Cognition in Patients with Post-Bariatric Hypoglycemia

Abstract

Objective:

Bariatric surgery, a highly effective treatment for obesity and associated comorbidities, may improve cognition and brain volume in parallel with cardiometabolic function. However, some post-bariatric individuals develop post-bariatric hypoglycemia (PBH), which can be frequent and severe. The impact of recurrent hypoglycemia on cognition in PBH is unknown. The objective of this study was to determine whether individuals with PBH display reduced cognitive function compared to post-surgical counterparts without hypoglycemia.

Methods:

Fourteen adults with a history of Roux-en-Y gastric bypass (RYGB), with hypoglycemia (PBH+, n=7) or without PBH (PBH-, n=7) completed assessments of memory, executive function, attention, and psychomotor speed.

Results:

Individuals with PBH (PBH+) displayed significantly decreased performance in category fluency (p <0.01), category switching (p <0.01), and category switching accuracy (p <0.01), compared to PBH-. Performance in the first (p =0.03) and third interval (p =0.045) of verbal fluency were significantly lower in PBH+ vs. PBH-. All other assessments did not differ.

Conclusion:

PBH+ individuals may be at greater risk for cognitive impairment compared to PBH- individuals, as suggested by the impaired semantic processing and cognitive flexibility, as well as greater difficulty initiating and sustaining word retrieval in PBH+.

Introduction

A growing body of research demonstrates that obesity is associated with impaired cognition. Body mass index (BMI) is inversely correlated with performance on assessments of executive function, attention, and processing speed¹, and is an independent risk factor for poor neurocognitive outcomes² and dementia³.

Bariatric surgery is highly effective for weight loss, remission and/or improvement in type 2 diabetes (T2D), and other components of metabolic syndrome⁴. While some studies show postoperative improvements in memory, including delayed and immediate verbal memory, executive function, and language^2,5, others failed to show improvement in cognition postoperatively^2,6.

Recurrent hypoglycemia after bariatric surgery, termed post-bariatric hypoglycemia (PBH), could modify post-operative cognition. PBH is increasingly recognized as a late (>1 year postoperative) complication. While estimates vary based on definition and ascertainment, self-reported symptoms occur in 36.4% of individuals post-RYGB, with 2.6–3.6% reporting severe hypoglycemia⁷. Severe hypoglycemia can cause neuroglycopenia, with acute changes in cognition, concentration, and attention requiring the assistance of others for resolution. Repeated hypoglycemia can occur multiple times daily, decreasing counterregulatory responses and hypoglycemia awareness, increasing risk for loss of consciousness, arrhythmias, seizures, coma, and death⁸.

Previous studies in patients with type 1 or type 2 diabetes (T1D, T2D) have demonstrated deleterious effects of recurrent hypoglycemia on brain structure and cognitive health. Recurrent severe hypoglycemia is associated with cortical atrophy⁹ and reduced executive function in T1D¹⁰, and increased risk of dementia in T2D¹¹. Thus, we tested the hypotheses that: (1) recurrent hypoglycemia in individuals with PBH (PBH+) is associated with impairment in cognition vs. age-based normative data, and (2) PBH+ have impaired performance on cognitive tests vs. post-RYGB individuals without hypoglycemia (PBH-).

Methods

Participants

Nineteen adults with history of RYGB (age 34–69 years) were recruited and enrolled in two groups based on presence (PBH+) or absence of hypoglycemia (PBH-) (Figure S1). Participants were recruited through participation in ongoing studies of PBH (NCT04428866, n=21; NCT03353415, n=1), in turn recruited via the Joslin Hypoglycemia Clinic, or with the assistance of Trialfacts, Research Match, and Craigslist.

Exclusion criteria included diabetes, traumatic brain injury, cerebrovascular accident, active treatment with diabetes medications (except acarbose prescribed for hypoglycemia), active alcohol/substance abuse, other systemic illnesses, or active depression (score >4 for Patient Health Questionnaire-2 (PHQ-2)). Participants were excluded if they reported similar cognitive assessments within the last year, to prevent practice effects. Eight participants were enrolled but subsequently excluded (Figure S1).

Procedures

The protocol was approved by Joslin Diabetes Center Committee on Human Studies (CHS). After informed consent was obtained, medical history and inclusion/exclusion criteria were reviewed, and assessments were completed: Edinburgh Hypoglycemia Symptoms Scale (EHSS)¹²; Dumping Syndrome Rating Scale (DSRS)¹³; Hypoglycemia Fear Survey (HFS)-II¹⁴; Gold Score (hypoglycemia awareness scale)¹⁵; and Beck Depression Inventory¹⁶. The neuropsychological battery was conducted within 2 weeks of screening. Capillary or venous glucose was analyzed to ensure glucose was >80 mg/dL at the start of testing; no participants experienced hypoglycemia before or during testing.

Cognitive Battery

The National Adult Reading Test (NART)¹⁷ was used to predict the Wechsler Adult Intelligence Scale Verbal, Performance, and Full-Scale Intelligence Quotients. Additional cognitive tests were performed as previously described¹⁸, including the Letter-Number Sequencing subtest from the Wechsler Memory Scale Third Edition (WMS-III) to assess working memory, the Rey Auditory-Verbal Learning Test (RAVLT) List A; with a delayed recall presented 30 minutes after the last trial to assess verbal and delayed memory, and grooved pegboard to assess motor speed and psychomotor processing. The Trail Making Number-Letter Switching condition, Color Word Interference Task, and Verbal Fluency Task (including letter fluency with letters F-A-S, category fluency with animals and boy names, and category switching) of the Delis Kaplan Executive Function System (D-KEFS)¹⁹ were used to assess executive function. A single rater administered all cognitive assessments.

Continuous Glucose Monitoring

Glycemic patterns were assessed with masked Dexcom G6 Professional Continuous Glucose Monitoring (CGM), over a 10-day period within 4 months prior to cognitive testing; 4 individuals concurrently wore their personal CGM device.

Statistical analysis

Raw scores for the D-KEFS and WMS-III Letter Number Sequencing tests were converted to scaled scores using age-based normative data in the respective examiner’s/scoring manual. All scaled scores had mean of 10 and standard deviation of 3, with greater score indicating better performance. Given the small sample size, non-parametric tests were utilized. Wilcoxon rank-sum tests were used to test differences between groups for continuous variables and dichotomous variables were assessed with Fisher’s Exact Test. Analyses were completed using R Studio. Unless otherwise specified, results are reported as median (IQR). P<0.05 was considered significant.

Results

To determine whether PBH was associated with modified cognition in a cross-sectional analysis, we performed cognitive testing in 7 individuals with PBH+ and 7 PBH- individuals. Sex, age, and race/ethnicity were similar between groups (Table 1). There were no differences in other factors potentially affecting cognition, including depression symptoms, alcohol or recreational drug use, smoking, levels of attained education and predicted IQ. Preoperative and current BMI and time since surgery were similar (Table S1, S2). There was a trend for PBH+ to have lost more weight (45.7% (15.2) of initial body weight), vs. 26.5% (27.4) in PBH- (p=0.05). History of micronutrient deficiencies, neuropathy, chronic fatigue, fibromyalgia, migraine headaches, prior T2D, learning disability, post-menopausal status, and postural orthostatic tachycardic syndrome did not differ between groups (Table S2).

Table 1. Participant Characteristics and Hypoglycemia History.

Median with IQR in parentheses is reported for continuous variables. Frequency with percentage in parentheses is reported for dichotomous variables, including sex, race, ethnic group, smoker status, recreational drug use, neuroglycopenia, hypoglycemia requiring assistance of others, experienced LOC, hx of seizure, and self-reported hypoglycemia pre-op. P values were assessed using Wilcoxon Rank Sum tests for continuous variables and Fisher’s Exact Test for dichotomous variables. P <0.05 reflect significance.

	PBH+ (n = 7)	PBH− (n=7)	p-value (PBH+ vs. PBH−)
Age (yr)	56 (8)	52 (9)	0.34
Female Sex—no. (%)	6 (85.7)	7 (100)	1.00
Race—no. (%)
White	7 (100)	6 (85.7)	1.00
African American	0	0
American Indian	0	1 (14.3)
Asian	0	0
Native Hawaiian/Pacific Islander	0	0
UNK/Not Reported	0	0
Ethnic Group – no. (%)
Hispanic	0	2 (28.6)	0.46
Non-Hispanic	7 (100)	5 (71.4)
UNK/Not Reported	0	0
Current Body Mass Index	31.4 (5.4)	32.9 (11.3)	0.44
Fat Percent	41.5 (11.9)	41.5 (12.7)	0.84
PHQ-2	1 (2.5)	0 (0)	0.06
Beck Depression Inventory	8 (9.5)	7 (7)	0.37
AUDIT-C	0 (1.5)	1 (1.5)	0.50
Alcohol Use (drinks per week)	0 (1.25)	0.25 (0.31)	0.50
Smoker – no. (%)	2 (28.6)	1 (14.3)	1.00
Recreational Drug Use– no. (%)	2 (26.8)	0	0.46
Education (yrs)	14 (2)	16 (4)	0.28
NART Predicted Full-Scale IQ	115 (6)	119 (3)	0.14
NART Predicted Verbal IQ	115 (6.5)	119 (3.5)	0.16
NART Predicted Performance IQ	114 (5)	116 (2)	0.13
Hypoglycemia History
Neuroglycopenia– no. (%)	7 (100)	0	<0.001
Hypoglycemia Requiring Assistance of Others– no. (%)	5 (71.4)	0	0.02
Experienced LOC– no. (%)	4 (57.1)	0	0.07
Median # of Episodes of Loss of Consciousness	2.9 (3.7)	0	0.03
Hx of Seizure– no. (%)	1 (14.3)	0	1.00
EHSS	64 (21.5)	19 (2)	<0.01
DSRS	28 (14.5)	10 (4.5)	0.05
HSF-II B Score	20 (13)	NA	NA
HSF-II W Score	41 (18)	NA	NA
Gold Score (n = 6)	2.5 (1)	NA	NA
Duration of hypoglycemia (yrs)	3 (3.775)	NA	NA
Self-reported Hypoglycemia Pre-op– no. (%)	2 (28.6)	0 (0)	0.46

All PBH+ individuals had a history of neuroglycopenia, with 5 of 7 reporting severe hypoglycemia requiring the assistance of others (Table 1). Of these, 4 had experienced loss of consciousness (LOC), and 1 had a hypoglycemia-associated seizure. As expected, PBH+ had a significantly greater EHSS, indicating greater symptom severity, vs. PBH- (64 (21.5) vs. 19 (2); p=0.002). PBH+ reported more symptoms of dumping syndrome (28 (14.5) vs. 10 (4.5; p=0.05). PBH+ reported a median score of 2.5 (1) for the Gold score, 20 (13) for the HSF-II B Score, and 41 (18) for the HSF-II W Score; PBH- reported no hypoglycemia and thus were not assessed with these measures. Hemoglobin A1c, fasting glucose, and capillary glucose prior to cognitive testing were normal and similar between groups, as expected. CGM analysis revealed significantly greater time with sensor glucose (SG) <54 mg/dL in PBH+ vs. PBH- (1.2% (1.7) vs. 0%). PBH+ were treated with medical nutrition therapy (100%), acarbose (43%), and diazoxide (29%), while 57% also used personal CGM (Table S1).

Analysis of broad cognitive domains (pre-morbid intelligence, memory, attention, executive function, and psychomotor speed) did not differ between groups (Table 2). However, PBH+ performed significantly worse than PBH- in category fluency (p<0.01), category switching (p<0.01), category switching accuracy (p<0.01), and during the verbal fluency first (p=0.03) and third intervals (p=0.045) (Figure 1).

Table 2. Cognitive Assessment.

Median with IQR in parentheses is reported for continuous variables. P values were obtained using Wilcoxon Rank Sum tests. P <0.05 reflect significance.

	PBH+ (n = 7)	PBH− (n=7)	p-value (PBH+ vs. PBH−)
Letter Number Sequencing Scaled Score	10 (2.5)	10 (1)	0.74
RAVLT Trial 1 Raw Score	6 (1.5)	5 (1.5)	0.65
RAVLT Trial 2 Raw Score	9 (1.5)	9 (3)	0.60
RAVLT Trial 3 Raw Score	11 (1.5)	10 (5)	1.00
RAVLT Trial 4 Raw Score	12 (2.5)	12 (2)	0.65
RAVLT Trial 5 Raw Score	12 (2.5)	12 (4)	0.75
RAVLT Immediate Recall Raw Score	50 (6)	48 (14.5)	0.90
RAVLT Delayed Recall Raw	9 (2)	10 (1.5)	0.24
Trails Scaled Score	12 (2)	11 (1.5)	0.51
Color Naming Scaled Score	10 (3)	9 (4)	0.30
Word Reading Scaled Score	11 (2)	9 (2.5)	0.30
Inhibition Scaled Score	10 (2)	11 (4)	0.95
Inhibition Switching Scaled Score	11 (1.5)	12 (3.5)	0.70
Letter Fluency Scaled Score	8 (3.5)	12 (2)	0.48
Category Fluency Scaled Score	10 (3)	14 (1)	<0.01
Category Switching Scaled Score	10 (2)	15 (4.5)	<0.01
Category Switching Accuracy Scaled Score	10 (2)	14 (3)	<0.01
Verbal Fluency First Interval Scaled Score	10 (3.5)	13 (3)	0.03
Verbal Fluency Second Interval Scaled Score	10 (5)	11 (2)	0.30
Verbal Fluency Third Interval Scaled Score	10 (1.5)	12 (2)	0.045
Verbal Fluency Fourth Interval Scaled Score	8 (6)	11 (1.5)	0.37
Dominant Hand Grooved Pegboard Time to Complete	74 (11)	76 (15)	1.00
Dominant Hand Grooved Pegboard Total Score	100 (10.5)	102 (15)	0.85
Non-dominant Hand Grooved Pegboard Time to Complete	86 (11)	70 (15)	0.18
Non-dominant Hand Grooved Pegboard Total Score	102 (18)	95 (15)	0.52

Figure 1. Verbal fluency measures are reduced in PBH+ vs. PBH−.

Scaled scores for letter fluency (LF), category fluency** (CF), category switching** (CS), category switching accuracy** (CSA), and verbal fluency (VF) first* and third* intervals.

*Indicates p<0.05 between PBH− and PBH+ groups.

**Indicates p<0.01 between PBH− and PBH+ groups.

Discussion

Previous studies have demonstrated that individuals who experience recurrent hypoglycemia have impairment in key cognitive domains. We now report that individuals who experience hypoglycemia after bariatric surgery demonstrate inferior cognitive performance compared to their PBH- counterparts, specifically in tests of verbal fluency. PBH+ had reduced performance in category fluency and category switching compared to PBH-, but no significant differences in letter fluency. Such impairments in category fluency, which targets semantic processing, but not letter fluency, which assesses lexical retrieval, suggest impairment of distinct neural pathways in PBH²⁰.

Differences in verbal fluency patterns were also observed between PBH+ and PBH- participants, with PBH+ demonstrating inferior performance in the first and third intervals, indicating difficulty in initiating and sustaining word retrieval¹⁹. However, all PBH+ verbal fluency scaled scores fell within the average range, while PBH- normative scores were above average.

While patients with PBH do not have diabetes, post-surgical glucose metabolism is not normal and is characterized by marked glycemic variability⁸. If and how this pattern – beyond severe hypoglycemia – may impact cognitive function remains unknown.

Limitations of this study include its small sample size, cross-sectional approach, and low power (increased risk of type 2 error). The battery of tests utilized was limited to ensure total assessment time remained under 1 hour, to reduce the impact of participant fatigue. This constrained the number of tasks and breadth of cognitive domains that were assessed. This assessment may have been less sensitive to detect changes in additional cognitive domains, compared to studies in long-duration T1D. It is possible that some PBH- individuals had some degree of undiagnosed hypoglycemia, thereby attenuating the ability to detect significant differences between groups.

While the clinical implications of our findings remain unclear, given that preoperative data are not available for each participant, they raise the possibility that recurrent hypoglycemia in patients with PBH may exert effects on some domains of cognitive performance. Further studies will be required to thoroughly evaluate this possibility, including comparisons to non-surgical controls, and baseline and post-surgical follow-up in a larger cohort to examine the frequency of severe hypoglycemic episodes in relation to long-term cognitive performance, and to identify additional predisposing factors including obesity duration, vascular disease and nutritional deficiencies. In the meantime, clinicians should be alert for symptoms of cognitive dysfunction in post-bariatric patients and consider neurocognitive testing in parallel with assessment of glycemic patterns in those individuals.

Study Importance.

• Hypoglycemia can be unpredictable and severe in some patients after bariatric surgery; however, little is known regarding the impact of recurrent hypoglycemia on long-term cognitive function. Our findings reveal individuals with post-bariatric hypoglycemia (PBH+) demonstrate inferior cognitive performance compared to their PBH- surgical counterparts, specifically in tests of verbal fluency.

• Clinicians should be alert for symptoms of cognitive dysfunction in PBH+ patients and consider neurocognitive testing in parallel with assessment of glycemic patterns, as well as other potential cognitive risk factors.