BACKGROUND The common but incompletely understood entity of malabsorption of food bound cobalamin is generally presumed to arise from gastritis and/or achlorhydria.
AIM To conduct a systematic comparative examination of gastric histology and function.
SUBJECTS Nineteen volunteers, either healthy or with low cobalamin levels, were prospectively studied without prior knowledge of their absorption or gastric status.
METHODS All subjects underwent prospective assessment of food cobalamin absorption by the egg yolk cobalamin absorption test, endoscopy, histological grading of biopsies from six gastric sites, measurement of gastric secretory function, assay for serum gastrin and antiparietal cell antibodies, and direct tests for Helicobacter pyloriinfection.
RESULTS The six subjects with severe malabsorption (group I) had worse histological scores overall and lower acid and pepsin secretion than the eight subjects with normal absorption (group III) or the five subjects with mild malabsorption (group II). However, histological findings, and acid and pepsin secretion overlapped considerably between individual subjects in group I and group III. Two distinct subgroups of three subjects each emerged within group I. One subgroup (IA) had severe gastric atrophy and achlorhydria. The other subgroup (IB) had little atrophy and only mild hypochlorhydria; the gastric findings were indistinguishable from those in many subjects with normal absorption. Absorption improved in the two subjects in subgroup IB and in one subject in group II who received antibiotics, along with evidence of clearing of H pylori. None of the subjects in group IA responded to antibiotics.
CONCLUSIONS Food cobalamin malabsorption arises in at least two different gastric settings, one of which involves neither gastric atrophy nor achlorhydria. Malabsorption can respond to antibiotics, but only in some patients. Food cobalamin malabsorption is not always synonymous with atrophic gastritis and achlorhydria, and hypochlorhydria does not always guarantee food cobalamin malabsorption.
- cobalamin malabsorption
- atrophic gastritis
- Helicobacter pylori
Abbreviations used in this paper
- egg yolk cobalamin absorption test
- maximum acid output
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Food cobalamin malabsorption, defined as the inability to absorb food bound or protein bound cobalamin while absorption of free cobalamin is intact, may be the most common form of cobalamin malabsorption. Evidence suggests that its frequency exceeds that of classical disorders of free cobalamin absorption such as pernicious anaemia.1 Nevertheless, the mechanisms responsible for food cobalamin malabsorption are incompletely understood.
Much clinical and laboratory evidence points to gastric dysfunction, especially loss of acid secretion, as a key factor.1-6Although the stomach's role is undoubted, selection may have influenced the strength of that association7 because most of the initial studies identified gastric surgery, gastric atrophy, or use of acid suppressing drugs as predisposing conditions and selected their study subjects from such populations.2 ,3 There has been no prospective systematic assessment of all the gastric correlates of food cobalamin malabsorption.
The present study was designed to address this gap in our understanding by examining gastric histology and gastric function in a blinded prospective fashion and comparing these characteristics with absorption results. The aim was to develop a profile of the stomach in food cobalamin malabsorption that could be compared concurrently with findings in subjects without malabsorption.
Most of the subjects were recruited during a community survey of cobalamin status in apparently healthy elderly subjects.8 Proportionately more subjects found to have low rather than normal cobalamin levels during the survey volunteered for the study, but their absorption status was not known beforehand. All participants were community dwellers and had no contraindications for the test procedures. Subjects with a history of gastrointestinal surgery (other than remote appendicectomy or cholecystectomy) or receiving acid suppressive medications were excluded. Although our initial focus was on volunteers >60 years old, six subjects below that age were subsequently accepted. The only subjects that were not recruited from a presumably healthy population were two patients who had been referred for testing because of mild cobalamin deficiency. The objective of the recruitment was to provide enough subjects with and without malabsorption for comparative analyses of gastric status, rather than to determine the prevalence rates of malabsorption. Median age of the subjects was 69 years (range 33–81), 11 were men and eight were women, and there were 12 whites, four blacks, two Latin Americans, and one Asian-American. The study was approved by the institutional review board at the Los Angeles County-USC Medical Center and each participant gave written informed consent.
The egg yolk cobalamin absorption test (EYCAT) method, modified from that of Doscherholmen and colleagues9 by omitting the dual isotope feature, has been previously described.10 In brief, 57Co-cyanocobalamin and unlabelled cobalamin were mixed with reconstituted lyophilised egg yolk in a Waring blender. The egg yolk preparation was scrambled in a Teflon coated skillet, and 12 or 13, 50 g portions were weighed out from this batch and stored at −20°C until use. Each portion contained about 0.8 μg of cobalamin (approximately 0.5 μCi of 57Co). On the day of testing, the fasted subject ate the scrambled egg yolk portion that had been reheated in a microwave oven; no other food or liquids were taken for four hours. The rest of the procedure was the same as in the standard Schilling test, with an intramuscular injection of 1000 μg of cyanocobalamin given two hours after the test meal. A 24 hour urine collection was done. The radioactivity excreted was compared with a 1 g egg yolk cobalamin standard from the original batch preparation.
Based on results in 168 normal and abnormal subjects, 24 hour excretion values below 1% indicate severe food cobalamin malabsorption and values ⩾2% represent normal absorption. Values of 1.00–1.99% represent probable or mild malabsorption. EYCAT reproducibility had been shown in the retesting of seven subjects within 1–6 months. Two subjects with severe malabsorption had repeat values of 0.43%/0.24% and 0.19%/0.17%/0.10%, respectively. Five subjects with normal absorption had repeat values of 4.06%/4.10%, 4.66%/5.38%, 2.67%/3.63%, 2.85%/3.21%, and 2.91%/2.25%, respectively.
After baseline blood samples were obtained, all fasted volunteers underwent the EYCAT. In order to rule out pernicious anaemia (defined gastroenterologically here as malabsorption of free cobalamin due to loss of intrinsic factor secretion) as the explanation for any abnormal EYCAT results, standard Schilling tests, which measure absorption of free cobalamin, were performed in all subjects with abnormal EYCAT results who agreed to undergo the test. Patients with pernicious anaemia absorb both free and food bound cobalamin poorly and usually have EYCAT results below 1%.2 ,10 Because a normal EYCAT result is presumptive proof that absorption of free cobalamin is also normal, Schilling tests were rarely done in patients with normal food cobalamin absorption. Other markers of pernicious anaemia, such as gastric intrinsic factor secretion, gastrin levels, and antibodies to intrinsic factor, were also assessed in most patients. As a result, two subjects with findings suspicious of pernicious anaemia were excluded from the study and are not presented here.
ENDOSCOPY AND BIOPSY
After absorption testing, each volunteer underwent upper gastrointestinal endoscopic examination. The procedure was performed by an investigator who was blinded to the subject's absorption test results. Biopsies were taken from standardised sites as shown in fig 1, using a large capacity pinch biopsy forceps with an 8 mm open span. The two biopsies on the greater curve of the body (mid body and 3 cm distal to it) were designed to sample the oxyntic mucosa where the folds are thickest. The biopsy of the mid body lesser curve was designed to help complete the overall picture of the status of the oxyntic mucosa, fully recognising that oxyntic glands in the lesser curve of the body are often thinner and gastritis, if present, may appear to be more severe.11
All biopsies were mounted mucosal side up on a plastic mesh and fixed in Bouin's solution for 2–6 hours before transfer to 70% alcohol and subsequent processing for paraffin embedding. Two slides each containing 15 serial sections at 4 μm were prepared, one for haematoxylin and eosin staining and the second for staining with haematoxylin and eosin-alcian blue, pH 2.5, to facilitate detection and grading of intestinal metaplasia. Four to five additional sections were stained with the Genta stain for identification ofH pylori.12
Biopsy specimens were coded, mixed, and evaluated by a single observer without knowledge of any of the other findings. A similarly blinded observer examined the specially stained tissue sections forH pylori.
Gastric inflammatory, epithelial, and glandular changes were scored separately in a similar fashion to that described previously.13 For inflammation, neutrophils and mononuclear cells were each graded separately with scores of 0–3, yielding a combined inflammatory cell density score ranging from 0 to 6. A score of 1 for neutrophils represented 1–2 extravascular foci of neutrophils in the lamina propria or the epithelium. A grade of 2 represented 3–4 such foci, and a grade of 3, >4 foci. Grade 3 generally reflected the presence of neutrophil collections in every high power field. For mononuclear cells, a grade of 1 represented a perceived mild increase, whereas grade 3 represented dense infiltrates creating an apparent increased space between adjacent pits (foveolae) with or without extension into the subjacent glands. The surface and pit epithelia were assessed separately on scales of 0–3, with an overall epithelium score of 0–6. The grades of mild to severe changes (scores 1–3) integrated the severity and diffuseness of epithelial abnormalities on serial sections. The abnormalities that were especially noted were mucin depletion with cytoplasmic basophilia, disturbed nuclear polarity, enlarged nuclei, and prominent nucleoli. Different gastric components were assessed separately. The scores for inflammation and epithelium were summed up in a “gastritis index score” of 0–12.
The presence or absence of atrophy in the oxyntic mucosa, especially from the greater curve, was a central question in this study. The oxyntic mucosa consists of very shallow pits and most of the thickness of the mucosa is occupied by glands (fig 2). An atrophy grade was given to help provide a simple “mind's eye” view of the appearance of the gastric mucosa in terms of the presence or absence of atrophy. A score of 0 represented no atrophy, a score of 1 represented non-atrophic inflammation (superficial gastritis), and scores of 2, 3, and 4 represented increasing degrees of atrophy based on a perception of loss of glands involving the top, middle, and bottom thirds of the mucosal gland zone, respectively. This loss could be associated with either inflammatory cells encroaching into, and seemingly replacing, glands or with intestinal metaplasia or both. Restated, grade 2 represents mild loss of glands and grade 4 marked loss of glands.
Intestinal metaplasia was graded 0–4 based on the estimate of the horizontal span involvement by quartiles. Grade 1 represents up to 25% intestinalisation of the horizontal span of the epithelium and grade 4 represents >75% involvement. As mentioned previously, the presence of full thickness mucosal intestinal metaplasia mandated a score of severe atrophy (grade 4) for either oxyntic or antral mucosa.
Antral glands occupy less than 50% of the overall thickness of the mucosa. Therefore, atrophy was graded as present (score 4) or absent. Because the foveolae and intervening lamina propria occupy most of the antral mucosal plane of thickness, scores for superficial non-atrophic gastritis were graded as 1 (mild), 2 (moderate), and 3 (severe).
GASTRIC SECRETORY STUDIES
Several days after endoscopy, gastric intubation was performed in the fasted subject without radiographic monitoring. Basal and pentagastrin stimulated (6 μg/kg) gastric aspirates were each collected for one hour by continuous suction. Volume and pH by glass electrode were measured, and maximum acid output (MAO) was calculated in mEq/h. Pentagastrin stimulated gastric juice was also neutralised and assayed for intrinsic factor content14 and pepsin concentration.15
HELICOBACTER PYLORI TESTING
In addition to the “gold standard” of histological assessment of H pylori, as described above, subjects were tested by 13C urea breath testing (Meretek Diagnostics, Nashville, Tennessee, USA).
Gastric aspirates for quantitative cultures were collected into a sterile container at the time of endoscopy via the biopsy channel of the endoscope before air insufflation, after discarding the initial aspirate of about 3 ml. An attempt was also made to obtain fluid from the duodenal lumen via a catheter inserted through the biopsy channel of the endoscope at the time of endoscopy. Cultures were considered positive when >105 organisms/ml were obtained. Because our sampling technique did not guarantee exclusion of air, anaerobic culture results are not included.
Data were tabulated and analysed by standard statistical methods using SAS for Windows software version 6.08 (SAS Institute, Cary, North Carolina, USA). These included Student's ttest, Wilcoxon test, logistic regression analysis, Fisher's exact test, and Spearman's and Pearson's correlations as appropriate.
Eleven of 19 subjects had food cobalamin malabsorption, although this should not be construed as a prevalence rate. Malabsorption was severe in six (group I; EYCAT <1.00%) and mild in five (group II; EYCAT 1.00–1.99%). The remaining eight subjects had normal absorption (group III; EYCAT ⩾2.00%). The findings in the three groups are shown in table 1.
Schilling test results were normal in all eight subjects tested, including the five subjects with severe malabsorption. None of the 19 subjects in any of the groups had anti-intrinsic factor antibodies. Further proof that pernicious anaemia (free cobalamin malabsorption due to lack of intrinsic factor) was not present in any of the 19 subjects was provided by normal intrinsic factor levels in 15 of 16 subjects who underwent gastric testing; the remaining subject (No 3) had subnormal, but not absent, intrinsic factor. He was not regarded as having pernicious anaemia because his normal Schilling test result showed that he did not malabsorb free cobalamin; indeed, the Schilling test result was again normal when he was retested six months later. Only two of the 19 volunteers had antiparietal cell antibodies, both of whom were in group I and had severe gastritis and achlorhydria (subject Nos 1 and 3); their normal Schilling test results ruled out a diagnosis of pernicious anaemia.
HISTOLOGICAL FINDINGS AND HELICOBACTER PYLORI STATUS (TABLE 2)
Atrophy grade (reflecting the severity of atrophy) for oxyntic biopsy sites E and F was significantly higher in group I than in group III (p=0.003). Mean (SD) overall grades were 2.8 (1.2) in group I compared with 1.2 (1.0) in group II and 0.8 (0.7) in group III (p=0.01).
However, the most striking aspect of the histological findings was that they differed among the six group I subjects despite the subjects' equally severe malabsorption (table 2). Subject Nos 1–3 were therefore designated subgroup IA because they had severe oxyntic gland atrophy and intestinal metaplasia (exemplified in fig 3). They also had achlorhydria (see below). In contrast, subject Nos 4–6 were assigned to subgroup IB because they did not have atrophy (exemplified in fig4). They were H pylori positive and their major histological scores, that is the gastritis index score and degree of corpus atrophy and intestinal metaplasia, were not significantly different from those of H pylori infected subjects without malabsorption (group III) or with only mild malabsorption (group II; exemplified in fig5).
Ten subjects nearly equally distributed among the three absorption groups were H pylori positive by histological stain (table 2) as well as 13C urea breath testing; the remainder were negative on both tests. Unlike the overlap in gastritis index scores between groups I and III (3.48 (3.16)v 2.91 (2.79); p=0.72), the scores were significantly different between H pyloriinfected and uninfected subjects (5.28 (2.20)v 0.79 (0.71); p=0.0003 for all six sites combined).
As reflected by the gastritis index scores in table 2, the severity of inflammation in the antrum was similar to that of the body (r=0.8, p<0.0001). However, only subject No 16 in group III, who had normal acid secretion, had antral atrophy (grade 4.0). Intestinal metaplasia was minimal, which suggests that the antral atrophy in subject No 16 (and the severe superficial gastritis in subjects Nos 5, 6, and 10) was focal rather than extensive.
GASTRIC SECRETION RESULTS (TABLE 1)
Pentagastrin stimulated gastric juice was examined in 15 of 22 subjects; the 16th subject (No 6) only allowed collection of her basal specimen which thus provided limited information. MAO, which did not correlate with age, sex, H pylori status, or gastric index scores, correlated inversely with corpus atrophy scores (p=0.001) and was lower in group I than in group III (1.86 (2.59)v 11.20 (12.53) mEq/h; p=0.047). Gastric pepsin concentrations, which also correlated inversely with corpus atrophy scores (p=0.02), were also lower in group I although the difference between groups I and III did not achieve statistical significance (278 (115) v 584 (93) U/ml; p=0.064).
Despite the significant overall differences, some overlap in MAO values was seen between groups I and III. Several subjects with normal absorption had lower MAO than two of the severely malabsorbing subjects (subject Nos 4 and 5). An overlap in gastric pepsin concentration was also noted. Indeed, subject No 13, whose absorption was normal, had the lowest pepsin concentration in the entire study.
The most striking gastric secretory finding was that subjects in group I could be divided into the same two distinct subgroups (IA and IB) as their histological and H pylori findings (table 2). All three subjects in subgroup IA had achlorhydria and low pepsin concentrations, as expected in gastric atrophy (table 1). Two of the three also had very high serum gastrin levels and positive antiparietal cell antibodies; the third (subject No 2) had normal gastrin levels presumably because he alone also had diffuse antral intestinal metaplasia and atrophy. In contrast, both subjects in subgroup IB whose stimulated gastric juice was available had normal pepsin secretion and only mild hypochlorhydria despite equally severe malabsorption; they also had either normal or only minimally elevated serum gastrin levels and did not have antiparietal cell antibodies (table 1). As mentioned previously, their histological findings were not unlike those in group III (table 2).
No bacterial growth was found in 12 of the 15 gastric juices and in five of the eight duodenal aspirates that were cultured. In total, five subjects, all with malabsorption, had positive cultures for either or both fluids (subjects Nos 1, 2, 3, 5, and 10). Alpha and gammaStreptococci,Neisseria,Xanthomonas, and/orPseudomonas species were identified among the different subjects.
RESPONSE TO ANTIBIOTICS
Five subjects with severe malabsorption (subject Nos 1–5) and one with mild malabsorption (subject No 10) were treated with oral antibiotics. These six subjects included all five who had positive aerobic bacterial cultures and three who were positive forH pylori. Tetracycline was given alone for one week or, in subject No 5, for two weeks together with metronidazole and bismuth subsalicylate.
On retesting 1–7 days after completion of antibiotics, the abnormal EYCAT results did not improve in the three subjects in subgroup IA, who were all achlorhydric and had severe gastric atrophy. All three non-responders were also H pylori negative but had positive gastric and duodenal aerobic bacterial cultures. In contrast, EYCAT improved in subjects Nos 4 (from 0.01% to 2.07%), 5 (from 0.41% to 1.45% and 1.26%), and 10 (from 1.34% and 1.69% to 5.40%); all three responders had much less severe histological changes than non-responders and had normal or borderline MAO,H pylori infection, and variable aerobic bacterial growth. All three subjects who improved also “cleared” their H pylori infection by histological and breath test criteria within one week of treatment. Interestingly,H pylori infection recurred in subject No 10 as his EYCAT returned towards his pretreatment values (1.87% and 2.22%). The responders and non-responders could not be distinguished by their pretreatment aerobic bacterial cultures, which had been positive in nearly all cases. None of the gastric biopsies repeated after antibiotic treatment showed significant changes compared with their histological findings before treatment, apart from disappearance of H pylori.
Food cobalamin malabsorption occurs when the vitamin cannot be released from its binding to proteins in food and fails to become available for intrinsic factor mediated absorption.1 The release process has been shown by in vivo and in vitro studies to require acid and pepsin and is presumed to occur in the stomach.1-6 Therefore, gastric dysfunction is viewed as the key to any malfunction. Indeed, food cobalamin malabsorption was first described in patients with gastric resection, vagotomy, or gastritis with achlorhydria.2 ,3 Subsequent studies showed that acid suppressive drugs often depressed food cobalamin absorption also.6 ,16 ,17
Nevertheless, this may be only a partial picture of the process and its disorders. The frequent selectivity for patients with gastric dysfunction in clinical studies of food cobalamin malabsorption, including most of the pioneering studies, may have contributed to an exclusive emphasis on achlorhydria. However, recurrent indications that achlorhydria alone may not explain all cases of malabsorption5 10 18–20 and the lack of systematic information on gastric status suggested that a prospective comparison of gastric histology and function was needed in subjects with and without food cobalamin malabsorption who were not selected for known gastric dysfunction a priori.
The present results show that even though subjects with severe malabsorption (group I) had, overall, significantly lower acid secretion and worse histological findings in the corpus than those with normal absorption (group III), there were important and frequent exceptions. There was an overlap in both gastric function and histology between individual members of the three groups. The overlap in gastric function was evident to the extent that three of eight subjects with normal absorption and one of two subjects with mild malabsorption had poorer acid secretion than several subjects with severe malabsorption (table 1). This variability shows that hypochlorhydria and low pepsin levels occur both with and without malabsorption and therefore are not always sufficient in themselves to induce food cobalamin malabsorption.
The most noteworthy finding was the identification of two distinct subsets of gastric characteristics within group I, the group with severe malabsorption. Subgroup IA had the severe gastric atrophy, achlorhydria, and diminished pepsin secretion that fit the dominant model of food cobalamin malabsorption in the literature. Two of these three individuals also had marked hypergastrinaemia and positive antiparietal cell antibodies.
In contrast, subgroup IB presented a different picture. Hypochlorhydria was mild, pepsin secretion was adequate, and gastric histology, especially in the greater curve, showed notably mild abnormalities. In these characteristics, the functional and histological status of the stomach in subgroup IB was virtually indistinguishable from that found in many of the subjects whose absorption was normal (group III). It was largely these subjects who were responsible for the overlaps mentioned previously.
The existence of subgroups IA and IB, even though the numbers of subjects were small, is important because it clearly indicates that atrophic gastritis and achlorhydria are not the sole settings in which severe food cobalamin malabsorption occurs. However, despite the suggestiveness of active, biopsy proven H pylori infection in all three subjects in subgroup IB but in none of the subjects in subgroup IA, the nature of the alternative mechanism or mechanisms remains unclear.
Suter and colleagues18 reported that antibiotics improved food cobalamin malabsorption. They proposed that upper gastrointestinal anaerobes play a role in the malabsorption, although their in vitro studies showed no direct effect of anaerobes. H pylori status was not studied. Close examination of response to antibiotics was not a primary intent of our study, but we confirmed that antibiotics can reverse food cobalamin malabsorption. Unlike Suter and colleagues,18 however, we found that such a response is not universal but occurs in only some patients. The presence of achlorhydria and severe atrophy (subgroup IA) appeared to preclude any response to antibiotics, whereas subjects with mild gastric changes and active H pylori infection improved (subgroup IB or group II). The presence of aerobic bacterial contamination did not seem to distinguish between responders and non-responders. However, anaerobic cultures were not obtained.
Antibody findings and urea breath test data have indicated a significant epidemiological association between H pylori infection and severe food cobalamin malabsorption.19 ,21 Based on those results and our present observations, and despite the fact that many infected subjects both here and in previous studies did not have malabsorption, the possibility that the response to antibiotics was due to suppression ofH pylori must be considered. Gastritis induced by H pylori is known to reverse with antibiotics, with increased acid secretion within one week.22 Nevertheless, differentiation between suppression of H pylori or bacteria as the operative antibiotic effect is not possible from our data. While suggestive, the findings permit no definitive conclusion about the role ofH pylori in food cobalamin malabsorption. Neither H pylori nor bacterial infection was specific to food cobalamin malabsorption in our subjects.
In summary, our data clearly demonstrate that severe food cobalamin malabsorption can occur in patients who do not have atrophic gastritis of the oxyntic mucosa or achlorhydria. One immediate practical conclusion is that food cobalamin malabsorption should never be automatically equated with atrophic gastritis, contrary to some proposals.20 Indeed, those investigators stretched the concept by ignoring absorption test results if they conflicted with gastric histology findings.20 Nor should serum markers of atrophic gastritis be substituted for direct tests of food cobalamin absorption, which has also been suggested23 despite evidence suggesting their limited reliability.10 ,24 ,25Such markers occurred frequently in subgroup IA but not in subgroup IB.
In addition, our preliminary findings with antibiotic therapy confirm the original suggestion by Suter and colleagues18 that food cobalamin malabsorption responds to antibiotics. However, this is only true for some patients. The presence of severe oxyntic gland gastric atrophy and achlorhydria may militate against improvement in absorption by antibiotic administration, regardless of microbial status. A role for H pylori as a factor in at least some patients with food cobalamin malabsorption is supported by our data; however, its explanation is not clear and many infected subjects retain normal absorption. The mechanisms that produce food cobalamin malabsorption are clearly multiple. The mechanisms that are unrelated to achlorhydria will now require more attention and careful scrutiny.
This study was supported by grant DK 32640 from the National Institutes of Health and by the NIH National Center for Research Resources of the GCRC grant MO1 RR-43. Computational assistance was provided by the NIH NCRR GCRC MO1 RR-43 CDMAS project. We thank Dajun Qian for help with statistical analysis, Susie Nakao and the nurses of the Clinical Research Center at LAC-USC Medical Center for help in studying the subjects, Dr Tom Kawada and other members of the Division of Nuclear Medicine for help in preparing and administering the egg yolk test, Jeanne M Howard for help with collecting data, Dr Myriam Marin-Sorenson for histological assistance, and Mo-Li Chen for technical assistance.
Abbreviations used in this paper
- egg yolk cobalamin absorption test
- maximum acid output