Gastric Surgery as a Long-term Risk Factor for Malignant Lesions of the Larynx FREE

THE TERM duodenogastroesophageal reflux refers to regurgitation of the duodenal contents through the pylorus into the stomach, with subsequent reflux into the esophagus. This reflux is common after total or partial gastrectomy, occurring in 5% to 35% of the subjects.¹ After partial gastrectomy, with removal of the pylorus and antrum, the duodenal contents can easily flow backward through the small gastric remnant into the esophagus. This mechanism may also be facilitated by an abnormal clearing capacity of the esophagus due to its abnormal anatomy after surgical resection. In most cases, the vulnerable epithelium is the esophagus² (experimentally, reflux of duodenal contents into the esophagus can cause not only Barrett esophagus, with subsequent adenocarcinomas, but also squamous cell carcinomas),³ but, in theory, the epithelium of the larynx and pharynx may also be damaged. In the past few years, reflux otolaryngologic manifestations have gained the interest of physicians,^4- 5 so that a long history of reflux is identified as a common risk factor in patients with carcinoma of the laryngopharynx who have never smoked.⁶

Partial gastrectomy (or any gastric surgery that leads to bile reflux) may therefore serve as an epidemiologic model for studying the effects of duodenal reflux on the larynx and pharynx. To this aim, the present study assessed the frequency of laryngeal malignancies or premalignancy among subjects who had undergone gastric surgery, in comparison with a control group of sex- and age-matched dyspeptic individuals.

During the past years, we have accumulated a database of all patients undergoing endoscopy at our institution. For this study, we recalled all patients who had had gastric resection 5 years or more previously and who, from January 1, 1997, to January 31, 2002, were consecutively referred to us for routine follow-up upper endoscopy for an otorhinolaryngologic evaluation. Patients who underwent gastric surgery less than 5 years earlier were excluded from the study. Patients with a Roux-en-Y reconstruction were also excluded. Of the initial database of 134 eligible patients with gastric resection 5 or more years previously, we excluded 11 individuals whom we were unable to contact, 7 who had died in the meantime, and 23 who did not meet the inclusion criteria of the study (7 subjects were too far away and 16 had actually had gastric surgery <5 years previously). Ninety-three patients (mean age, 68 ± 16 years; 80 men and 13 women) were therefore eligible for this study.

From the same outpatient setting, 93 sex- and age-matched patients, without gastric resection, who consecutively underwent upper endoscopy for dyspeptic symptoms were recalled and recruited as a control group. Each control subject was enrolled prospectively after each matched patient with gastric resection. All patients included in the study gave their informed consent.

The 2 study populations underwent indirect laryngoscopy. In the event of any suggestive lesions, videolaryngoscopy with biopsy for histologic examination or direct vocal cord decortication was performed. Other nonmalignant laryngeal lesions, such as vocal cord edema, posterior laryngitis, and interarytenoid area edema, were all classified as benign.

The clinical history of patients and endoscopic findings were obtained and recorded in a prospective fashion; the same type of information was acquired for all subjects. Patients with previous laryngeal malignancy were therefore investigated with regard to their habits before the development of the laryngeal lesion. A questionnaire was used to obtain a history from each patient regarding his or her social habits. Regarding the use of tobacco, patients were classified as nonsmokers (persons with no history of smoking habit), former smokers (persons who had smoked but had quit before our evaluation), and active smokers (persons who were smoking at the time of our evaluation). Information about the quit date, number of cigarettes smoked per day, and duration of smoking over time were also recorded. To quantify the smoking habit, we considered the pack-year smoking history (1 pack-year = 25 cigarettes per day for 1 year). The alcohol intake was recorded as the mean number of standard drinks consumed per day (1 standard drink = 12 g of absolute alcohol).⁷ Patients were subdivided into 3 groups: nondrinkers, including subjects who were teetotalers or who occasionally drank small quantities of alcohol; social drinkers, men or women who habitually drank 3 or fewer drinks per day or 2 or fewer drinks per day, respectively; and heavy drinkers, men or women who habitually drank more than 3 drinks per day or more than 2 drinks per day, respectively.⁸

Each patient received a validated questionnaire⁹ to evaluate typical symptoms of reflux disease (heartburn, regurgitation, and chest pain) as not present and present and scored as follows: 0, no symptoms; 1, mild; 2, moderate; and 3, severe.

Crude and adjusted odds ratios (ORs) and relative 95% confidence intervals (CIs) were calculated. The Wilcoxon-Mann-Whitney test was used to compare distribution of possible confounding factors between groups. Multivariate analysis was performed with logistic regression with forward selection method (P<.1 was used as inclusion criteria). All analyses were performed with Stata software (Stata Corp, College Park, Tex). All tests for significance were 2-tailed.

Of 93 individuals with gastric resection who were enrolled in this study, 17 had a vagotomy and antrectomy with gastroduodenostomy reconstruction (Billroth I [BI]), 65 patients had a vagotomy with gastrojejunostomy reconstruction (Billroth II [BII]), and 11 had a total gastrectomy. Eighty-nine patients (96%) had been operated on because of complications of peptic ulcer disease and 4 (4%) because of gastric cancer. Twenty-two subjects underwent gastric resection less than 10 years earlier, 20 patients from 10 to 20 years earlier, and 51 patients more than 20 years earlier. The mean time since gastric surgery was 26 years (range, 5-42 years).

Of 93 patients with gastric resection, 7 patients (8%) had current or previous laryngeal malignancies or current precancerous mucosal changes (PMCs). The time range since gastric resection in these patients was 22 to 42 years. In particular, 3 patients (3%) had previously undergone vocal cordectomy for squamous cell carcinoma (SCC) of the larynx, whereas laryngoscopy detected areas of leukoplakia on the true vocal cords in 2 patients (2%) and on the false vocal cords in the other 2. Histologic examination of the patients with areas of leukoplakia, after vocal cord decortication, showed moderate epithelial dysplasias in 3 cases and severe dysplasia in 1 case. Clinical characteristics of the 7 subjects with SCC or PMC are shown in Table 1.

Among the individuals in the control group, laryngoscopy diagnosed leukoplakia on the vocal cord in 1 subject, who reported reflux symptoms and had both smoking and alcohol habits. Histologic examination after vocal decortication showed mild dysplasia in this subject.

Alcohol (social and heavy drinkers) and smoking (former and current smokers) habits were not statistically different between individuals with gastric resection and the other group (P = .49 and P = .16, respectively).

The risk of developing an SCC or PMC in patients with gastric resection vs control patients was not statistically significant (crude OR, 7.00; 95% CI, 0.87-55.77; Fisher exact test, P = .06). The adjusted OR (with sex, age, and alcohol [yes or no] and smoking [yes or no] history included in the regression model) was 9.88 (95% CI, 1.01-97.31; likelihood ratio χ² = 28.77; P<.001). The adjusted OR (with sex, age, and alcohol [yes or no] and smoking [considering the pack-years] history included in the regression model) was 30.15 (95% CI, 1.67-550.85; likelihood ratio χ² = 33.34; P<.001). The adjusted OR (with sex, age, and alcohol [nondrinkers, social drinkers, and heavy drinkers] and smoking [nonsmokers, former smokers, and current smokers] history included in the regression model) was 9.15 (95% CI, 1.03-81.15; logistic regression χ² = 10.77; P<.04). Of the covariates included in the analysis, only the pack-years were statistically significant in the last model (OR, 3.40; 95% CI, 1.73-5.21).

All subjects with laryngeal SCC or PMC had had a BII gastric resection.

There was a significant increased prevalence of benign laryngeal lesions in patients with gastric resection vs the control group. The difference in the frequency of benign findings between patients who had undergone gastric resection more than 20 years earlier as compared with no more than 20 or less than 20 years earlier and between gastric resection performed 10 or less than 10 years as compared with more than 10 years earlier were both not significant. In addition, we did not find any significant difference in the prevalence of benign lesions between patients with various types of gastric resection.

Finally, we did not find any significant differences between subjects with gastric resection and the control group regarding the presence of hiatal hernia and clinical history of reflux (Table 2). The difference in the presence of esophagitis was, on the contrary, significant (P<.001).

This study, although small, is the first, to our knowledge, that suggests the development of laryngeal malignancy or premalignancies as long-term complications of gastric surgery (8% of cases among subjects with previous gastric resection). The limited statistical power of this study does not allow definitive conclusions to be drawn, and the association that we describe is worthy of further investigation with large sample sizes in larger collaborative studies. However, our results support the hypothesis that a chronic reflux of duodenal contents (containing bile acids and salts, and pancreatic enzymes), present in excess after gastric surgical resection, may promote the development of laryngeal malignant changes by creating chronic stimulation and inflammation on the squamous epithelium of the larynx,¹⁰ with possible subsequent neoplastic degeneration. The great relative distance of the larynx from the stomach could explain the long time (>20 years in all 7 subjects with SCC or PMC in this study) needed for the reflux to become harmful.

The foregoing hypothesis is compatible with the significantly higher frequency of benign laryngeal lesions we found among patients who had had gastric resection than in the control group. It is possible, in fact, that the benign abnormalities are related to the reflux and may represent index lesions caused by chronic stimulation. In addition, the hypothesis is consistent with the significantly higher frequency of esophagitis found among patients with gastric resection as compared with the control group (Table 2).

It is also possible that the high percentage of laryngeal malignancy or premalignancy found among patients with gastric resection reflects the fact that we enrolled patients referred to us for upper endoscopy. In theory, these patients may have had more symptoms (including those caused by reflux) than other patients with gastric resection who did not require endoscopy. In any case, this observation is also consistent with our hypothesis.

Duodenogastroesophageal reflux may be harmful because of several mechanisms. In particular, trypsin can play a key role, since, when not inactivated by pepsin, it remains active for 1 hour at various pH levels.¹¹ The proteolytic properties of this enzyme may promote detachment of the surface cells from the epithelium, presumably by digesting the intercellular substances and surface structures that contribute to the maintenance of cohesion between cells.^12- 14 In addition, evidence exists in animal models that unconjugated bile acids appear to augment the damaging effects of trypsin at pH 7.^15- 16 On the other hand, bile acids have been thought to damage mucosal cells by their detergent property and solubilization of the mucosal lipid membranes.¹⁷ Alternatively, bile acids gain entrance across the mucosa because of their lipophilic state, causing intramucosal damage primarily by disorganizing membrane structure or interfering with cellular function.^17- 18

A possible criticism of our study is that patients who underwent gastric resection may also be those with a history of heavy smoking and alcohol habits and with subsequently more severe complications of peptic ulcer disease and, therefore, candidates for gastric resection. This observation may be partly true, since tobacco and alcohol are known carcinogens and their use is a risk factor for the development of laryngeal and pharyngeal malignancies. The ideal population to investigate our hypothesis should thus be composed of nonsmokers and nondrinkers so as to directly correlate gastric resection with neoplastic lesions. However, in the present study, 1 of the 7 patients with SCC or PMC had never smoked and had never been an alcohol drinker. In addition, 2 patients had ceased to smoke after the gastric resection, developing an SCC or PMC 27 and 32 years later. Furthermore, the control population of this study comprised outpatients referred for dyspeptic symptoms to the same clinical outpatient setting (gastrointestinal endoscopy) and who were characterized by smoking and drinking habits similar to those of patients who had undergone gastric resection.

Moreover, a prerequisite for any complication of reflux is that it occurs to an extent beyond the tolerance of the epithelium. Smoking and alcohol abuse may both decrease this tolerance, as well as being associated with a higher incidence of reflux disease.^19- 22 A chronic insult on the laryngeal mucosa, represented by duodenal reflux, may therefore (as already hypothesized for acid reflux^23- 25) constitute a further risk factor or cofactor for the onset of laryngeal carcinogenesis.

In our series, all subjects with laryngeal SCC or PMC had had a BII gastric resection, although the statistical analysis with respect to the subjects with other types of surgery is not available (for this purpose, larger sample sizes of study populations will be needed). However, the finding is consistent with recent evidence, based on spectrophotometric assessment of biliary reflux (bilimetry), reporting higher levels of biliary reflux in gastric resection with BII reconstruction vs other types of surgical approach (BI and Roux-en-Y reconstruction).²⁶ Interestingly, higher mean pH values and N-nitrous compound concentrations were found in patients with BII resection as compared with both patients with BI gastrectomies and normal control subjects,²⁷ whereas N-nitrous compound levels were also recorded in patients with more severe histologic changes in gastric stump carcinogenesis.²⁸ Higher levels of polyamines (actively involved in cell proliferation) were also observed after BII gastric resection compared with subjects never operated on.²⁹

In conclusion, our findings show for the first time the possibility of onset of precancerous or neoplastic laryngeal lesions as long-term complications of gastric resection. It should also be pointed out that, of the 7 patients identified as having SCC or PMC, 4 patients were found to have premalignant lesions solely on the basis of their history of gastric resection. Although the present study appears to be worthy of further evaluation on larger patient populations, with a possible assessment of the presence and quality of reflux compounds, we believe that a periodic and careful multidisciplinary (otolaryngological and gastrointestinal) evaluation in long-term survivors of gastric resection may contribute to early diagnosis of laryngeal disorders.

Corresponding author: Giovanni Cammarota, MD, Università Cattolica del Sacro Cuore, Policlinico "A. Gemelli," Istituto di Medicina Interna e Geriatria, Largo A. Gemelli, 8, 00168 Roma, Italia (e-mail: gcammarota@rm.unicatt.it).

Accepted for publication December 22, 2002.