Prognostic value of positron emission tomography/computed tomography in transplant-eligible newly diagnosed multiple myeloma patients from CASSIOPEIA: the CASSIOPET study

¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography (PET)/computed tomography (CT) is a reliable imaging technique for evaluating and monitoring multiple myeloma (MM) patients with a prognostic value for progression-free survival (PFS) and overall survival.^1-5 PET/CT-positive features at diagnosis have indeed been found to correlate with poorer outcomes.^3,6 Here, we report the first results of CASSIOPET, a companion study of CASSIOPEIA (clinicaltrials gov. Identifier: NCT02541383),⁷, ⁸ which evaluated the prognostic value of baseline PET/CT on PFS. Study design and eligibility criteria for the CASSIOPEIA trial have been previously published.⁷ Briefly, transplanteligible patients with newly diagnosed MM (NDMM) were randomized 1:1 to receive four 28-day, pre-autologous stem cell transplant (ASCT) induction cycles and two 28-day post-ASCT consolidation treatment cycles of daratumumab plus bortezomib/thalidomide/dexamethasone (D-VTd) or bortezomib/thalidomide/dexamethasone (VTd) in CASSIOPEIA Part 1. The primary endpoint, stringent complete response, was evaluated 100 days after ASCT. In CASSIOPEIA Part 2, patients achieving a partial response (PR) or better 100 days post-ASCT underwent a second 1:1 randomization to observation or maintenance therapy with intravenous daratumumab 16 mg/kg every 8 weeks for up to 2 years.

Among patients randomized in CASSIOPEIA, those eligible for inclusion in CASSIOPET had received a PET/CT scan ≤6 weeks before randomization in CASSIOPEIA. Patients were excluded if they were unable to access or undergo PET/CT investigation, had uncontrolled diabetes, or had received steroids ≤12 hours before the PET/CT scan. All patients provided written informed consent.

The primary endpoint of the CASSIOPET study is PFS from the second CASSIOPEIA randomization. This PFS analysis for CASSIOPEIA was recently reported.⁸ CASSIOPET analyses reported here evaluate the prognostic value of PET/CT at baseline on PFS from the first CASSIOPEIA randomization, PFS differences between baseline PETnegative versus PET-positive patients in each treatment arm, and the effect of daratumumab on PET/CT negativity at post-consolidation.

PET/CT scans were performed at baseline and postconsolidation (day 100 [±7 days] post-ASCT). All patients had fasted for ≥6 hours before the PET/CT scan. No dexamethasone was to be administered ≤12 hours before the PET/CT scan. Blood glucose levels were measured before ¹⁸F-FDG injection with a preferred glycemia level ≤150 mg/dL. No insulin was administered ≤2 hours before ¹⁸F-FDG injection, and no oral contrast was given. Whole-body imaging was performed 55 to 75 minutes after the ¹⁸F-FDG injection. First scout and low-dose CT data (head to feet) were obtained, followed by PET data acquisition, image reconstruction, and analysis.

Acquired imaging data were uploaded to a central electronic repository system (KEOSYS, Saint-Herblain, France) and analyzed using the IMAGYS platform. Five-point Deauville scores (range, 1-5)⁹ were applied to bone marrow (BM), bone focal lesions (FL), extramedullary disease (EMD), and paramedullary disease (PMD). Localization of the most intense ¹⁸F-FDG uptake was identified, and the maximal standardized uptake value (SUV_max) was calculated. Bone SUV_max was defined as the hottest value between BM, FL, and PMD. PET images were interpreted (blinded to patient treatment) by an independent team of nuclear-medicine physicians with extensive MM experience. PET/CT scan assessments did not include separate assessments of CT scans; thus, patients may have been PET-negative but could still display lytic lesions in the CT scan.

PET–complete response (PET-CR) was defined as an uptake of less than or equal to the mediastinal blood pool in all localizations. PET–unconfirmed CR (PET-uCR) was defined as an uptake between the mediastinal blood pool and liver. PET-PR was defined as a decrease in the number and/or activity of BM, FL, EMD, or PMD but persistence of lesions with uptake above liver activity or BM uptake above liver activity. Patients with PET–stable disease (PET-SD) had no significant modification of FL, EMD, or PMD compared with baseline. Patients with PET–progressive disease (PET-PD) had a new lesion (FL, EMD, or PMD) compared with baseline. PET/CT-positive patients were defined as patients with PET-PR and PET-SD. PET/CT-negative patients were defined as patients with PET-CR and PET-uCR. Clinical response was assessed according to International Myeloma Working Group criteria.¹⁰

The prognostic effect of including explanatory PET/CT variables on PFS was assessed using Cox regression models. Seven baseline PET/CT characteristics were chosen based on expert knowledge: PET positivity, presence of FL, BM infiltration, PMD, EMD, FL SUV_max, and bone SUV_max.

The prognostic effect of each of the seven baseline PET characteristics was estimated using a univariable Cox model in addition to the prognostic effect of known prognostic factors: serum lactate dehydrogenase (LDH) levels, serum β₂ microglobulin concentration, cytogenetic risk, and International Staging System (ISS) disease staging.

The prognostic effect of each of the seven baseline PET characteristics was then estimated adjusting for the treatment group and revised ISS (r-ISS). These covariates were chosen based on expert knowledge without statistical covariate selection procedures due to the relatively small number of PFS events. Adjustment for treatment group accounts for the randomized design of CASSIOPEIA, of which CASSIOPET is an ancillary study. r-ISS is the current stratification score for myeloma patients and combines ISS (which includes serum b₂ microglobulin and serum albumin), cytogenetic risk, and serum LDH level into a single variable.

A final multivariable Cox model was constructed including the seven baseline PET characteristics and adjusting for treatment and r-ISS. At the time of analysis, 20 PFS events were observed in the D-VTd group and 34 in the VTd group. Thus, the third multivariable Cox model results are exploratory and should be interpreted cautiously due to the low ratio of events per variable.

Proportional hazards and log-linearity of effects were assessed. No statistically significant violations of the proportional hazard assumption were detected at the customary 5% P value threshold using the Schoenfeld residuals. No violation of the log-linearity assumption was detected using P splines. The presence of multicollinearity was assessed using the variance inflation factor; no value exceeded 2 for all PET/CT characteristics.

The log-rank estimator with Kaplan–Meier representation was used to describe PFS. Baseline and post-consolidation PET/CT negativity rates were compared between treatment groups using the chi-square test, odds ratios, and 2-sided 95% confidence intervals (CI). The role of interactions between baseline PET positivity and treatment in the PFS distribution could not be assessed, as zero PFS events were observed in the D-VTd PET-negative group, leading to a hazard ratio (HR) of 0 with a non-estimable variance using classical statistical tests.

The primary results of CASSIOPEIA Part 1 have been reported (median follow-up, 18.8 months).⁷ The current analysis of CASSIOPET was performed using patient data with a median follow-up time of 29.2 months. Of 1,085 patients enrolled in CASSIOPEIA, 268 (D-VTd, n=137; VTd, n=131) had assessable baseline PET; 184 (D-VTd, n=101; VTd, n=83) patients were also PET-evaluable post-consolidation (Figure 1). Baseline characteristics of patients with assessable baseline PET were similar to those in the overall CASSIOPEIA trial (Online Supplementary Table S1). At baseline, 54 patients (20%) were PET-negative and 214 (80%) were PET-positive.

PFS was better for baseline PET-negative versus PET-positive patients (hazard ratio [HR]: 0.42, 95% CI: 0.18-0.97, P=0.0365; Figure 2A). The 12- and 18-month PFS rates were higher in patients who were PET-negative (12- and 18-month rates, 100%) versus PET-positive (12-month rate, 93%; 18-month rate, 87%) at baseline. When stratified by treatment group, PFS was better among patients who were PET-negative versus PET-positive in the D-VTd arm. However, PFS was not significantly different in the VTd arm (Figure 2B). By univariable analysis, baseline PET characteristics associated with PFS were PMD (P<0.001), EMD (P=0.034), FL (P=0.047), FL SUV_max (P=0.043), and bone SUV_max (P=0.021). All these characteristics, except for FL, remained prognostic factors when adjusting for treatment arm and r-ISS (Table 1). A multivariable analysis including all PET/CT characteristics and adjusting for treatment arm and r-ISS showed that PMD (HR: 3.16, 95% CI: 1.60-6.28) and EMD (HR: 2.32, 95% CI: 1.04-5.19) remained independently associated with a higher risk of relapse or death (Table 1).

Figure 1.STROBE flow chart for CASSIOPET. ASCT: autologous stem cell transplant; MRD: minimal residual disease; PET: positron emission tomography; PET/CT: positron emission tomography/computed tomography. ^aBaseline PET–evaluable patients were defined as patients with assessable baseline PET acquired before the first dosing date. ^bPost-consolidation PET–evaluable patients included patients with assessable day 100 post-ASCT PET data and positive baseline PET but excluded patients with a date of PET/CT post-consolidation >±90 days from the date of the day 100 MRD assessment. ^c13 patients had unevaluable baseline PET/CT but evaluable post-consolidation PET/CT and were included in the post-consolidation analysis.

Figure 2.Progression-free survival outcomes by baseline positron emission tomography/computed tomography status in CASSIOPET. (A) Progression-free survival (PFS) for baseline positron emission tomography (PET)-negative patients versus PET-positive patients and (B) PFS for baseline PET-negative patients versus PET-positive patients by treatment group. Baseline PET assessments were performed prior to the first dose of study drug, and PFS was based on time from first randomization. CI: confidence interval; D-VTd: daratumumab plus bortezomib/thalidomide/dexamethasone; HR: hazard ratio; NE: not estimable; VTd: bortezomib/thalidomide/dexamethasone. ^aBased on a log-rank test.

Of the 184 patients with post-consolidation PET measurements, 118 (64%) were assessed as PET-CR and 47 (26%) as PET-uCR (Online Supplementary Table S2). Seventeen (9%) patients were assessed as PET-PR and two (1%) as PET-SD. Overall, 165 (90%) patients were PET-negative and 19 (10%) were PET-positive. The rates of PET negativity were high and similar between the D-VTd (90%) and VTd (89%) groups.

Results of the CASSIOPET study presented here confirm that baseline PET/CT findings have a prognostic value for PFS. PFS was indeed better for baseline PET-negative versus PET-positive NDMM patients, including patients treated with daratumumab. The presence of PMD, EMD, FL, and the FL SUV_max and bone SUV_max were associated with shorter PFS. When adjusting for treatment arm and classical NDMM (r-ISS) prognostic score, PMD and EMD had independent prognostic value. PET-CR post-consolidation rates were high and similar in both D-VTd and VTd groups.

PET/CT is negative in approximately 10-20% of symptomatic MM patients. This study shows that PET/CT negativity, even if considered as false-negative for disease detection, could be considered for its prognostic value. Rasche et al. demonstrated that ¹⁸F-FDG PET/CT may be considered ineffective for approximately 11% of patients due to low expression of the hexokinase 2 enzyme.¹¹ However, another study of 90 NDMM patients receiving novel agents during induction therapy showed that low hexokinase 2 expression associated with PET/CT negativity correlated with relatively better prognosis versus PET/CT-positive patients.⁴ Baseline PET/CT-negative patients may thus represent a less aggressive subgroup of MM patients, associated with better outcomes in the setting of quadruplet therapy and ASCT.

This prospective study demonstrates PMD as an independent prognostic factor in MM. Previous prospective studies have shown the prognostic value of EMD, SUV_max, and FL number.^2,5,6,12,13 However, these studies neither described nor assessed PMD as a potential prognostic biomarker. In the prospective IMAJEM study that demonstrated the prognostic value of EMD, EMD was detected at a similar percentage (7.5%) as in CASSIOPET (5-11%), but PMD was considered as FL.² The independent prognostic value of PMD shown here is consistent with data from Rasche et al., indicating the presence of large focal lesions as a strong independent poor prognosis factor in NDMM.^14,15

Spatial heterogeneity can limit the sensitivity of risk classification based on cytogenetics and gene expression profiling because these tests are based on cells obtained from a single BM biopsy. Rasche et al. have shown that high-risk genomic alterations can be present in focal lesions, yet absent in other locations.¹⁴ Combined with the results of other studies,^2,5 several PET/CT characteristics could be defined as possible high-risk biomarkers and used to define high-risk patients at the initial diagnosis of symptomatic MM.

The IMAJEM study² used background liver uptake to define PET/CT negativity, similar to the CASSIOPET study, and was recommended in the recent standardization by Zamagni et al.⁵ Regardless of the differing efficacies and regimens, both studies support the prognostic value of PET/CT.

Table 1.Univariable and multivariable analyses of the prognostic value of baseline positron emission tomography (PET) characteristics on progression-free survival based on all patients with PET measurements at baseline (54/268 progression-free survival events).

In conclusion, baseline PET/CT findings appear to have a prognostic value for PFS. Longer follow-up in CASSIOPEIA Part 2 will provide additional insight.

Footnotes

• Received September 20, 2021
• Accepted October 11, 2022

Correspondence

F. KRAEBER-BODÉRÉ - francoise.bodere@chu-nantes.fr

Disclosures

SZ served in a consulting or advisory role for Celgene, Bristol Myers Squibb, Janssen, Takeda, Oncopeptides, and Sanofi; and received research funding from Takeda and Janssen. AP received honoraria from Amgen, Celgene, Janssen, Sanofi, and Takeda. CH received honoraria from Celgene, Janssen, Amgen, and Takeda; and had travel, accommodations, or other expenses paid or reimbursed by Celgene, Janssen, and Amgen. TF received honoraria from, served in a consulting or advisory role for, and served on a speakers bureau for Janssen. XL received honoraria from and served in a consulting or advisory role for AbbVie, Amgen, Bristol Myers Squibb, Celgene, Gilead, Incyte, Janssen, Karyopharm, Merck, Mundipharma, Novartis, Roche, and Takeda. KB served in a consulting or advisory role for and received honoraria from Amgen, Celgene, Janssen, and Takeda. LK received honoraria from and served in a consulting or advisory role for Amgen, Janssen, Celgene, and Takeda; and had travel, accommodations, or other expenses paid or reimbursed by Amgen and Janssen. M-DL received honoraria from; served in a consulting or advisory role for; and had travel, accommodations, or other expenses paid or reimbursed by Amgen, Janssen, AbbVie, and Celgene. MCM received honoraria from Gilead; served in a consulting or advisory role for Celgene, Amgen, Takeda, Janssen-Cilag, Servier, and Bristol Myers Squibb; and received research funding from Celgene. PS received honoraria and research funding from Amgen, Celgene, Janssen, Karyopharm, and Takeda; and received researching funding from Skyline. LP and TK are employed by Janssen. FR, CdB, and JV are employed by Janssen and hold stock options from Johnson & Johnson. PM received honoraria from and served in a consulting or advisory role for Celgene, Janssen, Amgen, Sanofi, and AbbVie. All other authors have no conflicts of interest to disclose.

Contributions

All authors contributed to the study design, study execution, data analysis, and manuscript writing. All authors provided a full review of the manuscript and are fully responsible for all content and editorial decisions, were involved in all stages of manuscript development, and have approved the final version.

Data-sharing statement

The data sharing policy of Janssen Pharmaceutical Companies of Johnson & Johnson is available at https://www.janssen.com/clinical-trials/transparency. As noted on this site, requests for access to the study data can be submitted through Yale Open Data Access (YODA) Project site at http://yoda.yale.edu.

Funding

Acknowledgments

The authors would like to thank the patients who participated in this study, the investigators who participated in this study, staff members at the study sites, staff members who were involved in data collection and analyses, the data safety monitoring committee, the IFM, HOVON, and the Janssen team. Editorial and medical writing support in the development of an earlier draft of the manuscript were provided by Austin Horton, PhD, of Lumanity Communications Inc., and were funded by Janssen Global Services, LLC.