ORIGINAL ARTICLE
A Laboratory Score in the Diagnosis of Autoimmune Atrophic Gastritis A Prospective Study Emanuela Miceli, MD,* Donatella Padula, MD,* Marco V. Lenti, MD,* Alessandra Gallia, MD,* Riccardo Albertini, MD,w Michele Di Stefano, MD,* Catherine Klersy, MD,z and Gino R. Corazza, MD*
Background: Several biomarkers have been proposed for the diagnosis of autoimmune atrophic gastritis (AAG), but at the present there is no appropriate testing strategy for the disease. Goals: The aim of this study was to develop and validate a laboratory score able to address the diagnosis of AAG in a general practice setting. Study: We prospectively evaluated a number of serum biomarkers (vitamin B12, mean corpuscular volume, hemoglobin, gastrin, and chromogranin A levels) in a case-control population and built 2 biochemical scores, the first with all the parameters [Global Score (GS)], and the second as the best statistical combination of them [Simple Score (SS)]. In the second phase we validated the score that proved to be more efficient on a random population referred to our center (Gastroenterology Outpatient Clinic). Results: Both models turned out to be reliable in detecting patients with suspected AAG, showing excellent accuracy [area under the receiver operating curve (AUC-ROC) 0.94; 95% confidence interval (CI), 0.91-0.97 for GS and AUC-ROC 0.93; 95% CI, 0.89-0.86 for SS]. The SS proved to be more convenient because of its accessibility and availability in a general setting and its low cost. The validation of the SS showed a sensitivity of 85.7% (95% CI, 57.2-98.2) and a specificity of 83.7% (95% CI, 74.2-90.89). Conclusions: Herein, we describe 2 nonexpensive and reliable score models, particularly the SS, that can be applied in daily medical practice for identifying patients potentially affected by AAG.
deficiency, and reactive hyperplasia of the neuroendocrine cells.1,2 This leads to vitamin B12 and iron malabsorption, with potentially severe clinical consequences.3–9 Of note, AAG is a preneoplastic condition, predisposing to the development of both gastric adenocarcinoma10 and type I carcinoid.11,12 Despite its clinical consequences and neoplastic potential, this condition is often misdiagnosed. The clinical pattern of AAG is proteiform: gastroenterological manifestations such as dyspepsia, malabsorption, bloating, and atrophic glossitis3–5,13–14 are common findings but not specific enough to guide the diagnosis. Upper gastrointestinal endoscopy (UGIE) is invasive and therefore reserved for selected patients; moreover, UGIE is not always completed by an appropriate collection of biopsy specimens.15 Although several biomarkers have been proposed to prompt the diagnosis, none of them has proved to be satisfactorily sensitive, specific, or reliable.16–26 In contrast, a screening test is needed15 as epidemiological data suggest that AAG is neither a rare disease,27,28 nor is it confined to the elderly population or only to the gastroenterological setting.29 The aim of this study was, therefore, to evaluate the diagnostic role of some hematochemical parameters, to find a useful score for the diagnosis of AAG in general practice and to verify its reliability.
Key Words: gastric atrophy, score, gastrin, validation, stomach
(J Clin Gastroenterol 2015;49:e1–e5)
MATERIALS AND METHODS Study Population
A
utoimmune atrophic gastritis (AAG) is characterized by the destruction of the gastric body and fundus mucosa, resulting in hypochlorhydria, intrinsic factor
Received for publication July 18, 2013; accepted January 22, 2014. From the *1st Department of Internal Medicine; wClinical Chemistry Laboratory; and zService of Biometry and Statistics, Fondazione IRCCS Policlinico San Matteo, University of Pavia, Pavia, Italy. All authors contributed to the conception and design of the studies. E.M., D.P., M.V.L., and G.R.C. drafted the initial paper. C.K. was responsible for the initial data analyses. All authors made contributions to developing the analysis plan and interpreting the data at subsequent stages. Successive drafts of the paper were corrected by G.R.C. for critical revision. The authors declare that they have nothing to disclose. Reprints: Gino R. Corazza, MD, 1st Department of Internal Medicine, Fondazione IRCCS Policlinico San Matteo, University of Pavia, Piazzale Golgi n.19, Pavia 27100, Italy (e-mail: gr.corazza@ smatteo.pv.it). Copyright r 2014 by Lippincott Williams & Wilkins
J Clin Gastroenterol
Volume 49, Number 1, January 2015
The study was performed between January 2007 and September 2011 in the Gastroenterology Outpatient Clinic of the Fondazione IRCCS Policlinico San Matteo of Pavia. We prospectively enrolled newly admitted patients with a first diagnosis of AAG (139 patients: 103 females, mean age = 58 ± 17 y; 36 males, mean age = 66 ± 16 y). The diagnosis was based on pathologic findings (severe atrophy of the body and fundus30 with antrum sparing, antral gastrin cell hyperplasia, and enterochromaffin-like cell hyperplasia2). The clinical findings of most of these patients have already been described in a previous study.29 Five-hundred and ten patients (330 females, mean age = 47 ± 16 y; 180 males, mean age = 47 ± 17 y) undergoing an UGIE over the same time span, and whose pathologic characteristic of the biopsy specimens excluded AAG, were enrolled as controls. The ratio of cases and controls was 1:5. In 26 AAG patients a previous Helicobacter pylori infection was detected by gastric biopsies, urea breath test, or stool antigen test and was successfully eradicated. None www.jcge.com |
e1
J Clin Gastroenterol
Miceli et al
of the 139 patients had evidence of H. pylori infection in the gastric biopsy performed at enrollment. The study was approved by the local ethics committee and each patient provided written informed consent.
Biochemical Tests Blood samples were collected from cases and controls at fasting. All the subjects were instructed to avoid taking proton pump inhibitors for at least 1 month before blood sampling. We evaluated the following laboratory parameters: hemoglobin level (Hb, abnormal if 98 fL), vitamin B12 level (abnormal if 120 pg/mL), and chromogranin A level (CgA, abnormal if >100 ng/mL). Complete blood counts were performed by a Cell-Dyn Sapphire (Abbott). Vitamin B12 in serum was assayed by an automated immunochemistry analyzer, Immulite 2000 XPi (Siemens Healthcare Diagnostics, NY); the method is a solid-phase, competitive chemiluminescent enzyme immunoassay. Plasma gastrin levels were determined by a chemiluminescent enzyme-labeled immunometric assay based on a ligand-labeled murine monoclonal capture antibody specific for gastrin and separation by an anti-ligand coated solid phase (Immulite 2000 XPi; Siemens Healthcare Diagnostics, NY). Serum CgA level was measured by a solid-phase 2-site immunoradiometric assay (Cisbio Bioassays, Bagnols sur Ceze Cedex, France). The analytical quality of the biochemical parameters listed was ensured by a 3-level internal quality control (performed daily or during every analytical run) and by participation in national and international external quality programs.
Volume 49, Number 1, January 2015
98 fL, gastrin > 120 pg/mL, and CgA > 100 ng/mL). Sensitivity, specificity, and overall diagnostic ability as measured by the area under the receiver operating curve (AUC-ROC) were computed. To build a diagnostic score to detect AAG, 2 multivariable logistic models were fitted including Hb, vitamin B12, MCV, gastrin, and CgA, or Hb, vitamin B12, MCV, and gastrin (all dichotomized as above), respectively, as independent variables, and with either a case or a control as the dependent variable. The choice of variables to be included in the model was decided a priori, based on the current knowledge. Model performance was measured by means of the c-statistics. Models were compared informally by inspecting the c-statistics, sensitivity, specificity, and predictive values and their 95% confidence intervals. The score was built as a linear combination of the included independent variables. The regression coefficients were rounded to the closest 0.5. The best cutoff for the score to detect AAG was identified by means of ROC curve analysis. The laboratory costs of all the parameters were evaluated. For the purpose of validation, the score was then applied to an independent cohort of 100 patients referred to our outpatient clinic between September 2011 and January 2012. Sensitivity, specificity, and AUC-ROC of the predicted diagnosis versus the actual diagnosis of AAG, with 95% confidence intervals (95% CI), were computed. Variable distribution and/or model assumptions were assessed graphically. Stata 12 (StataCorp, College Station, TX) and Medcalc 12 (MedCalc Software, Mariakerke, Belgium) were used for computation.
RESULTS
Statistical Analysis Data are described as mean and SD if continuous and as counts and percent if categorical. Cases and controls were compared with the Mann-Whitney U test and the Fisher exact test, respectively. For the purpose of the analysis, the biochemical parameters were dichotomized according to their conventional cutoff for pathology and coded 0/1 for normal/abnormal (Hb 120 pg/mL Chromogranin A (ng/mL) Chromogranin A > 100 ng/mL
Case
Controls
Sensitivity
Specificity
ROC Area
(N = 139) (21%)
(N = 510) (89%)
P
(95% CI)
(95% CI)
(95% CI)
11.4 ± 2.3 81 (58%) 245.8 ± 176.5 686 (62%) 93.4 ± 13.6 57 (41%) 14.7 ± 3.1 67 (48%) 668.0 ± 732.6 117 (84%) 184.7 ± 227.7 85 (61%)
13.4 ± 1.3 64 (13%) 445.5 ± 185.9 54 (11%) 88.6 ± 6.9 20 (4%) 12.6 ± 1.5 80 (10%) 62.1 ± 54.2 24 (5%) 33.3 ± 32.4 18 (4%)
< 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001
58 (49-66)
87 (84-90)
0.73 (0.68-0.77)
62 (53-70)
89 (86-91)
0.76 (0.72-0.80)
41 (32-49)
96 (94-97)
0.68 (0.64-0.73)
48 (39-56)
90 (87-92)
0.69 (0.65-0.74)
84 (77-89)
95 (93-96)
0.90 (0.87-0.93)
61 (52-69)
96 (94-97)
0.79 (0.75-0.83)
*Mean ± SD for continuous variables; N (%) for categorical variables. CI indicates confidence interval; Hb, hemoglobin; MCV, mean corpuscular volume; RDW, red cell distribution width; ROC, receiver operating curve.
e2 | www.jcge.com
r
2014 Lippincott Williams & Wilkins
J Clin Gastroenterol
Volume 49, Number 1, January 2015
A Laboratory Score for Autoimmune Gastritis
sensitivity was generally low. Correspondingly, the overall diagnostic performance as measured by the AUC-ROC ranged from modest (0.68, MCV) to high (0.90, gastrin).
Logistic Models On the basis of the 2 logistic models fitted, 2 different scores were built, labeled Global Score (GS) and Simple Score (SS), respectively, with the latter nested into the former, as described below.
“Global Score” This model was built as a linear combination of Hb, MCV, gastrin, vitamin B12, and CgA, dichotomized according to their conventional cutoff for pathology, as described in the Methods section. The computed probability of AAG, given the possible values of the GS, is shown in Figure 1. The score ranged from 0 to 9: specifically, a patient with all his/her laboratory findings within normality would have a score of 0, whereas patients with all his/her scores outside normality would have a score of 9. The algorithm for calculating the GS is shown in Figure 2. The optimal cutoff to discriminate AAG subjects from non-AAG subjects was found to be a GS > 2.5. The corresponding sensitivity was 88.4% and specificity was 94.1%. Positive and negative predictive values were 80.3% and 96.8%, respectively, given the prevalence of AAG of 21% (by design). The AUC-ROC was 0.94.
“Simple Score” In the next phase of the study, we tried to simplify the GS by including only laboratory parameters, which were available for primary care physicians involved in the diagnosis of AAG. This “simplified” score, SS, included only the Hb, MCV, and gastrin values and was calculated as shown in Figure 2. The score ranged from 0 to 6.5. The probability of AAG, given the possible values of the SS, is shown in Figure 3. The best cutoff to discriminate AAG subjects and nonAAG subjects was an SS > 1.5. The corresponding sensitivity was 85.6%, whereas specificity was 95.3%. Positive and negative predictive values were 83.2% and 96.0%, respectively, given the prevalence of AAG of 21% (by design). The AUC-ROC was 0.93.
1
Prob of AAG (95%CI)
.9 .8 .7 .6 .5 .4 .3 .2 .1 0 0
1
2
3
4 5 6 Global Score
7
8
9
10
FIGURE 1. Probability of autoimmune atrophic gastritis (AAG) based on GS (95% CI). CI indicates confidence interval; GS, Global score. r
2014 Lippincott Williams & Wilkins
Model Comparison Both models turned out to be reliable in detecting patients with suspected AAG, showing excellent and comparable accuracy [AUC-ROC (c-statistic) 0.94; 95% CI, 0.91-0.97 for GS and 0.93; 95% CI, 0.89-0.86 for SS], a feature that is essential for a clinically nuanced disorder such as AAG. Correspondingly, sensitivities, specificities, and predictive values were all high (> 80%) and of comparable magnitude, both when assessing the overall logistic model and while using the identified cutoffs of 2.5 and 1.5 for GS and SS, respectively, to classify patients. This similar effectiveness becomes relevant both for clinical practice, as the Simple Score uses only 3 parameters and for economic aspects. In our center, calculation of the GS showed the laboratory cost per patient to be 21.5 Euros (approximately 28 US Dollars), whereas with the SS it was 7 Euros (approximately 9 US Dollars).
Validation of the Simple Score For the purpose of validation, the SS was then applied to an independent cohort of 100 patients referred to our outpatient clinic between September 2011 and January 2012 mainly because of dyspeptic symptoms or bowel abnormalities (classified according to Rome III diagnostic criteria for functional gastrointestinal disorders31). Fourteen patients were diagnosed with AAG, and their median SS was 4 (25th-75th, 4-5.5), whereas in non-AAG patients it was 0 (25th-75th, 0-1.5). On the basis of the 1.5 cutoff identified as optimal for detecting AAG previously, 74 patients were categorized as non-AAG and 26 as AAG. As a result, 2 patients of the 100 were underdiagnosed, whereas 14/100 were overdiagnosed. However, the overall diagnostic accuracy was still high, with an AUC-ROC of 0.85 (95% CI, 0.74-0.95) as compared with 0.93 of the testing sample; sensitivity and specificity were 85.7% (95% CI, 57.2-98.2) as compared with the original 85.6%, and 83.7% (95% CI, 74.2-90.8), confirming the good performance of the SS.
DISCUSSION The aim of this case-control study was to build and evaluate the usefulness of a score based on hematological parameters for the diagnosis of AAG. The extensive and unspecific clinical pattern of AAG may delay UGIE, which, because of its invasiveness and its costs, cannot be considered as an ideal screening test. Therefore, we first assessed the hematological profile of the cases and the controls and then created 2 scoring systems that were able to discriminate AAG subjects. According to the literature, anemia is a common feature of AAG and it can result from different pathways, from iron to vitamin B12 malabsorption or both. In our series, anemia was found in 58% of the cases and in only 13% of the controls. Among all the studied parameters, hemoglobin was certainly the least specific as anemia is an extremely common finding in the general population. With regard to macrocytosis, this can be detected in several conditions such as drug or alcohol abuse, liver diseases, malabsorption, folate deficiency, and strict vegetarian diet. The MCV is a routine examination and should, therefore, always be critically evaluated, even in the absence of anemia.3 Vitamin B12 deficiency is another common feature of AAG, being present in 62% of cases versus 11% of controls. The lack of cyanocobalamin characterizes a late www.jcge.com |
e3
J Clin Gastroenterol
Miceli et al
Volume 49, Number 1, January 2015
Global Score* GS = 1.5*Hb + 0.5*MCV + 3.5*Gastrin + 2*Vitamin B12 + 1.5*Chromatogranin
Optimal cut-off to detect AAG: GS>2.5
Simple Score* SS = 1.5*Hb + 1*MCV + 4*Gastrin
Optimal cut-off to detect AAG: SS>1.5
*Codes to be used for score calculation 1
Hb 98 fl
Gastrin >120 pg/ml
Vitamin B12 100 ng/ml
0
Hb ≥12 g/dl
MCV ≤98 fl
Gastrin ≤120 pg/ml
Vitamin B12 ≥240 pg/ml
Chromogranin A ≤100 ng/ml
FIGURE 2. Algorithm for calculation of GS and SS. AAG indicates autoimmune atrophic gastritis; GS, Global score; Hb, hemoglobin; MCV, mean corpuscular volume; SS, Simple Score.
systemic stage of the disease, known as pernicious anemia that emerges when stores of vitamin B12 are depleted, usually in a time span between 6 and 10 years. Hypergastrinemia and incremented CgA, because of neuroendocrine cell reactive hyperplasia, are common findings in AAG. In the second phase of the study, we built 2 AAG scoring models, based on laboratory findings, and we compared their performance. The SS, which had a similar diagnostic accuracy to the GS, showed a slight advantage, given the smaller number of parameters to be evaluated, and therefore the fewer samples to be taken. This in turn involves lower costs for society (7 Euros per patient for the SS, 21.5 Euros for the GS). Moreover, the parameters used for the construction of the SS were selected because they 1
Prob of AAG (95%CI)
.9 .8 .7 .6 .5 .4 .3 .2 .1 0 0
1
2
3 4 Simple Score
5
6
7
FIGURE 3. Probability of autoimmune atrophic gastritis (AAG) based on SS (95% CI). CI indicates confidence interval; SS, Simple Score.
e4 | www.jcge.com
were easily accessible and were also available in nongastroenterological settings. The validation of the SS on an independent population confirmed its usefulness to identify patients affected by AAG. The selected cutoff of 1.5 yielded high diagnostic accuracy, with an AUC-ROC of 0.85. In particular, the sensitivity was as high as in the testing sample, ensuring that AAG patients would be correctly identified in any setting and transferred to specialized Gastroenterology centers to undergo UGIE for confirmation. Conversely, specificity was slightly lower than in the testing sample, although still above 85%. It should be noted that we decided not to evaluate antibodies, neither anti-parietal cell antibodies (PCA) nor intrinsic factor (IF) antibodies. The decision was based on cost-efficiency reasoning, these tests being very expensive, and on their limited availability. Moreover, it has been demonstrated that PCA are unspecific14,26 and they are not constantly detectable during the natural history of AAG,32,33 whereas anti-IF antibodies are not sensitive enough to be considered as a screening test32,34 and, unlike PCA, their occurrence seems to be related only to the advanced stage of the disease, in particular when hematological alterations appear.26 It should be noted that the prevalence of AAG found in our validation series is very high; one reason for this is because our setting is a Gastroenterology tertiary referral clinic, whereas in the general population we expect it to be lower. Further studies are required to validate this score in an external unselected population. The size of the study population for the validation study is relatively small, giving larger confidence intervals. We must also stress that all the subjects used in the score derivation phase were asked to avoid taking proton pump inhibitors for a certain time to avoid interactions with the laboratory test, and this can be considered a selection bias that does not exist in the general r
2014 Lippincott Williams & Wilkins
J Clin Gastroenterol
Volume 49, Number 1, January 2015
population. To avoid this bias, we imposed no limitations for the group used for the validation phase. Because of the unspecific clinical presentation and the invasiveness of UGIE, AAG is an underdiagnosed disease. However, it is burdened by severe systemic complications and high morbidity. Hence, there is the need to find elements that can lead not only gastroenterologists but also general practitioners toward the diagnosis. We propose 2 scoring models that can indicate a diagnosis of AAG in a simple, noninvasive, low-cost way. The biomarkers used are easily available even for general practitioners. The models also proved to be extremely reliable. From these results, it appears that these scores are eligible for use in extensive screening programs to select candidate patients for endoscopy. This is particularly true for the SS, which has a better cost-benefit. Endoscopy should be recommended in those patients who are categorized as positive by the proposed algorithm, to detect AAG early and prevent clinical manifestations.
A Laboratory Score for Autoimmune Gastritis
15. 16. 17.
18.
19.
20.
REFERENCES 1. Strickland RG, Mackay IR. A reappraisal of the nature and significance of chronic atrophic gastritis. Am J Dig Dis. 1973;18:426–440. 2. Torbenson M, Abraham SC, Boitnott J, et al. Autoimmune gastritis: distinct histological and immunohistochemical findings before complete loss of oxyntic glands. Mod Pathol. 2002;15:102–109. 3. Toh BH, Van Driel IR, Gleeson PA. Pernicious anemia. N Engl J Med. 1997;337:1441–1448. 4. Hershko C, Ronson A, Souroujon M, et al. Variable hematologic presentation of autoimmune gastritis: age-related progression from iron deficiency to cobalamin depletion. Blood. 2006;107:1673–1679. 5. Stabler SP, Allen RH, Savage DG, et al. Clinical spectrum and diagnosis of cobalamin deficiency. Blood. 1990;76:871–881. 6. Shorvon SD, Carney MW, Chanarin I, et al. The neuropsychiatry of megaloblastic anaemia. Br Med J. 1980;281: 1036–1038. 7. Van Oijen MG, Sipponen P, Laheij RJ, et al. Gastric status and vitamin B12 levels in cardiovascular patients. Dig Dis Sci. 2007;52:2186–2189. 8. Rede´en S, Ryberg A, Petersson F, et al. Homocysteine levels in chronic gastritis and other conditions: relations to incident cardiovascular disease and dementia. Dig Dis Sci. 2010;55:351–358. 9. Betterle C, Dal Pra C, Mantero F, et al. Autoimmune adrenal insufficiency and autoimmune polyendocrine syndromes: autoantibodies, autoantigens, and their applicability in diagnosis and disease prediction. Endocr Rev. 2002;23:327–364. 10. Correa P. Human gastric carcinogenesis: a multistep and multifactorial process. Cancer Res. 1992;52:6735–6740. 11. Burkitt MD, Pritchard DM. Pathogenesis and management of gastric carcinoid tumours. Aliment Pharmacol Ther. 2006;24:1305–1320. 12. Kokkola A, Sjo¨blom SM, Haapiainen R, et al. The risk of gastric carcinoma and carcinoid tumours in patients with pernicious anaemia. A prospective follow-up study. Scand J Gastroenterol. 1998;33:88–92. 13. Lahner E, Annibale B. Pernicious anemia: new insights from a gastroenterological point of view. World J Gastroenterol. 2009;15:5121–5128. 14. Soykan I, Yakut M, Keskin O, et al. Clinical profiles, endoscopic and laboratory features and associated
r
2014 Lippincott Williams & Wilkins
21. 22. 23. 24. 25. 26.
27. 28.
29. 30.
31. 32. 33. 34.
factors in patients with autoimmune gastritis. Digestion. 2012;86:20–26. Sipponen P, Graham DJ. Importance of atrophic gastritis in diagnostics and prevention of gastric cancer: application of plasma biomarkers. Scand J Gastroenterol. 2007;42:2–10. Carmel R. Pernicious anemia. The expected findings of very low serum cobalamin levels, anemia, and macrocytosis are often lacking. Arch Intern Med. 1988;148:1712–1714. Iijima K, Abe Y, Kikuchi R, et al. Serum biomarker tests are useful in delineating between patients with gastric atrophy and normal, healthy stomach. World J Gastroenterol. 2009;15: 853–859. Sipponen P, Ranta P, Helske T, et al. Serum levels of amidated gastrin-17 and pepsinogen I in atrophic gastritis: an observational case-control study. Scand J Gastroenterol. 2002;37: 785–791. Va¨a¨na¨nen H, Vauhkonen M, Helske T, et al. Non-endoscopic diagnosis of atrophic gastritis with a blood test. Correlation between gastric histology and serum levels of gastrin-17 and pepsinogen I: a multicentre study. Eur J Gastroenterol Hepatol. 2003;15:885–891. Korman MG, Strickland RG, Hansky J. Serum gastrin in chronic gastritis. BMJ. 1971;2:16–18. Checchi S, Montanaro M, Pasqui L, et al. Serum ghrelin as a marker of atrophic body gastritis in patients with parietal cell antibodies. J Clin Endocrinol Metab. 2007;92:4346–4351. Borch K, Axelsson CK, Halgreen H, et al. The ratio of pepsinogen A to pepsinogen C: a sensitive test for atrophic gastritis. Scand J Gastroenterol. 1989;24:870–876. Kekki M, Samloff IM, Varis K. Serum pepsinogen I and serum gastrin in the screening of severe atrophic corpus gastritis. Scand J Gastroenterol. 1991;26:109–116. Oksanen A, Sipponen P, Miettinen A, et al. Evaluation of blood tests to predict normal gastric mucosa. Scand J Gastroenterol. 2000;35:791–795. Toh BH, Chan J, Kyaw T, et al. Cutting edge issues in autoimmune gastritis. Clin Rev Allergy Immunol. 2012;42: 269–278. Lahner E, Norman GL, Severi C, et al. Reassessment of intrinsic factor and parietal cell autoantibodies in atrophic gastritis with respect to cobalamin deficiency. Am J Gastroenterol. 2009;104:2071–2079. Carmel R. Prevalence of undiagnosed pernicious anemia in the elderly. Arch Intern Med. 1996;156:1097–1100. Telaranta-Keerie A, Kara R, Paloheimo L, et al. Prevalence of undiagnosed advanced atrophic corpus gastritis in Finland: an observational study among 4,256 volunteers without specific complaints. Scand J Gastroenterol. 2010;45:1036–1041. Miceli E, Lenti MV, Padula D, et al. Common features of patients with autoimmune atrophic gastritis. Clin Gastroenterol Hepatol. 2012;10:812–814. Dixon MF, Genta RM, Yardley JH, et al. Classification and grading of gastritis. The updated Sydney System. International Workshop on the Histopathology of Gastritis, Houston 1994. Am J Surg Pathol. 1996;20:1161–1181. Rome Foundation. Guidelines-Rome III diagnostic criteria for functional gastrointestinal disorders. Gastrointestin Liver Dis. 2006;15:307–312. Andres E, Serraj K. Optimal management of pernicious anemia. J Blood Med. 2012;3:97–103. Davidson RJL, Atrah HI, Sewell HF. Longitudinal study of circulating gastric antibodies in pernicious anaemia. J Clin Pathol. 1989;42:1092–1095. Khan S, Del-Duca C, Fenton E, et al. Limited value of testing for intrinsic factor antibodies with negative gastric parietal cell antibodies in pernicious anaemia. J Clin Pathol. 2009;62:439–441.
www.jcge.com |
e5
A Laboratory Score in the Diagnosis of Autoimmune Atrophic Gastritis A Prospective Study Emanuela Miceli, MD,* Donatella Padula, MD,* Marco V. Lenti, MD,* Alessandra Gallia, MD,* Riccardo Albertini, MD,w Michele Di Stefano, MD,* Catherine Klersy, MD,z and Gino R. Corazza, MD*
Background: Several biomarkers have been proposed for the diagnosis of autoimmune atrophic gastritis (AAG), but at the present there is no appropriate testing strategy for the disease. Goals: The aim of this study was to develop and validate a laboratory score able to address the diagnosis of AAG in a general practice setting. Study: We prospectively evaluated a number of serum biomarkers (vitamin B12, mean corpuscular volume, hemoglobin, gastrin, and chromogranin A levels) in a case-control population and built 2 biochemical scores, the first with all the parameters [Global Score (GS)], and the second as the best statistical combination of them [Simple Score (SS)]. In the second phase we validated the score that proved to be more efficient on a random population referred to our center (Gastroenterology Outpatient Clinic). Results: Both models turned out to be reliable in detecting patients with suspected AAG, showing excellent accuracy [area under the receiver operating curve (AUC-ROC) 0.94; 95% confidence interval (CI), 0.91-0.97 for GS and AUC-ROC 0.93; 95% CI, 0.89-0.86 for SS]. The SS proved to be more convenient because of its accessibility and availability in a general setting and its low cost. The validation of the SS showed a sensitivity of 85.7% (95% CI, 57.2-98.2) and a specificity of 83.7% (95% CI, 74.2-90.89). Conclusions: Herein, we describe 2 nonexpensive and reliable score models, particularly the SS, that can be applied in daily medical practice for identifying patients potentially affected by AAG.
deficiency, and reactive hyperplasia of the neuroendocrine cells.1,2 This leads to vitamin B12 and iron malabsorption, with potentially severe clinical consequences.3–9 Of note, AAG is a preneoplastic condition, predisposing to the development of both gastric adenocarcinoma10 and type I carcinoid.11,12 Despite its clinical consequences and neoplastic potential, this condition is often misdiagnosed. The clinical pattern of AAG is proteiform: gastroenterological manifestations such as dyspepsia, malabsorption, bloating, and atrophic glossitis3–5,13–14 are common findings but not specific enough to guide the diagnosis. Upper gastrointestinal endoscopy (UGIE) is invasive and therefore reserved for selected patients; moreover, UGIE is not always completed by an appropriate collection of biopsy specimens.15 Although several biomarkers have been proposed to prompt the diagnosis, none of them has proved to be satisfactorily sensitive, specific, or reliable.16–26 In contrast, a screening test is needed15 as epidemiological data suggest that AAG is neither a rare disease,27,28 nor is it confined to the elderly population or only to the gastroenterological setting.29 The aim of this study was, therefore, to evaluate the diagnostic role of some hematochemical parameters, to find a useful score for the diagnosis of AAG in general practice and to verify its reliability.
Key Words: gastric atrophy, score, gastrin, validation, stomach
(J Clin Gastroenterol 2015;49:e1–e5)
MATERIALS AND METHODS Study Population
A
utoimmune atrophic gastritis (AAG) is characterized by the destruction of the gastric body and fundus mucosa, resulting in hypochlorhydria, intrinsic factor
Received for publication July 18, 2013; accepted January 22, 2014. From the *1st Department of Internal Medicine; wClinical Chemistry Laboratory; and zService of Biometry and Statistics, Fondazione IRCCS Policlinico San Matteo, University of Pavia, Pavia, Italy. All authors contributed to the conception and design of the studies. E.M., D.P., M.V.L., and G.R.C. drafted the initial paper. C.K. was responsible for the initial data analyses. All authors made contributions to developing the analysis plan and interpreting the data at subsequent stages. Successive drafts of the paper were corrected by G.R.C. for critical revision. The authors declare that they have nothing to disclose. Reprints: Gino R. Corazza, MD, 1st Department of Internal Medicine, Fondazione IRCCS Policlinico San Matteo, University of Pavia, Piazzale Golgi n.19, Pavia 27100, Italy (e-mail: gr.corazza@ smatteo.pv.it). Copyright r 2014 by Lippincott Williams & Wilkins
J Clin Gastroenterol
Volume 49, Number 1, January 2015
The study was performed between January 2007 and September 2011 in the Gastroenterology Outpatient Clinic of the Fondazione IRCCS Policlinico San Matteo of Pavia. We prospectively enrolled newly admitted patients with a first diagnosis of AAG (139 patients: 103 females, mean age = 58 ± 17 y; 36 males, mean age = 66 ± 16 y). The diagnosis was based on pathologic findings (severe atrophy of the body and fundus30 with antrum sparing, antral gastrin cell hyperplasia, and enterochromaffin-like cell hyperplasia2). The clinical findings of most of these patients have already been described in a previous study.29 Five-hundred and ten patients (330 females, mean age = 47 ± 16 y; 180 males, mean age = 47 ± 17 y) undergoing an UGIE over the same time span, and whose pathologic characteristic of the biopsy specimens excluded AAG, were enrolled as controls. The ratio of cases and controls was 1:5. In 26 AAG patients a previous Helicobacter pylori infection was detected by gastric biopsies, urea breath test, or stool antigen test and was successfully eradicated. None www.jcge.com |
e1
J Clin Gastroenterol
Miceli et al
of the 139 patients had evidence of H. pylori infection in the gastric biopsy performed at enrollment. The study was approved by the local ethics committee and each patient provided written informed consent.
Biochemical Tests Blood samples were collected from cases and controls at fasting. All the subjects were instructed to avoid taking proton pump inhibitors for at least 1 month before blood sampling. We evaluated the following laboratory parameters: hemoglobin level (Hb, abnormal if 98 fL), vitamin B12 level (abnormal if 120 pg/mL), and chromogranin A level (CgA, abnormal if >100 ng/mL). Complete blood counts were performed by a Cell-Dyn Sapphire (Abbott). Vitamin B12 in serum was assayed by an automated immunochemistry analyzer, Immulite 2000 XPi (Siemens Healthcare Diagnostics, NY); the method is a solid-phase, competitive chemiluminescent enzyme immunoassay. Plasma gastrin levels were determined by a chemiluminescent enzyme-labeled immunometric assay based on a ligand-labeled murine monoclonal capture antibody specific for gastrin and separation by an anti-ligand coated solid phase (Immulite 2000 XPi; Siemens Healthcare Diagnostics, NY). Serum CgA level was measured by a solid-phase 2-site immunoradiometric assay (Cisbio Bioassays, Bagnols sur Ceze Cedex, France). The analytical quality of the biochemical parameters listed was ensured by a 3-level internal quality control (performed daily or during every analytical run) and by participation in national and international external quality programs.
Volume 49, Number 1, January 2015
98 fL, gastrin > 120 pg/mL, and CgA > 100 ng/mL). Sensitivity, specificity, and overall diagnostic ability as measured by the area under the receiver operating curve (AUC-ROC) were computed. To build a diagnostic score to detect AAG, 2 multivariable logistic models were fitted including Hb, vitamin B12, MCV, gastrin, and CgA, or Hb, vitamin B12, MCV, and gastrin (all dichotomized as above), respectively, as independent variables, and with either a case or a control as the dependent variable. The choice of variables to be included in the model was decided a priori, based on the current knowledge. Model performance was measured by means of the c-statistics. Models were compared informally by inspecting the c-statistics, sensitivity, specificity, and predictive values and their 95% confidence intervals. The score was built as a linear combination of the included independent variables. The regression coefficients were rounded to the closest 0.5. The best cutoff for the score to detect AAG was identified by means of ROC curve analysis. The laboratory costs of all the parameters were evaluated. For the purpose of validation, the score was then applied to an independent cohort of 100 patients referred to our outpatient clinic between September 2011 and January 2012. Sensitivity, specificity, and AUC-ROC of the predicted diagnosis versus the actual diagnosis of AAG, with 95% confidence intervals (95% CI), were computed. Variable distribution and/or model assumptions were assessed graphically. Stata 12 (StataCorp, College Station, TX) and Medcalc 12 (MedCalc Software, Mariakerke, Belgium) were used for computation.
RESULTS
Statistical Analysis Data are described as mean and SD if continuous and as counts and percent if categorical. Cases and controls were compared with the Mann-Whitney U test and the Fisher exact test, respectively. For the purpose of the analysis, the biochemical parameters were dichotomized according to their conventional cutoff for pathology and coded 0/1 for normal/abnormal (Hb 120 pg/mL Chromogranin A (ng/mL) Chromogranin A > 100 ng/mL
Case
Controls
Sensitivity
Specificity
ROC Area
(N = 139) (21%)
(N = 510) (89%)
P
(95% CI)
(95% CI)
(95% CI)
11.4 ± 2.3 81 (58%) 245.8 ± 176.5 686 (62%) 93.4 ± 13.6 57 (41%) 14.7 ± 3.1 67 (48%) 668.0 ± 732.6 117 (84%) 184.7 ± 227.7 85 (61%)
13.4 ± 1.3 64 (13%) 445.5 ± 185.9 54 (11%) 88.6 ± 6.9 20 (4%) 12.6 ± 1.5 80 (10%) 62.1 ± 54.2 24 (5%) 33.3 ± 32.4 18 (4%)
< 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001
58 (49-66)
87 (84-90)
0.73 (0.68-0.77)
62 (53-70)
89 (86-91)
0.76 (0.72-0.80)
41 (32-49)
96 (94-97)
0.68 (0.64-0.73)
48 (39-56)
90 (87-92)
0.69 (0.65-0.74)
84 (77-89)
95 (93-96)
0.90 (0.87-0.93)
61 (52-69)
96 (94-97)
0.79 (0.75-0.83)
*Mean ± SD for continuous variables; N (%) for categorical variables. CI indicates confidence interval; Hb, hemoglobin; MCV, mean corpuscular volume; RDW, red cell distribution width; ROC, receiver operating curve.
e2 | www.jcge.com
r
2014 Lippincott Williams & Wilkins
J Clin Gastroenterol
Volume 49, Number 1, January 2015
A Laboratory Score for Autoimmune Gastritis
sensitivity was generally low. Correspondingly, the overall diagnostic performance as measured by the AUC-ROC ranged from modest (0.68, MCV) to high (0.90, gastrin).
Logistic Models On the basis of the 2 logistic models fitted, 2 different scores were built, labeled Global Score (GS) and Simple Score (SS), respectively, with the latter nested into the former, as described below.
“Global Score” This model was built as a linear combination of Hb, MCV, gastrin, vitamin B12, and CgA, dichotomized according to their conventional cutoff for pathology, as described in the Methods section. The computed probability of AAG, given the possible values of the GS, is shown in Figure 1. The score ranged from 0 to 9: specifically, a patient with all his/her laboratory findings within normality would have a score of 0, whereas patients with all his/her scores outside normality would have a score of 9. The algorithm for calculating the GS is shown in Figure 2. The optimal cutoff to discriminate AAG subjects from non-AAG subjects was found to be a GS > 2.5. The corresponding sensitivity was 88.4% and specificity was 94.1%. Positive and negative predictive values were 80.3% and 96.8%, respectively, given the prevalence of AAG of 21% (by design). The AUC-ROC was 0.94.
“Simple Score” In the next phase of the study, we tried to simplify the GS by including only laboratory parameters, which were available for primary care physicians involved in the diagnosis of AAG. This “simplified” score, SS, included only the Hb, MCV, and gastrin values and was calculated as shown in Figure 2. The score ranged from 0 to 6.5. The probability of AAG, given the possible values of the SS, is shown in Figure 3. The best cutoff to discriminate AAG subjects and nonAAG subjects was an SS > 1.5. The corresponding sensitivity was 85.6%, whereas specificity was 95.3%. Positive and negative predictive values were 83.2% and 96.0%, respectively, given the prevalence of AAG of 21% (by design). The AUC-ROC was 0.93.
1
Prob of AAG (95%CI)
.9 .8 .7 .6 .5 .4 .3 .2 .1 0 0
1
2
3
4 5 6 Global Score
7
8
9
10
FIGURE 1. Probability of autoimmune atrophic gastritis (AAG) based on GS (95% CI). CI indicates confidence interval; GS, Global score. r
2014 Lippincott Williams & Wilkins
Model Comparison Both models turned out to be reliable in detecting patients with suspected AAG, showing excellent and comparable accuracy [AUC-ROC (c-statistic) 0.94; 95% CI, 0.91-0.97 for GS and 0.93; 95% CI, 0.89-0.86 for SS], a feature that is essential for a clinically nuanced disorder such as AAG. Correspondingly, sensitivities, specificities, and predictive values were all high (> 80%) and of comparable magnitude, both when assessing the overall logistic model and while using the identified cutoffs of 2.5 and 1.5 for GS and SS, respectively, to classify patients. This similar effectiveness becomes relevant both for clinical practice, as the Simple Score uses only 3 parameters and for economic aspects. In our center, calculation of the GS showed the laboratory cost per patient to be 21.5 Euros (approximately 28 US Dollars), whereas with the SS it was 7 Euros (approximately 9 US Dollars).
Validation of the Simple Score For the purpose of validation, the SS was then applied to an independent cohort of 100 patients referred to our outpatient clinic between September 2011 and January 2012 mainly because of dyspeptic symptoms or bowel abnormalities (classified according to Rome III diagnostic criteria for functional gastrointestinal disorders31). Fourteen patients were diagnosed with AAG, and their median SS was 4 (25th-75th, 4-5.5), whereas in non-AAG patients it was 0 (25th-75th, 0-1.5). On the basis of the 1.5 cutoff identified as optimal for detecting AAG previously, 74 patients were categorized as non-AAG and 26 as AAG. As a result, 2 patients of the 100 were underdiagnosed, whereas 14/100 were overdiagnosed. However, the overall diagnostic accuracy was still high, with an AUC-ROC of 0.85 (95% CI, 0.74-0.95) as compared with 0.93 of the testing sample; sensitivity and specificity were 85.7% (95% CI, 57.2-98.2) as compared with the original 85.6%, and 83.7% (95% CI, 74.2-90.8), confirming the good performance of the SS.
DISCUSSION The aim of this case-control study was to build and evaluate the usefulness of a score based on hematological parameters for the diagnosis of AAG. The extensive and unspecific clinical pattern of AAG may delay UGIE, which, because of its invasiveness and its costs, cannot be considered as an ideal screening test. Therefore, we first assessed the hematological profile of the cases and the controls and then created 2 scoring systems that were able to discriminate AAG subjects. According to the literature, anemia is a common feature of AAG and it can result from different pathways, from iron to vitamin B12 malabsorption or both. In our series, anemia was found in 58% of the cases and in only 13% of the controls. Among all the studied parameters, hemoglobin was certainly the least specific as anemia is an extremely common finding in the general population. With regard to macrocytosis, this can be detected in several conditions such as drug or alcohol abuse, liver diseases, malabsorption, folate deficiency, and strict vegetarian diet. The MCV is a routine examination and should, therefore, always be critically evaluated, even in the absence of anemia.3 Vitamin B12 deficiency is another common feature of AAG, being present in 62% of cases versus 11% of controls. The lack of cyanocobalamin characterizes a late www.jcge.com |
e3
J Clin Gastroenterol
Miceli et al
Volume 49, Number 1, January 2015
Global Score* GS = 1.5*Hb + 0.5*MCV + 3.5*Gastrin + 2*Vitamin B12 + 1.5*Chromatogranin
Optimal cut-off to detect AAG: GS>2.5
Simple Score* SS = 1.5*Hb + 1*MCV + 4*Gastrin
Optimal cut-off to detect AAG: SS>1.5
*Codes to be used for score calculation 1
Hb 98 fl
Gastrin >120 pg/ml
Vitamin B12 100 ng/ml
0
Hb ≥12 g/dl
MCV ≤98 fl
Gastrin ≤120 pg/ml
Vitamin B12 ≥240 pg/ml
Chromogranin A ≤100 ng/ml
FIGURE 2. Algorithm for calculation of GS and SS. AAG indicates autoimmune atrophic gastritis; GS, Global score; Hb, hemoglobin; MCV, mean corpuscular volume; SS, Simple Score.
systemic stage of the disease, known as pernicious anemia that emerges when stores of vitamin B12 are depleted, usually in a time span between 6 and 10 years. Hypergastrinemia and incremented CgA, because of neuroendocrine cell reactive hyperplasia, are common findings in AAG. In the second phase of the study, we built 2 AAG scoring models, based on laboratory findings, and we compared their performance. The SS, which had a similar diagnostic accuracy to the GS, showed a slight advantage, given the smaller number of parameters to be evaluated, and therefore the fewer samples to be taken. This in turn involves lower costs for society (7 Euros per patient for the SS, 21.5 Euros for the GS). Moreover, the parameters used for the construction of the SS were selected because they 1
Prob of AAG (95%CI)
.9 .8 .7 .6 .5 .4 .3 .2 .1 0 0
1
2
3 4 Simple Score
5
6
7
FIGURE 3. Probability of autoimmune atrophic gastritis (AAG) based on SS (95% CI). CI indicates confidence interval; SS, Simple Score.
e4 | www.jcge.com
were easily accessible and were also available in nongastroenterological settings. The validation of the SS on an independent population confirmed its usefulness to identify patients affected by AAG. The selected cutoff of 1.5 yielded high diagnostic accuracy, with an AUC-ROC of 0.85. In particular, the sensitivity was as high as in the testing sample, ensuring that AAG patients would be correctly identified in any setting and transferred to specialized Gastroenterology centers to undergo UGIE for confirmation. Conversely, specificity was slightly lower than in the testing sample, although still above 85%. It should be noted that we decided not to evaluate antibodies, neither anti-parietal cell antibodies (PCA) nor intrinsic factor (IF) antibodies. The decision was based on cost-efficiency reasoning, these tests being very expensive, and on their limited availability. Moreover, it has been demonstrated that PCA are unspecific14,26 and they are not constantly detectable during the natural history of AAG,32,33 whereas anti-IF antibodies are not sensitive enough to be considered as a screening test32,34 and, unlike PCA, their occurrence seems to be related only to the advanced stage of the disease, in particular when hematological alterations appear.26 It should be noted that the prevalence of AAG found in our validation series is very high; one reason for this is because our setting is a Gastroenterology tertiary referral clinic, whereas in the general population we expect it to be lower. Further studies are required to validate this score in an external unselected population. The size of the study population for the validation study is relatively small, giving larger confidence intervals. We must also stress that all the subjects used in the score derivation phase were asked to avoid taking proton pump inhibitors for a certain time to avoid interactions with the laboratory test, and this can be considered a selection bias that does not exist in the general r
2014 Lippincott Williams & Wilkins
J Clin Gastroenterol
Volume 49, Number 1, January 2015
population. To avoid this bias, we imposed no limitations for the group used for the validation phase. Because of the unspecific clinical presentation and the invasiveness of UGIE, AAG is an underdiagnosed disease. However, it is burdened by severe systemic complications and high morbidity. Hence, there is the need to find elements that can lead not only gastroenterologists but also general practitioners toward the diagnosis. We propose 2 scoring models that can indicate a diagnosis of AAG in a simple, noninvasive, low-cost way. The biomarkers used are easily available even for general practitioners. The models also proved to be extremely reliable. From these results, it appears that these scores are eligible for use in extensive screening programs to select candidate patients for endoscopy. This is particularly true for the SS, which has a better cost-benefit. Endoscopy should be recommended in those patients who are categorized as positive by the proposed algorithm, to detect AAG early and prevent clinical manifestations.
A Laboratory Score for Autoimmune Gastritis
15. 16. 17.
18.
19.
20.
REFERENCES 1. Strickland RG, Mackay IR. A reappraisal of the nature and significance of chronic atrophic gastritis. Am J Dig Dis. 1973;18:426–440. 2. Torbenson M, Abraham SC, Boitnott J, et al. Autoimmune gastritis: distinct histological and immunohistochemical findings before complete loss of oxyntic glands. Mod Pathol. 2002;15:102–109. 3. Toh BH, Van Driel IR, Gleeson PA. Pernicious anemia. N Engl J Med. 1997;337:1441–1448. 4. Hershko C, Ronson A, Souroujon M, et al. Variable hematologic presentation of autoimmune gastritis: age-related progression from iron deficiency to cobalamin depletion. Blood. 2006;107:1673–1679. 5. Stabler SP, Allen RH, Savage DG, et al. Clinical spectrum and diagnosis of cobalamin deficiency. Blood. 1990;76:871–881. 6. Shorvon SD, Carney MW, Chanarin I, et al. The neuropsychiatry of megaloblastic anaemia. Br Med J. 1980;281: 1036–1038. 7. Van Oijen MG, Sipponen P, Laheij RJ, et al. Gastric status and vitamin B12 levels in cardiovascular patients. Dig Dis Sci. 2007;52:2186–2189. 8. Rede´en S, Ryberg A, Petersson F, et al. Homocysteine levels in chronic gastritis and other conditions: relations to incident cardiovascular disease and dementia. Dig Dis Sci. 2010;55:351–358. 9. Betterle C, Dal Pra C, Mantero F, et al. Autoimmune adrenal insufficiency and autoimmune polyendocrine syndromes: autoantibodies, autoantigens, and their applicability in diagnosis and disease prediction. Endocr Rev. 2002;23:327–364. 10. Correa P. Human gastric carcinogenesis: a multistep and multifactorial process. Cancer Res. 1992;52:6735–6740. 11. Burkitt MD, Pritchard DM. Pathogenesis and management of gastric carcinoid tumours. Aliment Pharmacol Ther. 2006;24:1305–1320. 12. Kokkola A, Sjo¨blom SM, Haapiainen R, et al. The risk of gastric carcinoma and carcinoid tumours in patients with pernicious anaemia. A prospective follow-up study. Scand J Gastroenterol. 1998;33:88–92. 13. Lahner E, Annibale B. Pernicious anemia: new insights from a gastroenterological point of view. World J Gastroenterol. 2009;15:5121–5128. 14. Soykan I, Yakut M, Keskin O, et al. Clinical profiles, endoscopic and laboratory features and associated
r
2014 Lippincott Williams & Wilkins
21. 22. 23. 24. 25. 26.
27. 28.
29. 30.
31. 32. 33. 34.
factors in patients with autoimmune gastritis. Digestion. 2012;86:20–26. Sipponen P, Graham DJ. Importance of atrophic gastritis in diagnostics and prevention of gastric cancer: application of plasma biomarkers. Scand J Gastroenterol. 2007;42:2–10. Carmel R. Pernicious anemia. The expected findings of very low serum cobalamin levels, anemia, and macrocytosis are often lacking. Arch Intern Med. 1988;148:1712–1714. Iijima K, Abe Y, Kikuchi R, et al. Serum biomarker tests are useful in delineating between patients with gastric atrophy and normal, healthy stomach. World J Gastroenterol. 2009;15: 853–859. Sipponen P, Ranta P, Helske T, et al. Serum levels of amidated gastrin-17 and pepsinogen I in atrophic gastritis: an observational case-control study. Scand J Gastroenterol. 2002;37: 785–791. Va¨a¨na¨nen H, Vauhkonen M, Helske T, et al. Non-endoscopic diagnosis of atrophic gastritis with a blood test. Correlation between gastric histology and serum levels of gastrin-17 and pepsinogen I: a multicentre study. Eur J Gastroenterol Hepatol. 2003;15:885–891. Korman MG, Strickland RG, Hansky J. Serum gastrin in chronic gastritis. BMJ. 1971;2:16–18. Checchi S, Montanaro M, Pasqui L, et al. Serum ghrelin as a marker of atrophic body gastritis in patients with parietal cell antibodies. J Clin Endocrinol Metab. 2007;92:4346–4351. Borch K, Axelsson CK, Halgreen H, et al. The ratio of pepsinogen A to pepsinogen C: a sensitive test for atrophic gastritis. Scand J Gastroenterol. 1989;24:870–876. Kekki M, Samloff IM, Varis K. Serum pepsinogen I and serum gastrin in the screening of severe atrophic corpus gastritis. Scand J Gastroenterol. 1991;26:109–116. Oksanen A, Sipponen P, Miettinen A, et al. Evaluation of blood tests to predict normal gastric mucosa. Scand J Gastroenterol. 2000;35:791–795. Toh BH, Chan J, Kyaw T, et al. Cutting edge issues in autoimmune gastritis. Clin Rev Allergy Immunol. 2012;42: 269–278. Lahner E, Norman GL, Severi C, et al. Reassessment of intrinsic factor and parietal cell autoantibodies in atrophic gastritis with respect to cobalamin deficiency. Am J Gastroenterol. 2009;104:2071–2079. Carmel R. Prevalence of undiagnosed pernicious anemia in the elderly. Arch Intern Med. 1996;156:1097–1100. Telaranta-Keerie A, Kara R, Paloheimo L, et al. Prevalence of undiagnosed advanced atrophic corpus gastritis in Finland: an observational study among 4,256 volunteers without specific complaints. Scand J Gastroenterol. 2010;45:1036–1041. Miceli E, Lenti MV, Padula D, et al. Common features of patients with autoimmune atrophic gastritis. Clin Gastroenterol Hepatol. 2012;10:812–814. Dixon MF, Genta RM, Yardley JH, et al. Classification and grading of gastritis. The updated Sydney System. International Workshop on the Histopathology of Gastritis, Houston 1994. Am J Surg Pathol. 1996;20:1161–1181. Rome Foundation. Guidelines-Rome III diagnostic criteria for functional gastrointestinal disorders. Gastrointestin Liver Dis. 2006;15:307–312. Andres E, Serraj K. Optimal management of pernicious anemia. J Blood Med. 2012;3:97–103. Davidson RJL, Atrah HI, Sewell HF. Longitudinal study of circulating gastric antibodies in pernicious anaemia. J Clin Pathol. 1989;42:1092–1095. Khan S, Del-Duca C, Fenton E, et al. Limited value of testing for intrinsic factor antibodies with negative gastric parietal cell antibodies in pernicious anaemia. J Clin Pathol. 2009;62:439–441.
www.jcge.com |
e5
