Article Text

Download PDFPDF

Limited value of testing for intrinsic factor antibodies with negative gastric parietal cell antibodies in pernicious anaemia
  1. S Khan1,2,
  2. C Del-Duca1,
  3. E Fenton1,
  4. S Holding2,
  5. J Hirst1,
  6. P C Doré2,3,
  7. W A C Sewell1,3
  1. 1Path Links Immunology, Scunthorpe General Hospital, Scunthorpe, UK
  2. 2Department of Immunology, Hull Royal Infirmary, Hull, UK
  3. 3Hull York Medical School, University of Hull, Hull, UK
  1. Dr S Khan, Path Links Immunology, Scunthorpe General Hospital, Cliff Gardens, Scunthorpe DN15 7BH, UK; sujoykhan{at}


Background: The appropriate testing strategy for diagnosing pernicious anaemia using gastric parietal cell (GPC) and/or intrinsic factor antibodies (IFA) is controversial. Intrinsic factor antibodies are found in only about 70% of cases. Indirect immunofluorescence screening for gastric parietal cell antibodies is more sensitive, labour intensive, and less specific.

Methods: The frequency of antibody positivity (IFA and/or GPC) was retrospectively examined in patients tested for both autoantibodies over a three-year period. It was investigated whether B12 levels were related to antibody status. These findings were validated in a prospective study of IFA in 91 GPC negative patients with low B12 levels.

Results: Of 847 samples identified in the retrospective study, 4 (0.47%) were positive for only intrinsic factor antibodies, 731 (86.3%) positive for GPC alone, and 112 (13.2%) for both. Student t test on log-transformed data showed B12 levels had no bearing on autoantibody status. 91 consecutive patients with low B12 levels were tested for both autoantibodies; all were negative for gastric parietal cell antibodies. Only one sample was positive for intrinsic factor antibody using the porcine intrinsic factor assay, but was negative by a human recombinant intrinsic factor-based ELISA.

Conclusions: This study provides evidence that testing for gastric parietal cell antibodies is an appropriate screening test for pernicious anaemia, with intrinsic factor antibodies reserved for confirmatory testing or in patients with other autoantibodies that mask the GPC pattern; B12 levels are not related to autoantibody status.

Statistics from

Circulating gastric parietal cell (GPC) autoantibodies are detected in about 90% of patients with pernicious anaemia1 and considered as the end stage of type A chronic atrophic gastritis.2 Antibodies to intrinsic factor are seen in 50–70% of patients and are viewed as a more specific marker of pernicious anaemia.3 The necessity of early detection of gastric parietal cell and/or intrinsic factor antibodies (IFA), with the need for regular follow-up of pernicious anaemia patients to watch for the development of additional autoimmune diseases has been recently highlighted.4 However, the optimal testing strategy remains unclear and considerable controversy still exists as to whether both indirect immunofluorescence testing (for GPC) and ELISA (for IFA) need to be done as a screening test for pernicious anaemia.4 5 It has been reported that as GPC antibodies tend to disappear, IFA levels increase,6 and that the presence of IFA may be influenced by patient age and race.7 8

The indirect immunofluorescence (IIF) test employs rodent stomach as substrate, and positive staining of the parietal cell cytoplasm indicates the presence of GPC antibodies.9 Although this test is labour intensive, it is sensitive and has the additional advantage of determining the presence of additional antibodies such as mitochondrial, liver kidney microsomal, and nuclear (antinuclear antibody), as liver and kidney tissues are examined along with stomach tissue. GPC antibodies are directed against the α- and β-subunits of the gastric H+/K+ ATPase, and three parietal cell autoantigens of 60–90 (β-subunit), 92 and 100–120 kDa have been identified.10 Murine models of autoimmune gastritis show features of chronic mononuclear cell infiltrate (pathogenic CD4+ Tcells reactive to α- and β-H+/K+ ATPase) in gastric mucosa and destruction of parietal and zymogenic cells,11 similar to that of human autoimmune gastritis.

GPC antibodies are more common in women than in men, and the prevalence increases with age. It is estimated that 16% of the healthy elderly population aged >60 years are positive for GPC antibodies.9 GPC antibodies coexist with autoimmune diseases such as type 1 diabetes mellitus,4 Sjögren syndrome,9 and autoimmune hypothyroidism,12 and are reported to be a useful marker of occult or latent pernicious anaemia in first-degree relatives with pernicious anaemia.13 The interval between development of autoantibodies and subsequent development of pernicious anaemia can be as long as two to three decades.9

We wanted to assess the utility of screening low B12 samples for GPC antibodies alone, rather than using IFA (or both) assays. We undertook a retrospective analysis of samples that had both GPC and IFA tests done over a three-year period (including B12 levels measured within a year), that was followed by a prospective study of samples with low B12 levels which were tested for both autoantibodies.


Our laboratories undertake reflex testing for GPC antibodies in samples found to have low B12 levels. This protocol was assessed as part of a laboratory validation process and so did not require ethical approval. A retrospective study of anonymised data identified patients tested for both autoantibodies over a three-year period. Results of B12 levels were recorded if requested within a year for that patient. Following the initial study, a prospective study of IFA in low B12 samples, negative for GPC antibodies, was carried out.

Anaemia was defined as a haemoglobin level of ⩽115 g/l in women and ⩽135 g/l in men; it was microcytic (mean corpuscular volume (MCV) <80 fl) or macrocytic (MCV >100 fl).

Serum B12 levels were assayed in the chemiluminescence immunoassay analyser Architect i 2000 (Abbott Diagnostics Laboratories, Abbott Park, Illinois, USA).

Gastric parietal cell antibodies by indirect immunofluorescence

GPC antibodies were detected by IIF using mouse liver, kidney and stomach (LKS) as a substrate (Bio-Diagnostics, Worcestershire, UK). Bound IgG was detected using antihuman IgG fluorescein isothiocyanate (FITC) at a screening dilution of 1 in 40. GPC antibodies were reported as negative or positive if there was cytoplasmic staining of parietal cells. This can be difficult to interpret with a brush-border pattern and is obscured in the presence of anti-mitochondrial antibodies.

Intrinsic factor antibodies using ELISA

IFA were detected by a solid phase enzyme immunoassay with highly purified intrinsic factor purified from porcine gastric mucosa as the antigen and performed as per the manufacturer’s instructions (EuroImmun, Lübeck, Germany). The recombinant human IFA assay was done according to the manufacturer’s instructions (Orgentec, Launch Diagnostics, Kent, UK).


Our three-year retrospective study identified 2067 samples that were tested for GPC and IFA, including B12 levels measured within a year of the test. Of these, 112 (13.2% of those with either autoantibody) were positive for both autoantibodies, 1220 were negative for both, 731 were GPC+/IFA− (86.3%), and only 4 (0.47%) were GPC−/IFA+ (see fig 1). Student t test on log-transformed data showed no significant difference in B12 level between any of the autoantibody groups.

Figure 1

B12 level is independent of autoantibody status. The (x) denotes the mean B12 level for each group and the notch in the box denotes the median B12 levels. Student t-test between GPC+/IFA− and GPC+/IFA+ showed p = 0.14 (unequal variance); and between GPC+/IFA− and GPC−/IFA+, p = 0.10 (unequal variance).

The four IFA+/GPC− cases had an age range of 56–87 years; B12 levels were <60–265 ng/l (normal range 180–1130). Two had undetectable B12 levels (<60 ng/l) and raised MCV values (136.6 fl and 135.5 fl); the former had haemolytic anaemia with anti-Lewis b antibody and the latter had rectal malignancy. Another patient had autoimmune hepatitis with low B12 level (118 ng/l) and positive IFA at diagnosis; he became negative for both autoantibodies after four years.

Having established that IFA+/GPC− patients are very rare, we undertook a prospective study to verify this finding. Ninety-one consecutive patients with low B12 levels and negative for GPC were tested for IFA using the porcine IF assay. Thirteen of these had evidence of macrocytosis (MCV range 101–112 fl); haemoglobin concentrations were 10.6–14.5 g/dl and B12 levels were <60–140 ng/l. Eight patients had low ferritin levels of 2.4–9.4 μg/l (normal range 10–204); one of these had a raised MCV (102 fl), but was negative for autoantibodies.

All 91 patients were negative for GPC antibodies, and 90 tested negative for IFA. The one IFA-positive sample, that tested positive on the porcine IF assay, was negative using a recombinant human IF antigen. This patient had haemoglobin of 148 g/l, MCV 99 fl and B12 <60 ng/l. No evidence of any other autoantibodies (smooth muscle, mitochondrial or liver–kidney microsomal antibody) was identified.


Our study demonstrates that both autoantibodies do not need to be tested simultaneously, as the finding of IFA alone is very rare. The small number of GPC−/IFA+ samples means that the results of any statistical tests should be interpreted with caution, but visual assessment of the distributions on the box and whisker plots confirms that the B12 results found in GPC−/IFA+ samples were similar to those in the other groups. We were unable to identify any specific features that could be used to guide the laboratory to test for IFA in the absence of GPC antibodies.

The final diagnosis of patients investigated in this study is unknown; hence the sensitivity/specificity of GPC+/−IFA cannot be determined. We included samples that had both autoantibodies, and also B12 levels measured to prevent an ascertainment bias from being introduced.

The incidence of B12 deficiency increases with age and is common in the elderly; other causes of low B12 levels include dietary B12 deficiency, post-surgical malabsorption, Crohn’s disease, coeliac disease, Whipple disease, chronic pancreatitis, and infection with Helicobacter pylori that can cause a severe form of cobalamin malabsorption.14 Iron deficiency anaemia can be the initial presentation of patients with H pylori induced atrophic body gastritis, who may progress to cobalamin depletion with age.15 We identified eight patients with low ferritin and B12 levels without any autoantibodies. It is possible that they either had H pylori with low ferritin and B12 levels, or had advanced autoimmune destruction of parietal cells such that no antigens were available to drive the autoimmune process.

GPC autoantibody testing is therefore the most appropriate means of screening for pernicious anaemia, with IFA testing as a more specific, but less sensitive test, being reserved for confirmatory testing. If this method of testing were adopted, it would reduce laboratory costs significantly (£2.20 (approx. €2.40 or $3.20) per well on the IFA kit compared to £0.62 (approx. €0.66 or $0.90) per well on mouse LKS slide, costs refer to reagents only). However, if IIF testing demonstrates heterophile or antimitochondrial antibodies, which can mask a true GPC antibody, they should be tested for intrinsic factor antibodies.

Take-home messages

  • Gastric parietal cell antibody detection using indirect immunofluorescence can be used as a screening test for pernicious anaemia.

  • Intrinsic factor antibodies should only be used as a confirmatory test, or if the immunofluorescence pattern is masked by other autoantibodies.

  • B12 levels had no bearing on autoantibody status.


We thank the laboratory staff at Path Links Immunology, Scunthorpe General Hospital for their support.



  • Competing interests: None declared.

Request Permissions

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.