A seventh gene joins the SCDO list.

PUBLISHED 2026-09-11 · EVIDENCE UPDATE

One case is a curiosity. Two is a pattern.

In 2018, a Dutch group reported a newborn boy, of North African descent, with ribs that were missing, fused, split, or undergrown on both sides, malformed vertebrae, and lungs too small to sustain him. He lived nine days. Every known spondylocostal dysostosis gene came back normal. Exome sequencing found a single explanation: he carried two broken copies of a gene called DMRT2, one from each unaffected parent. The paper ended, honestly, with a question mark. One child, one variant, no second case anywhere in the literature. A gene with a mouse phenotype that matched, but in humans, still a hypothesis.

Eight years passed. Then, in the February 2026 issue of the American Journal of Medical Genetics, a team from Tel Aviv published a second infant. Firstborn to related parents in Israel. Nearly 37 weeks at delivery, in respiratory distress from the first minutes, intubated in the NICU. The same family of rib malformations. A homozygous frameshift in DMRT2, this time a single deleted letter, c.129delC, predicted to truncate the protein at amino acid 75. It appears in neither gnomAD nor a local database of more than 28,500 exomes. He died at three months, of respiratory failure after severe infections.

Why the second case settles the gene, and unsettles the phenotype.

Two variants, one start-loss and one frameshift, both predicted to abolish the protein, in families with no shared ancestry, both producing the severe costovertebral pattern, is exactly the convergence human genetics relies on. The 2018 question mark can come off. DMRT2 joins DLL3, MESP2, LFNG, HES7, TBX6, and RIPPLY2 on the list, and the authors' practical recommendation is concrete: skeletal-dysplasia gene panels should now include it.

But the second baby brought something the first did not. On imaging and at surgery, he had no thymus. His lymphocyte counts were profoundly low. He suffered severe CMV pneumonitis and overwhelming bacterial infections in his three months. The team went back through his exome looking for another cause, some immunodeficiency gene they had missed, and found nothing that explained it. Which raises, cautiously, a possibility no one has associated with SCDO before: that DMRT2, already known from mouse work to be expressed in the dermomyotome of each forming somite, may also matter for immune development, perhaps through the shared embryonic neighborhood of somite-derived connective tissue and the thymus. The authors flag it as a phenotypic expansion, not a conclusion. One patient cannot prove a mechanism. Two genes' worth of caution belong in any sentence about it.

What it changes for a family reading this site.

Small, specific things. If your child has an SCDO-pattern skeleton on imaging and the panel came back negative, ask whether it was built before 2026 and whether it includes DMRT2; older panels will not. If your child has the SCDO pattern and also unusual, repeated, or severe infections, this paper is the reason to say both facts in one sentence to your immunologist, and to mention the word thymus. The terminology fork (STD versus SCD), the prenatal signs, and the care path stand as written: a new gene refines the diagnostic workup, it does not move the clinical fundamentals. Our genetics page will be updated to count seven genes where it lists six.

What it does not change.

Reproductive counseling still rests on the same arithmetic: both published families fit autosomal recessive inheritance, carrier parents, 25 percent recurrence risk in each pregnancy, both children were boys but there is no evidence yet of a sex effect. Neither case implies anything about milder DMRT2 phenotypes surviving to adulthood; neither has been found. If someone markets a "DMRT2 test" to worried parents on the strength of two case reports, that is commerce, not medicine.