Pre-Transplant Autonomic Nervous System Malfunction Is a Powerful Predictor for Survival after Allogeneic Hematopoietic Cell Transplantation

Background: Autonomic nerve function can be affected by various factors such as patient activity of daily living, disease status, treatment history and/or comorbidity. Indeed, autonomic nervous system dysfunction, as indexed by heart rate variability (HRV), has shown prognostic value for various disease outcomes, but its influence on outcomes following allogeneic hematopoietic cell transplantation (allo-HCT) is unclear. We therefore conducted a prospective observational study to determine the impact of autonomic nervous system dysfunction on outcomes following allo-HCT.

Methods: From July 2011 to July 2013, we prospectively registered patients who had received allo-HCT in our institute. We also applied well-established examination of HRV as a surrogate maker of the autonomic nervous system. HRV data were collected from 24-hour ambulatory ECG recordings within 28 days of the starting day of the conditioning regimen for allo-HCT. Obtained data were analyzed with RR data analysis software (MemCalc/CHIRAM version 3, GMS Co., LTD., Tokyo, Japan). The following parameters were employed as markers of HRV: standard deviation of NN interval (SDNN) and the squares of the differences between adjacent normal-to-normal RR intervals (r-MSSD) in time domain analysis, and low-frequency (LF) and high-frequency (HF) power in frequency domain analysis. Δ Akaike's information criterion (Δ AIC) was used to compare the HRV components-added model to the basic model by the Pre-transplantation Assessment of Mortality (PAM) score, HCT-specific comorbidity index (HCT-CI), or disease risk index (DRI) in a Cox proportional hazards model.

Results: We registered 112 consecutive allo-HCTs (103 patients) who agreed to participate in this study. The median age of the patients was 45.5 years (range: 18-66) and the median follow-up period for survivors (76 patients) was 961 days (range: 159-1654). Diagnoses included acute myeloid leukemia (n=55, 49%), acute lymphoid leukemia/lymphoblastic lymphoma (n=13, 12%), mixed phenotype acute leukemia (n=3, 3%), myelodysplastic syndrome (n=15, 13%), adult T cell leukemia/lymphoma (n=7, 6%), aggressive lymphomas (n=6, 5%), indolent lymphoma/chronic lymphoid leukemia (n=3, 3%), chronic myeloid leukemia/other myeloproliferative neoplasms (n=5, 4%), aplastic anemia (n=2, 2%), and others (n=3, 3%). Stem cell sources consisted of HLA-matched related donors for 18 transplants (16%), mismatched related donors for 24 transplants (22%), matched unrelated donors for 32 transplants (29%), mismatched unrelated donors for 8 transplants (7%), and cord blood for 30 transplants (27%). Myeloablative conditioning was employed for 84 transplants (75%).

Median value of LF, HF, SDNN, and r-MSSD was 409 msec² (range: 27-1279), 103 msec² (range: 5-1309), 122 msec (range: 48-231), and 20 msec (range: 4-95), respectively.From univariate analysis, the reduction of LF, HF, SDNN, or r-MSSD was significantly associated with decreased probability of OS (hazard ratio (HR): 0.7 for LF per 100 msec² (p<0001), 0.5 for HF per 100 msec² (p=0.001), 0.2 for SDNN per 100 msec (p=0.004), and 0.6 for r-MSSD per 10 msec(p=0.01), respectively). Of these four HRV components, the LF-added model for PAM, HCT-CI, and DRI had the highest values of Δ AIC (16.5, 22.2 and 11.4, respectively). From the analysis with stratification into four LF groups divided equally by the number of transplants (n=28), two-year OS was 92.9, 84.5, 59.7, and 33.2%, respectively (p <0.0001). LF solely represented a better discriminating variable for the prediction of survival. Additionally, from bivariate analysis for OS on each LF-added model to PAM, HCT-CI, or DRI, decreased LF was found to be an independent predictor for lower OS (adjusted HR for LF per 100 msec²: 0.7 (p=0.0003), 0.7 (p<0.0001), or 0.7 (p=0.001), respectively).

Conclusions: Pre-transplant autonomic nervous system dysfunction is a powerful predictor of survival likely because it reflects the comprehensive physiological or pathological status of the patient. Of note, LF was a powerful predictor for survival and statistically independent from well-known prognostic indexes or scoring systems such as PAM, HCT-CI, and DRI.