Characteristics of CHF patients who underwent ASV therapy
Characteristics of 115 patients at the onset of ASV therapy are shown in Table 1. Mean age was 64.7 ± 12.7 years, male gender was predominant—78.3 %, the proportion of patients with dilated cardiomyopathy (DCM) was 37.4 %, and patients having a disease duration of ≥1 year accounted for 67.8 %. At onset, more than 80 % of patients received diuretics and beta blockers, and approximately 80 % of patients were medicated with angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers. Despite the fact that patients had already undergone the sufficient treatment of their heart failure (HF), the combined proportion of patients with NYHA class III and IV HF was as high as 43.2 %, mean LVEF was 37.9 %, mean CTR was 56.7 %, and median plasma BNP concentration was 312.8 pg/mL. Therefore, the majority of patients on ASV therapy were found to have severe CHF. All patients underwent ASV therapy providing end-expiratory pressure, minimum pressure support, and maximum pressure support. The default values, the number of patients who used the device under default settings, and the range for these pressures were, respectively, as follows: 5, 3, and 10; 81, 112, and 105; and 3–8, 3–4, and 8–12 cmH2O.
Table 1 Baseline characteristics of enrolled patients, ASV-continued patients, and ASV-discontinued patients
Patient disposition
Patient disposition is shown in Fig. 1. The retrieval rate of the case report form on 115 patients who were analyzed for efficacy was 100 %; the attending physician had discontinued ASV therapy at his/her discretion in 28 patients (24.3 %) of them (ASV-discontinued patients). The most predominant reason for discontinuation of ASV therapy was “impatience” (20 patients, 71.4 %), followed by “economic reason” (4 patients, 14.3 %), “improvement in HF” (2 patients, 7.1 %), and “deterioration of HF” (2 patients, 7.1 %). In contrast, the attending physician had continued ASV therapy at his/her discretion in 87 patients (75.6 %) of them (ASV-continued patients), 13 of whom died due to the progression of HF. The proportions of patients were 40.9 % (47/115) to 73.9 % (85/115), who were analyzed for seven investigation items: vital signs [body weight, pulse rate, systolic blood pressure (SBP), and diastolic blood pressure (DBP)], echocardiography, BNP, renal function test, symptoms of CHF, chest X-ray, and hematology.
Among 115 patients who were analyzed for efficacy, 18 died within 1 year after the onset of ASV therapy: 16 died due to the spontaneous deterioration of HF or to lethal arrhythmias, 1 to suicide, and 1 to ileus. It was the attending physician who had found no causality between ASV therapy and death at his/her discretion.
Sleep study at the onset of ASV therapy
The status of conducting the sleep study at the onset of ASV therapy is shown in Fig. 2a. Patients, who underwent the study, were assessed for the severity of SDB by means of the apnea–hypopnea index (AHI). Consequently, SDB was present in 50.4 % (58/115) of patients. The percentages of patients with mild, moderate, and severe SDB were 12.2 % (14/115), 17.4 % (20/115), and 20.9 % (24/115), respectively. Patients with CHF who were complicated by moderate or severe SDB accounted for 38.3 % (44/115) of patients. On the other hand, the proportions of patients whose SDB severity was not assessed because the sleep study was not performed and of patients who were not complicated by SDB were 38.3 % (44/115) and 11.3 % (13/115), respectively. Namely, ASV therapy was not applied for the objective of treating SDB in at least 61.7 % of patients.
Subgroup analyses of variables for SDB, HF, and hemodynamics between ASV-discontinued patients and ASV-continued patients
The results from the stratified analysis on AHI distributions in the subgroups of ASV-continued and -discontinued patients are shown in Fig. 2b. The mean pre-ASV values of AHI were 24.0 ± 21.3 and 28.8 ± 19.2/h in ASV-continued patients and ASV-discontinued patients, respectively, with no statistically significant difference (P = 0.386). However, the combined proportions of NYHA class III and IV HF patients (P = 0.047), LVEF (P = 0.001), and CTR (P = 0.010) at the onset of ASV therapy were significantly higher, lower, and greater, respectively, in ASV-continued patients than in ASV-discontinued patients (Table 1).
NYHA classification
Figure 3a indicates pre- and post-ASV therapy changes in the combined proportion of patients with NYHA class III and IV. Time-course changes before and after ASV therapy in the proportion are shown in Fig. 3b. Eighty-one patients showed no significant difference in the proportion during 1 year before the onset of ASV therapy. After ASV therapy, however, the proportion significantly decreased (P = 0.012) from the pre-ASV value of 43.2 % to the post-ASV value of 23.4 %. Figure 3b indicates time-course changes by treatment week in the proportion. The statistical analysis of the proportions between weeks −52, −26, and −12 and onset revealed a significant increase (P = 0.018) from 33.3 to 41.2 % only between week −52 and onset. In contrast, the statistical analysis of the proportions between onset and weeks 12, 26, and 52 disclosed significant decreases (P = 0.018, P = 0.008, and P < 0.001) against the onset value at week 12 and later. Thus, the distribution patterns of NYHA classes also favorably changed after ASV therapy (Table 2; P = 0.001) and at weeks 26 and 52 of treatment (Table 3; P = 0.008 and P < 0.001, respectively). The abovementioned improvements in NYHA class were highly likely to be attributable to ASV therapy because patients showing improvements in NYHA class had a significantly greater number (P < 0.001) of days of ASV therapy (frequency × duration).
Table 2 Variables before and after ASV therapy
Table 3 Variables at baseline and ASV therapy weeks
Echocardiography
The mean pre- and post-ASV values of LVDd and LVDs were 62.7 and 60.4 mm, as well as 52.7 and 48.8 mm, respectively; therefore, both variables decreased significantly (P = 0.029 and P = 0.002, respectively). In association with these changes, LVEF increased significantly (P = 0.009) from the mean pre-ASV value of 34.9 % to the mean post-ASV value of 39.8 % (Fig. 4; Table 2). Furthermore, statistical analysis by treatment week revealed that LVEF tended to increase at weeks 12 and 26 of treatment and increased significantly (P = 0.035) at week 52 of treatment. LAD remained unchanged after ASV and between baseline and each treatment week (Table 3).
Vital signs
Body weight and pulse rate changed neither in the comparison of the pre- and post-ASV values nor in statistical analysis by treatment week. After ASV therapy, SBP and DBP decreased significantly against the pre-ASV values (P = 0.035 and P = 0.044, respectively). However, SBP showed no significant change in statistical analysis by treatment week. Furthermore, DBP showed a significant decrease (P = 0.043) against the onset value only at week 52 of treatment (Tables 2, 3).
Other examinations
After ASV therapy, BNP, Hb, eGFR, and CTR showed no statistically significant difference against the pre-ASV values. However, statistical analysis by treatment week revealed a significant decrease (P = 0.017) in BNP only at week 12 of treatment from the median onset value of 388.4 pg/mL to the median post-ASV value of 263.1 pg/mL, as well as significant decreases in CTR at weeks 12, 26, and 52 of treatment from the onset values (58.0 %), 54.2, 54.2 and 54.1 %, respectively (P = 0.003, P = 0.019, and P = 0.010, respectively) (Tables 2, 3).
Stratified analyses of the combined proportions of NYHA class III and IV HF patients and of LVEF
The results of the statistical analyses on the combined proportions of NYHA class III and IV HF patients and LVEF, which were found after ASV therapy using the AHI value of 15/h as the cutoff value, are shown in Fig. 5. Between the subgroup of patients with AHI (≥15/h) and the subgroup of patients with AHI (<15/h), no significant difference was found in the combined proportion of NYHA class III and IV HF patients (Fig. 5a) or LVEF (Fig. 5b). However, patients with a pre-ASV value of AHI (<15/h) tended to have severe HF as compared to patients with a pre-ASV value of AHI (≥15/h). This finding is not concordant with previous clinical studies that have recognized an association between SDB severity and HF severity. We speculate that this discordance is attributable to the fact that in Japan ASV therapy is conducted in expectation of SDB improvement and CHF patient’s hemodynamic improvement. Furthermore, SDB severity was not related to continuity of ASV.
Multivariate logistic regression analysis
The results of the multivariate logistic regression analysis on independent variables associated with LVEF improvement (any numerical increase from the pre-ASV value), with and without AHI as a background factor, are shown in Tables 4 and 5, respectively.
Table 4 Logistic regression analysis of patients’ background factors associated with LVEF improvement when not including AHI
Table 5 Logistic regression analysis of patients’ background factors associated with LVEF improvement when including AHI
The backward selection method was used to analyze background factors [gender, age (≥65 years), underlying heart disease (DCM, ischemic heart disease, hypertensive heart disease, valvular heart disease, and others), diabetes mellitus, hypertension, chronic kidney disease (CKD), atrial fibrillation, medications [e.g., diuretics, beta blockers, and angiotensin receptor blockers (ARBs)], and LVEF at onset], and variables to perform the multivariate logistic regression analysis were selected. Consequently, age (≥65 years), CKD, and LVEF at onset (≤40 %) were selected as independent variables for the dependent variable—LVEF improvement. The Wald χ
2 value was significant (P < 0.05). Age showed a positive correlation [regression coefficient 1.850; odds ratio 6.362; 95 % confidence interval (CI) 1.076–37.632; P = 0.041] with LVEF improvement, while both CKD and LVEF at onset showed negative correlations (regression coefficient −2.842; odds ratio 0.058; 95 % CI 0.008–0.423; P = 0.005; and regression coefficient −0.066; odds ratio 0.936; 95 % CI 0.887–0.988; P = 0.017) (Table 4).
To determine the presence or absence of any contribution of AHI to the dependent variable, furthermore, the multivariate logistic regression analysis including AHI as an additional background factor was conducted. Consequently, AHI was found not to have any contribution to the dependent variable although the number of analyzable patients reduced to 28 from 48 (regression coefficient −0.887; odds ratio 0.412; 95 % CI 0.019–9.020; P = 0.573) (Table 5).