PFS is dependent on BM immune cell composition. (A) Kaplan-Meier plot shows PFS according to high vs low GLR (1-year PFS rate, 80% vs 50%, respectively; (hazard ration [HR], 0.28; 95% confidence interval [CI], 0.12-0.65); log-rank test, P < .01). ROC curve is used to determine the optimal GLR cutoff for predicting disease progression. The AUC was 0.649 (95% CI, 0.493-0.806). (B) Forest plot shows the results of multivariate Cox regression analysis including GLR, age (<65 vs >65 years), sex, ISS, LDH levels, and ASCT. (C) Kaplan-Meier plot shows PFS according to high vs low GTL ratio (1-year PFS rate, 80% vs 52%, respectively; HR, 0.29 [95% CI, 0.12-0.67]; log-rank test, P < .01) for patients with MM. ROC curve is used to determine the optimal GTL ratio cutoff for predicting disease progression. The AUC was 0.645 (95% CI, 0.489-0.802). (D) Forest plot shows the results of multivariate Cox regression analysis including GTL ratio, age (<65 vs >65 years), sex, ISS, LDH levels, and ASCT. (E) Patients with MM were divided into 2 groups according to the first-line treatment that they received: those who received daratumumab and those who did not. (F-G) Kaplan-Meier plots show PFS according to high vs low GTL ratio and the first-line therapy regimen, both in the daratumumab-treated group (F) (1-year PFS rate: low ratio/with daratumumab, 45%; high ratio/with daratumumab, 82%; HR, 4.26 [95% CI, 1.57-11.5]; log-rank test, P < .01) and daratumumab-untreated group (G) (1-year PFS rate: low ratio/without daratumumab, 77%; high ratio/without daratumumab, 67%; HR, 2.78 [95% CI, 0.50-15.5]; log-rank test, ns). All graphs depict results derived from PCST tube samples. AUC, area under the curve; ns, not significant; RD, lenalidomide-dexamethasone; VMP, bortezomib-melphalan-prednisone; VTd bortezomib-thalidomide-dexamethasone.