. Multiple myeloma and infections: a gene, Janssen, Takeda, and Amgen and has received research funding population-based study on 9253 multiple myeloma patients. from Takeda and Amgen. Michele Cavo has received honoraria from Haematologica. 2015;100:107–13. Amgen, Bristol-Myers Squibb, Celgene, Janssen, and Takeda. Juan 5. Ying L, YinHui T, Yunliang Z, Sun H. Lenalidomide and the risk José Lahuerta has received fees for a consulting/advisory role from of serious infection in patients with multiple myeloma: a Celgene, Janssen, and Takeda. Olivier Allangba has received fees for a systematic review and meta-analysis. Oncotarget. 2017;8: consulting/advisory role from Novartis and has received travel, 46593–600. accommodations, and expenses from Takeda, Pﬁzer, Celgene, Amgen, 6. Teh BW, Harrison SJ, Worth LJ, Thursky KA, Slavin MA. and Roche. Eileen Boyle has received fees for a consulting/advisory Infection risk with immunomodulatory and proteasome inhibitor- role from Celgene. Aurore Perrot has received honoraria and fees for a based therapies across treatment phases for multiple myeloma: a consulting/advisory role from Celgene, Janssen, Takeda, and Bristol- systematic review and meta-analysis. Eur J Cancer. Myers Squibb. Philippe Moreau has received honoraria from Celgene, 2016;67:21–37. Takeda, Novartis, Amgen, and Janssen-Cilag and has received fees for 7. Benboubker L, Dimopoulos MA, Dispenzieri A, Catalano J, a consulting/advisory role from Celgene, Takeda, Novartis, Amgen, Belch AR, Cavo M, et al. Lenalidomide and dexamethasone in and Janssen. Murielle Roussel has received research funding from transplant-ineligible patients with myeloma. N Engl J Med. Amgen, Celgene, and Janssen. Mohamad Mohty has received honor- 2014;371:906–17. aria from Celgene, Janssen, Bristol-Myers Squibb, Takeda, Novartis, 8. Facon T, Dimopoulos MA, Dispenzieri A, Catalano JV, Belch A, and Amgen; fees for consulting/advisory role from Celgene, Janssen, Cavo M, et al. Final analysis of survival outcomes in the phase 3 Bristol-Myers Squibb, Takeda, Novartis, and Amgen; speakers’ FIRST trial of up-front treatment for multiple myeloma. Blood. bureau fees from Janssen and Sanoﬁ; research funding from Sanoﬁ; 2018;131:301–10. and travel, accommodations, expenses from Sanoﬁ, JAZZ, Novartis, 9. San Miguel J, Weisel K, Moreau P, Lacy M, Song K, Delforge M, Janssen, and Amgen. Alexandre Civet has received fees for a con- et al. Pomalidomide plus low-dose dexamethasone versus sulting/advisory role from Celgene. Bruno Costa is an employee of high-dose dexamethasone alone for patients with relapsed and and owns stock in Celgene. Antoine Tinel is an employee of and owns refractory multiple myeloma (MM-003): a randomised, open- stock in Celgene. Yann Gaston-Mathé is an employee of IntegraGen, label, phase 3 trial. Lancet Oncol. 2013;14:1055–66. has received fees for a consulting/advisory role from Celgene, and has 10. Wang M, Dimopoulos MA, Chen C, Cibeira MT, Attal M, received travel, accommodations, expenses from Celgene. Thierry Spencer A, et al. Lenalidomide plus dexamethasone is more Facon has received fees for a consulting/advisory role from Amgen, effective than dexamethasone alone in patients with relapsed or Celgene, Janssen, Karyopharm, PharmaMar, and Takeda and has refractory multiple myeloma regardless of prior thalidomide received speakers’ bureau fees from Amgen, Celgene, Janssen, and exposure. Blood. 2008;112:4445–51. Takeda. Andrew Belch, Philippe Rodon, Jan Van Droogenbroeck, 11. Dimopoulos M, Spencer A, Attal M, Prince HM, Harousseau JL, Lugui Qiu, Ann Van de Velde, Jae Hoon Lee, Salomon Manier, Dmoszynska A, et al. Lenalidomide plus dexamethasone for Michel Attal, and Jean Yves Mary declare that they have no conﬂict of relapsed or refractory multiple myeloma. N Engl J Med. interest. 2007;357:2123–32. 12. Weber DM, Chen C, Niesvizky R, Wang M, Belch A, Stadtmauer EA, et al. Lenalidomide plus dexamethasone for relapsed multiple Open Access This article is licensed under a Creative Commons myeloma in North America. N Engl J Med. 2007;357: Attribution 4.0 International License, which permits use, sharing, 2133–42. adaptation, distribution and reproduction in any medium or format, as 13. Palumbo A, Hajek R, Delforge M, Kropff M, Petrucci MT, long as you give appropriate credit to the original author(s) and the Catalano J, et al. Continuous lenalidomide treatment for newly source, provide a link to the Creative Commons license, and indicate if diagnosed multiple myeloma. N Engl J Med. 2012;366:1759–69. changes were made. The images or other third party material in this 14. Fisher LD, Lin DY. Time-dependent covariates in the Cox article are included in the article’s Creative Commons license, unless proportional-hazards regression model. Annu Rev Public Health. indicated otherwise in a credit line to the material. If material is not 1999;20:145–57. included in the article’s Creative Commons license and your intended 15. Akaike H. Factor analysis and AIC. Psychometrika. use is not permitted by statutory regulation or exceeds the permitted 1987;52:317–32. use, you will need to obtain permission directly from the copyright 16. Fine JP, Gray RJ. A proportional hazards model for the sub- holder. To view a copy of this license, visit http://creativecommons. distribution of a competing risk. J Am Stat Assoc. org/licenses/by/4.0/. 1999;94:496–509. 17. Rajkumar SV, Dimopoulos MA, Palumbo A, Blade J, Merlini G, Mateos MV, et al. International Myeloma Working Group upda- References ted criteria for the diagnosis of multiple myeloma. Lancet Oncol. 2014;15:e538–548. 1. Nucci M, Anaissie E. Infections in patients with multiple 18. Perri RT, Hebbel RP, Oken MM. Inﬂuence of treatment and myeloma in the era of high-dose therapy and novel agents. response status on infection risk in multiple myeloma. Am J Med. Clin Infect Dis. 2009;49:1211–25. 1981;71:935–40. 2. Augustson BM, Begum G, Dunn JA, Barth NJ, Davies F, 19. Rayner HC, Haynes AP, Thompson JR, Russell N, Fletcher J. Morgan G, et al. Early mortality after diagnosis of multiple Perspectives in multiple myeloma: survival, prognostic factors and myeloma: analysis of patients entered onto the United Kingdom disease complications in a single centre between 1975 and 1988. Medical Research Council trials between 1980 and 2002—Med- Q J Med. 1991;79:517–25. ical Research Council Adult Leukaemia Working Party. J Clin 20. Pratt G, Goodyear O, Moss P. Immunodeﬁciency and immu- Oncol. 2005;23:9219–26. notherapy in multiple myeloma. Br J Haematol. 2007;138:563–79. A predictive model for risk of early grade ≥ 3 infection in patients with multiple myeloma not. . . 1413 21. Teh BW, Harrison SJ, Allison CC, Slavin MA, Spelman T, Worth 23. Palumbo A, Rajkumar SV, San Miguel JF, Larocca A, Niesvizky LJ, et al. Predicting risk of infection in patients with newly R, Morgan G, et al. International Myeloma Working Group con- diagnosed multiple myeloma: utility of immune proﬁling. Front sensus statement for the management, treatment, and supportive Immunol. 2017;8:1247. care of patients with myeloma not eligible for standard autologous 22. Teh BW, Harrison SJ, Worth LJ, Spelman T, Thursky KA, Slavin stem-cell transplantation. J Clin Oncol. 2014;32:587–600. MA. Risks, severity and timing of infections in patients with 24. Terpos E, Kleber M, Engelhardt M, Zweegman S, Gay F, multiple myeloma: a longitudinal cohort study in the era of Kastritis E, et al. European Myeloma Network guidelines for the immunomodulatory drug therapy. Br J Haematol. management of multiple myeloma-related complications. 2015;171:100–8. Haematologica. 2015;100:1254–66.
Leukemia – Springer Journals
Published: Apr 26, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera