We performed a retrospective study to investigate the usefulness of immunohistochemical stains for the diagnosis of Helicobacter pylori (HP). We reviewed 200 consecutive gastric biopsy specimens, as well as immunohistochemical stains for HP. Of the biopsy specimens, 32 were positive for HP by immunohistochemical staining; of those, HP was seen on H&E stains in 29 cases (91%). The number of high-power fields required to detect HP on H&E-stained slides ranged from 1 to 25 (mean, 5.75). Combined significant (2+ or 3+) acute and chronic inflammation had a specificity of 98% and a negative predictive value of 97%. Our results show that, in our institution, HP can be seen relatively easily with H&E staining in the majority of cases; however, a small number of cases with significant inflammation can be missed if stains are not used.
More than 25 years ago, Marshall and Warren1 characterized a curved, gram-negative, flagellated bacillus found in biopsy specimens from the antral mucosa, ultimately named Helicobacter pylori (HP). The ensuing years have produced roughly 30,000 reports ranging from HP’s association with lymphoma to its unusual fatty acid substitution in lipids and lipopolysaccharides to the usefulness of barrier-born pigs as animal models for the study of HP induced gastritis! Ridiculous? Perhaps, but this amount of literature also illustrates the shared fascination and intrigue of many people for these bacteria. Accurate diagnosis of HP involves the combined knowledge, effort, and improved technology of laboratories, endoscopists, and pathologists. Yet, a set of eyes and a microscope are still considered a convenient and practical way of making this diagnosis.
Since the discovery of HP, pathologists have used many diagnostic techniques while searching for a more sensitive and specific detection method. Among these competing techniques are immunohistochemical studies and special stains such as Giemsa and Warthin-Starry. For years, differing arguments have been developed about the need for and usefulness of these ancillary techniques. Some laboratories preemptively stain all gastric biopsy specimens, while others favor a more “judicious” application of ancillary studies. We retrospectively analyzed the use of immunohistochemical staining at our institution and reviewed the literature to assess the usefulness of preemptive staining of gastric biopsy specimens for the identification of HP.
Material and Methods
This was a retrospective study of 200 consecutive gastric biopsy specimens obtained from our files using the key words “gastritis” and/or “Helicobacter pylori” in the diagnosis line, diagnosis comment, or clinical history fields. We excluded gastric resections from our study. The H&E-stained slides, along with any accompanying special stains, were included in this study.
H&E staining was initially performed on biopsy material that had been formalin fixed and paraffin embedded. Progressive H&E staining was performed using the Leica Jung Autostainer XL (Leica Microsystems, Buffalo Grove, IL). Paraffin was removed using xylene followed by ethyl alcohol dehydration. The sections were rehydrated with tap water and stained with Gill hematoxylin (Gill et al2) for 4 minutes. After another tap water wash, Scott tap water (Luna3) was applied for 40 seconds. After an ethyl alcohol rinse, the eosin-phloxine stain (Luna3) was applied for 35 seconds. The sections were then dehydrated in ethyl alcohol, cleared with xylene, and coverslipped using a Richard-Allan mounting medium (Richard-Allan Scientific, Kalamazoo, MI). We performed immunohistochemical stains for HP (B-0471, dilution 1:200; DAKO, Carpinteria, CA) for all biopsy specimens in which no immunohistochemical stain had been initially ordered (82 cases).
The H&E-stained and accompanying immunohistochemically stained slides were independently evaluated by 2 pathologists (S.B.S. and S.A.Q.). We recorded the type and degree of inflammation on a 3-digit scale (1, mild; 2, moderate; 3, severe) Table 1. We graded neutrophils within the biopsy specimens as follows: 1, rare neutrophils (difficult to find); 2, small clusters of neutrophils (<10); and 3, more diffuse infiltration or crypt abscesses. We graded lymphocytes as follows: 1, rare lymphoid aggregate or scattered lymphocytes within the lamina propria; 2, multiple lymphoid aggregates; and 3, more diffuse infiltration and/or lymphoid follicles with germinal centers. We graded plasma cells as follows: 1, rare scattered plasma cells confined to the superficial lamina propria; 2, small aggregates extending into the glandular portion of the mucosa; and 3, diffuse (full-thickness) infiltration. In addition, the presence or absence of ulcer or erosion, atrophy, reactive change, and intestinal metaplasia was assessed. The H&E-stained slides were evaluated using dry-field microscopy, and the 40× objective was the highest power available. A minimum of 20 high-power fields (HPFs) per biopsy specimen were examined for the presence of HP, and the number of HPFs needed to view HP on H&E stains and immunostains was assessed.
Two-by-two contingency tables were populated to compare histologic features with the “gold standard” (positive identification of HP organisms via microscopy and immunohistochemical staining). Sensitivity, specificity, positive predictive value, and negative predictive value were calculated from the tables with respect to each histopathologic feature or combination thereof. The statistical significance of this categorical data was assessed using the χ2 test. For sample sets with insufficient power to generate valid results, the Fisher exact test was used. A P value less than or equal to .05 was considered statistically significant for both test methods. In addition, the average number of HPFs required to detect the organism on H&E-stained slides was calculated.
A total of 196 biopsy specimens from 181 patients with a mean age of 41.1 years fit our criteria. There were 106 (58.6%) females and 75 (41.4%) males in the study with ages ranging from 1 to 82 years (mean, 41.1 years; median, 47 years). Of the cases, 41 (20.9%) were originally signed out as chronic active gastritis and the rest as chronic gastritis.
On review, 29 biopsy specimens (29/196 [14.8%]) were positive for HP by routine H&E staining and confirmed as positive by immunohistochemical staining. Immunohistochemical studies also revealed 3 additional positive cases that were not initially identified by H&E staining; in 2 of these, HP was subsequently seen on review Image 1A and Image 1B. No cases were thought to be definitely positive on H&E and found to be negative by immunohistochemical studies. Using immunohistochemical results as the gold standard, the sensitivity and specificity of H&E review for identifying HP organisms were 91% and 100%, respectively. Of 32 HP+ biopsy specimens, 30 (94%) showed at least a moderate degree of chronic inflammation (lymphocytes and/or plasma cells) and 29 (91%) showed some degree of active inflammation. Taken individually, the most sensitive (94%) and the most specific (88%) histologic findings in our study were moderate to severe plasmacytic infiltration and moderate to severe neutrophilic infiltration, respectively. When these 2 histologic findings combined with moderate to severe lymphocytic infiltration were present in the same level, the specificity was 98%. Absent or mild plasmacytic, lymphocytic, and neutrophilic infiltration had the lowest sensitivity (3%).
In our review, we found 6 (3.1%) of 196 cases demonstrating lymphoid follicles with germinal centers. Of those 6 biopsy specimens, 5 (83%) were positive for the HP organism. The number of fields required to detect HP on H&E-stained slides ranged from 1 to 25 (mean, 5.75; median, 4). The organism was seen in 10 or fewer HPFs in 25 (86%) of the 29 cases positive by H&E. With H&E used as a reference, our results show that immunohistochemical staining has 100% positivity and sensitivity. Figure 1 and Table 2 summarize our results.
HP infection is the most common cause of chronic active gastritis.4 Other causes of chronic gastritis include autoimmune gastritis, Crohn disease, lymphocytic gastritis, and other infections.5 Endoscopic biopsies for “gastritis” constitute a considerable proportion of a pathologist’s daily workload; this factor, along with HP’s known association with gastric adenocarcinoma and lymphoma, make detection of the organism important for patient care.6 The prevalence of HP has been steadily declining, which is believed to be due to improved sanitation, but may also be due to wide availability of therapy.7 Overall, about 16% of the cases in our study were positive for HP. However, it is important to recognize that our cases were selected for gastritis, which gave us a more focused sample of inflamed gastric biopsy specimens, but also most likely led to overrepresentation of the organism in our patient sample.
Several methods and techniques, including culture, rapid urease test, H&E stains, special stains, and immunohistochemical stains, are available to pathologists and clinicians for the detection of HP in gastric biopsy specimens. Immunohistochemical staining is considered by many as the most sensitive and specific method of staining and also the method with the lowest rate of interobserver variation.8–10 However, the necessity for routine special stains and/or immunohistochemical stains has been debated in recent years. It is definitely easier to find HP by using special stains, but is it really cost-effective? A recent study by Wang et al6 confirmed what many pathologists already knew: routine special stains, specifically immunohistochemical stains, are not cost-effective or necessary. What that study failed to address, given its limited sample size, was the fatigue factor of looking at too many biopsy specimens. In our study, we correlated histomorphologic changes commonly observed with active HP infection. In addition, we investigated the ease of identifying the organism, when present, on H&E-stained slides.
As expected, in most of the positive cases, HP induced a chronic active inflammatory response.11 Cutler et al12 concluded that the absence of chronic antral inflammation was the best method to exclude infection. We found that the most sensitive and most specific histologic changes were the presence of moderate to severe plasmacytic inflammation (94%) and the presence of moderate to severe neutrophilic infiltrate (88%). The presence of concurrent moderate to severe plasmacytic, lymphocytic, and neutrophilic inflammation increased the specificity to 98%. In other words, it would be exceedingly unlikely to find a patient with HP infection who does not have at least moderate plasmacytic infiltration (negative predictive value, 98%). Conversely, it would be unlikely for a patient without HP to have moderate to severe active and chronic inflammation. In our study, all 3 cases in which HP was not observed by H&E staining but was subsequently identified by immunohistochemical staining had moderate to severe chronic active gastritis.
Overall, in the majority of cases, the most expedient and least expensive test for identifying HP in gastric biopsy specimens was microscopic examination of H&E-stained slides. The sensitivity and specificity of observing the organism using this method were 91% and 100%, respectively. The value of a properly performed H&E stain compared with other special stains in the identification of HP has been discussed in other articles.8,9,13,14 Our findings agree with those of others who have found that a properly done H&E stain might be as effective as special stains.13,15 Most important, when the pathology reports for the study cases were reviewed, we found that the pathologists who had originally signed out the cases identified HP in all positive cases (with or without a stain). In other words, all of the additional stains we performed in this study (82 biopsy specimens) did not identify any unidentified case of HP-induced gastritis.
Intestinal metaplasia is a common finding in all forms of gastritis, and its prevalence increases as the disease progresses Image 1C and Image 1D. In general, HP adheres only to the native gastric epithelium and is not found on areas of intestinal metaplasia.7,16 In our study, intestinal metaplasia was observed in 17 of 196 biopsy specimens, 4 of which were positive for HP. None of the biopsy specimens that were positive for HP and intestinal metaplasia had the organism attached to the metaplastic epithelium. Of note, 2 of 3 biopsy specimens positive for HP by immunohistochemical studies but not identified by H&E staining had intestinal metaplasia present. This finding may support the observation that HP is difficult to detect in cases with extensive intestinal metaplasia.11 Intestinal metaplasia, along with ulceration, erosion, and reactive changes, demonstrated no statistically significant association with the presence of HP. Lymphoid follicles with germinal centers are not only characteristic but also a hallmark of the diagnosis of chronic HP gastritis. In fact, some authors believe that if sufficient biopsy specimens are examined, these lymphoid follicles will be found in 100% of HP+ cases.5,7 Our study shows that the presence of lymphoid follicles with germinal centers correlates highly with the presence of HP, despite the small number of cases involved.
One of the arguments for the routine use of special stains is that “the use of special stains abrogates the need for a prolonged search for the bacteria on routine H&E-stained slides.”14 Our study and those by others6,9,13 show that a “prolonged search” is seldom required. First, the typical histologic appearance of HP gastritis is moderate to severe chronic active gastritis, which is readily recognizable at low power. Second, most cases positive for HP have a significant number of organisms on the luminal surface of the foveolar epithelium Image 1E and Image 1F. Finally, we showed that the average number of HPFs needed to identify the organisms is 6 (5.75, rounded). In fact, the organism was seen in 10 or fewer HPFs in the majority of cases. We add that neither of the pathologists who evaluated the slides has any special training in gastrointestinal pathology. In addition, there was no significant difference between the resident and faculty member in identifying the organisms.
On the other hand, it is fair to say that identifying the organism on immunostain was much easier and less time-consuming. In our study, the organism could be seen from low power in many cases using the immunostain, and only a limited number of HPFs were required to confirm its presence (a maximum of 3 HPFs). In an extremely busy practice, this reduced time and effort can significantly improve the efficiency and productivity of pathologists.
This study did not look at clinical outcomes and the significance of missing a case of HP on a patient’s health. It is also imperative to note that different laboratories have different H&E staining techniques with variable sensitivities and specificities for identifying HP. Our study included only cases from our laboratory, and our findings may not accurately apply to other institutions.
The need for special stains for the detection of HP is institution- and laboratory-dependent. Our study shows that, in our institution, the routine use of special stains is not necessary for the identification of HP because the organism is readily identifiable in the majority of cases with H&E staining. A more comprehensive study including clinical findings, prior treatment, other clinical tests, outcome data, and cost analysis would be helpful in making a better case for routinely performing or not performing the special stains. We recommend the use of special stains for biopsy specimens with moderate to severe chronic active or inactive gastritis in which HP is not identified by H&E staining, for posttreatment biopsy specimens, and in cases in which structures “suspicious” but not definitive for HP are seen on H&E stains. Immunohistochemical staining is highly sensitive and specific for HP, and lymphoid follicles with germinal centers strongly suggest the presence of HP.