A study published in the Archives of Dermatological Research demonstrates how a large national population cohort can shed light on rare disease risks that have previously been understudied. Using the National Institutes of Health (NIH) All of Us Research Program database, investigators from Weill Cornell Medicine found that individuals with congenital ichthyosis have higher odds of several skin cancers compared to matched controls, a finding the authors suggest could inform clinical monitoring.
The research was led by Dr. Shari R. Lipner, Professor of Clinical Dermatology at Weill Cornell Medicine, a NORD Rare Disease Center of Excellence, together with co-authors Kaya L. Curtis and Steven Zeldin. The study demonstrates that meaningful rare disease research can be conducted using population-scale data resources.
A rare condition, an unanswered question
Congenital ichthyoses (CI) are a rare and heterogeneous group of keratinization disorders characterized by defective keratinocyte differentiation, resulting in chronic scaling, skin barrier disruption, and inflammation. Though case reports had suggested a possible link between CI and increased risk of skin cancer, no large-scale study had ever quantified that risk relative to the general population.
Several biological mechanisms have been proposed to explain why CI patients might be at elevated cancer risk, including chronic inflammation and impaired skin barrier function. Yet without a large comparison group, these hypotheses remained difficult to evaluate empirically.
Dr. Lipner described what drew her team to this question: “Congenital ichthyoses are rare, genetically driven skin barrier disorders. Because of their rarity, important questions, including ‘what are the long-term cancer risks?’ are challenging to answer with traditional, single-center datasets. While case reports have long suggested a possible link between ichthyosis and cutaneous malignancies, this risk has not been quantified at a population level.
The All of Us dataset offered a way to move beyond anecdote. Its scale and national scope made it possible to identify a sufficiently large cohort of patients with congenital ichthyosis and compare them to well-matched controls. Our motivation was to take a clinical observation and test it rigorously in a large dataset.”
A population-scale approach to a rare disease question
To study this association, the team conducted a nested case-control study using the All of Us database. They identified 198 participants with congenital ichthyoses and matched them 1:12 to 2,376 controls. Using multivariate logistic regression, they assessed the odds ratios for congenital ichthyoses and skin cancer associations.
The scale and diversity of All of Us made this study possible. On using the platform, Dr. Lipner noted: “The process of accessing and working with the All of Us database was relatively straightforward. Coding allowed us to match congenital ichthyosis patients with controls based on demographics and controlling for confounding factors. The cloud-based environment with Jupyter notebooks allowed for computationally intensive tasks, such as multivariate regression and quantile regressions with 10,000 bootstrap replications, to run without being bottlenecked by local hardware.”
Elevated risk across multiple cancer types
After controlling for potential confounders, congenital ichthyosis was significantly associated with actinic keratosis (AK) (OR = 3.65), a pre-malignant lesion, meaning CI patients were more likely to have AK than matched controls. CI was also associated with melanoma (OR = 2.39) and basal cell carcinoma (BCC) (OR = 1.90). Squamous cell carcinoma (SCC) did not reach statistical significance. The absence of a statistically significant SCC association stands in contrast to earlier case literature, which had identified SCC as the most commonly reported skin cancer in CI. The authors suggest this may reflect underdiagnosis in clinical practice: SCCs can be challenging to recognize against the backdrop of the scaling, redness, and thickening that characterize CI skin, as well as differences in cohort composition and data source limitations.
Age at malignancy diagnosis was also examined. Median ages at diagnosis for CI participants were 66.4 years for AK, 74.6 years for BCC, 74.2 years for melanoma, and 71.6 years for SCC, broadly similar to controls, though generally older than ages reported in prior case literature, which the authors attribute to reporting bias toward younger, more striking cases in smaller published series.
From rare disease findings to broader research and clinical impact
Based on their results, the authors recommend routine skin cancer surveillance in CI patients, pending additional studies with larger cohorts and narrower confidence intervals. Dr. Lipner reflected on where this work is headed: “This work propelled us to perform additional research projects. We extended the analysis into a separate large dataset (TriNetX), where we corroborated the data from our All of Us project. We also performed an additional case-control study utilizing the All of Us database, studying the association of congenital ichthyosis with cutaneous infections. We found that congenital ichthyosis patients had higher odds of onychomycosis, cutaneous fungal infections, verruca, and tinea pedis.”
Studies like this one highlight how population-scale data can begin to fill gaps in rare disease research. Many rare diseases lack dedicated registries or large prospective research cohorts, meaning that the majority of clinical recommendations are built on expert opinion and small case series. Population databases such as All of Us cannot replace disease-specific studies or clinical trials, but they offer a matched general population comparator, at scale, across a diverse and nationally representative cohort. That makes it possible to move some clinical observations toward evidence-based, actionable recommendations that can inform screening and treatment guidance.
Limitations of the current study include the relatively small number of CI participants, the inability to stratify by ichthyosis subtype, and the exclusion of pediatric patients. The All of Us database’s policy of masking cohort sizes when counts fall below 20 also imposed some reporting constraints on rare outcomes. The study’s strengths include its large, matched control group, diverse participant population, and rigorous case-control design.
Accessing All of Us Data for Rare Disease Research
All of Us is poised to build one of the world’s most robust health databases, with future releases expected to include data from more than 747,000 participants available in the Curated Data Repository Version 9 (CDRv9), including whole genome sequences from over 535,000 participants and continued expansion of pediatric enrollment. This growing resource will enable scientists to better understand health across diverse populations and generations, advancing opportunities for discovery in rare disease research and beyond.
All of Us data is already enabling new research on rare conditions. Scientists are using the program’s genomic and health history data to uncover rare gene‑disease links more reliably than before. Researchers are studying neurofibromatosis, cystic fibrosis, autoimmune diseases, Ehlers-Danlos syndrome, sickle cell disease, pulmonary arterial hypertension, and many other conditions.
Researchers interested in conducting similar analyses can apply for access to the All of Us Researcher Workbench at workbench.researchallofus.org. The Workbench provides cloud-based tools for analyzing genomic data, electronic health records, physical measurements, and survey responses from a cohort that continues to grow.
This blog post is funded by the Division of Engagement and Outreach, All of Us Research Program, National Institutes of Health. Pyxis Partners Award Number: 1OT2OD038104-01
