Professor Dr H.-G. Dammann, Klinische Forschung Hamburg, Robert-Koch-Str. 26, D-20249 Hamburg, Germany. E-mail: email@example.com
Background : The most frequently reported adverse events associated with acetylsalicylic acid intake are minor gastrointestinal complaints. Galenic modifications, such as buffered formulations with or without ascorbic acid, may improve the benefit–risk ratio by decreasing the local mucosal side-effects of acetylsalicylic acid.
Aim : To assess endoscopically-proven gastrointestinal lesions and the amount of gastric microbleeding of four different buffered and plain acetylsalicylic acid formulations, one containing paracetamol.
Methods : A randomized, four-fold cross-over study was performed in 17 healthy subjects who underwent serial oesophago-gastro-duodenoscopy before and after each course of 4-day dosing. Gastric aspirates were collected for the determination of haemoglobin concentrations to detect microbleeding.
Results : Buffered acetylsalicylic acid plus ascorbic acid yielded the lowest Lanza score, the lowest increase in the number of mucosal petechiae and the lowest increase in the amount of gastric microbleeding. Subjects receiving acetylsalicylic acid plus paracetamol plus caffeine showed the highest Lanza score of all treatments, and a considerably greater sum of petechiae in the oesophagus, stomach and duodenum compared with those receiving buffered acetylsalicylic acid plus ascorbic acid.
Conclusions : The trial confirms that buffering of acetylsalicylic acid improves local gastric tolerability. Acetylsalicylic acid in combination with ascorbic acid shows significantly fewer gastric lesions and the lowest increase in gastric microbleeding compared with the other tested formulations.
The use of acetylsalicylic acid (ASA) for the treatment of pain has been established in self-medication for more than 100 years. A maximum daily intake of 3 g, divided into several doses, is recommended for short-term use. In 2002, the Food and Drug Administration initiated an evaluation of the most commonly available analgesic monosubstances for non-prescription use (ASA, naproxen, ibuprofen, paracetamol) and stated that all were safe and effective when used appropriately.1 A meta-analysis of 72 randomized studies with 6550 patients by the Cochrane Institute on the efficacy and safety of ASA did not reveal any severe gastrointestinal events.2 Minor gastrointestinal complaints, which are the most commonly reported side-effects associated with cyclo-oxygenase inhibitors, including ASA,3–6 were not evaluated in this setting. Such findings support the hypothesis that the risk of severe gastrointestinal haemorrhage with ASA seems to be significantly increased only with long-term use of higher dosages as a consequence of systemic effects.7,8
It has been postulated that the gastric damage caused by ASA is due to a local effect on the mucosa.9 This explains why galenically modified ASA formulations, such as enteric-coated ASA formulations,10 buffered ASA solutions11 and buffered chewable tablets,12 all of which have become widely available, are believed to have the potential to minimize minor gastrointestinal lesions.
Several hypotheses have attempted to explain the underlying mechanisms of mucosal injury by ASA, including: ASA accumulation in mucosal cells, alteration of the permeability of the cell membrane due to ion trapping and back-diffusion of H+ ions from the gastrointestinal lumen;13 inhibition of prostaglandin E2 synthesis of the gastric mucosa by ASA, leading to stimulation of gastric acid and pepsin secretion, and to lesions of microvessels;10 and the formation of free radicals due to ASA activation of neutrophils. The latter has only recently been confirmed by the demonstration of a protective effect of the antioxidant ascorbic acid (vitamin C) in ASA-induced duodenal mucosal lesions.14
Other strategies to minimize the gastric damage caused by ASA involve the use of a combination of ASA with substances that interfere with the pathogenetic process of mucosal lesions, including a combination of ASA with antacids,15 H2-blockers,15 prostaglandins, such as rioprostil16 or misoprostol,17 or paracetamol,18,19 the latter being assumed to stimulate endogenous prostaglandin synthesis.
In order to assess the potential influence of different kinds of ASA formulations (plain vs. buffered) and the presence and absence of ascorbic acid and paracetamol, we examined, in a cross-over design, the effects of four commercially available ASA formulations (plain ASA, buffered ASA, buffered ASA plus ascorbic acid, plain ASA plus paracetamol plus caffeine) on the mucosa in the upper gastrointestinal tract in healthy subjects.
Healthy male and female subjects with a gastric pH < 3 and an age between 18 and 45 years were enrolled in one centre in Germany. Although a much wider age group of subjects self-medicate with ASA, this age group was chosen because of the availability and consent of the subjects to participate in the study. The subjects tested negative for human immunodeficiency virus and hepatitis B surface antigen and were devoid of gastritis or lesions of the gastric mucosa on oesophago-gastro-duodenoscopy. Subjects with known allergies or hypersensitivities, asthma or existing Helicobacter pylori infection (as assessed by 13C breath test) were not included in the study. Women of child-bearing potential not following a medically accepted contraceptive method were not considered for enrolment. Subjects with high xanthine derivative consumption, alcohol abuse and regular medication intake [e.g. antacids, H2-antagonists, proton pump inhibitors, pirenzepine, misoprostol, non-steroidal anti-inflammatory drugs (NSAIDs)] were also excluded from the study. Haemoglobin, liver function tests and bilirubin were within clinically acceptable ranges.
An independent ethics committee relevant to the study site approved the study protocol. Each subject gave written informed consent.
This was a prospective, investigator-blind, randomized, four-fold cross-over study enrolling 17 subjects. The study was performed in 1999 and assessed the gastric tolerability of four different buffered and plain ASA medications, which are listed in Table 1.
Table 1. Medications taken in a four-fold cross-over design
Buffered ASA plus ascorbic acid effervescent tablet (Aspirin plus C, Bayer Vital, Germany)
400 mg ASA/
3 times daily
Plain ASA tablet (Aspirin, Bayer Vital, Germany)
2 tablets 500 mg
3 times daily
Plain ASA plus paracetamol plus caffeine tablet (Thomapyrin Schmerztabletten, Boehringer Ingelheim, Germany)
250 mg ASA/
50 mg caffeine
3 times daily
The investigator was blind to the medications. Subjects were hospitalized for the duration of each study period. Medications were administered three times daily for 4 days in each treatment period with a total of 10 drug administrations per study period. The periods were separated by 10-day washout periods. During each study period, oesophago-gastro-duodenoscopy was performed on the morning before the first intake of medication and on the fourth day, 60 min after the last intake of medication. All endoscopies were performed by the same investigator. The investigator documented the number of petechiae as well as haemorrhagic and non-haemorrhagic erosions in the oesophagus, stomach and duodenum. The findings were classified using the adapted lesion score developed by Lanza.20 Photographs documented all relevant pathological findings. Following endoscopy and 1.5 h after the intake of medication, gastric aspirates were collected for the determination of haemoglobin concentrations to detect microbleeding21,22 via orogastric intubation with a 16 F gauge Salem sump tube (Sherwood, Belgium). The pH was measured with a glass electrode (Corning, Denmark).
The primary efficacy criterion was the sum of haemorrhagic and non-haemorrhagic erosions of the gastrointestinal mucosa following the intake of 10 doses of each study medication compared with baseline in the valid case population, i.e. all subjects completing the four treatment periods according to the study protocol. In order to achieve optimal power, the sum of erosions observed in the oesophagus, fundus, body, antrum and duodenum was taken for assessment. The primary efficacy criterion was tested for differences between the treatment groups using Friedman analysis of variance by ranks.
The secondary efficacy criteria were: (i) the number of haemorrhagic and non-haemorrhagic erosions in each of the regions (oesophagus, fundus, body, antrum and duodenum), tested for differences between the treatment groups using Friedman analysis of variance by ranks; (ii) the comparison of erosions using the adapted Lanza lesion score, separated by each region and analysed descriptively; (iii) the number of petechiae; and (iv) the extent of microbleeding (µL/10 min) of the gastric mucosa, using analysis of covariance followed by Scheffe's test and contrasting ASA plus ascorbic acid with all other treatments. Quality and incidence rates of adverse events, vital signs and laboratory examinations were the basis for the safety evaluation.
Based on previously published study results,10 a sample size of 16 subjects was considered as appropriate for the four-fold cross-over design of this study.
Seventeen subjects received study medication and were valid for intention-to-treat and safety analyses. Table 2 shows the demographic and physical characteristics of the enrolled subjects.
Homogeneity testing did not reveal relevant differences in demographic data between the subjects of the four treatment sequences. One subject took an additional NSAID during the course of the study, was excluded from the efficacy analysis and was replaced by another subject. Seventeen subjects were valid for intention-to-treat analysis; 16 subjects completed the four treatment periods according to the study protocol and were valid cases.
Sum of mucosal erosions (primary efficacy criterion)
Table 3 summarizes the results of the primary efficacy criterion (sum of all erosions observed in the oesophagus, stomach and duodenum) for both the valid case and intention-to-treat populations.
Table 3. Sum of mucosal erosions (oesophagus, stomach, duodenum)
The mean sum of erosions in the valid case population ranged from 0 to 0.19 ± 0.75 at baseline; at the end-point, the range was 0.13 ± 0.50 − 1.38 ± 2.90 in the four treatment groups. Friedman analysis revealed no significant differences between the four treatment groups (P = 0.456). The lowest mean sum was observed with buffered ASA plus ascorbic acid, followed by plain ASA and buffered ASA. Subjects receiving plain ASA plus paracetamol plus caffeine had the highest mean sum of erosions. The median number of erosions at baseline and at the end-point was zero in all groups, indicating that there were only a minority of subjects with positive findings. The outcomes were similar in the intention-to-treat population; again, no statistically significant difference was found between the four ASA formulations.
The assessment of mucosal lesions was performed using the adapted Lanza lesion score: 0, normal mucosa; 1, petechiae; 2, 1–5 erosions; 3, 6–10 erosions; 4 > 10 erosions.20 The results are summarized in Table 4. Homogeneity testing did not show any differences in the baseline values between the groups.
Table 4. Lanza score (total for oesophagus, stomach and duodenum)
Most mucosal lesions were observed in the gastric corpus and antrum and only a few in the duodenum and gastric fundus. No lesions were seen in the oesophagus.
The Lanza score was highest in subjects receiving plain ASA plus paracetamol plus caffeine (mean 2.5 ± 1.2), followed by plain ASA (mean 2.3 ± 1.3) and buffered ASA (mean 2.0 ± 1.5). Buffered ASA plus ascorbic acid yielded the lowest Lanza score (mean 0.5 ± 1.0) of all treatments. Friedman analysis showed an overall significance of P < 0.003. Individual comparisons of Lanza scores between the treatment groups using the Wilcoxon test showed that buffered ASA plus ascorbic acid was significantly superior to the other treatments (vs. buffered ASA, P < 0.005; vs. plain ASA plus paracetamol plus caffeine and plain ASA, P < 0.001).
Petechiae were a subset of the parameters included in the Lanza score. Table 5 displays the sum of petechiae observed in the four treatment groups. Homogeneity testing did not show any differences in the baseline values between the groups.
Table 5. Sum of petechiae (oesophagus, stomach and duodenum)
Most petechiae were found in the gastric corpus and antrum. The oesophagus was devoid of petechiae and only a few were observed in the duodenum. There were no significant differences in the number of petechiae in the gastric fundus between the four treatment groups.
Petechiae were most frequent in subjects receiving plain ASA (mean 74.2 ± 103.6), followed by plain ASA plus paracetamol plus caffeine (mean 41.4 ± 40.5) and buffered ASA (mean 13.8 ± 14.1). Buffered ASA plus ascorbic acid yielded the lowest sum of petechiae (mean 1.7 ± 4.6) of all treatments. Friedman analysis showed an overall significance of P < 0.001. Individual comparisons of total sums using the Wilcoxon test showed that buffered ASA plus ascorbic acid was significantly superior to the other treatments (P < 0.001).
Figure 1 displays a synopsis of summary data on petechiae, erosions and Lanza scores.
Gastric mucosal bleeding
Baseline values of mucosal bleeding ranged from 1.1 µL/10 min (buffered ASA) to 3.2 µL/10 min (plain ASA), which represents a nearly three-fold baseline difference (see Table 6). At the end-point, the lowestmean amount of bleeding was observed with buffered ASA plus ascorbic acid (4.0 ± 7.4 µL/10 min), followed by plain ASA plus paracetamol plus caffeine (6.4 ± 4.1 µL/10 min) and buffered ASA (7.1 ± 6.0 µL/10 min). Subjects receiving plain ASA had the highest mean amount of bleeding (14.3 ± 12.0 µL/10 min).
Table 6. Gastric mucosal bleeding (µL/10 min; valid case population; n = 16)
An adjustment of the least square means based on log-transformed data analysis of covariance was performed to account for baseline variation and the skewed distribution of end-point values. The analysis of covariance yielded significant treatment effects (P = 0.025). A more detailed analysis using Scheffe's test showed that buffered ASA plus ascorbic acid was superior to all other treatments (P = 0.01).
Figure 2 displays the data on gastric mucosal bleeding.
The medications were generally well tolerated. Fifteen subjects experienced 27 adverse events, which were mostly of mild or moderate intensity; all of them resolved. The most frequently reported adverse events were epigastric pain and headache. There were no serious adverse events and no dropouts due to adverse events. Forty-eight per cent of adverse events occurred during washout periods. Forty-one per cent of adverse events were unrelated to the study drug.
For the primary end-point of the study, i.e. sum of erosions in the gastrointestinal mucosa, subjects receiving ASA plus paracetamol plus caffeine had the highest mean number of mucosal lesions; however, the differences between the four ASA formulations did not reach statistical significance. Gastric mucosal lesions were observed mainly in the antrum and were not correlated with the total amount of ASA (daily dose between 1500 and 3000 mg). These findings contrast with a previously published US study, in which plain ASA caused significant dose-dependent gastric injury when compared with placebo and enteric-coated ASA formulations.10 The difference between the study results may be explained by the low number of subjects in this study or the different definitions of erosions and petechiae in Germany and the USA.
Of all the treatments, buffered ASA plus ascorbic acid yielded the lowest Lanza score, the lowest increase in the number of mucosal petechiae and the lowest increase in the amount of gastric microbleeding compared with baseline. Subjects receiving ASA plus paracetamol plus caffeine showed the highest Lanza score of all treatments and a considerably greater sum of petechiae in the oesophagus, stomach and duodenum compared with buffered ASA plus ascorbic acid. All differences were statistically highly significant. Buffered ASA without ascorbic acid demonstrated less petechiae compared with the plain ASA formulation. The findings indicate that, of all the ASA formulations tested, buffered ASA plus ascorbic acid caused fewer gastrointestinal lesions.
The total daily dose of ASA administered in the four formulations varied between 1500 and 3000 mg, the total dose of buffered ASA plus ascorbic acid amounting to 2400 mg daily.
One may speculate about the potential role of ascorbic acid in contributing to the improved gastrointestinal tolerability. An intriguing pathogenetic aspect of the multifactorial process of ASA-induced mucosal changes in the upper gastrointestinal tract is the formation of free radicals due to the activation of neutrophils by ASA. In addition, the protective effect of the antioxidant ascorbic acid in ASA-induced gastric mucosal lesions has been demonstrated.23
Two aspects merit attention: paracetamol as a combination partner with ASA and the total daily ASA dose. Fixed-dose combinations of paracetamol and ASA, with lower doses of each substance than those administered with the respective monotherapies, are believed to improve the safety profile, in particular to reduce the number of gastrointestinal events, while providing the same level of pain relief.18,19 The data from this study do not confirm this assumption, but rather seem to prove the opposite. The Lanza score and number of petechiae in the oesophagus, stomach and duodenum were significantly higher with the ASA plus paracetamol plus caffeine formulation than with ASA plus ascorbic acid alone, even though the total amount of ASA was higher with ASA plus ascorbic acid (ASA 2400 mg/day) than with ASA plus paracetamol plus caffeine (ASA 1500 mg/day). It may even be hypothesized that paracetamol may increase the potential for ASA to have adverse effects on the gastrointestinal mucosa or, by itself, may not be as devoid of gastrointestinal adverse events as previously believed. The latter is supported by comparative pharmaco-epidemiological data from Italy, in which ASA showed the lowest number of gastrointestinal adverse events when compared with five other analgesics, including paracetamol. Subjects reported gastrointestinal adverse events with ASA and paracetamol in 2.1% and 7.6% of cases, respectively.24 Another study in patients with tension-type headache did not reveal clinically relevant differences between ASA 500 and 1000 mg, paracetamol 500 and 1000 mg and placebo with regard to the number of gastrointestinal adverse events, which were of the order of 5.8–6.3%, 3.6–3.8% and 2.7%, respectively.25
In conclusion, the trial confirms that buffering of ASA improves the local gastric tolerability. ASA in combination with ascorbic acid shows significantly fewer gastric lesions and the lowest increase in gastric microbleeding compared with the other tested formulations. Of special interest is the fact that the combination of ASA with paracetamol and caffeine shows a higher sum of petechiae than each of the tested buffered formulations, although it contains only half the amount of ASA.
This study was sponsored by Bayer Vital GmbH, Leverkusen, Germany.