Elizabeth Phillips

Persistent human-adapted microbes can act as immunological "keystones," organizing host defense across tissues and shaping vulnerability under immune perturbation. More generally, tissue immunity is calibrated by persistent niche-resident organisms that tune compartment-specific thresholds of cytotoxicity and peripheral tolerance; keystone organisms represent the apex subset with multi-niche scope. Here we operationalize keystone organisms as pathogens whose containment requires coordinated engagement of multiple immune arms and whose residence is structured across anatomical niches. Using 18 curated immunological and evolutionary traits across 43 organisms, unsupervised analyses resolved four reproducible archetypes and identified a compact keystone set dominated by persistent herpesviruses and Mycobacterium tuberculosis. We then translated the clinical literature into a pathogen×immune-perturbation×niche tensor capturing where and when each organism emerges under defined immune deficits. We quantified "diagnostic breadth" with two complementary summaries: immune breadth (diversity of perturbations associated with emergence) and niche breadth (diversity of anatomical sites). Clinical emergence patterns perfectly separated trait-defined keystones from all other organisms and highlighted expanded niche breadth as the primary discriminator, whereas immune breadth showed no significant group separation. Finally, a mechanistic model integrating barrier disruption, latent reservoir activation, and tissue-resident immune control predicted clinical emergence from first principles-without fitting parameters to individual pathogens-and ranked true emergences 2.9-fold above chance among its highest-confidence predictions. Together, these results link evolutionary adaptation to clinically readable patterns of reactivation, motivate archetype-aware surveillance under immunosuppression, and provide a framework for immunogen design that prioritizes conserved, functionally constrained targets. Because the clinical tensor is literature-curated and sparse, "perfect separation" refers to keystone-vs-other discrimination within this dataset and is not a claim of universal out-of-sample performance.

Importance; Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis (SJS/TEN) are rare, potentially fatal adverse drug reactions. As the use of immune checkpoint inhibitors (ICIs) expands, their role as direct inducers or synergistic contributors to SJS/TEN remains incompletely characterized.

Objective: To determine whether ICIs are independent risk factors for SJS/TEN, evaluate their interactions with known culprit drugs, and assess their impact on latency and mortality.

Design: Cross-sectional analysis of adverse event reports submitted to the U.S. Food and Drug Administration Adverse Event Reporting System (FDA FAERS) between January 2013 and December 2023, sanitized and de-duplicated. Logistic regression and Cox models were used to assess predictors of SJS/TEN development, mortality, and latency.

Setting: Global pharmacovigilance reports submitted to FAERS.

Participants: A total of 17,495 unique and de-identified patients reported SJS/TEN, of 13,986,839 total reports.

Exposures: Suspected causative drugs, including ICIs.

Main Outcomes and Measures: Primary outcomes were the adjusted odds of developing SJS/TEN, time-to-event (TTE) of reaction/drug latency, and all-cause mortality. Depending on the analysis, covariates included age, sex, number of concomitant drugs, cancer diagnosis, and specific drug exposures.

Results: Of 17,495 SJS/TEN cases (median age 53 years, 37.6% male), 970 (5.5%) had ICI exposure and 653 (3.7%) listed an ICI as the primary suspect. ICI exposure was associated with developing SJS/TEN (adjusted OR, 6.69; 95% CI, 6.19-7.23) while controlling for age, exposure to strong and weak culprits, number of concomitant drugs, and cancer diagnosis. ICI increases SJS-TEN risk among patients exposed to allopurinol (OR, 4.35; 95% CI, 3.12-6.06) and TMP-SMX (OR, 5.68; 95% CI, 4.05-7.95) with the same covariates. Among patients with small-molecule-induced SJS/TEN, mortality was strongly associated with ICI exposure (particularly exposure to multiple ICI, OR, 7.31; 95% CI, 3.09-17.27). Among all SJS/TEN cases, ICI exposure was associated with delayed onset, compared to cancer patients not exposed to ICI and non-cancer patients (median 20 vs 14 vs 13 days; P < .0001).

Conclusions and Relevance: ICIs are associated with increased SJS/TEN risk, both independently and in combination with known culprit drugs, and may delay disease onset. These findings support increased vigilance in prescribing known culprits alongside ICI.

Co-trimoxazole is a leading global cause of severe cutaneous adverse drug reactions (SCAR) including Stevens-Johnson syndrome/toxic epidermal necrolysis (SJS/TEN) and drug reaction with eosinophilia and systemic symptoms (DRESS). Co-trimoxazole-induced SCAR are associated with HLA class I alleles including HLA-B*13:01 and HLA-B*38:02 in Southeast Asian (SEA) populations. However, the global generalizability of these associations is unknown but critical for population-appropriate risk stratification and diagnosis.
To determine HLA risk factors associated with co-trimoxazole-induced SJS/TEN and DRESS in populations from the United States (US) and South Africa (SA).
We performed high-resolution HLA typing on dermatologist-adjudicated co-trimoxazole-induced SCAR patients in the US (n=63) and SA (n=26) compared to population controls. Peptide binding and docking analyses were performed using MHCcluster2.0 and CB-Dock2.
In a multiple logistic regression model, HLA-B*44:03 (Pc<0.001, OR: 4.08), HLA-B*38:01 (Pc<0.001, OR: 5.66), and HLA-C*04:01 (Pc=0.003, OR: 2.50) were independently associated with co-trimoxazole-induced SJS/TEN in the US. HLA-B*44:03 was also associated with co-trimoxazole-induced DRESS in SA (Pc=0.019, OR: 10.69). Distinct HLA-B variants with shared peptide binding specificities (SPBS) and HLA-C*04:01 identified 94% and 78% of co-trimoxazole-induced SJS/TEN and DRESS in the US, respectively. The SEA risk allele HLA-B*13:01, with SPBS to HLA-B*44:03, was identified in just 1/63 US SCAR patients.
HLA alleles with SPBS to SEA-related risk alleles including HLA-B*44:03 (SPBS with HLA-B*13:01) and HLA-B*38:01 (SPBS with HLA-B*38:02) but also HLA-C*04:01 predisposed to co-trimoxazole-induced SCAR in the US and SA. These findings provide biological plausibility and strategies for global risk prediction and diagnosis of co-trimoxazole-induced SCAR.
HLA alleles including HLA-B*13:01 and HLA-B*38:02 are risk factors for co-trimoxazole-induced SCAR in Asian populations. However, the generalizability of these associations to other global populations is unknown but critical for population-appropriate risk stratification and diagnosis.
HLA alleles with shared peptide binding specificities (SPBS) to Asian-related risk alleles including HLA-B*44:03 (SPBS with HLA-B*13:01) and HLA-B*38:01 (SPBS with HLA-B*38:02) but also HLA-C*04:01 predisposed to co-trimoxazole-induced SCAR in the US and South Africa.
HLA alleles previously associated with co-trimoxazole-induced SCAR do not identify risk across populations. However, HLA alleles with SPBS provide biological plausibility and strategies for global and population-appropriate clinical risk stratification and diagnosis of cotrimoxazole-induced SCAR.

Importance
Severe cutaneous adverse reactions (SCARs), including Stevens-Johnson syndrome/toxic epidermal necrolysis (SJS-TEN), drug reaction with eosinophilia and systemic symptoms (DRESS), acute generalized exanthematous pustulosis (AGEP), and generalized bullous fixed drug eruption (GBFDE), are rare but life-threatening drug hypersensitivity syndromes. Due to their low incidence and diagnostic complexity, large-scale characterization of SCAR is challenging.

Objective
To characterize the demographics, causative agents, trends, latency, and phenotypic overlap of SCAR using a large-scale, sanitized pharmacovigilance dataset from FAERS (FDA Adverse Event Reporting System).

Design
Cross-sectional study of spontaneous adverse event reports. Cases were drawn from the U.S. Food and Drug Administration Adverse Event Reporting System (FDA FAERS) from January 2004 to December 2023 and subjected to sanitization and deduplication. Disproportionality analysis was used to characterize causative agents. Machine learning (random forest classifiers) was used to analyze predictors of drug latency and mortality.

Setting
Global pharmacovigilance reports submitted to FAERS.

Participants
A total of 56,683 deduplicated SCAR reports were identified, representing 0.33% of reports during the study period.

Exposures
Suspected causative drugs, including both small molecules and biologics.

Main outcomes and measures
Main outcomes included the frequency and distribution of SCAR syndromes, reporting trends over time, latency from drug start to reaction onset, drug-specific disproportionality (PRR, ROR, IC), and co-reporting between SCAR types and related conditions.

Results
A total of 56,683 unique SCAR reports were identified, including SJS-TEN (28,871), DRESS (22,444), AGEP (6,183), and GBFDE (150). We identified 237 drugs with significant disproportionality for SCAR overall. Co-reporting between SCARs was significantly enriched (p < 1e-200), suggesting overlapping phenotypes. Latency varied by drug and syndrome (median: GBFDE 3 days, AGEP 4 days, SJS-TEN 12 days, DRESS 20 days).

Conclusions and relevance
SCAR syndromes display distinct but overlapping phenotypes, with variable latency and diverse causative agents. These findings, based on the largest SCAR dataset to date, highlight the need for improved classification frameworks and molecular validation. Large-scale pharmacovigilance, integrated with genomic and histopathologic data, will be critical to improving diagnosis, mechanistic understanding, and clinical management of SCAR.

Background
Large language models (LLMs) have emerged as transformative technologies, revolutionizing natural language understanding and generation across various domains, including medicine. In this study, we investigated the capabilities, limitations, and generalizability of Generative Pre-trained Transformer (GPT) models in analyzing unstructured patient notes from large healthcare datasets to identify immune-related adverse events (irAEs) associated with the use of immune checkpoint inhibitor (ICI) therapy.

Methods
We evaluated the performance of GPT-3.5, GPT-4, and GPT-4o models on manually annotated datasets of patients receiving ICI therapy, sampled from two electronic health record (EHR) systems and seven clinical trials. A zero-shot prompt was designed to exhaustively identify irAEs at the patient level (main analysis) and the note level (secondary analysis). The LLM-based system followed a multi-label classification approach to identify any combination of irAEs associated with individual patients or clinical notes. System evaluation was conducted for each available irAE as well as for broader categories of irAEs classified at the organ level.

Results
Our analysis included 442 patients across three institutions. The most common irAEs manually identified in the patient datasets included pneumonitis (N=64), colitis (N=56), rash (N=32), and hepatitis (N=28). Overall, GPT models achieved high sensitivity and specificity but only moderate positive predictive values, reflecting a potential bias towards overpredicting irAE outcomes. GPT-4o achieved the highest F1 and micro-averaged F1 scores for both patient-level and note-level evaluations. Highest performance was observed in the hematological (F1 range=1.0-1.0), gastrointestinal (F1 range=0.81-0.85), and musculoskeletal and rheumatologic (F1 range=0.67-1.0) irAE categories. Error analysis uncovered substantial limitations of GPT models in handling textual causation, where adverse events should not only be accurately identified in clinical text but also causally linked to immune checkpoint inhibitors.

Conclusion: The GPT models demonstrated generalizable abilities in identifying irAEs across EHRs and clinical trial reports. Using GPT models to automate adverse event detection in large healthcare datasets will reduce the burden on physicians and healthcare professionals by eliminating the need for manual review. This will strengthen safety monitoring and lead to improved patient care.

Importance:
Seasonal variation in adverse drug reactions has clinical and mechanistic implications for understanding disease mechanisms and risk mitigation strategies. Stevens-Johnson Syndrome/Toxic Epidermal Necrolysis (SJS/TEN) is a life-threatening mucocutaneous reaction with high morbidity and mortality, which may have a seasonal component.

Objective:
To determine whether the reporting of SJS/TEN to the U.S. Food and Drug Administration Adverse Event Reporting System (FAERS) follows a seasonal pattern, incorporating both traditional seasonal analyses and time-series modeling.

Design:
Cross-sectional, population-based analysis of FAERS reports from January 2010 to December 2019. Seasonal differences were assessed using Kruskal-Wallis tests and Seasonal-Trend Decomposition using Loess (STL). Seasonal autoregressive integrated moving average (SARIMA) models were used to counterfactually forecast SJS/TEN and comparator conditions during the COVID-19 pandemic, assessing changes in reporting.

Setting:
Population-based analysis of spontaneous adverse event reports submitted to FAERS.

Participants:
All deduplicated FAERS reports with complete event dates from 2010 to 2019 were included. SJS/TEN cases were identified using standardized MedDRA terms. Comparator analyses of known seasonal conditions – photosensitivity reactions, influenza, and respiratory syncytial virus (RSV) – served as positive controls.

Exposures:
Drug exposures as recorded in FAERS.

Main Outcomes and Measures:
The primary outcome was the monthly and seasonal proportion of unique SJS/TEN reports, normalized using all FAERS reports during a particular interval as the denominator. Seasonality strength was quantified from STL decomposition (range 0–1). SARIMA models were applied to pre-COVID data to counterfactually forecast trends from March 2020 to December 2023. Forecast accuracy was evaluated using mean squared error (MSE), root mean squared error (RMSE), and residual diagnostics.

Results:
Among 5,900 SJS/TEN cases reported from 2010–2019, no significant monthly or seasonal variation was detected (p > 0.05), and seasonality strength was low (0.163). Positive controls (influenza, RSV, photosensitivity) showed expected strong seasonality. SARIMA forecasts indicated a mild increase in SJS/TEN reporting during the pandemic, compared to its previous declining trend. Influenza and RSV dropped below predictions during the pandemic, while photosensitivity remained relatively consistent.

Conclusions and Relevance:
SJS/TEN reporting to FAERS does not exhibit apparent seasonality, in contrast to positive controls. Time-series modeling confirmed these findings and highlighted the relative stability of SJS/TEN reporting during the pandemic compared to respiratory viruses.

Background: Electronic health records (EHR) are increasingly used for studying multimorbidities. However, concerns about accuracy, completeness, and EHRs being primarily designed for billing and administration raise questions about the consistency and reproducibility of EHR-based multimorbidity research.

Methods: Utilizing phecodes to represent the disease phenome, we analyzed pairwise comorbidity strengths using a dual logistic regression approach and constructed multimorbidity as an undirected weighted graph. We assessed the consistency of the multimorbidity networks within and between two major EHR systems at local (nodes and edges), meso (neighboring patterns), and global (network statistics) scales. We present case studies to identify disease clusters and uncover clinically interpretable disease relationships. We provide an interactive web tool and a knowledge base combing data from multiple sources for online multimorbidity analysis.

Findings: Analyzing data from 500,000 patients across Vanderbilt University Medical Center and Mass General Brigham health systems, we observed a strong correlation in disease frequencies ( Kendall’s ι− = 0.643) and comorbidity strengths (Pearson π = 0.79). Consistent network statistics across EHRs suggest a similar structure of multimorbidity networks at various scales. Comorbidity strengths and similarities of multimorbidity connection patterns align with the disease genetic correlations. Graph-theoretic analyses revealed a consistent core-periphery structure, implying efficient network clustering through threshold graph construction. Using hydronephrosis as a case study, we demonstrated the network’s ability to uncover clinically relevant disease relationships and provide novel insights.

Interpretation: Our findings demonstrate the robustness of large-scale EHR data for studying complex disease interactions. The alignment of multimorbidity patterns with genetic data suggests the potential utility for uncovering shared etiology of diseases. The consistent core-periphery network structure offers a strategic approach to analyze disease clusters. This work also sets the stage for advanced disease modeling, with implications for precision medicine.

Funding: VUMC Biostatistics Development Award, UL1 TR002243, R21DK127075, R01HL140074, P50GM115305, R01CA227481

Stevens-Johnson syndrome and toxic epidermal necrolysis (SJS/TEN) is a rare but life-threatening cutaneous drug reaction mediated by human leukocyte antigen (HLA) class I-restricted CD8+ T-cells. To obtain an unbiased assessment of SJS/TEN cellular immunopathogenesis, we performed single-cell (sc) transcriptome, surface proteome, and TCR sequencing on unaffected skin, affected skin, and blister fluid from 17 SJS/TEN patients. From 119,784 total cells, we identified 16 scRNA-defined subsets, confirmed by subset-defining surface protein expression. Keratinocytes upregulated HLA and IFN-response genes in the affected skin. Cytotoxic CD8+ T-cell subpopulations of expanded and unexpanded TCRαβ clonotypes were shared in affected skin and blister fluid but absent or unexpanded in SJS/TEN unaffected skin. SJS/TEN blister fluid is a rich reservoir of oligoclonal CD8+ T-cells with an effector phenotype driving SJS/TEN pathogenesis. This multiomic database will act as the basis to define antigen-reactivity, HLA restriction, and signatures of drug-antigen-reactive T-cell clonotypes at a tissue level.