About

Allan I. Pack, M.B.Ch.B., Ph.D., FRCP, is the John Miclot Professor at the University of Pennsylvania's Perelman School of Medicine. He is a fellow of the Institute on Aging and a member of several research institutes at the university, including the Institute for Neurological Sciences, the Center for Neurobiology and Behavior, and the Institute for Diabetes, Obesity and Metabolism. Dr. Pack's research focuses on the genetics and genomics of sleep and its disorders, conducting studies in Drosophila and mice, and translating these findings to humans. His laboratory investigates the genetic determinants of sleep homeostasis, molecular mechanisms of sleepiness, and sleep promotion through both hypothesis-driven and discovery science techniques, including behavioral sleep studies, molecular analysis, and gene expression profiling. In addition to his research, Dr. Pack is dedicated to research training and directs three NIH-funded training grants. Clinically, he specializes in sleep disorders, with a particular emphasis on the diagnosis and management of obstructive sleep apnea, and is recognized internationally for his expertise in this area.

Research topics

• Intensive care medicine
• Medicine
• Genetics
• Biology
• Physical therapy
• Internal medicine
• Psychology
• Neuroscience
• Endocrinology
• Psychiatry
• Pathology
• Computational biology
• Gerontology
• Environmental health

Selected publications

• Cell-specific variant-to-gene mapping identifies conserved neural and glial regulators of sleep

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-04-09

  articleOpen access

  Summary Excessive daytime sleepiness (EDS) is a heterogeneous phenotype with little known of its genetic basis. Large-scale genome-wide association studies (GWAS) have reported genomic loci associated with EDS, though since most of these are non-coding, the causal gene(s) underlying the association are not known. Additionally, the cell types in which these genes exert their effects on sleep have not been functionally explored in vivo . Here, we employed a chromatin-based variant-to-gene mapping approach to first implicate candidate effector genes at EDS GWAS loci in human-derived neural and glial cell lines. Subsequent cell type-specific RNAi knockdown of orthologous genes using neural and glial GAL4 drivers in Drosophila confirmed cell-specific regulation of sleep by these GWAS-implicated effector genes. Among these, ruby (ortholog to AP3B2 ), a component of the AP-3 vesicular trafficking complex emerged as a robust sleep regulator. Targeted knockdown in flies localized ruby function to astrocyte-like glia, where loss of ruby increased sleep duration. The conserved role of ruby/ ap3b2 was validated in zebrafish where CRISPR-mediated loss increased daytime sleep. Together, these findings show that physical variant-to-gene mapping predicted cell-type–specific gene function for complex sleep traits and revealed ruby/AP3B2 as a conserved glial regulator of sleep and arousal. This work provides a generalizable framework for connecting non-coding GWAS variants and their corresponding effector genes to identify novel and highly conserved regulators of sleep.

  PublisherDOI

• Cross-species evidence for a developmental origin of adult hypersomnia with loss of synaptic adhesion molecules beat-Ia/CADM2

  Nature Communications · 2026-01-12 · 2 citations

  articleOpen access

  Idiopathic hypersomnia (IH) is a poorly understood sleep disorder characterized by excessive daytime sleepiness despite normal nighttime sleep. Combining human genomics with behavioral and mechanistic studies in fish and flies, we uncover a role for beat-Ia/CADM2, synaptic adhesion molecules of the immunoglobulin superfamily, in excessive sleepiness. Neuronal knockdown of Drosophila beat-Ia results in sleepy flies and loss of the vertebrate ortholog of beat-Ia, CADM2, results in sleepy fish. We delineate a developmental function for beat-Ia in synaptic elaboration of neuropeptide F (NPF) neurites projecting to the suboesophageal zone (SEZ) of the fly brain. Brain connectome and experimental evidence demonstrate these NPF outputs synapse onto a subpopulation of SEZ GABAergic neurons to stabilize arousal. NPF is the Drosophila homolog of vertebrate neuropeptide Y (NPY), and an NPY receptor agonist restores sleep to normal levels in zebrafish lacking CADM2. These findings point towards NPY modulation as a treatment target for human hypersomnia. Cross-species studies show that loss of the synaptic adhesion molecules beat-Ia/CADM2 causes excessive sleepiness by disrupting development of wake-promoting NPF/NPY brain circuits. Targeting NPY signaling may offer treatment for human hypersomnia.

  PublisherOA PDFDOI

• 0674 Short-Term Reliability of Obstructive Sleep Apnea Symptom Subtypes

  SLEEP · 2025-05-01

  articleOpen access

  Abstract Introduction Over 10 years ago, a seminal study using cluster analysis in the Icelandic Sleep Apnea Cohort (ISAC) identified subtypes of patients with obstructive sleep apnea (OSA) defined as excessively sleepy (presenting with multiple sleepiness complaints), disturbed sleep (characterized by insomnia-related complains), and minimally symptomatic. These subtypes have since been shown to generalize in clinical and population samples worldwide. An important step towards clinical translation is understanding the short-term (e.g., weeks) reliability of these subtypes. We describe results from an ongoing study to answer this question across sites in the Sleep Apnea Global Interdisciplinary Consortium (SAGIC). Methods We used a prospective test-retest study of subjects with newly diagnosed or suspected OSA prior to initiating therapy. Participants completed a questionnaire on OSA symptoms twice (at least two weeks apart). A patient’s symptom subtype at each timepoint was determined based on questionnaire responses using a prediction model derived in existing SAGIC patients. To understand short-term reliability of subtypes, we calculated the percent agreement and simple kappa coefficients. Kappa values indicate slight (0.00-0.20), fair (0.21-0.40), moderate (0.41-0.60), substantial (0.61-0.80), and almost perfect (0.81-1.00) reliability. Results Our sample included 116 participants from 4 SAGIC sites: Ohio State (n=43), Kansas (n=7), Sydney (n=37) and Perth (n=29). The sample was 50% male, 55.2±13.9 years-old, had a BMI of 31.8±6.8 kg/m2, and an AHI of 22.3±18.5 events/hour. Questionnaires were completed an average (±SD) of 18.8±7.8 days apart. At the first visit, 51 (44.0%) patients had the disturbed sleep subtype, 42 (36.2%) were minimally symptomatic, and 23 (19.8%) were excessively sleepy. At the second visit, 47 (40.5%) had the disturbed sleep subtype, 43 (37.1%) the minimally symptomatic and 26 (22.4%) the excessively sleepy. Overall, 84.5% (95% CI: 77.9%, 91.1%) of patients fell within the same subtype at both timepoints. The Kappa (95% CI) was 0.759 (0.657, 0.853), indicating substantial short-term reliability. Conclusion While recruitment is ongoing across all SAGIC sites, this early analysis suggests substantial reliability of symptom subtypes over a short timeframe (without treatment). This further supports the potential clinical utility of the symptom subtype framework in advancing more personalized patient care. Support (if any)

  PublisherOA PDFDOI

• 1281 Association Between Sleep Duration and Psychomotor Vigilance Test Metrics

  SLEEP · 2025-05-01

  articleOpen accessSenior author

  Abstract Introduction Previous studies suggest an association between short and long sleep duration and adverse health outcomes, including impaired cognitive function. Using sleep diary and psychomotor vigilance test (PVT) data, we evaluated the relationship between sleep duration and vigilance/alertness. Methods De-identified data from adults enrolled in the Stanford Technology Analytics and Genomics in Sleep (STAGES) study was obtained from the National Sleep Research Resource. Data included demographics, questionnaires, polysomnography, sleep diaries, and a 3-minute PVT. Average sleep duration per night was calculated for each patient using sleep diary entries over two weeks. The primary outcomes from the PVT were the mean reciprocal of the response time (mean RRT) and the number of 355 msec lapses. Analysis of covariance (ANCOVA) was used to compare PVT outcomes among five groups based on average sleep duration: < 6 (n=33), 6-7 (n=100), 7-8 (n=272), 8-9 (n=254), and ≥ 9 (n=106) hours. Covariates included age, sex, body mass index (BMI), apnea-hypopnea index (AHI), Patient Health Questionnaire-9 (PHQ-9), time PVT was taken, and educational level. Results A total of 765 patients [44% male; age 45.2±14.6 years; BMI 31.3±9.10 kg/m2; AHI 16.2±22.0 events/hour; mean±SD] were included. The mean sleep duration was 7.88±1.12 hours/night. In covariate-adjusted analyses, both mean RRT (p=0.005) and number of 355 msec lapses (p=0.001) significantly differed based on average sleep duration. Slower response speed and more lapses occurred in patients with short (< 6 hours; mean [95% CI] of 3.16 [3.05, 3.27] 1/s and 13.6 [9.50, 17.8] lapses) and long (≥ 9 hours; 3.14 [3.08, 3.21] 1/s and 14.7 [12.3, 17.1] lapses) sleep duration, compared to those sleeping 6-7 (3.30 [3.24, 3.35] 1/s and 8.89 [7.28, 10.5] lapses), 7-8 (3.24 [3.20, 3.27] 1/s and 11.1 [9.92, 12.4] lapses) or 8-9 (3.24 [3.20, 3.27] 1/s and 10.5 [9.38, 11.6] lapses) hours. Conclusion These data suggest that individuals who report either short (< 6 hours) or long (≥ 9 hours) sleep duration have worse alertness and vigilance based on PVT. This is consistent with the “U-shaped” relationship reported between sleep duration and other health outcomes in prior studies. Support (if any) NIH grant P01 HL160471

  PublisherOA PDFDOI

• 375 Glycogen storage disease (GSD) type IV, perinatal neuromuscular type: A rare case and new discoveries on evaluation

  The American Journal of the Medical Sciences · 2025-01-15

  articleOpen access1st authorCorresponding
  PublisherDOI

• New initiative in <i>SLEEP</i> : Perspectives in sleep science and sleep medicine

  SLEEP · 2025-10-09

  articleOpen accessSenior author
  PublisherOA PDFDOI

• Abstract P2105: Polygenic risk scores for obstructive sleep apnea relying separately on BMI- adjusted and -unadjusted genetic associations reveal separate pathways of cardiovascular disease risk

  Circulation · 2025-03-11

  article

  Background: OSA is a heterogeneous disease, with obesity a significant risk factor in many but not all cases of OSA, via increased airway collapsibility, reduced lung volumes, and possibly body fat distribution. Research question: We sought to develop PRSs that summarize the genetic liability to OSA that include and exclude obesity related pathways, and to study the associations of these PRSs with OSA comorbid cardiometabolic and CVD outcomes. Approach: Using 1.2 million race/ethnic diverse samples from the Million Veteran Program, FinnGen, TOPMed, All of Us (AoU), Geisinger’s MyCode, MGB Biobank, and the Human Phenotype Project (HPP), we developed, selected, and assessed PRSs for OSA, relying on genome wide association studies both adjusted and unadjusted for BMI: BMIadjOSA and BMIunadjOSA PRS. We tested their associations with cardiometabolic and CVD outcomes in AoU. Results: In association with OSA, adjusted odds ratios (ORs) per 1 standard deviation of the PRSs ranged from 1.38 to 2.75, all statistically significant (Figure). The associations of BMIadjOSA and BMIunadjOSA PRSs with CVD outcomes in AoU shared both common and distinct patterns. For example, BMIunadjOSA PRS was associated with type 2 diabetes, heart failure, and coronary artery disease, but the associations of BMIadjOSA PRS with these outcomes were statistically insignificant with estimated OR close to 1. In contrast, both BMIadjOSA and BMIunadjOSA PRSs were associated with hypertension and stroke. Sex stratified analyses revealed that BMIadjOSA PRS association with hypertension was driven by data from females: females had OR=1.1, p-value=0.002, but males OR=1.01 and statistically insignificant. OSA PRSs were also associated with dual-energy X-ray absorptiometry (DXA) body fat measures. In BMI adjusted analysis, BMIadjOSA PRS was associated with higher visceral adipose tissue (VAT) proportion of total body fat mass (TFM), with lower proportion of gynoid fat mass out of TFM, higher proportion of android fat mass out of TFM, and lower gynoid to android fat mass. In females only, the PRS was associated with higher VAT to SAT ratio (Figure). Conclusions: Distinct components of OSA genetic risk are related to obesity and body fat distribution, and may influence clinical outcomes. These may explain differing OSA risk and associations with cardiometabolic and CVD morbidities between sex groups.

  PublisherDOI

• Positive airway pressure therapy and cardiovascular events in obstructive sleep apnoea: an observational clinical cohort study

  Sleep Medicine · 2025-12-23 · 1 citations

  articleOpen access

  Studies support the short-term benefit of continuous positive airway pressure (CPAP) therapy on cardiometabolic risk in adults with obstructive sleep apnoea (OSA). Evidence is limited on the benefits of CPAP for preventing acute major adverse cardiovascular events (MACE). This study aimed to assess the association between CPAP use and incidence of MACE in a longitudinal clinical cohort of adults with OSA at a large U.S. healthcare system. Adults with OSA (apnoea-hypopnoea index [AHI]≥5) were identified from Kaiser Permanente Southern California between 2018 and 2020 (N=34,782). MACE was defined as first occurrence of myocardial infarction, stroke, unstable angina, heart failure or cardiovascular death, using validated electronic health record algorithms. CPAP use (h/night) was based on daily telemonitoring data. Inverse probability of treatment weighted Cox proportional hazards models stratified by OSA severity (mild [5≤AHI<15]), moderate-severe [AHI ≥15]), were used to assess associations between CPAP use and MACE. Among individuals with moderate-severe OSA, those using CPAP <4 h/night (HR [95% CI] = 0.53 [0.35-0.82]; p=0.004) or ≥4 h/night (HR [95% CI] = 0.46 [0.27-0.77]; p=0.004) had lower MACE incidence compared to those not using CPAP. Increased CPAP use (in hours) was associated with lower MACE incidence in moderate-severe OSA (HR [95% CI] = 0.90 [0.82-0.98]; p=0.021). In individuals with mild OSA, CPAP use was not associated with lower MACE incidence. CPAP use was associated with lower MACE incidence in adults with moderate-severe OSA. Treatment of moderate-severe OSA may have a positive impact on prevention of MACE. • Participants with moderate-severe obstructive sleep apnea (OSA) not using continuous positive airway pressure (CPAP) were two times more likely to experience major adverse cardiovascular events (MACE) than those using CPAP • Increased CPAP use (in hours) was associated with lower MACE incidence in moderate-severe OSA • In individuals with mild OSA, CPAP use was not associated with lower MACE incidence

  PublisherDOI

• Respiratory Inductance Plethysmography to Quantify Changes in Ventilation in Obstructive Sleep Apnea

  IEEE Transactions on Biomedical Engineering · 2025-10-07 · 3 citations

  articleOpen access

  Background and objective: The study aims to determine whether respiratory inductance plethysmography (RIP) signals can be used to quantify changes in ventilation and provide advanced obstructive sleep apnea (OSA) severity metrics. This approach seeks to address limitations in current airflow-based OSA measures, particularly those relying on nasal pressure, which may be compromised by oral breathing. Methods: Adult patients with OSA (N = 89, 68Male:21Female) completed in-laboratory polysomnography (PSG) allowing for RIP-based ventilation estimates to be compared against a gold standard oronasal-pneumotach (normalized ventilation %eupnea). Concordance was assessed on three levels: 1) individual breath ventilation, 2) individual respiratory event depth (percentage reduction in ventilation from local average), and 3) patient-specific OSA severity in terms of average event depth and ventilatory burden (average event depth x average event duration x event rate). To address overestimation of RIP ventilation during obstruction, we developed and applied a calibration and linearization method (“RIP correction”). Concordance analysis evaluated median bias for both small (130%eupnea), along with bias and intraclass correlation coefficient (ICC) calculation for events and patient-specific measures. Results: For individual breaths (N = 495,631), RIP correction reduced overestimation bias for small breaths from 12 to 2%eupnea. For individual events (N = 34,497), RIP correction reduced mean bias for event depth estimates from 9 to 1%eupnea. For patient-specific analysis underestimation of average event depth was attenuated from 9 to 4%eupnea and for ventilatory burden, from 275 to 116%eupnea min/hr. Additionally, RIP correction improved ICC for event depth and patient-level traits. Conclusion: RIP signals, with appropriate processing, enable quantification of advanced ventilation-based OSA metrics, addressing concerns that airflow-based measures may be affected by breathing route.

  PublisherDOI

• Mendelian randomization in SLEEP: avoiding pitfalls with MR-SLEEP guidelines

  SLEEP · 2025-03-11 · 5 citations

  articleOpen access
  PublisherOA PDFDOI

Recent grants

• NIH Grant R01HL060756

  NIH · $2.7M · 2005

• Approaches to Genetic Heterogeneity of Obstructive Sleep Apnea

  NIH · $3.2M · 2016–2022

• NIH Grant P01AG017628

  NIH · $30.5M · 2019

• NIH Grant R01HL111725

  NIH · $1.7M · 2017

• NIH Grant T32AG000256

  NIH · $1.7M · 2005

Frequent coauthors

• Brendan T Keenan

  University of Pennsylvania

  547 shared

• Diane C Lim

  University of Miami

  418 shared

• Rajath Elias Soans

  373 shared

• James Shackleford

  Drexel University

  373 shared

• Greg Maislin

  265 shared

• Samuel T. Kuna

  Philadelphia VA Medical Center

  211 shared

• Þórarinn Gíslason

  National University Hospital of Iceland

  176 shared

• Indira Gurubhagavatula

  University of Pennsylvania

  134 shared

Education

• Other

  University of Glasgow, Scotland

  1967

• Ph.D., Mathematical models of lung function

  University of Glasgow, Scotland

  1976