Characterization of laryngopharyngeal reflux in patients with premalignant or early carcinomas of the larynx




An association between laryngopharyngeal reflux (LPR) and laryngeal carcinoma has been suggested, but remains unproven. The current pilot study was performed to determine the incidence of LPR among patients with early laryngeal carcinomas or dysplasia and to examine the associations between levels of LPR and histologic stage, smoking status, the symptom of heartburn, and body position during reflux episodes.


Behavioral and 24 hour pH monitoring data were prospectively acquired and analyzed for 40 previously untreated adults with dysplasia, T1or T2 laryngeal carcinomas.


Eighty-five percent of patients had LPR. No significant association was shown between the level of LPR and histologic stage or smoking status. Heartburn did not predict LPR. The incidence of LPR in the upright body position was 91%, compared with 9% in the supine position, among patients with LPR.


When compared to available normative data, the current findings show a high incidence of LPR in patients with premalignant and early laryngeal cancer. These findings highlight the need for a matched-control study evaluating LPR as a potential predisposing factor for laryngeal carcinoma. Cancer 2003;97:1010–4. © 2003 American Cancer Society.

DOI 10.1002/cncr.11158

An association has been suggested between laryngeal carcinoma and laryngopharyngeal reflux (LPR), the retrograde propulsion of stomach acid to the level of the larynx and pharynx.1, 2 Despite publication of several clinical reports, there is no consensus on whether LPR is associated with squamous cell carcinoma of the larynx, nor is there data establishing LPR as a definitive risk factor for this malignancy. Furthermore, no definitive criteria and little information are available to assist clinicians in managing patients with both LPR and potentially malignant laryngeal lesions. As the first step in an ongoing investigation of the relationship between LPR and laryngeal carcinogenesis, we determined the incidence and pattern of presentation of LPR in patients with premalignant lesions or early carcinomas of the larynx.

The mutagenic effects of tobacco carcinogens that lead to genetic abnormalities in the multistep development of laryngeal carcinoma are well described.3 In addition, smoking decreases lower esophageal sphincter tone, delays gastric emptying, and stimulates gastric acid secretion, thereby increasing the potential for gastroesophageal reflux (GER) and LPR.2 The reflux of gastric contents may act as a cocarcinogen to potentiate the mutagenic effects of tobacco.2 In nonsmokers, acid reflux may act as an independent risk factor for laryngeal carcinoma, similar to the role of gastroesophageal reflux in the development of Barrett esophagus and esophageal neoplasia.4

Currently, there is no consensus regarding the appropriate methodology to detect or quantify LPR, the indications for treatment, and the optimal assessment of treatment response.1, 5–7 Although 24 hour ambulatory pH monitoring is a well-established method for evaluating GER, it has not yet attained the same level of validation for LPR. The level of normal reflux into the laryngopharynx remains controversial.5, 8, 9 As a result, data characterizing the incidence and pattern of LPR remain sparse and the signs and symptoms of LPR that may distinguish it from GER have not been identified.

The aim of the current pilot study was to determine, using objective methodology and criteria, the incidence of LPR in patients with early (T1 or T2) laryngeal carcinomas or premalignant lesions (dysplasia). In addition, we looked for associations between levels of LPR and histologic stage, smoking status, the symptom of heartburn, and body position during reflux events. These data will form the foundation for future investigations of the potential role of LPR in laryngeal carcinogenesis and its impact on the management of early neoplasia.


The current study was approved by The University of Texas M.D. Anderson Cancer Center institutional review board, and all patients provided written informed consent to participate. Forty previously untreated patients presenting with moderate to severe laryngeal dysplasia or early laryngeal carcinoma (stage T1 or T2) were recruited to participate. Patients who had been previously biopsied and/or treated for GER were eligible for inclusion.

Data Acquisition

Standard 24 hour dual-probe monitoring was performed using a Zinetics 24 single-use internal reference pH catheter with dual sensors and an 18 cm interprobe distance (Medtronics Functional Diagnostics, Inc., Shoreview, MN). The pH probe was placed under direct visualization using a flexible fiberoptic endoscope as described by Vincent et al.8 The catheter distance measured at the level of the nares was recorded for confirmation of stability following the 24 hour monitoring period. The catheter was secured to the nasal dorsum with adhesive. The pH data were recorded at four second intervals throughout the monitoring period on a Digitrapper III ambulatory pH monitor (Medtronics Functional Diagnostics).

During the 24 hour monitoring period, patients kept a diary of all oral intake except water and other specific information, such as meal times, intervals spent in the supine position, etc. for later correlation with reflux events. Participants were asked to refrain from taking antireflux medications, including over-the-counter antacids, gastric-acid pump inhibitors, or histamine blockers, for a minimum of 72 hours before examination. The patients were instructed to maintain their normal routine during the 24 hour study period. All patients tolerated the procedure well and completed the 24 hour study without experiencing adverse effects.

The data were uploaded to a personal computer using the Polygram for Windows with the Esophagram Reflux Analysis Module version 2.05 software package (Medtronic Functional Diagnostics, Inc., Skovlunde, Denmark) for analysis after completion of data acquisition. The subjects' diaries were used to verify and confirm all start and end times for marked events and to insert any events that had been recorded in the diaries but not on the monitor.

The treating head and neck surgeon performed indirect laryngoscopy to identify and record signs of laryngeal reflux. The clinical signs of reflux used in the current study included laryngeal erythema, edema, granulomas or ulcers, pachydermia of the interarytenoid mucosa, and subglottic edema.

Scoring of Reflux Episodes

An LPR episode was defined as a decrease in pH to a level of 4.0 or below in the pharynx or upper esophagus. Each episode was confirmed by a similar decrease in the pH level detected by the distal sensor immediately before the occurrence of proximal reflux. We used normative data published by Vincent et al.8 to define three levels of LPR in the current study. Vincent et al. reported a majority of normal patients (52%, 12/23 patients) had at least one reflux episode (mean = 2.0, median = 1.0, 95th percentile = 6.9). Forty-eight percent of normal subjects had a RAI (reflux area index) of 0 (95th percentile = 6.3). The RAI is a measure of the area under the pH curve below 4.0, corrected for the duration of the study for each individual.8 On the basis of this data, we divided the study population into three subgroups according to the total number of reflux episodes (TRE) and the RAI. Each patient was assigned to levels 0, 1, or 2, according to their pH monitoring results. Patients in level 0 subgroup had no evidence of reflux as defined by 0 TRE and a RAI of 0. All other patients were categorized as having LPR and were assigned to level 1 or 2 based on their TRE or RAI, whichever was higher. Patients in level 2 subgroup had a higher than 7 TRE and/or an RAI > 6.3. These values represent reflux levels above the 95th percentile reported for normal subjects. Level 1 subgroup was made up of patients who exhibited reflux episodes that fell below the 95th percentile reported for normal subjects. The criteria for the three subgroups are shown in Table 1.

Table 1. Inclusion Criteria and Patient Distribution for Laryngopharyngeal Reflux
LPR LevelCriteriaaDistribution
  • LPR: laryngopharyngeal reflux; TRE: total number of reflux episodes; RAI: reflux area index.

  • a

    Patients were categorized into LPR levels based on either their TRE or their RAI, whichever was higher.

2> 7and/or> 6.32255

Data Analysis

To avoid erroneous inclusion of reflux events while the patients were eating, all events that occurred during meal intervals and within two minutes after the recorded meal end time were excluded from analysis. Each level of LPR (0, 1, or 2) was analyzed and compared with respect to four parameters: 1) histologic stage, 2) smoking status (nonsmoker, former smoker, or current smoker), 3) the symptom of heartburn, and 4) body position (upright or supine) during episodes of reflux. Supine position was defined as a reclining position at an angle < 45 degrees.

Associations among categoric variables were tested using the Fisher exact test. Ninety-five percent confidence intervals for point estimates of proportions were exact. All analyses were conducted using SAS statistical software (SAS Institute, Inc., Cary, NC).


Of the 40 adult patients who participated in the study, 27 (68%) were men and 13 (32%) were women. The median age was 59 years (range, 29-73 years). All patients had histologically confirmed moderate to severe dysplasia or early carcinoma (T1 or T2) of the larynx. Twenty-six patients (65%) had premalignant lesions, one of which was supraglottic and 25 of which were glottic. Fourteen (35%) had early carcinomas of the larynx, 3 of which were supraglottic and 11 of which were glottic. Fourteen patients (35%) had never smoked, 15 (37%) were former smokers (i.e., had stopped smoking at least one year before accrual), and 11 (28%) were current smokers (i.e., had stopped smoking less than one year before accrual or were smoking at the time of accrual). Table 2 shows the characteristics of the study population.

Table 2. Characteristics of 40 Patients with Laryngopharyngeal Reflux
Histologic diagnosis  
Smoking status  

The overall incidence of LPR in this group of patients was 85% (95% confidence interval, 70–94%). Only 15% of patients did not show evidence of LPR (level 0). Thirty percent of the patients had level 1 LPR, and 55% had level 2 LPR. Table 1 shows the distribution of the patient population. A description of patient characteristics at each level of LPR is summarized in Table 3 according to histologic stage, smoking status, the symptom of heartburn, and body position at the time of reflux.

Table 3. Data Summary for All Variables and Levels of Laryngopharyngeal Reflux
CharacteristicLevel 0 (n = 6)Level 1 (n = 12)Level 2 (n = 22)
n% of leveln% of leveln% of level
Smoking status      
Body position      
 Upright > Supine  11922091
 Upright = Supine  1800
 Upright < Supine  0029

The level 0 group comprised six patients who had no evidence of LPR (TRE and RAI values of 0). Five of these patients had premalignant lesions, while one had carcinoma. Three patients had never smoked, one was a former smoker, and two were current smokers. Four patients (67%), without objective evidence for LPR, reported symptoms of heartburn, while two patients (33%) were asymptomatic.

Of the 12 patients in level 1, 6 had premalignant lesions and 6 had cancerous lesions. Four of the 12 patients had never smoked, 5 were former smokers, and 3 were current smokers. Eight of the 12 patients (67%) in level 1 were asymptomatic for heartburn, while 4 patients (33%) reported periodic symptoms.

Twenty-two patients had level 2 LPR. Fifteen patients had premalignant lesions, while seven had early laryngeal carcinoma. Seven patients had never smoked, nine were former smokers, and six were current smokers. Five (23%) of the 22 patients were asymptomatic, while 17 (77%) reported symptoms of heartburn.

We evaluated the relationship between LPR and histologic diagnosis and smoking status using the Fisher exact test. Our results, presented in Table 4, show no statistically significant associations between the histologic stage and the level of LPR (P = 0.38). Of the 26 patients with premalignant lesions, 21 (81%) had objective evidence of LPR, of whom 15 (71%) had level 2 LPR. Of the 14 patients with early carcinoma of the larynx, 13 (93%) had objective evidence of LPR, yet only 7 of these 13 (54%) presented with level 2 LPR. Similarly, we found no significant association between smoking status and the level of LPR (P = 0.88).

Table 4. Cancer Diagnosis, Smoking Status, and Laryngopharyngeal Reflux Level
CharacteristicnLPR level, n (%)P valuea
  • LPR laryngopharyngeal reflux.

  • a

    Fisher exact test.

Histologic diagnosis    0.38
 Precancer265 (19.2)6 (23.1)15 (57.7) 
 Cancer (T1, T2)141 (17.1)6 (42.9)7 (50.0) 
Smoking status    0.88
 Never143 (21.4)4 (28.6)7 (50.0) 
 Former151 (6.7)5 (33.3)9 (60.0) 
 Current112 (18.2)3 (27.3)6 (54.5) 

We also analyzed the relationship between reflux and body position. Overall, the frequency of reflux occurrences was greater in the upright position than in the supine position in 91% of the 34 patients with LPR (95% confidence interval, 76–98%).

Thirty-six patients (90%) had two or more signs of reflux as indicated by their surgeon. Interestingly, four patients (10%) had no clinical evidence of reflux on laryngoscopic examination, yet one had level 1 and three had level 2 LPR as documented by 24 hour pH monitoring.


Clinicians are increasingly aware of the deleterious effects to the larynx caused by reflux of gastric contents. Although some authors have suggested an association between LPR and laryngeal carcinogenesis, there is insufficient data to either prove or disprove this hypothesis. Future investigations of LPR as a predisposing factor for laryngeal carcinoma require accurate, objective data characterizing the incidence and presentation of reflux in patients at risk for development of this malignancy.

We determined the incidence of LPR among 40 patients with early cancer or premalignancy of the larynx. In this group of patients, 85% showed objective evidence of LPR. In comparison, 52% of normal subjects were found to have LPR in the study reported by Vincent et al.8 Importantly, 55% of the current patients with premalignant and early malignant laryngeal lesions were above the 95th percentile of the normal range of LPR reported by those authors. Only 15% of the current patients had a TRE and an RAI of 0, compared to 48% of normal subjects who had an RAI of 0.8

The results of the current study support the findings of other investigators that LPR is a common occurrence among patients diagnosed with carcinoma of the larynx. Copper et al.10 reported a 62% incidence of LPR documented with 24 hour pH monitoring in 24 patients with head and neck carcinoma. The current study, however, specifically examined LPR in patients with dysplasia and early cancer of the larynx; a population where the presence of LPR may be more clinically relevant. We found no significant association between histologic stage and incidence of LPR. However, of those patients who did have LPR, more of those with premalignant lesions (71%) had level 2 LPR compared to those with early carcinoma (54%).

While several investigators have suggested a strong relationship between the use of tobacco and the development of GER,2, 11 the current results did not show a significant relationship between smoking and LPR. However, the small numbers enrolled in the current pilot study prevented meaningful subset analysis of the three smoking categories. Other studies12 did not find strong associations between reflux and tobacco use. Thus, in the current study, the presence of LPR did not correlate with the histologic stage of the laryngeal lesion or the smoking status of the patient. Although the small sample size prevents us from making definitive conclusions, these results may imply that any effect of LPR on laryngeal carcinogenesis is independent of other variables.

Although alcohol use is a potentially important risk factor for both GER and laryngeal carcinoma, to our knowledge, there is no definitive data available on the potential association between alcohol use and LPR in the process of laryngeal carcinogenesis. We did not include alcohol use in the current analysis because the reporting of alcohol use was nonuniform in this study, making it difficult to quantify in a statistically meaningful way. However, we acknowledge that alcohol use may contribute to the development of LPR and/or laryngeal carcinoma and plan to analyze this variable in the next phase of the current study.

The current results support a distinction between the presentations of LPR and those reported for GER. LPR occurred most commonly while the patient was in the upright position, unlike GER, which more often occurs when the patient is supine. Furthermore, while heartburn is one of the cardinal symptoms of GER,1, 12, 13 we found no significant relationship between the occurrence of heartburn and LPR in our patients. In fact, four of the six patients with level 0 LPR complained of heartburn, even though pH probe testing did not reveal any evidence of laryngopharyngeal reflux. The current results suggest that heartburn may not be a reliable indicator of LPR, as others have also reported.1, 8, 13 In addition, the reliability of clinical signs as indicators of LPR requires further investigation with larger sample sizes. The current results suggest that aggressive evaluation for LPR may be indicated in patients with premalignant and early laryngeal carcinomas whether or not they report heartburn or have laryngeal signs suggestive of reflux.

The current data also show that LPR episodes in patients with laryngeal carcinoma or dysplasia occur primarily while patients are in the upright position. In the current study, 91% of patients who showed evidence of LPR were in the upright position. This result corroborates the findings of other investigators that LPR, in contrast to GER, occurs predominantly while the patient is in the upright rather than the supine position.9 Therefore, common treatment recommendations, such as elevating the head while in the supine position, may not be effective in reducing or preventing the occurrence of LPR in this population. Clear differentiation of the pathophysiologic mechanisms of GER from those of LPR is needed to determine the most effective treatment regimens for each problem.


To our knowledge, this is the first study that has scientifically evaluated a population of patients with laryngeal dysplasia and early carcinoma using objective data from 24 hour pH monitoring based on methodologically defined levels of LPR. Eighty-five percent of patients in the current pilot study showed one or more episodes of LPR, 55% of whom experienced levels of reflux above the 95th percentile reported for a group of normal subjects.8 Thus, when compared to the available normative data, the current findings suggest a high incidence of LPR in patients with early carcinogenic changes in the larynx. These findings clearly highlight the critical need for future investigations with larger sample sizes and matched controls to objectively define the parameters of pathologic reflux into the laryngopharynx and the potential role of LPR as a predisposing risk factor for the development of laryngeal carcinoma.


The authors thank Kate O'Suilleabhain for editorial assistance and. Janet Hampton for help in preparing the article.