Cameron Lesions in Patients with Hiatal Hernias: Prevalence, Presentation, and Treatment Outcome

Abstract

Background and study aims

Cameron lesions, as defined by erosions and ulcerations at the diaphragmatic hiatus, are found in the setting of gastrointestinal (GI) bleeding in patients with a hiatus hernia (HH). The study aim was to determine the epidemiology and clinical manifestations of Cameron lesions.

Patients and methods

This is a retrospective cohort study evaluating consecutive patients undergoing upper endoscopy over a two year period. Endoscopy reports were systematically reviewed to determine presence or absence of Cameron lesions and hiatus hernia. Inpatient and outpatient records were reviewed to determine prevalence, risk factors, and outcome of medical treatment of Cameron lesions.

Results

Of 8260 upper endoscopic examinations, 1306 (20.2%) reported a HH. When categorized by size, 65.6% of HH were small (<3 cm), 23.0% moderate (3-4.9 cm) and 11.4% were large (≥5 cm). Of these, 43 patients (mean age 65.2 years, 49% female) had Cameron lesions, with a prevalence of 3.3% in the presence of HH. Prevalence was highest with large HH (12.8%). On univariate analysis, large HH, frequent NSAID use, GI bleeding (both occult and overt) and nadir hemoglobin level were significantly greater with Cameron lesions compared to HH without Cameron lesions (p≤0.03). Large HH size and NSAID use were identified as independent risk factors for Cameron lesions on multivariate logistic regression analysis.

Conclusions

Cameron lesions are more prevalent in the setting of large HH and NSAID use, can be associated with GI bleeding, and can respond to medical management.

Background

Cameron lesions describe single or multiple gastric erosions and/or ulcerations typically visualized at the level of the diaphragmatic hiatus on upper endoscopic examination. Though often found incidentally, Cameron lesions may represent a source for obscure gastrointestinal (GI) bleeding in patients with hiatal hernias (HH) (1). Historically, chronic blood loss has been the main reported complication of HH (2), and Cameron et al were the first to describe an association between HH and occult GI bleeding (1). Acute GI bleeding can also occur, and overall bleeding rates have been reported to be as high as 58% (3). Cameron lesions can be missed and potentially are underreported (4-8).

The aim of this retrospective case control study was to describe the epidemiology and clinical manifestations of Cameron lesions. Additionally, we aimed to evaluate long term outcomes following medical and surgical management of patients with Cameron lesions.

Methods

Consecutive adult patients (>18 years old) found to have HH on upper endoscopy at our tertiary care center between May 2008 and May 2010 were retrospectively identified from the institutional endoscopic database (Provation MD, Minneapolis, MN). Patients were excluded if procedure documentation was incomplete. The clinical electronic medical record was interrogated and correlated with endoscopic findings on each patient undergoing upper endoscopy during the study period. Demographic, clinical and endoscopic parameters in patients with Cameron lesions were compared to a cohort of patients with HH but without Cameron lesions, identified over the same time frame. The study protocol and interrogation of the electronic medical record was approved by the Institutional Review Board at Washington University School of Medicine.

Electronic upper endoscopy procedure notes and endoscopic images were carefully scrutinized by investigators not involved in the clinical care of the patients. Procedure details, indications for the procedure, size of HH, and ancillary endoscopic findings (including but not limited to esophagitis, stigmata of recent bleeding and biopsy details if relevant) were extracted. In the point and click report system used for our endoscopic reports, the choice for HH size does not force a choice between axial and paraesophageal HH. Our endoscopists have not reported the presence of a paraesophageal HH on a consistent or systematic fashion; therefore for the purpose of this report, only axial lengths of HH are reported. . HH were classified by size [small (<3cm), moderate (3-4.9 cm) and large (≥5cm)]. The endoscopic database was then cross-referenced with the institution’s electronic medical record. Both inpatient and outpatient electronic records were interrogated. The following data was extracted: baseline demographics, clinical characteristics, indication for EGD, type of GI bleed (overt or occult), nadir hemoglobin, medications including proton pump inhibitor (PPI) use, treatment recommendations and outcome at follow up. Overt bleeding was defined as evidence of visible blood loss, either melena or hematochezia. Occult bleeding was defined as iron deficiency anemia and/or positive fecal occult blood test with no evidence of visible blood loss. In patients with multiple procedures, only details from the index endoscopy were extracted. Aborted procedures or reports with incomplete data recording were excluded.

For the purpose of this study, patients were diagnosed with Cameron lesions if both of the following findings were recorded at time of endoscopy: a) a HH was identified, and b) single or multiple erosions/ulcerations were reported at the level of the diaphragmatic hiatus. The endoscopic diagnosis and descriptions of erosions and ulcers within HH were corroborated with images taken during the endoscopy (Figure 1). Patients with clear evidence of a Cameron lesion were included as study cases while the control group consisted of all patients who were identified as having a HH without evidence of a Cameron lesion.

Figure 1.

Cameron lesions on endoscopy. Cameron lesions are typically seen as a mucosal erosion or ulcer, located at the diaphragmatic hiatus within a HH. The left panel shows a shallow Cameron lesion with a clean base. The right panel demonstrates a Cameron lesion with a visible vessel and bloody effluent, treated endoscopically with thermal therapy for hemostasis.

Data are reported as mean (standard deviation, SD) unless otherwise indicated. Grouped continuous variables were compared using two-tailed Student’s t tests. Intergroup comparisons between categorical variables were made using the Chi-squared test and Fisher’s exact test. Logistic regression was performed to examine predictors of the presence of Cameron lesions within identified axial hiatal hernias and a priori demographic and clinical factors. A p value < 0.05 was required for statistical significance. All analyses were performed with SPSS, version 19.0 (SPSS, Chicago, IL).

Results

A total of 8260 upper endoscopic procedures were performed during the 24 month study period. After evaluating for duplicate procedures and excluding aborted or incompletely documented procedures, a total of 6469 unique patients undergoing upper endoscopic procedures were identified. Of these, 1306 patients (20.2%, age 61.1 ± 15.4 years, 56.1% female) were identified as having HH. Classifying HH by size, there were 857 (65.6%) small HH, 300 (23.0%) moderate sized HH, and 149 (11.4%) large HH. Within this cohort, 43 patients [mean age: 65.2 (14.1) years, 49% females] had Cameron lesions identified at endoscopy, with a prevalence of 0.66% among all upper endoscopic procedures, and 3.3% among all patients with a HH. There was a gradient of prevalence of Cameron lesions when segregated by HH size. Of the 857 small HH, 10 (1.2%) had a Cameron lesion. Among 300 moderate sized HH, 14 (4.7%) had a Cameron lesion. Large HH had the highest prevalence: 19 of 149(12.8%) had a Cameron lesion (Figure 2A, p <0.001 across groups). Upon breakdown of the proportion of Cameron lesions by HH size, 23.2% were identified in small HH, 32.6% in moderate HH, and 44.2% in large HH (Figure 2B).

Figure 2.

Prevalence of Cameron lesions among patients with a hiatus hernia (HH). There is a gradient of prevalence of Cameron lesions, the highest in patients with a large HH (p<0.001 across groups). A. Proportion of patients with a Cameron lesion among all patients with the corresponding size HH. B. Proportion of patients with the HH of the corresponding size among patients with a Cameron lesion.

Further analyses were performed within the cohort of 1306 patients with HH. There were no age and gender differences between those with and without Cameron lesions (Table 1). However, patients with Cameron lesions had a significantly higher proportion of large HH (44.2% vs. 10.3%, p< 0.001) and more frequent NSAID use (46.5% vs. 9.9%, p< 0.001) when compared to HH patients without Cameron erosions. Further, nadir hemoglobin was lower, and the proportion with GI bleeding was higher in the Cameron lesion group (Table 1). When the cohort of patients with large HH (n=149, 19 with Cameron lesions) was further analyzed, those with Cameron lesions had a significantly higher prevalence of GI bleeding (73.7% vs. 31.5%, p<0.001) and a higher proportion with NSAID use at presentation (42.1% vs. 15.1%, p=0.005).

TABLE 1.

Demographic and Clinical Characteristics

	Cameron lesions n=43	HH without Cameron lesions n=1263	P value
Mean age (SD), yrs	65.2 (14.1)	61.0 (15.4)	0.08
Gender (% female)	21 (49)	712 (56)	0.36
HH size (%)
Small (≥ 2cm)	10 (23.2)	847 (67.1)	<0.001
Moderate (3-4cm)	14 (32.6)	286 (22.6)	0.13
Large (≥ 5cm)	19 (44.2)	130 (10.3)	<0.001
GI bleeding (%)	33 (76.7)	372 (29.5)	<0.001
Overt	18 (41.9)	243 (19.2)	<0.001
Occult	15 (34.9)	129 (10.2)	<0.001
Hemoglobin nadir (SD) (g/dL)	8.3 (2.5)	9.4 (2.6)	0.03
On PPI at presentation (%)	16 (37)	603 (47)	0.13
On NSAID at presentation	20 (46.5)	121 (9.9)	<0.001
Underwent surgical repair after HH diagnosis (%)	8 (18.6)	50 (4.0)	<0.001
Mean (SD) follow-up duration , yrs.	12.7 (6.7)	4.6 (9.6)	<0.001

HH: hiatus hernia; GI: gastrointestinal; PPI: proton pump inhibitor; NSAID: nonsteroidal anti-inflammatory drug

The most common clinical presentation of Cameron lesions was GI bleeding (76.7%), overt bleeding in 18 patients (41.9%) and occult bleeding in 15 (34.9%) patients. In a bivariate logistic regression, Cameron lesions were more commonly associated with overt bleeding in this setting (OR 3.0, 95% CI 1.62-5.62, p<0.001). Concomitant peptic ulcer disease was found in 12 (27.9%) patients in the Cameron lesion cohort. In the setting of overt GI bleeding, high-risk bleeding stigmata were found in 8 (18.6%), and included: pigmented material at ulcer base (2, 4.7%), non-bleeding visible vessel (2, 4.7%), overlying clot (3, 7.0%) and active bleeding (1, 2.3%). Endoscopic therapy was performed in 6 patients, with hemostatic clip placement, epinephrine injection, and/or heater probe application employed alone or in combination. None of the 15 patients who presented with occult bleeding had high-risk endoscopic stigmata identified. The majority (41, 95%) of patients with GI bleeding from Cameron lesions were managed medically, with PPI alone or combined with iron supplementation. Six (31.6%) patients with a Cameron lesion within the large HH cohort underwent surgical fundoplication. During an observation period of an average of 12.7 (6.7) months, there was no documented recurrence of GI bleeding in any of the patients diagnosed with Cameron lesions.

Factors independently associated with Cameron lesions were further evaluated. A multivariable logistic regression model was created containing demographic and clinical variables including age, gender, hiatal hernia size, PPI use, and NSAID use as dependent variables, and Cameron lesions as the outcome variable. The overall model was statistically significant (likelihood Chi-square=68.27, df=5, p<0.001) with a C statistic of 0.810 (Table 2). In this model, NSAID use (OR 7.06, 95%CI 3.64-13.69, p <0.001) and large HH size (OR 6.37, 95% CI 3.16-12.85, p <0.001) were independent, significant risk factors for Cameron lesions (Table 2). In order to assess statistically whether NSAID use and size of the hiatal hernia had an interactive effect, a regression model containing the HH size, NSAID use and HH*NSAID use was evaluated. This model demonstrated that the interaction term was not significant [exp (B) =0.422, p=0.202].

TABLE 2.

Risk factors for Cameron Lesions: Results of Multivariable Logistic Regression

	Odds Ratio	95% CI	P value
Age	1.01	0.988-1.03	0.387
Male gender	1.44	0.751-2.78	0.270
Large HH	6.37	3.16-12.85	<0.001
Not on PPI at diagnosis	1.87	0.952-3.66	0.069
NSAID use	7.06	3.64-13.69	<0.001

Discussion

In this large retrospective study, we report the prevalence of Cameron lesions in 3.3% of patients with HH identified on endoscopy, with a higher prevalence in patients with large HH. In our cohort, larger HH size, and NSAID use were independently significant risk factors for Cameron lesions, with no evidence of a significant interaction. Overt GI bleeding was the most common presenting symptom. In the majority, therapy with oral PPI is effective in treating GI bleeding related to Cameron lesions.

Cameron lesions can be easily missed and may be underreported, as diagnosis requires a high index of suspicion, familiarity with endoscopic appearance, and a careful endoscopic examination of the HH. Visualization involves adequate insufflation and distention, both antegrade and retrograde examinations, and inspection of the gastric mucosa on both sides of the diaphragmatic hiatus (8). Given these limitations, the prevalence of Cameron lesions is not well established. The literature suggests that the prevalence of these lesions is known to vary with HH size, with the highest prevalence occurring in those with a large HH. Cameron lesions have been reported in up to 5% of patients with HH and up to 13.7% of patients with a large HH (8). We report a similar prevalence of 3.3% in HH patients. There is a gradient of prevalence within this cohort, the highest in large HH and least in small HH. Further, we report that the overall prevalence of Cameron lesions is less than 1% in patients undergoing upper endoscopy.

Cameron lesions have traditionally been thought of as the mucosal sequela of mechanical trauma from the HH (1, 8, 9). Indeed ours and other studies have identified large HH size as a major risk factor for Cameron lesions and iron deficiency. To our knowledge this study is the first to identify that NSAID use as an independent risk factor for Cameron lesions. This observation leads us to speculate that mechanical trauma at the level of the diaphragmatic hiatus combines with mucosal injury (e.g. from luminal and mucosal factors, including gastric acid and NSAID use) in some cases, leading to Cameron lesion formation. Indeed, this dual hit hypothesis, of mechanical trauma in combination with mucosal damage from other factors being required for formation of Cameron lesions is supported by available literature (1, 3, 8, 10, 11). Thus, medical therapy consisting of antisecretory agents and perhaps withholding NSAIDs when appropriate, likely addresses non-mechanical factors that participate in the pathogenesis and thereby promoting the healing of Cameron lesions (8, 10-12). It is also possible that HH configuration changes with time, such that different areas of the stomach are under mechanical strain, allowing healing and re-injury of different areas. However, if mechanical trauma was the sole pathologic factor leading to Cameron lesions, healing might not be expected without surgical correction of the HH. It is important to recognize that large axial hiatus hernias typically can be associated with a paraesophageal component, especially in women; men generally tend to have sliding axial hiatus hernias with superimposed gastroesophageal reflux. Symptomatic paraesophageal and axial hiatus hernias could be associated with a risk for torsion and ischemia, for instance, and surgical correction remains a major consideration in these instances.

Cameron lesions typically are identified in the context of GI bleeding, both occult and overt bleeding. Our findings concur with the literature, some of which have suggested that over half the patients with Cameron lesions come to medical attention for GI bleeding (3). Presentation with GI bleeding in Cameron lesions correlated with a significantly lower nadir hemoglobin, similar to findings by Yakut et al (13). With respect to treatment outcomes, we observed that GI blood loss due to Cameron lesions resolved with medical therapy (PPI) in all patients with occult GI bleeding, while those with overt blood loss underwent endoscopic hemostasis to supplement medical therapy. While six patients did undergo surgical correction of HH during follow up, surgery was performed due to refractory reflux symptoms rather than recurrent GI bleeding. Therefore, our findings align with other reports that acid suppressive therapy should be the primary treatment offered in the setting of GI bleeding especially in small axial hernias and those without a paraesophageal component, with surgical correction of HH considered for symptomatic mixed (axial and paraesophageal) or large axial hernias, and for those with refractory symptoms (8, 10-12, 14).

The primary limitation of this study relates to its retrospective design. A prospective study following each patient with Cameron lesions would provide greater clarity in our appreciation of the natural history and treatment outcomes, and we recognize that our study is unable to make definitive conclusions in this regard. For instance, there was no follow-up data on nine patients with Cameron lesions. This may have affected our medical and surgical therapy response rates. Further, since Cameron lesions are difficult to identify during endoscopy, it is quite possible that a bleeding presentation prompted a more detailed examination of the stomach, thereby affecting incidence rates especially with bleeding presentations. Additionally, we did not have data on iron studies or H. pylori status. However, prior studies have concluded that there is no correlation between H. pylori and Cameron lesions (13) or H. pylori and HHs (15). Lastly, these data reflect our referral population within a large academic medical center. The prevalence may not be representative of a community-based gastroenterology practice.

In conclusion, our data show that Cameron lesions can participate in overt and occult GI bleeding in patients with HH. The prevalence of Cameron lesions is highest in patients with large (>5 cm) hernias, and those using NSAIDS. Thus, in individuals with those clinical features, careful examination of the hiatus hernia is particularly warranted. In our retrospective cohort, medical antisecretory management was successful in managing GI bleeding events related to Cameron lesions; in cases of overt bleeding, endoscopic interventions appear to have high rates of success.