Factors associated with lingual tonsil hypertrophy in Canadian adults

Lingual tonsil hypertrophy can play a major role in OSA and in difficult intubations, yet little attention has been paid to the etiology. There appears to be a complex interplay with laryngopharyngeal reflux (LPR) emerging as a strong potential contributor to LTH and, subsequently, OSA. Previous authors have also demonstrated an association between LTH and OSA, BMI, age and smoking, but there has not been general agreement [1, 7]. This study adds further support to an association between a positive reflux finding score and BMI with lingual tonsil hypertrophy, but does not support a history of childhood tonsillectomy or age being associated with lingual tonsil size in adults.

Sung et al. demonstrated a correlation between BMI, reflux finding score and lingual tonsil hypertrophy in OSA patients [1]. A trend was also seen for a negative correlation with age, but was not statistically significant. This association has also been supported by Friedman’s group in Chicago, where a statistically significant association was found between increasing LTH and decreasing age, RSI >10 and positive smoking status [7]. Interestingly, they did not, however, find an association with BMI, PPI use or allergy medication use. Our study failed to identify any trend or correlation between age and LTH. Part of the discrepancy may be related to the scoring scale used by the respective authors. In this study, the LTH scoring scale described by Sung et al. was used because Friedman’s scale was not yet published at the time of study [8]. Previous methods of measurement using both CT and MRI have also been described, but given the expense are not justified for routine evaluation in a universal-payer system such as in Canada [1]. Both Sung and Friedman’s scales have demonstrated similarly good inter-rater agreement. The kappa for Sung’s grading scale was reported to be 0.73, while Friedman’s grading scale had a reported kappa of 0.78 for the video assessment and 0.87 for live assessment. An advantage of Friedman’s scale is the use of video and various positions of the tongue, which appears to allow for greater inter-rater agreement and consistency and may be a source of some of our inter-rater disagreement. Sung’s group did, however, demonstrate good correlation with measurements on MRI suggesting validity of their scale and the use of standardized photography for grading [1]. Lastly, although lingual tonsil hypertrophy has been observed in pediatric patients with prior adenotonsillectomy, a post-mortem study examining 497 corpses found only 16 (3.2%) had LTH. Of those, 6 (37.5%) had evidence of previous tonsillectomy versus 119 (23.9%) of the whole study sample, but formal statistical analysis was not performed [9].

It is likely there is a complex interplay between obesity, OSA, LPR and LTH. DelGaudio et al. demonstrated increasing severity of LTH with more severe nasopharyngeal reflux on PH probe testing, but also found that those with mild LTH had a BMI that was 8 points lower than those in the moderate and severe groups. They did not assess for OSA in the studied population [10]. It is known that OSA creates negative intra-thoracic pressure that can exacerbate reflux, but other evidence suggests that a vasovagal reflex arc may be triggered by refluxate [11]. Two previous studies have demonstrated that treatment of reflux can help in the treatment of OSA. Friedman demonstrated an average reduction in AHI from 38 to 29 in patients with a negative pH study after treatment with proton pump therapy [12]. Senior found that the apnea index decreased by 31% and respiratory disturbance index decreased by 25% with treatment with omeprazole and lifestyle modifications after one month of therapy in patients with confirmed reflux on pH probe testing [13].

Limitations of this study include the cross-sectional nature of the study and lack of confirmation of OSA or LPR by objective testing. A recent study by Chang et al. examining the reliability of the RFS score among general otolaryngologists found only fair agreement and would suggest it is not reliable among non-expert users [14]. This has been the case with other studies as well [4]. Unfortunately, a more reliable endoscopic grading tool does not yet exist and use of it may have contributed to error in this study. The sample size of this study also potentially risks a type II error but no trend was present in the tonsillectomy and LTH data, making this unlikely. The p-value of the previously mentioned cadaveric study examining tonsillectomy and LTH was calculated using Fisher’s exact test to be 0.25, with a sample size of almost 500 [9]. With no previously published studies finding an association between adult LTH and childhood tonsillectomy, establishing the required sample size a priori was not possible. We performed a sample size calculation assuming that tonsillectomy would produce an LTH difference similar to what was seen in this study with the RFS and BMI. This gave a sample size of 68 with a beta of 0.8. A further refinement of this study would have involved collecting specific data regarding the age at which childhood tonsillectomy was performed.