Common Diseases in Clinical Cohorts — Not Always What They Seem, 2025, Fedik Rahimov, Ph.D et al

[Mij]

Abstract

Background
Misdiagnosis or underdiagnosis of rare diseases in patients with diagnoses of common diseases can lead to delayed or inappropriate treatments, thereby complicating the management of both rare and common conditions. Despite advances in molecular diagnostic techniques, the effect of rare diseases on the diagnosis of common diseases in research and clinical trials has not been comprehensively investigated.

Methods
We used exome- and genome-sequencing data from participants in the U.K. Biobank, a research study, and five clinical trials involving patients who had received a primary diagnosis of multiple sclerosis, inflammatory bowel disease, or atopic dermatitis to assess the incidence of monogenic rare diseases that often manifest with clinical symptoms overlapping with those of these common diseases.

Results
We identified 153 U.K. Biobank participants who carried a rare variant that contributes to a molecular diagnosis of a monogenic disorder — 53 of 1850 (2.86%) with a diagnosis of multiple sclerosis, 75 of 6681 (1.12%) with a diagnosis of inflammatory bowel disease, and 25 of 998 (2.50%) with a diagnosis of atopic dermatitis. We replicated the findings regarding such rare disease–causing variants in two independent cohorts — one including patients with a diagnosis of multiple sclerosis, and the other patients with a diagnosis of inflammatory bowel disease — who had undergone genome sequencing for research and for clinical trials, respectively. By combining genome and transcriptome analyses, we showed that molecular diagnosis can potentially elucidate mechanisms of inadequate response to therapeutic intervention.

Conclusions
Our study shows the value of systematic genome sequencing in understanding the phenotypic heterogeneity of common diseases and identifying failure to diagnose rare diseases and highlights the benefits of deep molecular phenotyping in clinical trials and patient care. (Funded by AbbVie and NIHR Cambridge Biomedical Research Centre.)
LINK

 

[Jaybee00]

"This is genuinely paradigm-shifting

1-3% of patients diagnosed with (relatively) common diseases – atopic dermatitis, inflammatory bowel disease, and multiple sclerosis – actually have much rarer monogenic conditions, revealed by genome sequencing

Given the obvious importance of a correct diagnosis, and the diminishing cost and time of genome sequencing, maybe it's time to move it to the frontline"


Do we think this might be an issue in ME/CFS?

 

[Hutan]

Ascertaining the root cause of a disease not only improves the ability to make a correct diagnosis but also is fundamental to the development of therapeutic or preventive measures. Patient recruitment for clinical trials typically depends on the clinical signs and symptoms of the disease under study, along with specific inclusion and exclusion criteria. Because of misdiagnosis or underdiagnosis, a small fraction of these patients might carry a rare variant capable of causing a monogenic disease that mimics the common disease under study.

In addition, patients affected by both common and rare diseases might have more severe or different coexisting conditions. Consequently, a therapeutic agent targeting the underlying pathogenesis of a disease may not elicit a response in patients with a distinct disease mechanism, thus potentially affecting the study outcome.

A UK Biobank cohort with data from electronic health records was a discovery cohort, evaluating people with MS, inflammatory bowel disease and atopic dermatitis. The folow-up cohorts consisted of people who had been diagnosed by health professionals as having multiple sclerosis or inflammatory bowel disease.

Therefore, we regarded the U.K. Biobank cohort as a discovery cohort and went on to test for the presence of monogenic disease–causing variants in follow-up cohorts comprising participants recruited for research (the Genuity Science cohort of patients with multiple sclerosis) and for the clinical trials in inflammatory bowel disease. These participants underwent more rigorous phenotyping than the U.K. Biobank participants and were thus evaluated as follow-up cohorts.

The Genuity Science cohort of persons with multiple sclerosis had undergone genome sequencing by Genuity Science. A total of 2025 participants in Ireland and 5450 in the United Kingdom were recruited prospectively from neurology clinics or retrospectively from existing biobanks. All were adults (≥18 years of age at the time of diagnosis), and their diagnosis was made by trained neurologists on the basis of the revised McDonald criteria of 2013.

An additional follow-up cohort comprised participants from five clinical trials of treatment for inflammatory bowel disease (Crohn’s disease or ulcerative colitis). ... At enrollment, all the participants had a diagnosis of moderate-to-severe, active Crohn’s disease or ulcerative colitis, made by trained gastroenterologists.

We evaluated 9529 participants with sequencing data from the U.K. Biobank (1850 with a diagnosis of multiple sclerosis, 6681 with a diagnosis of inflammatory bowel disease, and 998 with a diagnosis of atopic dermatitis), 7475 participants from Ireland and the United Kingdom with a diagnosis of multiple sclerosis in the Genuity Science cohort, and 1480 participants from five clinical trials in inflammatory bowel disease (580 with Crohn’s disease and 900 with ulcerative colitis) — a total of 18,484 participants

The study was funded by AbbVie, a biopharmaceutical company:

Our study was funded by AbbVie, which participated in the design and conduct of the study, the interpretation of data, and the review and approval of the manuscript for submission.

 

[Hutan]

Assembly and analysis of gene panels

Monogenic disease genes were compiled with the use of broad definitions that capture the most commonly recognized clinical symptoms of multiple sclerosis, inflammatory bowel disease, and atopic dermatitis. The PhoRank gene-ranking algorithm, implemented in VarSeq software, version 2.2.5 (Golden Helix), was used to build three gene panels, one for each disease. This algorithm ranks genes on the basis of their relevance to user-specified phenotypes as defined according to the Human Phenotype Ontology (HPO, a standardized vocabulary of phenotypic abnormalities encountered in human diseases and Gene Ontology biomedical terminology. For each of the common diseases, we used two HPO terms — HP:0002715 (abnormality of the immune system) and HP:0000707 (abnormality of the nervous system) for multiple sclerosis, HP:0002715 and HP:0004386 (gastrointestinal inflammation) for inflammatory bowel disease, and HP:0002715 and HP:0000951 (abnormality of the skin) for atopic dermatitis.

This is interesting, I think. For example for multiple sclerosis, they found all the monogenic disease genes relevant to 'Abnormality of the immune system' and 'abnormality of the nervous system'. They then looked for those disease genes in the study cohorts.

The remaining 1673 genes (for multiple sclerosis), 1044 genes (for inflammatory bowel disease), and 1500 genes (for atopic dermatitis) were used for pathogenic variant screening. For each of the three gene panels, we queried the exome-sequencing data in the discovery cohort (U.K. Biobank) and the follow-up cohorts for rare variants with allele frequencies of less than 1% in the entire dataset

Allele frequencies were calculated with the SeqArray R package, version 1.35.7. Next, we attempted to assign pathogenicity to these variants and determine their associations with known monogenic diseases in accordance with variant-classification guidelines recommended by the American College of Medical Genetics and Genomics (ACMG). Variant pathogenicity was assessed with VarSeq software, version 2.2.5.

It might be relatively easy to replicate the method with genetic data from people with ME/CFS, perhaps using the same screen as that used for multiple sclerosis. I wonder if these researchers would be interested in doing that? Failing that, perhaps another research group might like to try.

I haven't read far enough into this paper to know how reliable this approach is. We have seen that companies are already offering genetics screening panels, probably built using a similar approach. I imagine a lot of us would be interested in a screening panel that could identify monogenic diseases that might look like ME/CFS, for peace of mind.

 

[Hutan]

UK Biobank results
e.g. Multiple Sclerosis

Among 1850 U.K. Biobank participants with a diagnosis of multiple sclerosis, we identified 52 rare pathogenic or likely pathogenic variants in 34 genes in 53 participants (2.86%) (Figure 1); these genes have been linked to a number of mendelian disorders (Table S6). Among the identified variants, 47 have been reported to be pathogenic in the literature. The remaining variants were novel, and we classified them as likely pathogenic on the basis of the ACMG criteria.

One participant was heterozygous for the pathogenic BICD2 p.Thr703Met variant, which is associated with congenital, autosomal-dominant spinal muscular atrophy. Two other participants carried variants causative of autosomal-dominant familial amyotrophic lateral sclerosis: p.Gly38Arg in ANXA11 and p.Thr622Ala in MATR3. The highest number of pathogenic variants, 11 in 13 unrelated persons, were found in NOTCH3, which is linked to the autosomal-dominant cerebral arteriopathy and subcortical infarct leukoencephalopathy (CADASIL). There have been numerous instances of CADASIL being misdiagnosed as multiple sclerosis on the basis of radiological findings of confluent white-matter changes.

Our findings indicate that some U.K. Biobank participants with rare diseases might have received a misdiagnosis of multiple sclerosis or that certain rare diseases were underdiagnosed in participants with a diagnosis of multiple sclerosis. These misdiagnoses and underdiagnoses may have been due to overlapping symptoms or the phenotypic variability of multiple sclerosis, which became clear when we compared the complete electronic health records of persons with rare disease–causing variants with those without such variants, uncovering cases with severe coexisting conditions not typically expected in patients with multiple sclerosis. For instance, we observed a substantial enrichment of dementia (odds ratio, 14.79; P=0.002) and convulsion (odds ratio, 3.54; P=0.002) among participants with rare disease–causing variants (Table S7). These findings show that in addition to misdiagnosis, rare diseases can also exacerbate common disease symptoms when they co-occur.

So, the UK Biobank analysis suggested that nearly 3% of people diagnosed with MS were mis-diagnosed or had undiagnosed co-morbidities that could be affecting symptoms and treatment response.

e.g. Inflammatory bowel disease

Sporadic inflammatory bowel disease is a complex condition influenced by genetic and environmental risk factors and typically shows polygenic inheritance. However, there are also inherited forms of inflammatory bowel disease, with more than 100 genes linked to monogenic disease. Many of these monogenic conditions do not respond to standard therapies and are associated with high morbidity and mortality. Therefore, an accurate diagnosis for these patients is crucial for providing effective medical care and determining their eligibility for clinical trials.

Among 6681 U.K. Biobank participants with diagnosis codes for inflammatory bowel disease, we identified 54 rare pathogenic or likely pathogenic variants in 30 genes in 75 participants (1.12%)

e.g. atopic dermatitis

Furthermore, in 2 participants, we identified two ultrarare, pathogenic PTPN11 variants (p.Gly60Ala and p.Gln256Arg) that were previously associated with Noonan’s syndrome, a rare disorder that can cause dry skin.

To date, at least 12 genes have been implicated in Noonan’s syndrome. To better understand the extent of underdiagnosis of Noonan’s syndrome in the general population, we screened for these 12 genes among the U.K. Biobank participants and conducted a comprehensive analysis of coexisting conditions using electronic health records from the carriers of pathogenic or likely pathogenic variants (see the Supplementary Methods). We identified 164 persons who were probably affected by Noonan’s syndrome and clinically related disorders; however, only 5 of the 73 participants for whom we had general practitioner records had a formal diagnosis of Noonan’s syndrome (Table S10). Most of the participants received a diagnosis of either related symptoms or different diseases (Table S11).

 

[Hutan]

Results in the better screened clinical cohorts

In the Genuity Science multiple sclerosis cohort, 88 participants (1.18%) had a potential molecular diagnosis for a rare disease (Table S12). Although there were four times as many participants in the Genuity Science cohort (7475 participants) as in the U.K. Biobank cohort (1850 participants), the proportion carrying pathogenic variants was less than half that in the U.K. Biobank cohort (2.86%). This difference might be related to electronic health record–based diagnosis in the U.K. Biobank cohort, as compared with targeted phenotyping in the Genuity Science cohort, in which the participants with suspected rare diseases were mostly excluded.

Among 1480 participants with sequencing data in the inflammatory bowel disease clinical trials cohort, we identified 70 (4.73%) who had a molecular diagnosis of a rare disease (Table S14). The higher fraction of rare diseases identified in this cohort than in the U.K. Biobank cohort (1.12%) may be attributed to the targeted recruitment of patients with moderate-to-severe inflammatory bowel disease, in whom the condition may have developed owing to highly penetrant, rare variants. Similar to the findings in the U.K. Biobank, the findings among participants in the clinical trials showed that a majority (39 of 70) carried pathogenic variants in TNFRSF13B, which suggests that common variable immunodeficiency caused by pathogenic TNFRSF13B variants is probably misdiagnosed as inflammatory bowel disease or primarily manifests with inflammatory bowel disease–like symptoms. Consistent with the hallmarks of TNFRSF13B-associated immunodeficiency, 3 of 5 participants with ulcerative colitis and 5 of 7 participants with Crohn’s disease showed either extraintestinal manifestations or recurrent infections.

Because the people in these cohorts were participating in clinical trials, the authors were able to see if the people with identified rare disease variants benefitted from trialled treatments at different rates to participants who didn't. In some cases, their rate of response was substantially worse.

This is a nice example of where there may have been an initial placebo response for reported symptoms in a person with a likely misdiagnosis, but there was no long term response and no benefit when the bowel was visually examined.

In support of the hypothesis that the distinct underlying molecular cause of the disease may have led to a lack of adequate remission in these cases, several signs of possible misdiagnoses were identified on retrospective medical history review of these participants. For instance, a female participant 60 years of age who had received a diagnosis of Crohn’s disease and participated in the trial of adalimumab was heterozygous for a likely pathogenic variant (p.Arg148His) in the smooth-muscle actin gene ACTG2, which is associated with the autosomal-dominant familial visceral myopathy.<a href="https://www.nejm.org/doi/full/10.1056/NEJMoa2405459?logout=true#core-collateral-r36" data-xml-rid="r36">36</a> Abdominal pain, bowel obstruction, or gastroparesis, common in this condition, may have led to a misdiagnosis of Crohn’s disease in this participant. Although the participant initially appeared to show clinical remission at week 12, she did not have endoscopic remission and did not show clinical or endoscopic remission by week 56.

 

[Hutan]

Although some of these rare diseases are incidental co-occurrences with the common diseases that we evaluated here, on the basis of phenotypic similarities and incidence, we conclude that many are probably misdiagnosed as a common disease. Furthermore, misdiagnosis of monogenic diseases can have a profound effect on family members too. Identifying these rare diseases allows for genetic counseling, which is crucial for guiding family members through cascade genetic testing and supporting their health decisions.

They make the point that genetic screening should be part of clinical trial evaluations. It can shed light on why some people respond and others don't. I guess better knowledge of genetics can potentially make small trials more useful than they otherwise might be.

 

[Hutan]

I'd love to see some genetic screening like this applied to cohorts of people with 'functional diseases'.

 

[Amw66]

Phenotyping with AI is also interesting .
My aunt and daughter are very similar re many traits .

 

[Sean]

I'd love to see some genetic screening like this applied to cohorts of people with 'functional diseases'.

Cue the 'we don't think it is a good idea as it might encourage patients with functional disorders to think they have a biological disorder' excuse not to check if the functional diagnosis is correct.

Narrator: It wasn't.