Following the WTC attack on 11 September 2001, a diverse population of workers were exposed to vast amounts of dust while working in rescue, recovery and debris removal.¹ The World Trade Centre Health Program (WTCHP) has performed long-term health surveillance on more than 90 000 responders for effects from these exposures.²

New and persistent upper airway (UA) diseases (rhinosinusitis, gastro-oesophageal reflux disease and obstructive sleep apnoea (OSA)) have been consistent findings in the WTC General Responder cohort (WTCGRC). A statistically significant association between time of arrival at the site and respiratory symptoms has been demonstrated.¹ In the Role of Nasal Pathology in WTC Responders (WTCSNORE) study, we documented a prevalence of 43% of chronic rhinosinusitis (CRS).³ Many responders did not consistently use respiratory protective equipment, exposing them to highly alkaline and corrosive dust, including calcium sulphate (gypsum) and calcium carbonate (calcite), known chemical irritants to UA mucus membranes and large airways, that may have led to progression to CRS.⁴

OSA is a chronic condition with recurrent episodes of partial or complete UA collapse during sleep, main risk factors being obesity and male gender. Prevalence of OSA in general populations ranges from 5% to 50%.^{5 6} The consequences of untreated OSA include excessive daytime sleepiness, impaired quality of life, depression and cardiovascular and cerebrovascular morbidity and mortality.^{7 8} In WTCSNORE, we recruited subjects who did not have a diagnosis of OSA and had no significant snoring prior to 9/11; we also oversampled current non-snorers to reduce selection bias from snorers seeking therapy. Despite this, there was a 75% prevalence of OSA, significantly higher than any other epidemiological cohort reported in the literature with similar characteristics including weight, snoring and sleepiness levels.^{3 5 6} We also found that CRS was an independent risk factor for OSA.³ To explain this association, we initially hypothesised that nasal inflammation and nasal congestion from CRS increase nasal resistance and compromise UA patency, thus providing a potential mechanism for OSA. However, we were unable to demonstrate a difference in OSA prevalence between subjects with high versus low nasal resistance⁹ suggesting other mechanisms that impair mechanoreflexes to negative inspiratory pressure such as neuropathy, and fibrosis may impact UA function and reduced UA sensitivity.^10-12 Both the afferent and efferent limbs of UA reflexes are needed to maintain UA patency. Loss of sensation would lead to failure of UA muscle activation and contribute to UA collapse. An impaired afferent limb of UA reflexes relating to CRS might reduce the ability to perceive and/or process UA loading contributing to failure of reflex muscle activation of the UA. Additionally, UA sensory dysfunction is reported in OSA.^{12 13}

We hypothesised that subjects with both CRS and OSA would have decreased sensation in the UA compared with control subjects with neither CRS nor OSA, while subjects with CRS without OSA, and OSA without CRS would have intermediate values. To test this hypothesis, we examined the role of UA sensory function (2-point discrimination (2PD) and vibration sensitivity threshold (VT)) in four groups of subjects: with/without OSA and with/without CRS.

We recruited 163 subjects (age 57.3±7.6 years, BMI 29.8±5.0 kg/m² (mean±std)). We had sleep study data on 157 subjects, and UA sensation testing data on 147. Final analyses with regard to UA sensation results were performed on the 147 subjects. 69% of the 157 subjects who had sleep studies had OSA and 49% of all 163 recruited subjects had CRS. The median AHI was 7.0/hour (range of 0–75); RDI 17.47/hour (range of 2–101); ESS 5 (range of 0–22) (see table 1).

When we examined subjects based on their OSA and CRS status (table 2), patients who had OSA and CRS had a higher BMI than without OSA and or CRS. Similarly, those with OSA and CRS were sleepier than those without either OSA or CRS. Of note, those with CRS alone manifested greater sleepiness than those with OSA alone. As expected, AHI, RDI and CRS scores in those with OSA and CRS had significantly higher values than those without OSA and/or CRS (table 2). VT and 2PD in the UA (table 2, figures 1 and 2) or in the control locations (hand, and lip) did not differ among the groups (table 2). No differences were noted controlling for BMI (data not shown). Very few of our subjects (n=21) had either high or very high levels of WTC exposure,¹⁷ and we did not find an association between exposure levels and impairment in either vibration or tactile sensation (data not shown).

Our study did not demonstrate differences in UA sensory perception in subjects with/without OSA and CRS. Kimoff^{11 12} and others^18-21 previously demonstrated an oropharyngeal sensory impairment and mucosal inflammation in OSA,^{18 22} with only slight reversibility with Continuous Positive Airway Pressure (CPAP).¹² Nerve demyelination in UA tissue²³ and changes in UA mucosal nerve endings consistent with a pattern of injury and repair²⁴ have also been reported. Their findings unlike our study support the presence of an UA sensory neuropathy in OSA, but do not distinguish between causal damage (as could happen from CRS) or changes secondary to OSA, as from mechanical trauma associated with snoring or vibration, tissue traction and from obstructive events, or oxidative stress related to hypoxiareoxygenation or systemic inflammation.^25-27

There were significant differences between subjects in our population compared with the population that Kimoff et al¹² studied, who were recruited from a sleep clinic, were younger (mean age 41), more obese (mean BMI 33) and had more severe OSA (mean AHI of 48) than ours. Additionally, the control values in their subjects for mean VT (2.2 vs 9.7) and median 2-PD (10.5 mm vs 13 mm) were much lower than in our population (better sensitivity) despite similar methodology. In fact, the values obtained for 2-PD in our overall population were in the range seen in the OSA population in the Kimoff study (online supplemental table 1). A study by Jobin et al²¹ using similar methods to Kimoff also showed impaired UA VT and 2-PD in those with OSA. These subjects also had very severe OSA (mean AHI of 57/hour) and were obese (mean BMI of 32.4 kg/m²). A study by Guilleminault et al¹⁸ showed higher values for 2-PD in OSA than in subjects with upper airway resistance syndrome (UARS) and controls and no differences between UARS and controls. Interestingly, control, snorers and UARS subjects had 2-PD values ranging from 1.0 to 2 mm and those with OSA ranging from 3 to 5 mm. Similar results were also observed by Jeong et al²⁰ in children using the same device with greater impairment noted in children with obesity.¹³ These values are much lower than the values obtained by Kimoff or us in controls but were obtained with a different device which may explain this result. A device similar to the one used by Guilleminualt and Jeong was used by Heiser et al,¹⁹ and revealed a range of 5±2.4 mm in controls and 11.5±5.4 mm in patients with OSA. Finally, similar to our study, a study examining vibration sensation in men and women without and with snoring and OSA showed no difference in vibration sensation in any of these groups in men.²⁸ However, the vibration sensation was tested with a different device than the one we used. These findings suggest that results could be very dependent on the methodology of measurement, the severity of OSA, BMI and sex. Indeed, a recent systematic review and meta-analysis of UA sensation in OSA found that for all of the studies that performed 2-PD and VT, data on validation using objective measures such as pharyngeal nerve conduction velocities, airflow stimuli and cough reflex were unavailable.²⁹ We ensured reliability and validity of our measurements across different researchers by performing repeated measurements in the same test subjects across multiple days.

Other methods of testing UA sensation, including airflow, chemical, gustatory, olfactory, temperature and neurosensation have also been studied in patients with OSA.^{19-21 28-32} Overall, while these studies reported a significant correlation between UA sensory impairment and increased OSA severity, the results were significantly heterogeneous.²⁹

There is less evidence of impaired UA sensation in patients with CRS. A loss of smell has been reported in patients with CRS and has been attributed to inflammation and release of cytokines.³³ Impairment of olfactory neuronal function from chronic inflammation has also been suggested in animal studies.³⁴

While our findings suggest that the presence of CRS and OSA has no effect on UA sensory tactile and vibration sensory perception among the participants, the reasons for these negative results may be due to the population studied.

The WTCGRC were exposed to potential neurotoxic heavy metals and complex hydrocarbons in WTC dust and fumes which provide biological plausibility for the hypothesis that WTC exposure increases the risk of neuropathy.³⁵ The latter may in turn explain the substantially higher values in both UA vibration and tactile sensory thresholds in our subjects, compared with the controls in the Kimoff study and the similarity of those values to those seen in subjects with OSA in their study (see online supplemental table 1).¹² Thus, this population may have impaired UA and sensory perception as a result of WTC-related exposures, even in the absence of CRS and OSA that may mask any effect from having either or both diseases. There may be a ceiling effect to 2-PD and vibration sensory impairment and therefore differences between those with and without OSA and or CRS may not have been observed.

That our results for the UA are likely accurate is confirmed by the absence of difference in the 2-PD test results in the lip and the hand between our controls and the Kimoff study controls (online supplemental table 1). Additionally, we saw no differences in the VT and 2-PD test results between the two sites where testing was performed by different researchers (online supplemental table 2). Finally, in our population, while there was a significant correlation between the 2-PD in the lip to the UA (correlation coefficient r=0.17, p<0.05), the correlations between those two sites or between the lip and the hand were much closer for VT (r=0.38, p<0.0001, and r=0.48, p<0.0001, respectively); VT was, however, higher in our study in the lip and hand and significantly higher in the UA compared with those found in the Kimoff study. This may have been due to the modification of the probe length in order to allow it to reach the posterior pharynx. While these findings suggest a possible impairment in vibration sensation in the WTC population involving the extremities and the UA, the modification of the probe length for VT sensation testing may have impacted this finding. There was, however, in addition an impairment to tactile sensation in the UA, suggesting the presence of impaired UA sensation in this population irrespective of the presence of OSA or CRS.

Peripheral neuropathy symptoms are reported in the WTC population and are associated with exposure intensity.^{35 36} Neurotoxins can affect primarily large myelinated fibres, thereby affecting vibration sense with relatively little impairment in the detection of tactile stimuli.³⁷ Thus, the greater impairment in vibration sense relative to 2-PD in our population as compared with the healthy population in Kimoff’s study may be due to exposure to neurotoxins from WTC dust. We, however, did not find an association between exposure levels to dust and impairment in either vibration or tactile sensation. This may have been because very few of our subjects (n=24) had high or very high levels of exposure subjects.