INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Previous studies have reported a high incidence of gastroesophageal reflux (GER) among asthmatics (1, 2) and have also suggested that the treatment of reflux may result in marked improvement of asthma symptoms and pulmonary function (3- 6). In awake adult asthmatics with a positive Bernstein test, intraesophageal acid infusion has been reported to trigger bronchoconstriction with subsequent return toward baseline function when these symptoms were relieved with antacid treatment (7). Nighttime GER has been advocated as a possible causative and precipitating factor of nocturnal worsening of asthma: it can trigger changes in respiratory pattern in asthmatic children when asthma is complicated by the presence of esophagitis (8). Converserly, other reports present conflicting data. In fact, in a study based on the use of noninvasive methods, Hughes and coworkers reported that reflux did not play any role in the production of nighttime asthma symptoms (9). Similarly, Ekstrom and coworkers reported that GER does not play any important role as an immediate trigger factor in nocturnal asthma (10). Therefore, the relationship between GER and nocturnal asthma is still controversial: the heterogeneity of results may be partly due to methodological limitations in previous studies, in particular the need of waking up the patient to perform the functional measures. In addition, the selection of the samples in terms of severity of GER disease may have influenced the possibility of detecting any relationship with airway reactivity: in fact, the only study based on a reliable method (i.e., continuous and simultaneous measurement of airflow resistance and esophageal pH monitoring in nocturnal asthmatics) and reporting negative results was carried out on patients with mild GER disease (11). The aim of the present study was to test this hypothesis by applying proper techniques to the evaluation of the relationship between GER and nocturnal bronchoconstriction in asthmatics with moderate to severe GER disease.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sample

Seven adult lifelong nonsmokers affected by nonseasonal asthma (4 male, 3 female), as defined according to American Thoracic Society (ATS) criteria (12), age 20 to 52 yr, were studied; all reported a personal history of heartburn, pain, and regurgitation suggesting GER disease. The severity and the frequency of these symptoms were classified according to Vigneri and coworkers (13). Only subjects showing a symptom score  12 were considered for the study.

Their main anthropometric and functional characteristics are presented in Table 1. All the subjects were affected by nocturnal asthma as demonstrated by a fall in peak expiratory flow (PEF) values  15% from bedtime to morning awakening and had not required changes in medications during the last month before the study. All the subjects reported awakenings associated with nocturnal asthma symptoms and, frequently, with GER-related symptoms. All of them were receiving regular inhaled corticosteroids and short-acting ₂-agonists on an "as needed" basis. None of them received theophylline or antacid, protonic pump inhibitors, H2-blockers, or prokinetics.

All patients gave written, informed consent, and approval for this study was granted by the institutional ethic committee.

Protocol

Each subject slept in the sleep laboratory and was submitted to a polysomnographic assessment. The following signals were continuously recorded both on paper (Hewlett-Packard 7758B; Hewlett-Packard Co., Waltham, MA) and on tape recorder (Hewlett-Packard 3968A) for further playback and analysis: electroencephalogram, electro-oculogram and chin electromyogram, for conventional sleep staging (14); bidirectional oronasal airflow () by a light tight-fitting face mask with a calibrated heated pneumotachograph (Fleisch no. 1) connected to a differential pressure transducer (MP-45 Validyne, range ± 5 cm H₂O; Validyne, Northridge, CA); volume by electronic integration of signal; transpulmonary pressure (PL) by a balloon-tipped catheter placed in the lower third of the esophagus connected to a calibrated differential pressure transducer (MP-45 Validyne, range ± 80 cm H₂O) referenced to the mask. The position of the esophageal catheter was adjusted following the occlusion technique proposed by Baydur and coworkers (15); supraglottic pressure (Psg) at supraglottic level by a second balloon-tipped catheter positioned at a distance of 15 to 17 cm from the nares and connected to a calibrated differential pressure transducer (MP-45 Validyne, range ± 5 cm H₂O) referenced to the mask; esophageal pH by intraesophageal glass electrode (Proxima-Light; SensorMedics, Milano, Italy) placed 5 cm above the esophagogastric junction. The electrode was previously calibrated using a neutral buffer and an acid buffer of pH 4. The data of a pH meter were recorded at a rate of 12 samples/min.

Analysis

Total lung resistance (RL) was computed breath-by-breath by the isovolume method using computer-based routines on a Digital Alpha Station (Digital Equipment Co., Maynard, MA): PL was referred to different flows at equal lung volumes (200 ml) (16); in this way, the elastic component is made constant, and any pressure change can be referred to flow-resistive properties. Supraglottic resistance (Rsg) was computed as the ratio between Psg and . Lower respiratory resistance (RLR) was derived breath-by-breath as the difference between RL and Rsg and expressed as cm H₂O · L¹ · s.

GER episodes were evaluated in terms of decreases in esophageal pH below 4. They were separately evaluated on the basis of duration: short GERs  5 min (SGER) and long GERs > 5 min (LGER). Because of its skewed distribution, the GER duration was expressed as natural log (lnGER duration).

RLR at the start of the study night were measured while the patients were awake and in supine position. RLR were then analyzed as means per minute along the whole study night.

For each GER episode we calculated: RLR value during the minute immediately preceding the GER onset (RLR_b); the area under the RLR curve (AUCR_LR) computed as the area under the RLR values in each GER episode; end RLR at the resolution of GER episode (RLR_e); RLR_e as the difference between RLR_e and RLR_b. AUCR_LR was treated as natural log (lnAUCR_LR) after transforming all negative values into positive ones by adding 3.5 since the minimum raw AUC value was 2.5.

Because it has been suggested that GER effects on airway resistances may last until 10 min after the resolution of GER episode (17), we also computed RLR 10 min after the resolution of episodes (RLR_10'). For the latter analysis, when the time interval between GER episodes was shorter than 10 min, more GER were pooled together: in this case, RLR_b was the RLR value immediately preceding the first GER; the RLR_10' was again calculated as the difference between RLR at 10 min after GER resolution and RLR_b. The GER duration was assumed as the sum of the durations of all pooled GER episodes within such an event.

Differences between means were evaluated by two-way analysis of variance (two-way ANOVA), to take into account both differences between variables and between subjects. In order to know whether each GER episode was associated with RLR increases within the individual, we removed the differences between subjects looking only at changes within. Accordingly, correlations between lnAUCR_LR, RLR_e, and RLR_10' as outcome variables and lnGER as predictor one were evaluated by analysis of covariance, introducing our data in a multiple linear regression model. Each subject was treated as a categorical factor using dummy variables (18). Moreover, to investigate the relationships between AUCR_LR and RLR_e as outcome variables and subject number, GER duration, wake-sleep state, and RLR_b as predictor ones we used a stepwise logistic multiple regression model. The AUCR_LR and RLR_e were dichotomized using a threshold corresponding to the 50th percentile of each variable distribution, whereas the independent variables were transformed in dummy variables as in Table 2. For each of the two outcomes (AUCR_LR and RLR_e), all correlates at univariate analysis were entered into a stepwise logistic regression analysis. Variables were considered to be independent predictors of the outcome if the odds ratio differed from 1 and the 95% confidence limits did not include 1. All the statistical procedures were performed by Systat (Systat Inc., Evanston, IL) and Stata (Stata Corporation, College Station, TX) software packages.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The median total experimental time (TET) for all the subjects was 423 min (range 301 to 502); the median total sleep time was 194 (range 60 to 296). Table 3 shows total experimental and sleep time for all the subjects, with median sleep stage duration. Rapid eye movement (REM) sleep was observed only in one subject (BON).

During the study night, a total of 101 GER episodes were found and analyzed in the overall sample: 72 were SGER (median 1.0 min) and 29 LGER (median 9.0 min, range 6 to 41). Median time spent in GER was 17.1% of TET (range 7.1 to 39.7).

The mean RLR at the beginning of study night (patients awake and in supine position) was 11.8 ± 3.7 cm H₂O · L¹ · s (Table 1). The overall trend of RLR over the time in these patients showed instantaneous increases when GER episodes occurred and a subsequent decrease after the resolution of GER episodes. Conversely, not all the increases in RLR occurred after a GER episode. Figures 1A and 1B present the RLR trend during the night for two representative subjects (BON and PRE).

View larger version (38K): [in this window] [in a new window]   Figure 1.   Trend of RLR over the nighttime for two representative subjects. RLR are shown as means per minute of time. Shaded vertical areas represents the duration of each GER episode.

Table 4 shows individual GER duration in absolute time and as percent of relevant sleep stages.

By the two-way ANOVA we found significant differences between RLR_e and RLR_b both in SGER and LGER (Table 5); moreover, both AUCR_LR and R_LR10' values were found to be significantly higher in LGER with respect to SGER (p < 0.0001 for both).

The analysis of covariance showed highly significant correlations between lnAUCR_LR, RLR_e, and RLR_10' as outcome variables versus lnGER duration as predictor one (Table 6, and Figures 2-4). A weak subject effect was noted only for the RLR_e versus lnGER duration relationship.

View larger version (15K): [in this window] [in a new window]   Figure 2.   Relationship between ln AUCRLR and ln GER duration, separately for each subject. The analysis of covariance showed a significant effect of ln GER duration (p < 0.0001) and no subject effect (p = 0.73). The overall slope was 1.105.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Relationship between RLRe (i.e., the difference between RLR at the resolution of GER episode and RLR value during the minute immediately preceding the GER onset) and ln GER duration, separately for each subject. The analysis of covariance showed a significant effect of ln GER duration (p < 0.0001) and a small subject effect (p = 0.021). The overall slope was 3.532.

View larger version (16K): [in this window] [in a new window]   Figure 4.   Relationship between RLR10' (i.e., the difference between RLR 10 min after the resolution of GER episodes and RLR value during the minute immediately preceding the GER onset) and ln GER duration, separately for each subject. The analysis of covariance showed a significant effect of ln GER duration (p = 0.01) and no subject effect (p = 0.14). The overall slope was 2.28.

The stepwise logistic regression analysis showed that no effect was produced by subjects' GER duration, wake-sleep state, and RLR_b on AUCR_LR and on RLR_e, the GER duration being the only significant predictor for RLR increase.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study, based on the direct, continuous, and simultaneous evaluation of both respiratory resistances and intraesophageal pH during nighttime, demonstrates a significant role of spontaneous GER episodes in provoking and sustaining nocturnal bronchoconstriction in adult asthmatics with moderate to severe GER disease. In fact, we found that: (1) GER episodes are associated with significant RLR increase; (2) RLR increases are strongly related with GER duration; (3) RLR increases tend to be maintained for a while after GER episode termination. These results are in agreement with previous studies suggesting that GER may be an important trigger mechanism in asthmatic children with nocturnal wheezing (8, 19). A further indirect evidence of the potential role played by GER in asthma has been provided by Harding and coworkers, who demonstrated the effect of long-term antacid therapy in improving asthma symptoms and pulmonary function (3).

Previous studies on the relationship between GER and nocturnal asthma reported controversial results probably owing to different factors. First, complete awakening of the patient is required when a functional index such as PEF is used to monitor nocturnal bronchoconstriction (10); the effect of sleep disruption on airway tone is unpredictable, and a lack of temporal correspondence is produced between acid stimulation on esophageal mucosa and the recorded modification of the functional index. Second, other indexes used to avoid sleep disruption (such as continuous transcutaneous monitoring of oxygen saturation) yield only indirect evidence of bronchoconstriction (8, 19). A third factor is the choice of the population sample: our results seem to demonstrate that the severity of GER disease is an important factor to produce a clear bronchoconstrictive effect, as previously reported (19).

The changes in airway patency found in our study (i.e., significant RLR increase strongly related with GER duration and rapid, even if not immediate, decline after GER termination) may explain the negative results obtained in studies based on measures of respiratory function not simultaneous with GER, and requiring the waking up of the patient (9, 10). These negative results may be further explained on the basis of the low sensitivity of methods used to detect bronchoconstriction. Moreover, the rapid response in terms of RLR changes supports more the hypothesis of a vagal reflex as pathogenetic mechanism originating from irritant esophageal mucosal receptors (20), than that of gastric content aspiration into the airways (21, 22).

The time course of changes in resistances with the development of RLR increases when GER episodes occurred (Figure 1) seems to lend some support to the effect of GER duration on the degree (Figures 2-4) of bronchoconstriction. This assumption is founded on the significant correlations between AUCR_LR and RLR_e with GER duration. Two-way ANOVA showed significant differences in AUCR_LR and RLR_10' values between GER episodes of different length: these results indicate that LGER have a significant effect on RLR increase and on its persistence after GER termination. A significant effect on RLR may also result from GER episodes lasting not more than 5 min, as shown by the significant increase in RLR_e in SGER, even if the bronchoconstrictor effect is lower owing to the shorter duration.

The importance of duration of GER is confirmed by the fact that the higher the lnGER duration, the more prolonged is the effect on airways 10 min after GER termination, as expressed by higher RLR_10' (Figure 4). This suggests on one hand a slower recovery of pulmonary function after acid is cleared from the esophagus (23), on the other a delayed bronchoconstrictor effect due to bronchial inflammation (late-phase reaction) reflecting inflammatory mediator release (21). The wake-sleep status seems to have no impact on the bronchoconstrictor effect of GER episodes as confirmed by the stepwise logistic regression analysis.

With regard to the effect of choice of population sample on the results, the only previous nocturnal investigation methodologically comparable to ours found no relationship between spontaneous or simulated GER and nocturnal increase in respiratory resistances, either in presence or in absence of esophagitis; the investigators concluded that GER produces only a poor contribution to nocturnal asthma worsening. However, that study was carried out on asthmatic patients affected by GER disease with a lower number of GER episodes during the night and a lower mean duration (11): a total of 11 GER episodes during the night were reported with a mean duration of 10.4 ± 2.3 min (range 1.1 to 24.0) and a mean total reflux time of 19.1 ± 5.7 min (84 min in our study) or 5.4% ± 1.4 of TET (17.1% in our study).

In conclusion, GER itself seems to be able to elicit a clinically relevant nocturnal bronchoconstriction in asthmatic subjects. Bronchoconstriction severity and duration are related to GER duration; in fact, longer GER episodes produced greater and more prolonged increases in RLR. Conversely, the finding of a minority of GER-unrelated bronchoconstriction events suggests that GER represents just one component of a host of factors whose complex interaction results in the clinical phenomenon of nocturnal asthma.