For long-term therapy for esophagitis, a physician can choose to follow patients and treat relapses over the short-term with acid-suppressing agents in full healing doses. Another strategy is to prescribe daily maintenance treatment. However, lower doses of currently available therapies, including omeprazole, ranitidine, famotidine, and cimetidine, have not been found to be as effective as full healing doses in maintaining healing [8-11].

Lansoprazole, a new proton pump inhibitor, effectively suppresses production of gastric acid over 24 hours when consumed once every morning [12]. Lansoprazole has been shown to be highly effective for treating erosive esophagitis, with healing rates similar to those of omeprazole [13]. In the belief that lansoprazole would provide effective maintenance treatment and to determine the lowest effective maintenance dose, we compared two doses of lansoprazole—the effective healing dose of 30 mg and a lower dose of 15 mg—with placebo in a yearlong study.

The first portion of the study design followed the standard approach to measuring the effectiveness of maintenance treatment in erosive esophagitis. Patients with healed esophagitis were treated until relapse or, if they remained healed, for as long 1 year. In the primary efficacy analysis, we compared the effect of lansoprazole with that of placebo on the time until endoscopically documented recurrence of erosion. However, the study differed from usual maintenance studies at the point at which patients had a documented relapse. After the first recurrence, patients were not immediately withdrawn from the study. Instead, patients who had relapse entered the second portion of the study and were treated with open-label lansoprazole, 30 mg/d for 8 weeks. If erosion rehealed with this treatment, the patient continued in the study and received double-blind therapy in a manner that matched the original randomization code. Patients could thus remain in the study after one or more recurrences, but they could remain no longer than 12 months. This portion of the study design was used to meet a secondary objective: to compare the number of times erosion recurred while patients received daily maintenance treatment (lansoprazole treatment groups) with the number of recurrences that developed while patients received active treatment only at the time of a diagnosed recurrence (placebo group).

Methods

Top Methods Results Discussion Author & Article Info References

Study Design

This multicenter study was conducted at 25 centers ( 20 patients/center) and enrolled 186 patients who had a recent history of erosive reflux esophagitis. Investigators enrolled patients who showed healing of erosive esophagitis at the end of an 8-week short-term study [14]. In addition, new patients (recruited primarily from the investigators' existing patient sample) who met entrance criteria similar to those in the short-term study and patients who were not healed at the end of short-term study received open-label lansoprazole, 30 mg/d for 8 weeks. At the end of this open-label lead-in phase, patients with endoscopically proven healing of erosive esophagitis entered the double-blind maintenance phase of the study. Healing was documented within 7 days before patients entered the maintenance study.

At entry into the 12-month maintenance study, patients were randomly assigned in such a way that the treatment groups at each site had similar numbers of patients. We used double-blind procedures and separate randomization schedules with consecutive numbering for each site to assign patients to receive placebo, 15 mg of lansoprazole, or 30 mg of lansoprazole in a single daily dose before breakfast. All medications were similar in appearance, taste, and directions for use. All study personnel (including participants, investigators, endoscopists, study monitors, pharmacists, statisticians, and pathologists) were blinded to treatment during the study.

Endoscopy was done after 1, 2, 3, 6, 9, and 12 months. In addition, at any time during the maintenance period, a patient who reported symptoms suggesting recurrence of esophagitis was examined by endoscopy. If an erosive recurrence, symptomatic or asymptomatic, was documented on endoscopy, the patient received open-label lansoprazole, 30 mg/d for 8 weeks. If repeated endoscopy showed healing after 8 weeks of open-label treatment, the patient returned to the originally assigned treatment. Patients could be treated in this manner as many times as necessary during the 12-month period. If erosions did not heal, the patient was dropped from the study.

Patients who completed the 12-month maintenance phase entered a 3-month post-treatment phase in which no acid-suppressive medication was administered. Patients had endoscopy after 1 month and at the end of this period.

Patients were given Gelusil tablets [Parke-Davis, Morris Plains, New Jersey] for relief of discomfort as needed. Patients were not allowed to take other acid-suppressive or ulcerogenic medications during any phase of the study except for 1) 325 mg or less of aspirin per day for cardiovascular indications or two doses of a nonsteroidal anti-inflammatory drug per week for any indication and 2) systemic or topical corticosteroids not exceeding 10 mg of prednisone (or equivalent) per day.

Evaluations were scheduled before randomization and took place monthly for the first 3 months and then at 6, 9, and 12 months. Each regular study visit included endoscopic assessment; evaluation of symptoms, Gelusil use, and adverse events; physical examination and recording of vital signs data; and clinical laboratory evaluations, including measurement of the fasting serum gastrin level. Gastric biopsies were done at each scheduled endoscopy.

To standardize the grading scale and biopsy procedures, endoscopists were trained at an investigator meeting. At each site, efforts were made to ensure that the same endoscopist would do all endoscopies for a particular patient. All biopsy specimens were evaluated at a central location under the direction of Drs. Walter Rubin, Barbara Atkinson, and Marian Haber at the Medical College of Pennsylvania in consultation with Dr. Juan Lechago from Baylor College of Medicine.

Patients

Patients enrolled in the study had completed an earlier study of short-term lansoprazole treatment for erosive esophagitis [14] or had received a diagnosis of erosive esophagitis and had never received lansoprazole therapy. Patients showed endoscopic evidence of grade 2 or higher esophagitis on the modified Savary-Miller scale (that is, at least one or more erosions proximal to the squamocolumnar junction) before receiving short-term healing treatment. Patients were required to have endoscopic evidence of healing within 7 days of entering the double-blind maintenance phase. "Healing" was defined as a return of the esophageal mucosa to grade 0 or grade 1 (that is, no evidence of erosion).

The investigational review board at each site approved the protocol and patient consent form for study participation. Patients were excluded from the trial if they were younger than 18 years of age or had not given informed consent.

Study End Points

The primary efficacy variable was the time to the first recurrence of erosive esophagitis, symptomatic or asymptomatic, as determined by endoscopy. Erosive recurrence was defined as grade 2 or higher. Other measures of efficacy during this period included maintenance of symptom relief, severity of daytime and nighttime heartburn, and frequency of Gelusil use. Throughout the 12-month maintenance period, we also analyzed the number of erosive relapses and the frequency of Gelusil use; at the final treatment visit, we analyzed the severity of symptoms. We did not measure baseline Gelusil use.

Safety variables included assessments of adverse events, measurement of clinical laboratory variables (including gastrin levels), physical examination, electrocardiogram, and vital signs data. Full-mucosal thickness biopsy specimens from the greater curvature were obtained in a subset of patients in the clinical trial. Slides were evaluated qualitatively for patterns of enterochromaffin-like cell hyperplasia according to the Solcia classification of gastric endocrine cell growth [15].

Statistical Analysis

We used the SAS software program (SAS Institute, Cary, North Carolina) to do all statistical analyses. Analyses were based on two-tailed tests, and significance was defined as a P value 0.05 or less.

We compared the percentage of patients whose erosive esophagitis remained healed among treatment groups. Because this analysis used data obtained until the first erosive recurrence only, it did not differ from what would have occurred had patients been dropped after the first recurrence. Probabilities of remaining healed after months 1, 2, 3, 6, 9, or 12 of maintenance therapy were estimated using a life-table method [16], in which it is assumed that patients who withdrew from the study would have the same rate of recurrence as those who did not withdraw. The P values in pairwise comparisons of maintenance rates among treatment groups were based on Cochran-Mantel-Haenszel tests, with time as the stratification factor [16]. Endoscopies done more than 5 days after the last day of dosing in any treatment period were not used in the analysis.

At each visit, the investigators classified the severity of daytime and nighttime heartburn as none, mild (symptom was easily tolerated), moderate (symptom interrupted usual activities), or severe (symptom interfered greatly with usual activities and may have been incapacitating). Patients were considered symptomatic if symptoms were rated moderate or severe. We estimated the probabilities of remaining asymptomatic during treatment using the same lifetable method described in the preceding paragraph [16]. We analyzed changes in symptom severity between entry into the maintenance study and the end of the maintenance study using Cochran-Mantel-Haenszel tests for ordered response categories stratified on the value at entry into the maintenance study. We excluded assessments that had been done during open-label treatment periods or more than 5 days after the last dosing period. Differences in average Gelusil use among treatment groups were compared using the Wilcoxon two-sample test.

We tabulated treatment-emergent adverse events by frequency and severity. The mean changes in laboratory values from baseline to the final treatment visit were analyzed by one-way analysis of variance. Percentage changes in laboratory values from baseline to final treatment visit were categorized and analyzed with Cochran-Mantel-Haenszel tests for ordered response categories; the baseline value (low, medium, or high) was used as the stratification variable. Fasting gastrin values were obtained from samples taken after a fast of at least 8 hours and were compared pairwise between treatment groups using the Wilcoxon two-sample test.

TAP Holdings, Inc., participated in the data collection and analysis of these results.

Results

Top Methods Results Discussion Author & Article Info References

Demographic Characteristics and Patient Disposition

Patient Data Sets

We enrolled 186 patients into the study: 58 new patients and 128 from the previous study. Thirteen patients were dropped during lead-in: Nine had remained unhealed after open-label treatment, and four had not completed the lead-in phase. Thus, 173 patients (57 placebo recipients, 60 recipients of 15 mg of lansoprazole, and 56 recipients of 30 mg of lansoprazole) entered the double-blind treatment phase. Two patients had no endoscopies during double-blind treatment, and one patient received additional acid-suppressive medication. One hundred seventy evaluable patients remained: 55 in the placebo group, 59 in the 15-mg lansoprazole group, and 56 in the 30-mg lansoprazole group.

Demographic Characteristics and Clinical Status

Of the 170 evaluable patients, 162 were white and 92 were men. They ranged in age from 19 to 80 years (mean, 45 years). One hundred forty-two patients had received lansoprazole immediately before the maintenance phase, and 28 had received ranitidine. Fifty-one percent of evaluable patients had grade 3 esophagitis before healing. Four patients (2 in each lansoprazole group) had concomitant Barrett esophagus. The demographic and clinical variables for each treatment group are summarized in Table 1. These baseline variables did not differ significantly across treatment groups.

During the maintenance phase, 23 patients took nonsteroidal anti-inflammatory drugs or aspirin until erosions recurred or until the final visit (6 placebo recipients, 5 recipients of 15 mg of lansoprazole, and 12 recipients of 30 mg of lansoprazole). Of these patients, 3 in each treatment group took more than two doses of nonsteroidal anti-inflammatory drugs per week or more than 325 mg of aspirin per day.

Premature Withdrawals

Forty-one of the 173 treated patients (24%) withdrew prematurely during the 1-year maintenance treatment phase. More than a third (37%) of placebo recipients withdrew prematurely compared with 18% of patients receiving 15 mg of lansoprazole and 16% of patients receiving 30 mg of lansoprazole. The most common reason for premature withdrawal was treatment failure; 12 patients withdrew for this reason (10 placebo recipients and 2 recipients of 30 mg of lansoprazole). In addition, 2 placebo recipients and 2 patients receiving 15 mg of lansoprazole had esophagitis symptoms sufficient to warrant withdrawal from the study. Six patients withdrew because of adverse events, and a seventh withdrew because of an unintended pregnancy; these patients are described in the section on safety results during treatment.

Efficacy Results until the First Recurrence of Erosive Esophagitis

Endoscopy Results

Patients were treated until erosion recurred or, if they remained healed, for as long as 1 year. Figure 1 shows the life-table analysis of patients who remained healed during the 12-month study period. Both lansoprazole doses were significantly superior to placebo in maintaining healing over the study period (P < 0.001), but no significant difference was seen between doses. Forty-five percent of placebo recipients remained healed at 1 month compared with 93% of patients receiving 15 mg of lansoprazole and 98% of patients receiving 30 mg of lansoprazole. By month 12, only 24% of placebo recipients remained healed compared with 79% of patients receiving 15 mg of lansoprazole and 90% of patients receiving 30 mg of lansoprazole.

View larger version (14K): [in this window] [in a new window]   Figure 1. Proportion of patients who remained healed (that is, no erosions seen on endoscopy) during the 1-year maintenance period. Lansoprazole significantly improved maintenance of healing rates compared with placebo (P < 0.001). Within 1 month, 45% of placebo recipients remained healed compared with 93% of patients receiving 15 mg of lansoprazole and 98% of patients receiving 30 mg of lansoprazole.

The analysis by esophagitis grade before healing Figure 2 indicated that none of the four placebo recipients with grade 4 esophagitis at study entry remained healed 2 months after stopping active treatment. In contrast, 7 of 10 lansoprazole recipients with grade 4 esophagitis at study entry remained healed throughout the maintenance period.

View larger version (17K): [in this window] [in a new window]   Figure 2. Proportion of patients who remained healed (that is, no erosions seen on endoscopy) during the 1-year maintenance period, by treatment group and baseline esophagitis grade before healing. No placebo recipients and 7 of 10 lansoprazole recipients with grade 4 esophagitis at baseline remained healed at 2 months. Lansoprazole was significantly superior to placebo in maintaining healing, even when data are stratified by baseline esophagitis grade. Percentages were calculated by life-table methods.

We considered the following concomitant factors in our analysis of the percentage of patients remaining healed: age, sex, antisecretory treatment received immediately before maintenance treatment, esophagitis grade before healing, presence of hiatal hernia, and treatment center. We did Cochran-Mantel-Haenszel tests in which time and each concomitant factor considered individually were used as stratification variables. In these analyses, lansoprazole remained significantly (P < 0.001) superior to placebo. No significant differences were seen between the two lansoprazole groups.

Only three patients were excluded from the evaluable patient analysis. Thus, the intention-to-treat analysis, which was based on all patients, resulted in almost identical findings; the greatest difference between the two types of analyses was only 1 percentage point. Significant treatment differences persisted at the same magnitude as in the evaluable analysis.

Symptom Results

Patients were considered asymptomatic if they reported no or mild daytime or nighttime heartburn. Most patients were asymptomatic at study entry. Lansoprazole recipients remained asymptomatic significantly longer than did placebo recipients (P < 0.001) (Figure 3). No significant differences were noted between the lansoprazole groups.

View larger version (15K): [in this window] [in a new window]   Figure 3. Proportion of patients who remained asymptomatic during the 1-year maintenance period. Lansoprazole (Lan) was significantly superior to placebo in controlling symptoms of daytime and nighttime heartburn (P < 0.001). At 12 months, 35% of placebo recipients were asymptomatic compared with 72% of patients receiving 15 mg of lansoprazole and 67% of patients receiving 30 mg of lansoprazole. Percentages were calculated by life-table methods; patient data were considered to be censored at the time of the first erosive recurrence. An asymptomatic state was defined as no moderate or severe daytime or nighttime heartburn.

Fifty-five percent of placebo recipients who were asymptomatic at baseline remained asymptomatic after 1 month of maintenance treatment compared with 92% of patients receiving 15 mg of lansoprazole and 96% of patients receiving 30 mg of lansoprazole. On the basis of the life-table analysis, the probability of remaining asymptomatic during the 12 months was 35% for placebo recipients, 72% for patients receiving 15 mg of lansoprazole, and 67% for patients receiving 30 mg of lansoprazole. The severity of heartburn was also significantly greater among placebo recipients. At the time of first recurrence (or month 12 if the patients remained healed throughout the study), 7% of placebo recipients, 3% of patients receiving 15 mg of lansoprazole, and none of the patients receiving 30 mg of lansoprazole reported severe daytime heartburn. The results were similar for nighttime heartburn. Consistent with these findings, lansoprazole recipients used Gelusil significantly less frequently during this period than did placebo recipients (P < 0.001).

Efficacy Results for the Entire Maintenance Period

Endoscopy Results

Table 2 lists the number of times that erosions recurred in each patient by treatment group during maintenance therapy. Forty-five percent of placebo recipients, 8% of patients receiving 15 mg of lansoprazole, and 5% of patients receiving 30 mg of lansoprazole had two or more erosive recurrences during maintenance therapy that required open-label lansoprazole treatment. During the 12 months, erosion recurred three or more times in 35% of placebo recipients and 2% of lansoprazole recipients.

Symptom Results

When we analyzed all patients throughout the 12-month maintenance period, we found that lansoprazole effectively controlled the symptoms of daytime and nighttime heartburn compared with placebo (P < 0.001). Although few patients (2% of placebo recipients, none of the patients receiving 15 mg of lansoprazole, and 2% of the patients receiving 30 mg of lansoprazole) reported severe daytime heartburn at the start of maintenance therapy, 14% of placebo recipients reported severe symptoms at the final visit compared with 2% of patients receiving 15 mg of lansoprazole and none of the patients receiving 30 mg of lansoprazole. The results were similar for nighttime heartburn.

Safety Results during Treatment

Six patients withdrew from the study because of adverse events: two placebo recipients (one because of bloating and constipation and the other, who was receiving 30 mg of open-label lansoprazole at withdrawal, because of abdominal pain, syncope, and depression), one patient receiving 15 mg of lansoprazole (because of gastric polyp), and three patients receiving 30 mg of lansoprazole (one because of diarrhea, one because of chest pain, and one because of myocardial infarction). A seventh patient, treated with 30 mg of lansoprazole, withdrew prematurely because of an unintended pregnancy; after an uncomplicated pregnancy, the patient delivered a healthy term infant.

For the comparison of the incidence of adverse events among treatment groups, it should be noted that the duration of total exposure to the doubleb-lind study medication was approximately 1.7-fold longer in the lansoprazole groups than in the placebo group; this difference can be attributed to the number of placebo recipients who had rapidly recurrent erosive esophagitis. Nevertheless, little difference was seen between placebo and lansoprazole administration in the number of reported adverse events considered to be related to treatment. Two placebo recipients reported constipation during double-blind treatment that was considered to be related to treatment (mean duration of study drug exposure, 188 days), and five lansoprazole recipients (all in the 30-mg group) reported diarrhea during double-blind treatment that was considered to be related to treatment (mean duration of study drug exposure, 318 days). In most cases, the diarrhea was mild or moderate, lasted a short time (usually less than 2 weeks), and resolved while the patient continued treatment. Only one patient, mentioned previously, withdrew prematurely because of diarrhea.

No clinically significant differences were seen between the active treatment groups and the placebo group for any laboratory variable except fasting serum gastrin levels. During continuous lansoprazole treatment (including short-term healing and maintenance), fasting serum gastrin levels increased approximately 55% during the first month and then tended to remain at the same level for the rest of the treatment period. The elevated levels returned to baseline levels within 1 month after cessation of lansoprazole treatment. Although median levels were higher in lansoprazole recipients than in placebo recipients (P < 0.001), they remained within the normal range (0 pg/mL to 100 pg/mL) at the end of maintenance treatment. Eleven patients had gastrin levels that exceeded 400 pg/mL during lansoprazole treatment.

Greater curvature biopsy specimens were obtained at study entry and the final visit from 112 patients. Among the 36 placebo recipients and 38 patients receiving 15 mg of lansoprazole who had biopsy results, all specimens were classified as normal at both time points. Among the 38 patients receiving 30 mg of lansoprazole who had biopsy results, 35 biopsy specimens were classified as normal at both time points. One female patient, from whom no biopsy specimen was obtained before she began receiving 30 mg of open-label lansoprazole for 8 weeks, had micronodular hyperplasia at the maintenance baseline visit and at all but one of her treatment visits. Her gastrin levels ranged from 42 pg/mL before treatment to 98 pg/mL at month 12. In the remaining two patients, one man and one woman, cellular findings were normal at baseline and indicated simple hyperplasia at the final visit; the accompanying gastrin levels for the two patients were 106 pg/mL and 150 pg/mL, respectively. No evidence of carcinoid tumors or dysplasia was seen in any patient during the course of the study.

No clinically significant differences between lansoprazole recipients and placebo recipients were seen in any vital sign index. No adverse trends were seen with review of changes in physical examination results, and no clinically significant electrocardiographic changes were noted.

Discussion

Top Methods Results Discussion Author & Article Info References

Our results show that lansoprazole is effective in preventing recurrence of erosive esophagitis. By 12 months, only 24% of the placebo recipients remained healed compared with 79% of patients receiving 15 mg of lansoprazole and 90% of patients receiving 30 mg of lansoprazole. Lansoprazole was also effective in preventing the recurrence of moderate and severe daytime and nighttime heartburn. By the end of the 12-month maintenance period, only about one third of the placebo recipients remained asymptomatic compared with more than two thirds of the lansoprazole recipients. Among patients with symptoms, 14% of the placebo recipients reported severe daytime heartburn compared with only 2% of the patients receiving 15 mg of lansoprazole and none of the patients receiving 30 mg of lansoprazole. The results were similar for nighttime heartburn.

We enrolled patients with esophagitis severity similar to that of patients studied in other trials. In our study, 59% of patients had grade 3 or 4 esophagitis before short-term healing. The recurrence rates in the placebo group were similar to those reported previously for the rapid recurrence of erosive esophagitis upon cessation of active treatment [3-7]. One month after active treatment was stopped, 45% of placebo recipients remained in remission, and by 12 months, only 24% of placebo recipients remained healed.

Given that lower doses of currently available acid-suppressive medications have not been found to be as effective in maintenance treatment as their full-healing doses [8-11], one of our major objectives was to determine whether a lower dose of lansoprazole, rather than the full healing dose, was needed to maintain healing in patients with erosive esophagitis. Lansoprazole, 30 mg, has been shown to be effective in producing rapid healing of all grades of esophagitis and to be superior to ranitidine in both endoscopically documented healing and symptom relief [14, 17, 18]. The 30-mg daily dose has been shown to be superior to the 15-mg dose in the short-term healing of erosive esophagitis [18]. In contrast, our study showed that both doses of lansoprazole were very effective in maintaining healing of erosive esophagitis for as long as 1 year and in maintaining relief of daytime and nighttime heartburn. No statistically significant differences were seen between the two doses in any of the efficacy analyses. The rates of maintaining healing and of preventing symptom recurrence were similar. Therefore, the lower 15-mg daily dose of lansoprazole should be an effective maintenance regimen for erosive esophagitis.

Our study was also designed to look beyond the first relapse after active treatment was stopped and to compare maintenance treatment with treatment of recurrences only. The results showed that nearly half (45%) of the placebo recipients had two or more erosive relapses during the maintenance year and that more than one third (35%) had three or more relapses. These findings suggest that the erosive relapse rate is so high that patients are frequently receiving full doses of acid-suppressing agents on an "as-needed" basis and are receiving significantly more antacids. A disadvantage to this treatment strategy is that, in clinical practice, patients probably often take other acid-suppressing agents to suppress symptoms without taking a dose adequate to ensure healing. In contrast, with maintenance lansoprazole treatment, only 4% to 7% of patients had two relapses during the maintenance year, and 2% had three relapses. Therefore, more than 90% of the patients receiving long-term therapy remained healed without erosions throughout almost the entire year.

A limitation of our study was that, because patients were treated for asymptomatic erosive changes, we could not determine the frequency of such changes that led to symptomatic recurrences. We thus could not ascertain how often patients would have received full healing courses if this decision had been made only on the basis of symptoms.

In our study, lansoprazole was as well tolerated as placebo. Diarrhea, which was the most frequently reported adverse event in patients who received lansoprazole, was usually transient and did not persist with continued use of lansoprazole. One patient withdrew from the study prematurely because of diarrhea.

As expected on the basis of previous reports of increased gastrin levels among patients treated with proton pump inhibitors [19-21], fasting serum gastrin levels in the lansoprazole groups were 1.5 times greater than those in the placebo group. However, the median levels remained within the normal range (0 pg/mL to 100 pg/mL) at the end of maintenance treatment, and levels returned to the pretreatment baseline levels within 1 month after lansoprazole was stopped. No evidence of carcinoid tumors or any dysplastic endocrine cell growths was seen in any patient during the study. These findings are consistent with the long-term clinical experience with omeprazole [22]. Solcia and colleagues [22] reviewed biopsy specimens from 443 patients receiving long-term omeprazole therapy (range of treatment duration, several months to 4 years). They concluded that there was no evidence of neoplastic, dysplastic, or true proliferative changes with long-term acid suppression therapy and therefore that there was no safety concern about gastric endocrine cells with proton pump inhibitors.

In conclusion, lansoprazole was shown to be significantly more effective than placebo in maintaining healing and controlling symptoms. Lansoprazole, 15 mg, did not differ significantly from the 30-mg dose. These findings suggest that an effective treatment regimen for esophagitis is lansoprazole, 30 mg/d for 8 weeks to heal erosive disease, followed by 15 mg/d for maintenance.

Dr. Lanza: Houston Institute for Clinical Research, Baylor College of Medicine, 7777 Southwest Freeway, Houston, TX 77074.

Dr. Avner: 1220 East 3900 South, Salt Lake City, UT 84124.

Dr. Haber: Department of Pathology and Laboratory Medicine, Medical College of Pennsylvania and Hahnemann University, 3300 Henry Street, Philadelphia, PA 19129.