Prognostic impact of cytogenetic abnormalities detected by FISH in AL amyloidosis with daratumumab-based frontline therapy Free

The addition of the anti-CD38 monoclonal antibody daratumumab (Dara) to bortezomib-cyclophosphamide-dexamethasone (VCD) in frontline therapy has dramatically improved the depth of hematologic response, organ response, and progression-free survival in patients with immunoglobulin light chain (AL) amyloidosis.^1,2 The phase 3 randomized ANDROMEDA-AL trial that led to the accelerated approval of Dara demonstrated a threefold increase in the rate of hematologic complete response (heme-CR) with Dara-VCD compared with VCD.¹ Nevertheless, ∼20% of newly diagnosed patients are unable to achieve a very good partial response (VGPR) or better with Dara-VCD, potentially requiring subsequent treatments.^1,3,4 With this major shift in the treatment landscape, there is an urgent need for reevaluation of baseline prognostic factors in the era of Dara-VCD. Cytogenetics by fluorescence in situ hybridization (FISH), assessed in CD138 selected cells, plays an important role in the prognostic evaluation of plasma cell disorders.^5,6 Although the prognostic impact of FISH cytogenetics in multiple myeloma is well defined,⁷ there is a dearth of literature on its prognostic impact in large homogeneously treated cohorts of patients with AL. In the pre-Dara era, t(11;14) was shown to be an adverse prognostic factor for hematologic response rate, hematologic event-free survival (heme-EFS), and overall survival (OS) in bortezomib- and immunomodulatory drug–treated newly diagnosed patients with AL.^8,9 However, in a secondary analysis of the ANDROMEDA trial, the addition of Dara led to a dramatic increase in heme-CR and VGPR or better rate compared with VCD alone in t(11;14) subgroup, implying that Dara might be able to abrogate the negative prognostic impact of t(11;14) abnormality.³ Additionally, Dara was shown to benefit all cytogenetic subgroups with respect to heme-CR rate, VGPR or better, and major organ deterioration–progression-free survival.³ However, FISH results were available in just 155 patients in the Dara-VCD arm, and the prognostic impact of all individual cytogenetic abnormalities on response rate and heme-EFS within the Dara-VCD arm was not reported. Notably, in a real-world study on relapsed/refractory AL from Germany, gain/amp(1q) (hereafter, referred to as +1q) and hyperdiploidy had an adverse prognostic impact on heme-EFS and OS in patients treated with Dara-VD or daratumumab-dexamethasone (Dara-D).¹⁰ 

The objective of our study was to define the prognostic impact of cytogenetic abnormalities by FISH on clinical outcomes in a homogeneously treated multicenter real-world cohort of patients with newly diagnosed AL amyloidosis.

This is a multicenter retrospective cohort study involving 7 institutions from 3 countries (Columbia University Irving Medical Center, Boston University, Mayo Clinic Rochester, Medical College of Wisconsin, Heidelberg University, San Matteo University Hospital of Pavia, and University of Washington). The study was approved by the respective institutional review boards of each institution, and all research procedures were conducted in accordance with the Declaration of Helsinki. The key inclusion criterion was all consecutive patients with newly diagnosed systemic AL amyloidosis (biopsy proven and type confirmed to be light chain by either mass spectrometry, immunohistochemistry, or immunofluorescence microscopy) who initiated therapy with Dara-VCD or Dara-VD on or before 31 December 2022. The exclusion criteria were as follows: (1) patients in whom Dara was added to VCD or VD backbone beyond Cycle #1 or >4 weeks from treatment initiation; and (2) patients enrolled in any clinical trial of fibril-directed therapy or clone-directed therapy as a part of frontline treatment. The choice of treatment (Dara-VCD vs Dara-VD) was as per the treating clinicians. Data on FISH cytogenetics at baseline were collected from individual institutions as per their respective cutoffs for defining positive vs negative status on individual probes. The key probes of interest were as follows: t(11;14), +1q, hyperdiploidy, del(13q) or monosomy 13, del(17p), and myeloma high-risk (HR) translocations (t[4;14], t[14;16], or t[14;20]). Disease staging was performed according to the European modification of Mayo 2004 staging system, Mayo 2012 staging system, and renal staging by Palladini et al.^11-13 Patients were followed as per institutional guidelines. The STROBE checklist for observational studies is shown in supplemental Appendix 1, available in the Blood website.

All patients with a baseline difference between involved and uninvolved free light chain (dFLC) of ≥2 mg/dL were evaluable for hematologic response assessment. For patients with baseline dFLC >5 mg/dL, hematologic response was assessed per standard consensus criteria.^14,15 For patients with a baseline dFLC of 2 to 5 mg/dL, low dFLC partial response (PR) was defined as posttreatment dFLC <1 mg/dL without meeting the criteria for heme-CR.^16-18 Patients who achieved a VGPR, low dFLC PR, or heme-CR were classified as “VGPR or better.” For heme-EFS, the following outcomes were considered as events: hematologic relapse or progression as per International Society of Amyloidosis criteria,¹⁹ start of a subsequent line of therapy, or death, whichever is earlier.^8,10 Cardiac organ response was assessed based on consensus criteria,¹⁴ with response defined as a decrease in N-terminal pro brain natriuretic peptide (NT-proBNP) by >30%, with an absolute decrease of at least 300 pg/mL in patients with baseline NT-proBNP >650 pg/mL. Renal organ response was based on the updated consensus criteria, with response defined as a decrease in proteinuria by ≥30% or below 0.5 g per 24 hours without a drop in creatinine clearance.¹³ Of note, the laboratory values for troponins and BNP/NT-proBNP and the various troponin assays were harmonized based on the conversion table by Muchtar et al.²⁰ 

The end points of interest were hematologic response and heme-EFS stratified by prespecified cytogenetic subgroups. The Kaplan-Meier method was used to calculate the probabilities of heme-EFS and OS, with log-rank test to compare groups. Reverse Kaplan-Meier method was used to calculate the median follow-up time.²¹ For heme-EFS, patients who did not experience any of the above-mentioned events were censored at their last follow-up. The Kruskal-Wallis and Pearson χ² tests were used to ascertain the differences between continuous and categorical variables, respectively. Patients who underwent high-dose melphalan followed by autologous hematopoietic cell transplantation (HDM-AHCT) as planned consolidation therapy after Dara-VCD/Dara-VD were censored on the day of transplant. For such patients, data on hematologic and organ response immediately before HDM-AHCT was considered as their best response. However, when HDM-AHCT was performed due to suboptimal response to frontline therapy, it was considered as an event for heme-EFS. Cox proportional hazards regression models were used for univariate analysis to identify prognostic factors for heme-EFS and OS. The following prognostic variables of interest were tested in univariate analysis: age, sex, ethnicity, race, New York Heart Association (NYHA) class, involved light chain isotype, immunoglobulin isotype, serum albumin, estimated glomerular filtration rate, 24-hour urine protein, dFLC, NT-proBNP, high-sensitivity (hs) troponin-T, bone marrow plasma cells (BMPC), t(11;14) status, +1q status, hyperdiploidy status, del(13q) status, del(17p) status, HR translocation status, >2 organs involved, and frontline therapy. Variables that were significant on univariate analysis (P < .05) were entered into multivariate analysis to identify independent prognostic factors for heme-EFS and OS. Results from univariate and multivariate Cox models were reported as hazard ratio with 95% confidence interval (CI). A P value <.05 was considered statistically significant. A power calculation was not performed, and the sample size was derived based on the timeline of patient selection, which was in turn driven by the regulatory approval of Dara at individual sites. Statistical analysis was performed on JMP 14.0 (SAS, Cary, NC).

Of 313 consecutive patients who received Dara-VCD/Dara-VD, 283 (90.4%) had available FISH data on ≥1 probe, 8 (2.6%) had a technical failure due to insufficient CD138 selected cells, and 22 (7.0%) did not have FISH data available at diagnosis. Hence, 283 patients from 7 centers with available FISH data on ≥1 probe were included in our study. The baseline clinical and demographic characteristics of the study cohort and patients in individual cytogenetic subgroups are shown in Table 1. The baseline clinical features in patients with (n = 283) vs without FISH data (n = 30) at diagnosis are shown in supplemental Appendix 2. The only significant differences between the 2 groups were a higher median dFLC and median BMPCs in patients with available FISH data vs those without. The median age of the study cohort was 66 years (interquartile range [IQR], 59-71), with 65% males and 6% Black/African Americans. Approximately one-third of patients presented with NYHA class III/IV symptoms. The median dFLC was 25 mg/dL (IQR, 10-63), with more patients (59%) presenting with dFLC ≥18 mg/dL. The median NT-proBNP at diagnosis was 2310 pg/mL (IQR, 525-6666), with ∼1 in 5 presenting with NT-proBNP >8500 pg/mL. Approximately 50% of patients had Mayo cardiac stage IIIa/IIIb, and 10.3% had renal stage III disease at diagnosis. The 2 most common organs involved were the heart (71%), followed by the kidneys (63%). Most patients in our cohort (93.3%) were treated with Dara-VCD, with the rest receiving Dara-VD. Other than a higher proportion of females in the Dara-VD group than in the Dara-VCD group (57.9 vs 33.3%, respectively; P = .035), there was no significant difference in baseline clinical features between patients receiving Dara-VCD vs Dara-VD (supplemental Appendix 3). A total of 30 of 283 patients (10.6%) had received HDM-AHCT as planned consolidation therapy after Dara-VCD/Dara-VD. The median time from treatment initiation to consolidation with HDM-AHCT was 5.7 months (IQR, 4.4-8.3). Additionally, 6 patients had received HDM-AHCT due to suboptimal response to Dara-VCD/Dara-VD and 1 for hematologic progression.

The 3 most common cytogenetic abnormalities were t(11;14) (53.4%), deletion(13q)/monosomy 13 (28.9%), and +1q (22.3%). Among 58 patients with +1q, a total of 45 had available data on copy number. Of these 45 patients, 37 (82.2%) had 3 copies [gain(1q)], and 8 (17.8%) had ≥4 copies [amp(1q)]. Given the significant impact of +1q on hematologic response and heme-EFS as shown below, we investigated baseline clinical features in patients with vs without +1q (supplemental Appendix 4). Patients with +1q had a greater frequency of lambda light chain isotype (90% vs 77%, respectively; P = .029), dFLC ≥18 mg/dL (69% vs 54%, respectively; P = .042), BMPC ≥10% (78% vs 63%, respectively; P = .166), and hyperdiploidy (31% vs 15%, respectively; P = .012) than patients without +1q. A Venn diagram with distribution of cytogenetic abnormalities by FISH is shown in Figure 1.

Of 269 patients who were evaluable for hematologic response assessment (ie, baseline dFLC ≥ 2 mg/dL), 240 had available data on best hematologic response. Among these 240 patients, 106 (44.2%) achieved heme-CR, and 181 (75.4%) achieved VGPR or better. Next, we investigated heme-CR and VGPR or better rate stratified by cytogenetic subgroups (Figure 2). The only cytogenetic subgroup with a significantly lower rate of heme-CR and VGPR or better was +1q, with the heme-CR rate for +1q vs no +1q being 30.2% vs 47.9%, respectively (P = .022), and the respective VGPR or better rates being 64.2% vs 79.0% (P = .033).

Among 165 patients who were evaluable for cardiac response (ie, cardiac involvement and baseline NT-proBNP >650 pg/mL) and had available data on response assessment, 93 (56.4%) achieved a cardiac response. The median time to best cardiac response was 10.3 months (IQR, 4.6-16.1). Among 118 patients who were evaluable for renal response (ie, renal involvement and 24-hour urine protein >0.5 g/d) and had available data on response assessment, 76 (64.4%) achieved a renal response. The median time to best renal response was 10.9 months (IQR, 5.98-17.98). There was no difference in cardiac or renal organ response rate among different cytogenetic subgroups (supplemental Appendix 5). Data on the depth of organ response by cytogenetic subgroups are shown in supplemental Appendix 6.

At a median follow-up of 19.5 months (95% CI, 17.6-20.5) from treatment initiation, the median heme-EFS of the study cohort was 49.6 months (95% CI, 27.6 to not reached [NR]), and the median OS was NR (95% CI, 52.6 months to NR). The estimated heme-EFS and OS at 2 years was 59% (95% CI, 52-66) and 82% (95% CI, 77-87), respectively. The Kaplan-Meier curves for heme-EFS and OS of the study cohort are shown in supplemental Appendix 7. There was no significant difference in heme-EFS or OS between patients receiving Dara-VCD vs Dara-VD (supplemental Appendix 8). Notably, 65 of 283 patients (22.97%) had received a subsequent line of therapy, with the most common reason for switching treatment being suboptimal response to Dara-VCD/Dara-VD as per the treating clinician (47/65 [72.3%]), followed by hematologic progression or relapse (13/65 [20.0%]), toxicity (3/65 [4.6%]), and other reasons (2/65 [3.1%]). Among 47 patients with suboptimal response per treating clinician, 37 of 43 with available data (86%) had a PR or no response to Dara-VCD/Dara-VD, with the remaining 6 having VGPR with no organ response (n = 3), dFLC >1 mg/dL (n = 1), and involved free light chain >2 mg/dL (n = 1). The proportion of patients requiring a subsequent line of therapy by cytogenetic subgroup is as follows: t(11;14) vs no t(11;14), 20.8% vs 25.4% (P = .365); +1q vs no +1q, 34.5% vs 17.7% (P = .008); hyperdiploidy vs no hyperdiploidy, 28.9% vs 20.1% (P = .211); del(13q) vs no del(13q), 23.9% vs 22.7% (P = .838); del(17p) vs no del(17p), 18.2% vs 23.4% (P = .680); and HR translocations vs no HR translocations, 18.8% vs 23.5%, respectively (P = .658). Among patients requiring a subsequent line of therapy, the median time to next treatment was 5.4 months (IQR, 3.4-8.9). The 3 most common subsequent lines of therapy were venetoclax based (n = 30), followed by daratumumab-pomalidomide-dexamethasone (Dara-Pom-Dex) (n = 9) and HDM-AHCT (n = 7).

When stratified by cytogenetic subgroups, a significantly worse heme-EFS was noted in patients with +1q and hyperdiploidy. The median heme-EFS in patients with vs without +1q was 14.3 months (95% CI, 7.3 to NR) and NR (95% CI, 27.6 to NR), respectively (P = .005). The median heme-EFS in patients with vs without hyperdiploidy was 20.0 months (95% CI, 7.5 to NR) and 51.2 months (95% CI, 51.2 to NR), respectively (P = .024). The Kaplan-Meier curves for heme-EFS stratified by individual cytogenetic subgroups are shown in Figure 3. On univariate analysis for heme-EFS (Table 2), NYHA class III/IV symptoms, dFLC ≥18 mg/dL, NT-proBNP >8500 pg/mL, hs–troponin-T ≥50 ng/L, BMPCs ≥10%, +1q, and hyperdiploidy were identified as significant adverse prognostic factors. Significant prognostic factors for heme-EFS on multivariate analysis (Table 3) were the following: NYHA class IV (vs class I; HR, 26.99; 95% CI, 4.33-168.17; P < .001), NT-proBNP >8500 pg/mL (vs ≤8500; HR, 2.35; 95% CI, 1.20-4.59; P = .012), and +1q (vs no +1q; HR, 2.02; 95% CI, 1.13-3.59; P = .016).

There was no significant difference in OS by cytogenetic subgroups. The Kaplan-Meier curves for OS by cytogenetic subgroups is shown in supplementary Appendix 9. On univariate analysis for OS (Table 3), NYHA class II/III/IV (vs class I), NT-proBNP >8500 pg/mL, hs–troponin-T ≥50 ng/L, and >2 organ involvement were identified as significant adverse prognostic factors. On multivariate analysis for OS (supplemental Appendix 10), NYHA class (II, III, or IV vs I) and NT-proBNP >8500 pg/mL (vs ≤8500 pg/mL) were significant adverse prognostic factors.

In this largest study till date, to our knowledge, on the prognostic impact of FISH abnormalities for clinical outcomes in AL amyloidosis in the era of Dara-based frontline therapies, +1q is predictive for an inferior hematologic response rate, with less than a third of patients achieving heme-CR. Furthermore, +1q is an independent adverse prognostic factor for heme-EFS. On the contrary, the adverse prognostic impact of t(11;14), as demonstrated in the bortezomib era,^8,9 has been neutralized with Dara-based frontline therapies, with no significant difference in hematologic response rate, heme-EFS, or OS in patients with vs without t(11;14). Advanced NYHA class and NT-proBNP >8500 pg/mL remained as independent adverse prognostic factors for heme-EFS and OS on multivariate analysis. Our study also reiterates the fundamental biological difference between AL amyloidosis and multiple myeloma, with the incidence of t(11;14) being approximately threefold greater and the incidence of high-risk abnormalities such as deletion(17p) or HR translocations being approximately half in the former.

Before the ANDROMEDA-AL trial, the prognostic impact of FISH cytogenetics in the context of Dara-based therapies in AL was primarily investigated in the relapsed/refractory setting. In a study from the Heidelberg University group on 168 consecutive patients with relapsed/refractory AL that received treatment with either Dara-D or Dara-VD, hyperdiploidy and +1q were associated with inferior heme-EFS with Dara-D, whereas multiple myeloma (MM) high-risk abnormalities were associated with inferior heme-EFS with Dara-VD on univariate analysis.¹⁰ However, none of the cytogenetic abnormalities were significantly associated with heme-EFS on multivariate analysis in that study. In another study on 107 consecutive patients with AL treated with Dara-based regimens from the Boston University group, +1q was identified as an independent adverse prognostic factor on multivariate analysis for both major organ deterioration–progression-free survival and OS.²² Notably, in a secondary analysis of the ANDROMEDA trial that reported on the outcomes stratified by FISH abnormalities, there was no substantial detriment in heme-CR rate among Dara-VCD–treated patients harboring +1q, with the heme-CR rate at data cutoff being 59% in this subgroup.³ Furthermore, the proportion of Dara-VCD–treated patients with +1q who received subsequent therapies (28%) was comparable with the overall Dara-VCD cohort (34%). It is important to note, however, that the median follow-up of this cohort was shorter at ∼11 months,¹ with most patients (63%) being Mayo 2004 stage I/II and a substantially lower median baseline NT-proBNP (1389 pg/mL) than our cohort (2412 pg/mL). Furthermore, the proportion of patients with +1q who had a baseline dFLC ≥18 mg/dL was substantially lower in ANDROMEDA (45%)³ than our cohort (69%), suggesting inherent differences in the patient population. Patients harboring +1q in our cohort had a higher clonal burden, as evidenced by a numerically greater proportion of patients with BMPCs ≥10%. Future studies should evaluate the immunophenotype of clonal plasma cells in AL amyloidosis to better understand the distribution of MGUS-like (monoclonal gammopathy of undetermined significance) vs intermediate vs MM-like phenotypes stratified by different cytogenetic subgroups.

Although hyperdiploidy is a favorable prognostic factor in myeloma,⁷ prior studies have shown an adverse prognostic impact in AL amyloidosis.^9,10 In a study from the Mayo Clinic, hyperdiploidy was shown to be an independent negative prognostic factor for OS in their overall cohort as well as among bortezomib-treated patients,⁹ despite no significant impact on the hematologic response rate. Furthermore, patients with hyperdiploidy had a higher baseline dFLC and BMPCs in their cohort. In line with prior data from the Heidelberg group,²³ our data also demonstrate that hyperdiploidy and +1q often coexist. Furthermore, although hyperdiploidy did have a negative prognostic impact on heme-EFS in univariate analysis, there was no independent prognostic impact on multivariate analysis. Long-term follow-up in the era of Dara-based frontline therapy will be necessary to define the impact of hyperdiploidy on OS.

Our study has limitations. First, although the overall sample size was large, the median follow-up of our cohort is short, and hence, data on OS remain immature, especially in the setting of reduced early mortality in the Dara era. Additionally, we were underpowered to analyze the prognostic impact of 1q copy number (gain vs amplification), and a larger cohort of patients with +1q, along with long follow-up, is needed to answer that question. Second, because FISH testing was performed at individual centers or outside laboratories, we could not implement a homogeneous cutoff for each probe. Furthermore, all FISH probes were not consistently checked for the entire cohort. Clinical trials in AL amyloidosis should ensure uniform FISH testing to accurately define the prognostic impact of individual cytogenetic abnormalities in the face of changing therapeutics. Future studies should also assess the frequency and prognostic impact of del(1p32) in AL amyloidosis and investigate the role of next-generation sequencing to detect clonal abnormalities (eg, TP53 mutation), especially in patients with low BMPC burden, in whom it can be challenging to perform the complete FISH panel. Third, given the lack of well-accepted criteria for hematologic progression/relapse necessitating subsequent therapies in AL amyloidosis, second-line therapy was chosen at the discretion of the treating clinician. Fourth, because the participating institutions are all referral centers for AL amyloidosis, the baseline characteristics of patients (eg, proportion with advanced cardiac disease) may not be fully representative of broader AL population. Fifth, data on minimal residual disease (MRD) was not available because it is not uniformly performed at prespecified time points in AL amyloidosis outside of clinical trials. Future studies should investigate the rates of MRD negativity and MRD kinetics in different cytogenetic subgroups because it will likely affect the rate of hematologic relapse/progression on long-term follow-up. Finally, we did not specifically exclude patients who met the diagnostic criteria for multiple myeloma at baseline. Although the overall distribution of cytogenetic abnormalities in our cohort mirrored prior AL cohorts (eg, high frequency of t[11;14]), it remains a possibility that a small proportion of patients with concomitant myeloma might have been included in the data set.

In conclusion, we have reported the most comprehensive analysis of the impact of FISH abnormalities on outcomes in the context of Dara-VCD/Dara-VD frontline therapy. Given the suboptimal hematologic response rate and heme-EFS in patients with +1q, clinical trials of novel agents such as bispecific antibodies in the frontline setting should be enriched with patients harboring +1q. On the contrary, t(11;14) no longer seems to be an adverse prognostic factor in AL in the Dara era. Furthermore, additional data on larger cohorts of patients with AL harboring HR translocations and deletion(17p) with longer follow-up should be generated to define the natural history of that group in AL amyloidosis.

The authors acknowledge the patients with AL amyloidosis and their caregivers.

M.A.G. was supported by a grant from the National Cancer Institute, National Institutes of Health Specialized Programs of Research Excellence (SPORE) in Multiple Myeloma (grant 5P50 CA186781-04).

Contribution: R.C., S.Z., and E. M. conceived the study design and performed statistical analysis. R.C. wrote the first draft of the manuscript. R.C., S.Z., U.H., A.P., A. D’Souza, G.C., P.M., and G.B. performed data extraction from electronic medical records at respective institutions. S.Z., U.H., D.B., M.A.G., A. Dispenzieri, A. D’Souza, S.K., A.P., A.C., G.C., P.M., G.P., V.S., S.O.S., S.L., and E.M. provided critical input in the study design, statistical analysis, and manuscript, and also performed diagnosis and treatment of patients included in this study. All authors read and approved the final manuscript.

Conflict-of-interest disclosure: R.C. reports consulting/advisory board fees from Janssen, Sanofi, and Adaptive Biotech. M.A.G. reports personal fees from Alnylam, Aptitude Health, Ashfield, Celgene, Ionis/Akcea, Janssen, Johnson & Johnson, Juno, Physicians Education Resource, Prothena, Research To Practice, Sanofi, and Sorrento; personal fees for data safety monitoring board from AbbVie; and payment for development of educational materials for i3Health and educational program development for i3Health. V.S. reports consultant and adviser fees and research funding from Caleum; is a consultant for Attralus and Pfizer; received research funding from Celgene, Karyopharm, Millennium-Takeda, Oncopeptides, Prothena, and Sorrento; received research funding from and is an adviser for Janssen; and is an adviser for AbbVie, Proclara, Protego, PharmaTrace, Prothena, Regeneron, and Telix. S.O.S. reports consultant/adviser fees, travel grant, honoraria, and research funding from Janssen and Prothena; research funding from Sanofi; honoraria from Pfizer and Takeda; adviser fees from Telix; and travel grants from Binding Site, Celgene, and Jazz. G.P. reports consultant/adviser fees and honoraria from Janssen, Protego, and Zentalis; and honoraria from Pfizer, Sebia, Siemens, and The Binding Site. S.L. reports consultant and/or adviser fees from Adaptive Biotechnologies, Alexion Therapeutics, Bristol-Myers Squibb (BMS), Caelum Biosciences, Janssen Pharmaceuticals, Karyopharm Therapeutics, Oncopeptides AB, GlaxoSmithKline (GSK), AbbVie, Janssen, Pfizer, and Takeda Pharmaceutical; research funding from Celgene Inc, Sanofi, and Zentali; honoraria from Clinical Care Options and Regeneron Pharmaceuticals; royalties/patents with Caelum Biosciences; and, additionally, has a patent CAEL-101 with royalties paid to Columbia University. A.C. reports consulting fees from BMS and Adaptive; and research funding from Adaptive Biotechnologies, Harpoon, Nektar, BMS, Janssen, Sanofi, and AbbVie. S.K. reports research funding from Celgene, Adaptive, Sanofi, AbbVie, Takeda, Janssen, Kite, Merck, MedImmune/AstraZeneca, Novartis, and Roche; advisory board membership with AbbVie, Celgene, Janssen, Takeda, Adaptive, Kite, and MedImmune/AstraZeneca; and membership on independent review committee for Oncopeptides. A. D'Souza reports institutional research funding from AbbVie, Caelum, Janssen, Novartis, Prothena, Sanofi, Takeda, and TeneoBio; advisory board fees from BMS, Pfizer, and Janssen; and consulting fees from Prothena and Janssen. E.M. reports consulting for Protego. P.M. reports honoraria for lectures from Pfizer; honoraria for lectures and advisory board fees from Jannsen-Cilag; and advisory board fees from Siemens. The remaining authors declare no competing financial interests.

Correspondence: Rajshekhar Chakraborty, Department of Medicine, Columbia University Medical Center, 168 Fort Washington Ave, New York, NY 10032; email: rc3360@cumc.columbia.edu.