Findings of Cognitive Impairment at High Altitude: Relationships to Acetazolamide Use and Acute Mountain Sickness

Study population

All of our subjects had recently and rapidly ascended by vehicle and foot to the two testing sites in the Annapurna region of the Nepalese Himalayas. Those adult native English-speaking subjects visiting the Himalayan Rescue Association aid station in Manang at 3520 m (11,548 ft) or at a temporary aid station in Thorong Phedi at 4550 m (14,927 ft) were offered enrollment into the study. Surveys were only offered to clinically stable subjects at the end of their medical visit after clearance from the health professional caring for them. Inclusion criteria included native English-speaking adults between the ages of 18 and 65 years trekking through Manang or Thorong Phedi. This included asymptomatic subjects who presented themselves to the clinic for education and prevention of illness. Subjects were not excluded for having symptoms of AMS or minor ailments, including joint pain, blisters, rashes, diarrhea, and cough. Exclusion criteria included a history of dementia, a lack of native fluency in English, recreational drug or alcohol use, and outside the age range. The discriminating power of the Qmci for detecting CI is influenced by age; specifically it is less accurate for subjects older than 75 years (O'Caoimh et al., 2014). We chose a more conservative threshold of 65 years to reduce the potential confounding influence of age on CI. All recruited subjects were trekkers who had acutely ascended within 1 week. No permanent residents of high altitude were recruited.

Study procedures

A convenience sample of subjects who presented to the aid stations was enrolled. A consent document with information about the study was presented to each volunteer and written consent was obtained. Information collected from each subject included age, years of education, history of dementia, fluency in English, rate of ascent, medication use, and self-rated high-altitude experience. Subjects were asked about prophylactic and treatment medications for high-altitude illness, analgesics, and any potentially sedating drugs during their trek through the Annapurna region. Medications for general medical problems (e.g., antihypertensives, oral contraceptives, thyroid replacement, etc.) were not recorded. Each subject enrolled was administered the LLS and eQmci by a single investigator. This study was approved (Protocol No. 2014P001803) by the Partners Institutional Review Board (Boston, MA) and the Nepali Health Research Council.

Screening tests

The LLS uses a 5-item scale, which surveys a subject's assessment of headache, dizziness or lightheadedness, fatigue or weakness, gastrointestinal distress, and difficulty sleeping. Each symptom is graded on a scale of 0 (not present) to 3 (severe). The presence of a headache (score at least 1 for headache) and a total score of ≥3 is required to diagnose AMS (Roach et al., 1993).

The Qmci consists of 6 subtests, including orientation, registration, clock drawing, delayed recall, verbal fluency, and logical memory. It has a median administration time of 4.24 minutes (O'Caoimh et al., 2013). The range of score is 0 to 100. Two modifications were made to better apply this screen in high-altitude wilderness settings. We removed the clock drawing task, given the practical concern of needing to remove gloves to perform the test in a very cold environment. The clock drawing test is useful to assess visuospatial cognition, but is clinically a less useful subtest when compared with other elements in the Qmci (O'Caoimh et al., 2013). Instead, to maintain a distracting task before the verbal recall exam, subjects were asked to count backwards from 100 by 7. A second modification was made in which proper names replaced common names for repetition and recall testing. Proper names were matched for frequency of occurrence in the English language. High altitude seems to have a more dramatic effect on the recall of proper names, whereas recall of common names is relatively resistant to hypoxia (Pelamatti et al., 2003). The remaining subtests were unchanged. Subjects with normal cognition have a median score of 75 (O'Caoimh et al., 2014). Subjects who score ≤67 out of 100 meet criteria for mild CI and those who score ≤53 out of 100 meet criteria for dementia (O'Caoimh et al., 2014). These cut-offs are influenced by age and education and are recommended for younger adults (≤75 years) with more education (≥high school). At this cut-off, the Qmci has an 86% sensitivity and 89% specificity for detecting the presence of mild CI. CI is not seen in a young, healthy, and educated population (D.W. Molloy, 2015. pers. comm.). Sex does not significantly affect cut-off scores (O'Caoimh et al., 2014).

Statistical analyses

Continuous variables were summarized using mean with SD and compared using two-sample t-tests, whereas categorical variables were summarized using frequency and percentage and compared using chi-square tests. The relationship between the two measures was summarized using a Spearman's correlation coefficient. Two-sided p ≤ 0.05 was considered as statistically significant. All analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC).

Lack of relationship between CI and AMS

The lack of correlation between the LLS and eQmci suggests that CI occurs independently of AMS. This is consistent with previous research (Virues-Ortega et al., 2004). Kramer et al. (1993) tested 20 climbers at 4360 m (14,304 ft) ascending Denali. While they demonstrated deficits in learning and retention in perceptual and memory tasks, they did not correlate with AMS. In one study that tested mental capacity, subjects with AMS were worse in pursuit aiming compared with subjects without AMS, but in the remaining battery of cognitive tests, no significant differences were found (Bian et al., 2015). It is possible that there is a difference in the aspects of cognitive deficits seen with and without AMS. In the study by Regard et al. (1991), subjects simulated ascent to 4500 m (14,734 ft) over 24 hours in a hypobaric chamber. Those with AMS were deficient in short-term memory, but improved in conceptual tasks, whereas those without AMS had improved short-term memory, but no improvement in conceptual tasks (Regard et al., 1991).

Acetazolamide and CI

In subjects taking acetazolamide, we found a high incidence of AMS and of CI. Acetazolamide has been associated with confusion (Swenson, 2014). In a randomized study of subjects at high altitude, subjects who received prophylactic acetazolamide (125 mg twice per day) had impaired neuropsychological measures of concentration, cognitive processing speed, reaction time, short-term memory, and working memory (Wang et al., 2013). In a similar study, subjects who took acetazolamide (500 mg slow release, once per day) actually showed improvement in certain psychological tests (White, 1984). In addition to higher daily acetazolamide dosing, this second study differed from the former in that subjects traveled on foot rather than being airlifted to a higher altitude, which may be a confounding factor. A more recent study found that at sea level, acetazolamide use is associated in a dose-dependent fashion with a spectrum of psychometric impairments, including poor concentration, imbalance, and slower reaction time responses (Collier, 2016). Our study did not differentiate between prophylactic or treatment doses, but a future study can be designed to match groups with regard to this variable. Carbonic anhydrase (CA) plays a role in signal processing, synaptic plasticity, memory, nerve conduction, and cerebral oxygenation and consumption (Brechue et al., 1997; Sun and Alkon, 2002; Wang et al., 2015). Given these findings, CA inhibition is likely responsible for the deficits in cognition. Whether CI arises from the drug-induced metabolic acidosis that partially relieves the magnitude of the respiratory alkalosis (hypocapnia) or from more direct effects in the CNS is not easily resolved, but both may be likely. At sea level only 50% of patients taking acetazolamide (250 mg, 4 times a day) have relief of common CNS side effects with concurrent bicarbonate supplementation sufficient to correct the metabolic acidosis (Lichter, 1981).

Limitations of the study

There are several limitations in this investigation. First, we made modifications to the eQmci (the clock drawing task was replaced by reciting serial 7 and the repetition and recall of names were changed from common to proper). Although the Qmci has been validated against the Montreal Cognitive Assessment and Standardized Mini-Mental State Exam, the two minor changes in the eQmci were not validated before testing. This potentially could have an effect on the cut-off values originally tested for the Qmci. Second, traveling to a distant, culturally different location may alter one's performance on a cognitive test. Third, given the relatively small subject recruitment, other trends may be seen with a more robust sample. Unfortunately, our recruitment was limited as a result of the devastating 2015 earthquake in Nepal, which resulted in termination of the study. Fourth, there may be several secondary influences on neuropsychological function, including fatigue, hypothermia, desynchronosis, upper respiratory infections, and exposure to a new environment. An individual diagnosis was not recorded for every patient; however, most presentations to the clinic that were unrelated to altitude were quite minor (e.g., blisters) and would not be expected to affect cognitive function. Fifth, there was neither a low-altitude control group, nor were our subjects tested at low altitude to examine baseline scores. Sixth, there were no measurements of arterial oxygenation saturation by pulse oximetry. In future studies, this would be a better indicator of cerebral hypoxic stress than the actual altitude. Seventh, there were no longitudinal data to determine if low eQmci scores correlate with the development of HACE. Future research is needed to define the clinical significance of mild CI and to determine whether it can it be used to predict those that will develop HACE or experience other clinically relevant sequelae. In addition, there were no longitudinal data on the discontinuation of acetazolamide. Eighth, subjects were studied at two different altitudes. However, the results were not significantly changed when data were limited to Manang, where most subjects were tested. Ninth, in a convenience sample, in which subject self-selection exists (researchers have no control or insight in subjects' decisions to continue to ascend or not), unmeasured confounding variables may exist that could reduce the validity of our conclusions.

Future studies

Regardless of AMS, screening of mild CI may be useful for early detection of impaired cognition and poor decisional judgment at high altitude. Our study, however, does not demonstrate any clinical significance of having mild CI diagnosed by a cognitive function test, since we did not measure rates of adverse health events or poor decision making, but this is a potential area of future research. Cognitive tests are sensitive and have the potential to screen for hypoxic impairment at early stages before subjects develop symptoms or demonstrate poor judgment (Stepanek et al., 2013). Other cognitive tests have been employed at high altitude such as clock drawing (Quigley and Zafren, 2016). The eQmci can be administered in <5 minutes, does not require any special equipment, and has multiple versions available to minimize a learning effect when administered to the same subject over different points in time (Cunje et al., 2007). Future longitudinal studies in trekkers, as well as validation of the eQmci and comparison to other cognitive tests, may be useful to investigate the clinical relevance of mild CI. Further study of secondary influences on CI at high altitude (e.g., exercise, medications, and sleep quality) is also needed. Finally, the comparison of CI with other objective measures (e.g., optic nerve sheath diameter, magnetic resonance imaging, and serum biomarkers) might help advance a better understanding of acute altitude-related CNS disease.

In summary, our data demonstrate only a weak correlation between the LLS and eQmci, but a surprising prevalence of CI, an objective measure of CNS dysfunction, at high altitude. CI was prevalent in subjects both with and without AMS. The presence of CI in otherwise asymptomatic trekkers raises concerns for unrecognized CI at altitude. Furthermore, we found that acetazolamide use, independent of the severity of AMS may also contribute to the presence of CI at high altitude. Further research using CI testing may help define objective measures of early CNS dysfunction at altitude.