Burget D.W., Chiverton S.G., Hunt R.H. Is There an Optimal Degree of Acid Suppression for Healing of Duodenal Ulcers? // Gastroenterology. 1990;99:345-351.
Is There an Optimal Degree of Acid Suppression for Healing of Duodenal Ulcers?
A Model of the Relationship Between Ulcer Healing and Acid Suppression
The optimal degree and duration of suppression of gastric acidity required for the healing of peptic ulcers has never been established. Although very potent inhibitors of acid secretion are now available, the need for this degree of suppression has not been shown, and there is a possibility of adverse effects because of pronounced acid inhibition. Therefore, a model has been constructed that defines the relationship between duodenal ulcer healing and antisecretory therapy. Acid suppression data were obtained directly from investigators as raw data from 24-hour studies of acid secretion. Twenty-one experiments from seven investigators provided 490 24-hour studies using 19 different treatment regimens. Healing data were collected from a metaanalysis of published clinical trials of duodenal ulcer healing. A total of 144 published trials in 14,208 patients provided healing data at several endoscopic endpoints for the 19 drug regimens for which acidity data were provided. Weighted least-squares polynomial regression analysis was used to define those parameters of antisecretory therapy that contributed most to duodenal ulcer healing and to define the shape of the response surface. A highly significant correlation (r=0.9814) was found between healing and the degree of acid suppression, the duration of acid suppression, and the length of therapy. The shape of the contour expressing this relationship shows that healing increases as the duration of suppression increases and as gastric pH increases. However, suppression that increased pH beyond 3.0 was not found to increase ulcer healing further. It is concluded that a longer duration of antisecretory effect and/or a longer duration of therapy are of greater importance than potency for duodenal ulcer healing.
Healing is a matter of time, but it is also sometimes a matter of opportunity.
Antisecretory treatment for peptic ulcer disease can be undertaken using a wide range of drugs showing variable potency and duration of gastric acid suppression. In a previous analysis, we reported a highly significant correlation between the suppression of intragastric acidity and the proportion of duodenal ulcers healed at 4 weeks (1). Less potent agents were able to achieve equivalent healing to the more potent agents if therapy was extended to 8 weeks. The goal of antisecretory therapy is to reduce gastric acid secretion to allow the regenerative mucosal mechanisms to heal the ulcer. The development of potent, long acting antisecretory drugs such as the hydrogen- and potassium-stimulated adenosine triphosphatase inhibitors and new high-affinity H2-receptor antagonists can almost totally abolish acid secretion (2) and are associated with the highest observed healing rates (3). However, such profound suppression has raised concerns about possible adverse effects (4). Therefore, it is not clear to what extent acid secretion must be reduced to promote healing while avoiding adverse effects. The extent of antisecretory therapy may be defined in terms of three parameters: degree of effect, duration of effect, and length of therapy. An increase in any or all of these parameters results in an increase in the proportion of ulcers healing at any time point. The aim of this analysis was to study the relationship between duodenal ulcer healing and these parameters for data derived from the administration of a variety of drugs in different doses to define the optimal range of acid suppression. Raw data were obtained from 24-hour studies of the effects of antisecretory drugs in patients, thus allowing a more detailed study of the relationship than has previously been possible.
Materials and Methods
Gastric Acidity Data Collection
A workshop was held for investigators known to have studied the pharmacological alteration of intragastric pH using similar study protocols in duodenal ulcer patients. Seven investigators agreed to attend the meeting or provide their raw data for analysis: R. H. Hunt (Canada), A.B.R. Thomson (Canada), T. Gledhill (U.K.), R. P. Walt (U.K.), R. E. Pounder (U.K.), J. G. Williams (U.K.), and D. B. Jones (Australia). All studies included were performed in male, healed-duodenal ulcer patients, were placebo controlled, and used intermittent aspiration of gastric contents with pH determination by glass electrode ex vivo. Twenty-one studies were provided, which included 18 different active therapeutic regimens plus placebo. Fourteen studies were blinded and randomized, three were randomized but not blinded, one was blinded but not randomized, and three were neither blinded nor randomized. A total of 490 individual patient 24-hour studies was obtained, providing over 11,500 data points.
Minor variations in experimental protocols such as sample timing and frequency and meal timing result in variations in the total number of data points collected in a 24-hour period. Therefore, data from different experiments were pooled after adjustment so that all studies were based on the same number of raw data points. In cases of studies that overlapped the start and stop times so that 25 hourly readings were taken, the final reading was deleted. In four experiments representing 107 individual 24-hour studies, only H+ activity was provided. In these studies, pH readings of greater than 5 had been assigned an H+ activity of 0. Data were converted back to pH by the usual formula pH = -log H+, and Inactivities of 0 were deemed to have a pH of 5.
All data for each drug regimen were pooled, and the total duration in hours above each threshold pH of the total 24-hour period was calculated for predefined pH thresholds from values of 1.0-5.0 in intervals of 0.2 units. Calculations were limited to pH thresholds below 5.0 because of the missing data above pH 5.0 for the studies described. This calculation provides 1956 summary data points from the 490 24-hour studies.
Clinical Trial Data Collection
The proportion of ulcers healed was extracted from a database of published trials of acute duodenal ulcer healing determined by metaanalysis. A fully recursive search of the literature was based on a preliminary on-line computer search of the Medline and Excerpta Medico, databases from 1966-December 1986, supplemented by a manual search of available reviews, proceedings, and abstracts from the meetings of the American Gastroenterological Association, the British Society of Gastroenterology, and the World and European Congresses of Gastroenterology, from 1984-1987. The bibliographies of published studies were examined to locate any further trials. All clinical trials of treatment of acute duodenal ulcers were selected for preliminary inclusion and evaluation and were numbered; a total of approximately 900 abstracts and papers was selected. Each study located was evaluated independently by two investigators using the following inclusion criteria: (a) endoscopically controlled for both ulcer diagnosis and assessment of healing; (b) placebo or standard therapy controlled; (c) patients blinded to the nature of therapy; and (d) internally consistent and complete presentation of results. When there was disagreement as to the suitability of a study for inclusion, a committee of three investigators (D.W.B., S.G.C., and R.H. H.) reviewed disputed studies for inclusion criteria. All trials meeting the entry criteria were stored in a computer database and sorted by criteria that included author, drug name and geographical study site; trials were examined further to exclude duplicate publications. The final database of clinical trials for the years 1970-1987 included 210 trials.
For the analysis described in this paper, those clinical trials performed with drug regimens for which the raw 24-hour intragastric pH data were available were extracted. A total of 144 published trials, 39 of which also contributed placebo data, was included in the study analysis. These 144 clinical trials represent the data from a total of 238 treatment arms in 14,208 patients.
The proportion of ulcers healing at each endoscopic evaluation time point provided between 1 and 12 weeks was calculated as the number healed/number evaluated. The "intention to treat" calculation could not be applied because many studies did not report the number of patients initially entered. Healing at time zero was assigned a value of zero for analysis.
To explore the relationship between healing and three predictor variables simultaneously, the technique of multiple linear regression is required. Individual univariate simple regression analyses are not capable of defining the interrelationships between more than two variables. When three variables are plotted against each other, the result is a surface in three dimensions. The shape of the surface is unlikely to be a flat plane, and curvature must be a feature of the analysis, which is determined empirically by polynomial regression. This allows the shape to be defined without making any assumptions about the exact mathematical relation.
A full third-order polynomial model of the relationship between the healing proportion (Y) and the independent variables weeks of therapy (W), pH target threshold (T), and hours of suppression above the target value (H) takes the form:
Each of the potential terms in the full equation was calculated from the raw data, and stepwise weighted multiple linear regression analysis was performed using all terms (5). Stepwise regression analysis selects from the complete set of 19 terms defined in the equation only those terms that significantly improve the correlation of the predicted model. Terms were included in the fitted model if the partial F upon entry was significant (P < 0.05), and removed if the partial F on removal would not change significantly (P > 0.10). Regression was weighted using the number of patients endoscopically evaluated for the data point. The validity of the assumptions of regression analysis was confirmed with residual analysis using partial regression and normal probability plots. All calculations were performed using the SPSS statistical package (SPSS Inc., Chicago, IL).
A total of 11,571 pH readings from a possible maximum of 11,760 (98.4%) was collected for the gastric acid suppression analysis. The reduced gastric acid suppression data are summarized in Table 1 and the clinical trial healing data in Table 2. The multiple regression analysis was based on all data shown in Tables 1 and 2.
The regression analysis results for the final best-fit model describing the relationship between duodenal ulcer healing and the degree of antisecretory effect, duration of effect, and length of therapy are listed in Table 3. Of the 19 potential terms in the full polynomial model (Equation 1), 12 are included in the model, and the remaining seven terms were not found to increase the correlation significantly by the stepwise regression technique. A highly significant multiple regression is observed with a multiple correlation coefficient of 0.98139 and an adjusted R2 value of 0.9621 (P < 0.0001). Residuals were normally distributed, and no outliers were noted that produced any influential effects on the model.
The shape of a four-dimensional model cannot be shown graphically; therefore, the equation was reduced by solving separately for 2, 4, and 8 weeks of therapy. The simplified equation can then be graphed as a three-dimensional surface plot as shown in Figure 1. In these plots, the increase in healing can be seen with the increase in degree and duration of suppression. Graphs are truncated at 100% healing, although the equation does predict higher rates. Interpretation of three-dimensional plots is difficult, and an improved graph for this purpose is the contour plot in which the proportion of ulcers healed is shown as contour lines that connect points of equal healing.
Figure 2 shows contour plots for the predicted relationship between healing and acid suppression at 2, 4, and 8 weeks of therapy. In Figure 2, the highest healing rates are shown in the top right corner of the plots where the duration and degree of acid suppression are greatest. As the degree of acid suppression defined by pH threshold increases, the healing rate for any particular duration of suppression increases. If the degree of suppression is increased above pH 3 where contours are almost horizontal, there is a progressively smaller therapeutic gain. Therefore, substantial further increases in healing are more easily achieved by increasing the duration of suppression.
Table 1. Acid Suppression Data Calculated From 24-Hour Studies of Gastric Acid Secretion
NOTE. Data given as the duration (hours) of the total 24-hour day during which the pH was increased above the defined pH threshold. Drug regimens abbreviated as follows: ANT, antacid, 150 mmol, 7 times a day; ClG, cimetidine, 200 mg, three times a day and 400 mg at bedtime; C2Q, cimetidine, 200 mg four times a day; C3Q, cimetidine, 300 mg, four times a day; C4B, cimetidine, 400 mg, twice a day; C4Q, cimetidine, 400 mg, four times a day; C8N, cimetidine, 800 mg at bedtime; E35B, enprostil, 35 ftg, twice a day; E70N, enprostil, 70 ng, at bedtime; F4N, famotidine, 40 mg, at bedtime; O10M, omeprazole, 10 mg, once a day, O20M, omeprazole, 20 mg, once a day; O30M, omeprazole, 30 mg, once a day; O60M, omeprazole, 60 mg, once a day; P50B, pirenzepine, 50 mg, twice a day; PLA, placebo; R15B, ranitidine, 150 mg, twice a day; R3E, ranitidine, 300 mg, with evening meal; R3N, ranitidine, 300 mg, at bedtime.
Table 2. Duodenal Ulcer Healing Data From Metaanalysis of Published Clinical Trials
NOTE. Healing data given as number healed per number evaluated at each time point.
a No. of studies with arms contributing to this drug regimen.
b No. of patients entered, or if not available, no. evaluated at first endoscopy.
c Drug codes same as in Table 1.
Table 3. Regression Analysis Statistics for the Model Dejined by Stepwise Regression
NOTE. Adjusted R2 = 0.9629 (P < 0.0001); standard error of the estimate = 145.36. Terms defined in equation 1 based on duration of therapy in weeks (W), predefined pH threshold (T), and duration of acid suppression in hours (H).
This study refines our previous analysis (1) which shows that the degree of acid suppression and length of treatment were important variables in duodenal ulcer healing. In the previous study the data show that for the H2-receptor antagonists, the suppression of nocturnal but not daytime acidity was critical. However, antacids and omeprazole showed daytime suppression to have a significant effect. Because the H2-receptor antagonists are largely ineffective during the daytime, it was hypothesized that any predictive relationship between nocturnal acid suppression and duodenal ulcer healing could be explained on the basis of total duration of acid suppression irrespective of the time of day during which the suppression was effective. Therefore, in this study, the total duration of acid suppression was examined irrespective of drug class or time of day.
Figure 1. Three-dimensional response surfaces calculated for fixed, predefined durations of duodenal ulcer therapy of 2 weeks (A), 4 weeks (B), and 8 weeks (C). Each surface shows the relationship between the predicted healing rate (HR, vertical axis) and the duration of acid suppression in hours (x axis) above predefined pH threshold (y axis)
Polynomial regression was used in this analysis to construct a theoretical "response surface" to define the relationship between the response of duodenal ulcer healing and three predictor variables. The purpose of the construct is to elucidate those factors that contribute to healing and to identify the extent of acid suppression beyond which no further increase in the proportion of ulcers healed is likely to be observed. The use of regression techniques has allowed the relationship between these factors to be explored.
The results show that there is a highly significant, predictable relationship between the healing of duodenal ulcers and the defined parameters of antisecretory therapy, namely the degree and duration of suppression of intragastric acidity and length of treatment. Either parameter of antisecretory effect only influences ulcer healing when the healing contour lines in Figure 2 are not parallel to the axis represented by that parameter. Figure 2 clearly shows that very little increase in healing results from different antisecretory drugs and/or doses that increase intragastric pH above 3 and that the total duration of suppression during the 24-hour period is of much greater importance.
Although this analysis is based on very large data sets, it is inevitable that there are fewer data at the extremes of the plot, making interpretation outside the region from which data were drawn more difficult. It is inappropriate to use regression analysis to make predictions by extrapolation beyond the region of the data. The purpose of this study, however, is to elucidate the general shape of the response surface of the relationship between acid suppression and ulcer healing and not to use the model to make predictions regarding any specific drug therapy. Knowledge of the relative importance of the parameters affected by antisecretory agents may aid in directing the development of new therapies. Other parameters of intragastric acidity such as peaks and nadirs in pH, meal-related pH, or the time during the 24-hour period in which suppression is achieved may be important but have not been addressed in this analysis. With the techniques used, any such additional parameters would lead to autocorrelation because they are all related measures of intragastric acidity.
Figure 2. Contour plots calculated for fixed, predefined durations of duodenal ulcer therapy of 2 weeks (A), 4 weeks (B), and 8 weeks (C). Each contour plot shows the predicted healing rate as "isolines" connecting points of equal healing indicated by the numerical labels on each line
The observation that acid suppression to pH levels above 3 is unnecessary has an intriguing similarity to the historical concept of "free acid" first described in 1824 by Prout (6). Free acidity has been variously defined as that acidity below a pH ranging from 2.8-3.5, and titration to the endpoint for Toepfer's reagent (7) was the usual method for describing whether free acid was present or not. This threshold was considered to have clinical significance until the increasing use of the pH meter allowed simpler determination of accurate gastric juice pH, and pH 7.0 has since become the more usual definition of anacidity. The plateau of healing above pH 3 observed in this study suggests that reducing acidity below the level of so-called free acidity is not required for duodenal ulcer healing. Furthermore, in a preliminary analysis of these data, the greatest correlation of duodenal ulcer healing with the duration of suppression occurred when the arbitrary pH threshold chosen was 3.0.
Acid secretion in humans is not essential to life, but it is important to the activation of pepsinogens and the initiation of protein digestion, and to facilitate dietary iron and calcium absorption. Gastric acid has an important primary bactericidal barrier to protect the small bowel from ingested bacteria (8,9). Low pH is principally responsible for the bactericidal properties of gastric juice (10). Exogenous bacteria introduced into the stomach at a pH of 3.0 or less are destroyed within 15 minutes, although some bactericidal activity is observed up to pH 4.0.
In addition, the presence of gastric acid modulates the secretion of gastrin through a feedback mechanism (11) that may be important in controlling the trophic effects of this hormone (12). This may be important because potent, prolonged acid suppression in animal studies is associated with hypergastrinemia and the development of enterochromaffinlike cell hyperplasia and carcinoid tumours (13,14).
The different pepsins in gastric juice differ in their pH optima for proteolytic activity (15), but total activity is greatest within the pH range of 1.8-3.2. At levels above this, there is a rapid and progressive decline in activity, thus providing a further contributory factor for ulcer healing when antisecretory drugs increase pH to 3 or greater.
There is an optimal range of acid suppression in which the healing of duodenal ulcers is achieved while preserving the physiological role of gastric acid. The results of this analysis have defined the upper limit of this range. In this sense, the optimal acid suppression therapy should aim to increase the intragastric pH to above 3 for a period of 18-20 hours per day for healing to take place within 3-4 weeks. Healing may still be effected, with shorter durations of effect, if the therapy is continued for 6 weeks or longer. The importance of time, both in the duration of suppression and the length of therapy, is shown in Figure 3. There is little rationale to increasing the potency of antisecretory agents, because anacidity for the purpose of healing duodenal ulcers may be defined as any pH above 3. To paraphrase and update Hippocrates: Healing is not only a matter of opportunity, but also a matter of time.
Figure 3. Contour plot of duodenal ulcer healing rate predicted by the duration of suppression above a fixed pH threshold of pH 3.0 (vertical axis) and duration of therapy (horizontal axis)
Received July 17,1990. Accepted February 2,1990.
Address requests for reprints to: R. H. Hunt, FRCP, FRCP(C), McMaster University Medical Centre, Room 4W8,1200 Main Street West, Hamilton, Ontario, Canada L8N 3Z5.
The raw 24-hour data were obtained at a workshop held July 23-24,1987, in Windsor, England. The support for this workshop by Glaxo Group Research Ltd., Greenford, England, is gratefully acknowledged, The authors also wish to thank Drs. A.B.R. Thomson, T. Gledhill, R. P. Walt, R. E. Pounder, J. G. Williams, and D. B. Jones for providing their raw data for this analysis.
Назад в раздел
Популярно о болезнях ЖКТ читайте в разделе "Пациентам"
Информация на сайте www.gastroscan.ru предназначена для образовательных и научных целей. Условия использования.