Nonselective β-Adrenergic Receptor Inhibitors Impair Hematopoietic Regeneration in Mice and Humans after Hematopoietic Cell Transplants

Abstract

Peripheral nerves promote mouse bone marrow regeneration by activating β2- and β3-adrenergic receptor signaling, raising the possibility that nonselective β-blockers could inhibit engraftment after hematopoietic cell transplants (HCT). We observed no effect of β-blockers on steady-state mouse hematopoiesis. However, mice treated with a nonselective β-blocker (carvedilol), but not a β1-selective inhibitor (metoprolol), exhibited impaired hematopoietic regeneration after syngeneic or allogeneic HCTs. At two institutions, patients who received nonselective, but not β1-selective, β-blockers after allogeneic HCT exhibited delayed platelet engraftment and reduced survival. This was particularly observed in patients who received posttransplant chemotherapy for graft-versus-host disease prophylaxis, which also accentuated the inhibitory effect of carvedilol on engraftment in mice. In patients who received autologous HCTs, nonselective β-blockers were associated with little or no delay in engraftment. The inhibitory effect of nonselective β-blockers after allogeneic HCT was overcome by transplanting larger doses of hematopoietic cells. Significance: Patients who receive allogeneic HCTs followed by posttransplant chemotherapy for graft-versus-host disease prophylaxis may be at risk of delayed engraftment and increased mortality if administered nonselective β-blockers after transplantation. Transient discontinuation of nonselective β-blockers or transitioning to β1-selective inhibitors after HCT may accelerate engraftment and improve clinical outcomes. See related commentary by Bhatia, p. 666.

Figure 1.

Carvedilol treatment impairs hematopoietic regeneration in mice after syngeneic and allogeneic BM transplants. Mice were treated with nonselective β-blocker, carvedilol, or vehicle control for 7 days before BM transplantation and 21 days after transplantation. Each panel shows data from four independent experiments, and each dot represents a different mouse. All data represent mean ± SD. A–L, 6 × 10⁵ C57BL/Ka BM cells were syngeneically transplanted into irradiated C57BL/Ka-Thy-1.2 recipients. At 21 days after transplantation, we analyzed WBC (A), RBC (B), and PLT (C) counts; BM (D) and spleen (E) cellularity; and the frequencies of HSCs, MPPs, and LSK cells in the BM (F–H) and spleen (I–K) of carvedilol-treated (red, n = 19) and vehicle-treated (black, n = 17) mice. L, Survival of carvedilol-treated (red, n = 44) and vehicle-treated (black, n = 18) mice over time after syngeneic transplantation. M–X, 6 × 10⁵ T cell–depleted LP/J BM cells were allogeneically transplanted into irradiated C57BL/Ka-Thy-1.2 recipients. At 21 days after transplantation, we analyzed WBC (M), RBC (N), and PLT (O) counts, as well as BM (P) and spleen (Q) cellularity and the frequencies of HSCs, MPPs, and LSK cells in the BM (R–T) and spleen (U–W) of carvedilol-treated (red, n = 17) and vehicle-treated (black, n = 18) mice. X, Survival of carvedilol-treated (red, n = 44) and vehicle-treated (black, n = 20) mice over time after allogeneic transplantation. Y and Z, The number of CD41⁺ megakaryocytes (Y) or Gr1⁺ myeloid cells (Z) per field of view was counted in BM sections from carvedilol-treated (red, n = 10) and vehicle-treated (black, n = 10) mice at 13 days after transplantation. The statistical significance of differences among treatments were assessed using the Student t tests followed by Holm–Sidak multiple comparisons adjustments (A–C and M–T), matched samples two-way ANOVAs followed by Holm–Sidak multiple comparisons adjustments (D, E, I–K, Y, and Z), Student or Welch’s t tests followed by Holm–Sidak multiple comparisons adjustments (F–H), Mann–Whitney tests followed by Holm–Sidak multiple comparisons adjustments (U–W), or log-rank Mantel–Cox tests (L and X). All statistical tests were two sided.

Figure 2.

Nonselective β-adrenergic receptor inhibitors impair hematopoietic regeneration in patients after autologous and allogeneic transplantation. A, Propensity matching for age in recipients who received autologous transplant at UTSW who were (n = 73), or were not (n = 798), treated with nonselective β-blockers. The time to neutrophil (B) and platelet (C) engraftment in matched groups (n = 73 and 219 after 1:3 matching) is shown. D, Propensity matching for age in patients who received autologous transplant at UTSW who were (n = 123), or were not (n = 675), treated with β1-selective inhibitors. The time to neutrophil (E) and platelet (F) engraftment in matched groups (n = 120 and 360 after 1:3 matching). Propensity matching for age (G) and conditioning regimen (H) in patients who received allogeneic transplant at UTSW who were (n = 14), or were not (n = 297), treated with nonselective β-adrenergic receptor inhibitors (RIC, NMA, vs. myeloablative conditioning). The time to neutrophil (I) and platelet (J) engraftment in matched groups (n = 14 and 42 after 1:3 matching). Propensity matching for age (K) and conditioning regimen (L) in patients receiving allogeneic transplant at UTSW who were treated with β1-selective inhibitors (n = 23) vs. no β-blocker (n = 274). The time to neutrophil (M) and platelet (N) engraftment in matched (n = 16 and 48 after 1:3 matching). The statistical significance of differences among groups was assessed using Student t tests. All data represent mean ± SD.

Figure 3.

The use of nonselective β-blockers is an independent risk factor for delayed hematopoietic regeneration after allogeneic transplantation. The risk factor coefficients (B) for increased time to neutrophil (A) and platelet (B) engraftment among patients undergoing autologous HCT at UTSW, neutrophil (C) and PLT (D) engraftment among patients undergoing allogeneic HCT at UTSW, neutrophil (E) and PLT (F) engraftment among patients undergoing autologous HCT at Vanderbilt, and neutrophil (G) and PLT (H) engraftment among patients undergoing allogeneic HCT at Vanderbilt. A generalized linear model was used, with covariates including age, β1-selective inhibitor use, and nonselective β-blocker use for all cohorts. For patients receiving allogeneic HCT, other covariates included conditioning regimen [myeloablative conditioning (MAC) or RIC/NMA]; underlying disease [for UTSW lymphoid or myeloid, for Vanderbilt acute myeloid leukemia (AML), chronic myeloid leukemia (CML), myeloproliferative neoplasm (MPN), or myelodysplastic syndrome (MDS)]; cell dose (CD34⁺ cells/kg); cell source [BM or peripheral blood (PB)]; donor matching [matched related/matched unrelated donor (MRD/MUD), mismatched unrelated donor (MMUD), or haploidentical (Haplo)]; acute GvHD; and myelosuppressive GvHD treatment [methotrexate (MTX) or posttransplant Cytoxan (PTCy)]. Patients at UTSW also included splenomegaly and CMV infection as covariates, and patients at Vanderbilt included CMV serostatus (low risk = donor−/recipient−; intermediate risk = donor+/recipient+ or −; high risk = donor−/recipient+). B reflects the number of additional days required for neutrophil or PLT engraftment per unit of each predictive variable, with units being per year for age and binary (yes/no) for all other variables. B ± 95% confidence interval is shown. The dashed vertical line represents B of 0.

Figure 4.

Nonselective β-blockers delay engraftment particularly when limiting numbers of cells are transplanted. A, Cell doses administered (CD34⁺ cells/kg) to patients receiving allogeneic HCT at UTSW and Vanderbilt. B and C, Correlation between cell dose administered and the time to PLT engraftment in patients who were (B), or were not (C), treated with nonselective β-blockers. D and E, Correlation between cell dose administered and the time to neutrophil engraftment in patients who were (D), or were not (E), treated with nonselective β-blockers. F and G, Correlation between cell dose administered and the time to PLT engraftment in patients who were treated with β1-selective inhibitors (F) vs. those not on any β-blocker (G). H and I, Correlation between cell dose administered and the time to neutrophil engraftment in patients administered β1-selective inhibitors (H) vs. those not on any β-blocker (I). Statistical significance was assessed using Student t tests (A) or bivariate analyses using Pearson correlation (B–I).

Figure 5.

The inhibitory effect of carvedilol on hematopoietic regeneration after allogeneic transplantation can be overcome by transplanting larger doses of BM cells. Mice were treated with carvedilol or vehicle for 7 days before and 21 days after transplantation. We transplanted 3 × 10⁵, 6 × 10⁵, 12 × 10⁵, or 24 × 10⁵ T cell–depleted LP/J BM cells into irradiated C57BL/Ka recipients. A–K, A total of eight to 11 recipients per treatment from three independent experiments. Each dot represents a different mouse. All data represent mean ± SD. A–C, WBC (A), RBC (B), and PLT (C) counts from carvedilol-treated (right) or vehicle-treated (left) mice 21 days after transplantation. D and E, Total BM (D) and spleen (E) cellularity. F–K, The frequencies of HSCs, MPPs, and LSK cells in the BM (F–H) and spleen (I–K). L, Survival of carvedilol-treated and vehicle-treated mice after transplantation. The numbers of mice per treatment are shown in each panel. The statistical significance of differences among treatments was assessed using two-way ANOVAs followed by Sidak multiple comparison adjustments (A–F, H, I, and K), Mann–Whitney tests followed by Holm–Sidak multiple comparisons adjustments (G and J), or log-rank Mantel–Cox tests followed by Holm–Sidak multiple comparisons adjustment (L). All statistical tests were two sided.

Figure 6.

Posttransplant chemotherapy increased the inhibitory effect of nonselective β-blockers on engraftment after allogeneic transplantation in humans and mice. The risk factor coefficients (B) for time to PLT engraftment among patients at Vanderbilt who received allogeneic HCTs and who did not (A) or did (B) receive myelosuppressive GvHD prophylaxis with methotrexate or posttransplant cyclophosphamide. A generalized linear model was used, with covariates including age; β1-selective inhibitor use; nonselective β-blocker use; conditioning regimen [myeloablative conditioning (MAC) or RIC/NMA]; underlying disease [acute myeloid leukemia (AML), chronic myeloid leukemia (CML), myeloproliferative neoplasm (MPN), or myelodysplastic syndrome (MDS)]; cell dose (CD34⁺ cells/kg); cell source [BM or peripheral blood (PB)]; donor matching [matched related/matched unrelated donor (MRD/MUD), mismatched unrelated donor (MMUD), or haploidentical (Haplo)]; acute GvHD (aGvGD); and CMV serostatus (low risk = donor−/recipient−, intermediate risk = donor+/recipient+ or −, high risk = donor−/recipient+). B reflects the number of additional days required for neutrophil or PLT engraftment per unit of each predictive variable, with units being per year for age and binary (yes/no) for all other variables. B ± 95% confidence interval is shown. The dashed vertical line represents B of 0. C–N, Mice were treated with carvedilol or vehicle for 7 days before and 21 days after allogeneic transplantation of 6 × 10⁵ T cell–depleted LP/J BM cells into irradiated C57BL/Ka recipients. Cyclophosphamide (25 mg/kg body mass) or vehicle (PBS) were injected intraperitoneally at 3 and 4 days after transplantation. Each panel shows data from three independent experiments. C–M, A total of 10 to 13 recipients per treatment. Each dot represents a different mouse. All data represent mean ± SD. C–E, WBC (C), RBC (D), and PLT (E) counts at 21 days after transplantation. F and G, Total BM (F) and spleen (G) cellularity. H–M, The frequencies of HSCs, MPPs, and LSK cells in the BM (H–J) and spleen (K–M). N, Survival of carvedilol-treated and vehicle-treated mice with or without posttransplant cyclophosphamide treatment over time after transplantation. Data are from a total of 15 to 48 mice per condition in three independent experiments. The statistical significance of differences among treatments was assessed with one-way ANOVA followed by Sidak multiple comparisons adjustments (C–K), Welch’s one-way ANOVA followed by Dunnett T3 multiple comparison adjustments (L and M), or log-rank Mantel–Cox tests followed by Holm–Sidak multiple comparisons adjustment (N). All statistical tests were two sided.

Figure 7.

The use of nonselective β-blockers after transplantation is associated with worse clinical outcomes. A and B, Kaplan–Meier estimates of OS in patients receiving allogeneic HCT at UTSW who were, or were not, taking nonselective β-blockers (A) or β1-selective inhibitors vs. no β-blocker (B). One- and two-year OS (in years) is shown. C, The HR for OS in patients receiving allogeneic HCT at UTSW was estimated using a Cox proportional hazards model. D and E, Kaplan–Meier estimates of nonrelapse mortality (NRM) in patients receiving allogeneic HCT at UTSW who were, or were not, taking nonselective β-blockers (D) or β1-selective inhibitors vs. no β-blocker (E). F, The HR for NRM in patients receiving allogeneic HCT at UTSW was estimated using a Cox proportional hazards model. G and H, Kaplan–Meier estimates of OS in patients receiving allogeneic HCT at Vanderbilt who were, or were not, taking nonselective β-blockers (G) or β1-selective inhibitors vs. no β-blocker (H). I, The HR for OS in patients receiving allogeneic HCT at Vanderbilt was estimated as in C. J and K, Kaplan–Meier estimates of NRM in patients receiving allogeneic HCT at Vanderbilt who were, or were not, taking nonselective β-blockers (J) or β1-selective inhibitors vs. no β-blocker (K). L, The HR for NRM in patients receiving allogeneic HCT at Vanderbilt was estimated as in F. OS and NRM were compared between groups using a log-rank test (A, B, D, E, G, H, J, and K). For the Cox proportional hazards models (C, F, I, and L), covariates included age, conditioning regimen, nonselective β-blocker use, and β1-selective inhibitor use. The HR ± 95% confidence interval is shown, and the dashed vertical line represents a HR of 1.0.