Marinac, Catherine R.

Abstract Accurate prediction of risk of progression from smoldering (SMM) to active multiple myeloma (MM) is paramount to individualized early therapeutic strategies with minimum risk of overtreatment. Current risk stratification models do not account for evolving biomarker trajectories. We assembled the largest cohort to date of 2,270 SMM patients from six international centers with longitudinal clinical and biological data to train and validate the PANGEA 2.0 risk models. Four evolving biomarkers were significantly associated with shorter time-to-progression: M-protein increase ≥0.2g/dL, involved:uninvolved serum free light chain ratio increase ≥20, creatinine increase >25%, and hemoglobin decrease ≥1.5g/dL. PANGEA 2.0 outperforms established models including the 20/2/20 and IMWG models by more accurately predicting progression (C-statistics=0.69–0.84), even without biomarker history (C-statistics=0.69–0.83) or recent bone marrow biopsy. PANGEA 2.0 is an easy-to-use, open-access tool (https://ghobrial.shinyapps.io/pangea_2_calculator) to improve and individualize SMM risk stratification. Validation tools are available to compare PANGEA 2.0 to established models (https://ghobrial.shinyapps.io/pangea_validation).

Abstract Blood-based multicancer early detection (MCED) tests represent a new approach for cancer detection. To gain insights into the utility of various approaches of evaluating a cancer signal detected (CSD) test result, we evaluated diagnostic journeys of a subset of 6,662 participants in PATHFINDER who had a CSD on both an initial and refined version of an MCED test that also provided a prediction of cancer signal origin (CSO). We sought to determine whether CSO prediction–guided diagnostic evaluations led to diagnostic resolution; whether participants with known risk factors for cancer beyond age alone and a negative initial diagnostic evaluation had a cancer diagnosis during study follow-up; the utility of whole-body imaging in reaching diagnostic resolution; and differences in the diagnostic journeys needed to reach diagnostic resolution for both true- and false-positive results. Of the 39 participants in this analysis, 82% (32/39) achieved diagnostic resolution after the initial evaluation, including 78% (25/32) who reached resolution specifically with a CSO prediction–directed workup. Eighteen percent (7/39) required additional evaluation for persistent clinical suspicion of cancer, all of whom achieved resolution (3 with and 4 without cancer). Whole-body imaging contributed to diagnostic resolution in only 49% of CSD cases. Approximately 90% of true- and false-positive cases had imaging tests; more true positives versus false positives (81.0% vs. 38.9%) had nonsurgical and/or surgical procedures. In conclusion, CSO prediction–directed evaluations enabled diagnostic resolution for most participants, although some with negative initial evaluations but persistent suspicion of cancer required additional testing. Prevention Relevance: MCED testing has the potential to increase detection of cancer at earlier stages. As MCED testing is a new technology, there are few data on the diagnostic journeys patients undergo following testing. We observed that for most patients, CSO prediction–directed workups were efficient, leading to diagnostic resolution after initial evaluation.

Abstract Introduction: With over 80, 000 new cases of non-Hodgkin's lymphoma (NHL) diagnosed annually, modifiable risk factors remain unclear. Circadian disruption, linked to cancers like breast and prostate, affects immune cells (e.g., natural killer cells and T-helper cells) by altering their trafficking and proliferation. Some studies (PMID: 30566672, 27611440) suggest long sleep may increase overall NHL risk, including a Mendelian randomization (MR) study (PMID: 32895918). However, no research has examined the causal relationship between sleep traits, including long sleep, and NHL risk, by subtypes. We hypothesize that sleep traits are causally associated with NHL subtypes in European populations, testing this using MR with data from the International Lymphoma Epidemiology Consortium (InterLymph). Methods: We conducted two-sample MR using genome-wide association studies (GWAS) summary data for sleep traits and NHL subtypes. Only independent, genome-wide significant (p < 5×10-8) Single Nucleotide Polymorphisms (SNPs) were selected as valid instruments, sourced from the UK Biobank. Sleep traits included chronotype (153 SNPs), insomnia (48 SNPs), sleep duration (78 SNPs), excessive daytime sleepiness (37 SNPs), short sleep (27 SNPs), and long sleep (8 SNPs). NHL subtype data were from InterLymph: follicular lymphoma (6, 508 cases / 64, 183 controls), mantle cell lymphoma (1, 169 cases / 61, 603 controls), Waldenstrom macroglobulinemia/lymphoplasmacytic lymphoma (WM/LPL: 1, 697 cases / 59, 333 controls), and chronic lymphocytic leukemia (8, 522 cases / 67, 653 controls). The primary analysis used inverse-variance weighted (IVW) random-effects, with MR-Egger for pleiotropy adjustment. Results: Using the IVW method, we did not identify any statistically significant causal association between sleep traits and the risk of NHL subtypes, but did yield several intriguing trends. Long sleep (sleep duration >= 9 h per night) showed a marginally increased NHL risk [Odds ratio (OR) range: 1.51 - 247.9] across each subtype. Short sleep (sleep duration < 7 h per night) showed a similar trend [OR range: 1.25 - 1.81], except for WM/LPL, where we noticed a negative trend. The confidence limit for long sleep was imprecise due to the small number of variants. We did not notice any significant directional horizontal pleiotropy or weak instrument bias. Conclusions: We found no conclusive evidence of a relationship between sleep traits and four NHL subtypes. However, we see suggestive trends of long and short sleep associated with higher risk of some NHL subtypes. These findings align with the previous studies, which showed long sleep was associated with higher risk of NHL overall. Future directions include expanding our analysis to additional NHL subtypes, generating polygenic risk scores, and conducting stratified analyses. Citation Format: Pankhil Shah, Brittany Crawford, Anwar Merchant, Brenda Birmann, Angelica Macauda, Michelle A. Hildebrandt, Aaron Norman, Neil E. Caporaso, Meredith Yeager, Michael Dean, Immaculata De Vivo, Lynn Goldin, Nicola J. Camp, Rosalie Griffin, Delphine Casabonne, Federico Canzian, Pelin Unal, Elad Ziv, Catherine R. Marinac, Alexandra Nieters, Stephen J. Chanock, Mitchell J. Machiela, Michael Conry, Charlie Zhong, Hanla A. Park, Simon Cheah, Jonathan N. Hofmann, Elizabeth E. Brown, Celine Vachon, Susan Slager, Sonja Berndt, Sophia S. Wang, Vijai Joseph, James McKay, Henrik Hjalgrim, Lara Sucheston-Campbell, Karl Smith-Byrne, Alyssa Clay-Gilmour. Association between sleep traits and risk of non-Hodgkin's lymphoma subtypes: a mendelian randomization study in the International Lymphoma Epidemiology (InterLymph) Consortium [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 7415.

Introduction Multiple Myeloma (MM) precursors Monoclonal Gammopathy of Undetermined Significance (MGUS) and Smoldering Multiple Myeloma (SMM) have variable risk of progression to active MM, and identifying which patients may progress remains a clinical challenge. Deep proteome profiling of peripheral blood (PB) plasma may advance non-invasive precursor disease staging, monitoring and characterization. Here, we performed plasma proteomic profiling on patients across the MM disease continuum as well as progressive and stable disease to provide insights into MM disease biology and identify protein-based high-risk disease features that may improve prognostication. Methods We performed high-throughput profiling of >5400 proteins using the Olink® Explore HT library and Proximity Extension Assay (PEA) technology. We profiled 529 PB plasma samples from 485 individuals, including MGUS (n=100), SMM (n=203), MM (n=100), and healthy donors (HDs) (n=82). Sequential samples from SMM-MM progressors (n=32) and stable SMM-SMM non-progressors (n=32) were also profiled, where precursor samples ranged 1.04-6.91 years (median 2.33 years) prior to MM progression. T-tests, ANOVAs, and linear mixed effect (LME) models were used to identify significant proteins across disease stages, progression status, and time. Results were adjusted for multiple testing using Benjamini-Hochberg procedure. A subset of 86 individuals (13 HDs, 14 MGUS, 41 SMM, 18 MM) also had single-cell RNA sequencing (scRNA-seq) performed on tumor and immune cells from paired PB/BM collected at the same timepoint to enable cellular mapping of signals detected in plasma. Results We successfully captured significantly dysregulated proteins including proteins highly expressed on the surface of plasma cells, such as BCMA, SLAMF7 and CD38, highlighting the utility of PEA technology to monitor soluble levels of clinically relevant targets. Given SMM patients are clinically heterogeneous and can resemble MGUS or MM, we aimed to improve discrimination of disease states using biological features obtained from the plasma proteome. We developed a classifier using an Elastic Net model for each stage, where plasma proteins such as SLAMF7, BCMA, TACI, FCRL5 and others had the highest importance scores for the model. We demonstrated 97% SMM/MM samples could be identified from healthy samples (AUC=0.82), indicating our classifier could reliably screen disease-related cases. Moreover, 84% of SMM cases were correctly classified as clinically defined SMM, while misclassified samples were labelled as MM cases. Interestingly, the misclassified patients had consistently rising M-spike levels during clinical follow-up but without a shift in clinical stage or risk status, suggesting the plasma proteome may provide earlier indication of evolving disease. Since proteomic information may also hold value for improved risk stratification, we evaluated protein levels in precursor stage timepoint samples from SMM-MM progressors and stable SMM non-progressors. We identified a prognostic five-protein signature that was significantly elevated in SMM-MM progressors, of which BCMA and TACI were top proteins, as well as proteins vital for calcium homeostasis and integrin-mediated cell adhesion. BCMA and TACI levels also had a strong positive correlation with BM plasma cell infiltration, which suggests these proteins may be useful surrogates of BM tumor burden during routine blood-based assessment of precursor patients. Lastly, integrative analysis of scRNA-seq of tumor and immune cells was used to elucidate cell origin level information of the prognostic signature proteins. Four proteins were highly expressed in malignant versus non-malignant plasma cells, suggesting our prognostic signature partially provides a readout of plasma cell biology. Meanwhile, one protein was constitutively expressed on most leukocytes, suggesting that the interplay of other cell types in the immune microenvironment involved in progression may also be reflected in the plasma. Conclusion We performed the first comprehensive analysis of the plasma proteome of MM and its precursor conditions. Overall, we developed a classifier that utilizes plasma proteins alone to accurately classify disease stages and identified a prognostic protein signature associated with progressive disease.

Introduction: Over 80,000 new cases of Non-Hodgkin's Lymphoma (NHL) are diagnosed annually, but modifiable risk factors are largely unknown. Circadian disruption has been identified as a risk factor for several cancers, including breast and prostate cancer, and affects immune cells such as Natural Killer (NK) cells and T-helper cells. Disruptions in circadian rhythms can lead to aberrant immune cell trafficking and proliferation. Despite this biological connection, evidence linking circadian disruptions to NHL remains inconclusive. Only single Mendelian Randomization (MR) study (PMID: 32895918) has explored the causal relationship between sleep traits and overall NHL risk, without examining NHL subtypes. Studies have shown distinct risk factor profiles among NHL subtypes. Therefore, this study aims to investigate the causal relationship between sleep traits and NHL subtypes in European populations using MR. Methods: We conducted a two-sample MR analyses using genome-wide association studies (GWAS) summary data on sleep traits and NHL subtypes. The GWAS summary statistics on sleep traits were obtained from publicly available UKBiobank datasets. The sleep traits analyzed included chronotype (N=449,734), insomnia (N=237,627), sleep duration (N=446,118), daytime sleepiness (N=452,071), short sleep (N=411,934), and long sleep (N=339,926). Only independent, genome-wide significant SNPs (p < 5×10-8) were considered as valid instruments. The NHL subtype-specific summary statistics came from the InterLymph including follicular lymphoma (FL: N=6,508 cases / 64,183 controls), mantle cell lymphoma (MCL: N=1,169 cases / 61,603 controls), and Waldenström macroglobulinemia/lymphoplasmacytic lymphoma (WM/LPL: N=1,697 cases / 59,333 controls). Our primary analysis used the random-effects inverse-variance weighted (IVW) method. In addition, we employed MR-Egger to adjust for directional pleiotropy. Heterogeneity in the MR results was evaluated with Cochran's Q statistic. Results: The IVW method did not identify any statistically significant causal association between sleep traits and the risk of NHL subtypes (FL,MCL,WM/LPL), but did yield several intriguing associations. Chronotype (early rising) showed a tendency towards a marginally an increased NHL risk across subtypes [Odds ratio (OR) range: 1.07 - 2.56]. Conversely, a decreased risk was noted for long sleep (≥ 9 hours vs. 7-8 hours) [OR range: 0.19 - 0.62], insomnia [OR range: 0.72 - 0.88], and short sleep (excludes MCL) [OR range: 0.12 - 0.72]. Confidence limit for long sleep was wide due to small number of variants (N=8). The MR-Egger regression analysis did not provide evidence of significant directional horizontal pleiotropy. The F-statistics for all sleep traits were greater than 10 indicating absences of weak instrument bias. Conclusions: Our study found no statistically significant associations between sleep traits and three NHL subtypes (FL, MCL, WM/LPL) in European populations. However, we observed potential influences of chronotype, long and short sleep, and insomnia on NHL risk. These findings align with and extend previous MR study by providing subtype-specific insights. Future directions include expanding our analysis to additional NHL subtypes, generating polygenic risk scores, and conducting stratified analyses using the InterLymph population. These efforts aim to further elucidate the complex relationship between sleep characteristics and NHL risk across diverse subtypes and populations.

Guideline-recommended screening programs exist for only a few cancer types. Although all these programs are understood to lead to reductions in cancer-related mortality, standard-of-care screening tests vary in accuracy, adherence and effectiveness. Recent advances in high-throughput technologies and machine learning have facilitated the development of blood-based multi-cancer cancer early detection (MCED) tests. MCED tests are positioned to be complementary to standard-of-care screening and they may broaden screening availability, especially for individuals who are not adherent with current screening programs and for individuals who may harbor cancers with no available screening options. In this article, we outline some key features that should be considered for study design and MCED test development, provide an example of the developmental pathway undertaken for an emerging multi-biomarker class MCED test and propose a clinical algorithm for an imaging-based diagnostic resolution strategy following MCED testing.

10608 Background: Cancer is one of the most feared diseases. Multi-cancer early detection (MCED) tests have the potential to expand the number of cancers that are detected at earlier stages, which could result in better clinical outcomes. However, behavioral impacts based on MCED test results could affect the utility and value of these tests. Methods: Two independent data sources evaluated the impact of MCED test results on behavioral intentions and perceptions: a survey of the general population and PRO in an MCED clinical trial. A web-based, 30-minute survey of the US general population was used to ascertain attitudes toward a hypothetical MCED test. Eligible respondents were adults aged 40-80 years without a cancer diagnosis in the previous 5 years. The survey assessed the perceived value of negative results (ie, no cancer signal detected [NCSD]) and the projected impact on post-test behaviors. The prospective PATHFINDER study (NCT04241796) of adults aged ≥50 years who underwent MCED cancer screening included PRO assessments. Post-test assessments included: "I am feeling relieved about my test result" from the adapted MICRA and “How likely are you to follow your healthcare provider’s cancer screening recommendations.” The consistency of sentiment after an NCSD MCED result from the online survey and clinical study was assessed. Results: Of the 1073 online survey respondents, 53% were female, 77% White, 14% Black/African American, and 13% Hispanic. Mean age was 58 years. The most common anticipated emotional impacts were relief (84%), happiness (72%), and confidence (69%). Respondents (66-70%) reported they would expect less worry, fear, stress, and anxiety after an NCSD MCED result. The majority (73%) reported they would maintain or improve health behaviors after an NCSD result. NCSD results were projected to lead to an increase in health-related behaviors, including adherence to recommended cancer screening (98%; 68% increase, 30% maintain current level) and regular doctor visits (46%). Of the 6529 PATHFINDER study participants who received NCSD results, 64% were female, 92% White. Mean age was 63 years. 80% of participants (n = 4685/5865) with NCSD felt relief "sometimes" or "often" after their MCED test result. Over 95% of participants (n = 5637/5863) with NCSD indicated they were "likely" or "very likely" to adhere to cancer screening recommendations in the future. Conclusions: Results from a general population survey and a prospective clinical study indicate that an NCSD result may be associated with positive cancer screening intentions and other preventive healthcare behaviors. An NCSD result may also positively impact psychological and emotional health. While these findings suggest that an NCSD result may provide additional value beyond traditional clinical benefits associated with cancer screening, future research is needed to delineate the behavioral impact following MCED screening. Clinical trial information: NCT04241796 .

Summary: While the current approach to precursor hematologic conditions is to “watch and wait,” this may change with the development of therapies that are safe and extend survival or delay the onset of symptomatic disease. The goal of future therapies in precursor hematologic conditions is to improve survival and prevent or delay the development of symptomatic disease while maximizing safety. Clinical trial considerations in this field include identifying an appropriate at-risk population, safety assessments, dose selection, primary and secondary trial endpoints including surrogate endpoints, control arms, and quality-of-life metrics, all of which may enable more precise benefit–risk assessment.

10520 Background: PATHFINDER (NCT04241796) was a prospective, interventional return of results study that enrolled 6,662 participants (pts) from 7 US institutions to assess clinical implementation of a blood-based multi-cancer early detection (MCED) test. Test results for those with a cancer signal detected also included a cancer signal origin prediction designed to help guide diagnostic evaluations. We analyzed whether the diagnostic evaluations in PATHFINDER were appropriately directed by cancer signal origin predictions, assessed the residual risk of cancer after negative initial evaluation, and examined the role of whole-body imaging (WBI, defined as PET/CT or CT chest/abdomen/pelvis) in diagnostic evaluation. Methods: Analysis included pts with a cancer signal detected result from the two different versions of an MCED test evaluated in PATHFINDER. Prespecified criteria blinded to study results were used to define expected diagnostic evaluations by cancer signal origin prediction. These criteria were then used to judge concordance between cancer signal origin predictions and the diagnostic evaluation actually performed. Definitive cancer or no cancer diagnosis (cancer status at end of study [EOS]) was assessed at 12 months after enrollment (EOS). Results: 39 pts had cancer signal detected results by both test versions. Diagnostic evaluations were consistent with cancer signal origin in 30/39 (77%) cases. Diagnostic resolution was achieved in 32/39 (82%) pts after initial evaluation, while 7/39 (18%) required additional workup. 25 of the 32 (78%) pts who achieved diagnostic resolution after initial evaluation had a cancer signal origin-guided workup. Additional evaluation of the 7 pts not achieving initial diagnostic resolution was based on persistent clinical suspicion of cancer due to prior cancer history (n = 1) or abnormal/equivocal findings on initial evaluation (n = 6). Additional workup led to diagnostic resolution in all 7 cases (3 with cancer and 4 with no cancer confirmed at EOS). WBI was ordered in 27 cases, of whom 9 had a prior cancer history, 10 had non-localizing cancer signal origin (hematologic or indeterminate), and 1 had both. WBI was equivocal or had no impact on diagnostic resolution in 8/27 (30%) cases. Conclusions: Most cases resulted in diagnostic resolution after cancer signal origin-directed diagnostic workup. In pts with a cancer history and negative initial evaluations, and those with equivocal initial evaluations, there was still a risk of cancer, suggesting the need for additional evaluation. WBI was used mostly in those with a cancer history or non-localizing cancer signal origins, and had no impact on diagnosis in almost one-third of pts. Clinical trial information: NCT04241796 .