Machine learning links immune-cell measures to resected lung cancer

Research into genes has shown that osteogenesis imperfecta (OI), a condition that makes bones break easily, is much more varied than doctors once thought. It was first grouped into a set of conditions based on symptoms, but researchers have now identified at least 23 types linked to more than 20 genes. A detailed review explains how modern genetic tests have connected broken bones, changes in bone shape and other problems with connective tissues to faults in collagen production, processing, movement inside cells, bone mineral formation and signals that guide bone-building cells. These findings support a move away from classifying OI mainly by its visible symptoms. Instead, doctors can use the genetic cause to help guide diagnosis and care. However, a person's genetic subtype does not fully predict how the condition will affect them.

Changes in one copy of the COL1A1 or COL1A2 gene account for about 80–85% of OI cases. These genes are linked to the classic Sillence types I through IV. COL1A1 provides instructions for making the pro-alpha1(I) chain, while COL1A2 provides instructions for making the pro-alpha2(I) chain. These chains join to form a three-stranded structure called type I procollagen, which is then processed into collagen fibers. Disease-causing changes in these genes generally cause either a shortage of collagen or the production of faulty collagen that interferes with normal collagen. The first type is called haploinsufficiency, and the second is called a dominant-negative effect. Nonsense and frameshift variants often cause cells to break down the affected genetic message, reducing the amount of normal collagen they make. Missense variants, especially those that replace important glycine building blocks in the repeating Gly-X-Y pattern, can prevent the three-stranded collagen structure from folding correctly. These faulty molecules may undergo too much chemical modification because they remain in the cell for too long. They can also build up in a part of the cell called the endoplasmic reticulum. This can trigger the cell's unfolded-protein response, interfere with bone-building cells called osteoblasts and lead to abnormal collagen fibers. A shortage of collagen usually causes the milder type I form, while dominant-negative changes are more often linked to the more severe types II, III and IV.

The wider classification of OI helps explain why people with similar fractures may need different kinds of care. Type V OI is caused by a recurring genetic change called c.-14C>T in a non-protein-coding region of IFITM5. This change allows protein production to start at a different point, adding five amino acids to a protein called BRIL, which is mainly found in bone. A 2025 study using single-cell RNA sequencing in mouse models found cells called “chondrocyte-like osteoblasts.” These cells showed markers of both bone-building and cartilage-building cells. This finding supports the idea that abnormal BRIL may direct developing skeletal cells toward making cartilage in the wrong places instead of forming normal bone. Type V OI can cause unusually large calluses where bones heal, calcium deposits in the membrane between the forearm bones, and dislocation of the radial head near the elbow. However, unusually large healing calluses occur in only about 65% of cases. Even people with the same recurring gene change can have very different outcomes, from nearly normal height and walking ability to severe growth problems and lower bone density. This shows that other factors, including modifier genes, biological sex and other influences, also affect the condition.

Type VI OI has a different cause. It results from harmful changes in both copies of SERPINF1, the gene that provides instructions for making pigment epithelium-derived factor (PEDF). Most affected people have no detectable PEDF in their blood, making its absence a highly accurate and sensitive marker for this subtype. The condition severely disrupts bone mineralization, leaving large amounts of bone material that has not hardened properly. Under a polarized-light microscope, the bone's layered structure shows a distinctive “fish-scale” pattern. Fractures often begin after the first year of life and may be followed by severe bone deformity, short stature and loss of the ability to walk independently. A study that looked back at people with non-COL1 types of OI found that bisphosphonate treatment consistently reduced fractures in type V, but responses in type VI varied. Some people had more fractures even though their bone mineral density improved. This difference shows that having more mineral in the bone does not necessarily mean the bone's structure is stronger or healthier. It also shows why people who share a broad diagnosis may respond differently to the same medicine.

Other types of OI develop when different steps in collagen production and processing go wrong. Changes in both copies of CRTAP, P3H1 or PPIB can weaken a group of proteins that adds a chemical group called 3-hydroxyl to procollagen. These changes cause types VII, VIII and IX, which are often severe or life-threatening. Common features include fractures around birth, extremely weak bones, shortening of the limbs near the trunk and unusual popcorn-shaped areas at the ends of developing bones. Severity can range from fatal to moderate. Changes in SERPINH1 or FKBP10 interfere with two helper proteins, HSP47 and FKBP65, that help collagen fold, move through cells and form strong links. Type X can involve imperfect tooth development, long-term lung disease and kidney problems. Type XI often causes joint contractures present at birth and can resemble a condition called Bruck syndrome type I. Changes in BMP1 interfere with the cutting of a section from the end of procollagen and can cause severe bone fragility even when bone mineral density is normal or high. Changes in KDELR2 prevent the cell from efficiently retrieving HSP47 from the Golgi apparatus, a structure that processes and transports proteins. This disrupts collagen-fiber formation. Type XXI can range from fractures before birth and worsening bone deformity to delays in brain development. These are different genetic diseases, not simply different forms of the same collagen problem.

Problems with the development and signaling of osteoblasts create another group of OI types. SP7 provides instructions for making Osterix, a protein that controls gene activity and works downstream of RUNX2. Osterix is needed for osteoblasts to develop properly. Harmful changes in both copies of SP7 usually cause severe bone fragility, poor growth and repeated fractures. However, a new missense variant that arises spontaneously can produce an abnormal version of the protein with a new effect. This can cause a condition involving rapid bone turnover, fragile bones, poor mineralization, excessive bone growth in the skull and early fusion of skull bones. Type XV OI is linked to changes in both copies of WNT1, which disrupt the Wnt/beta-catenin signaling pathway. Symptoms range from moderate disease to worsening bone deformity, reduced mineralization of the skull, short stature, drooping eyelids and underdevelopment of parts of the brainstem and cerebellum. Changes in MESD interfere with the folding and transport of the Wnt co-receptor proteins LRP5 and LRP6. This causes type XX OI, which may include fractures before birth, very low bone density, underdeveloped lungs and early respiratory failure. Other newly identified types involve CREB3L1, which helps move collagen; MBTPS2, which supports protein processing inside cells; SPARC, which helps regulate the material around cells; TENT5A, which helps keep messenger RNA stable; CCDC134, which is involved in RAS/MAPK signaling; and PHLDB1, which is involved in AKT signaling.

Genetic patterns also differ among Asian populations, making it difficult to interpret diagnostic findings in the same way everywhere. In Chinese groups studied, COL1A1 and COL1A2 changes account for about 76–90% of genetic diagnoses, and type I is the most common clinical form. Among 74 Chinese families with recessive OI, WNT1 accounted for 40.54% of cases, SERPINF1 for 29.7% and FKBP10 for 13.5%. This pattern differs from those often reported in Western populations. In South Korea, changes in collagen genes account for about 87.4% of confirmed cases. People with glycine substitutions had a mean height Z-score of −5.94, compared with −1.66 in those without these changes. Their first fractures also occurred earlier, at an average of 11.12 months rather than 23.74 months. Among affected people older than 14, 62.5% had experienced more than 10 fractures over their lifetimes. In a group of 91 unrelated Vietnamese patients, COL1A1 and COL1A2 changes accounted for 59.4% of cases. A P3H1 splice-site variant was found in 85.7% of families with recessive OI and 10.3% of the full group. In an Indian group referred for testing, changes in one copy of a collagen gene explained only 47.6% of cases. Recessive OI accounted for as much as 48% of cases, and recurring WNT1 and FKBP10 changes suggested that some smaller populations may share genetic variants inherited from common ancestors, a pattern known as a founder effect.

As researchers identify more genetic causes, testing methods are becoming more detailed. Gene panels or whole-exome sequencing are recommended as first tests, but broad genetic sequencing does not automatically find every important change. Separate tests for copy-number changes and structural variants are needed to find deletions or duplications of parts of genes that standard testing methods may miss. Chromosome testing, known as karyotyping, can identify rearrangements in which chromosome sections change places without any overall gain or loss of genetic material. Long-read sequencing or whole-genome sequencing can help identify the exact breakpoints of complex changes. Sanger sequencing remains useful when checking a known family variant or the characteristic IFITM5 change in someone suspected of having type V OI. Experts interpret variants using guidance from ACMG/AMP and ClinGen. When a genetic result is uncertain, RNA studies, minigene reporter tests and collagen electrophoresis can provide additional evidence about how the change affects gene activity or collagen. This step-by-step approach combines DNA sequencing, analysis of chromosome and gene structure, and biochemical tests instead of relying on DNA sequence results alone.

Bisphosphonates remain the best-established medicines for treating children with OI, although their effects differ by drug and genetic subtype. Intravenous pamidronate and neridronate generally increase bone mineral density and reduce fractures. Evidence for bisphosphonates taken by mouth is less convincing. A large randomized trial of alendronate found that it increased mineral density in the lower spine but did not significantly reduce long-bone fractures compared with a placebo. Common side effects include temporary fever and muscle pain after the first intravenous treatment, and stomach or digestive discomfort with oral medicines. Newer medicines work in different ways. Denosumab blocks RANKL, a signal that activates bone-breaking cells called osteoclasts. Small studies in children have reported increases in bone density and growth, but stopping treatment can cause bone breakdown to speed up again and calcium levels in the blood to rise dangerously. In one study comparing treatments in children, 31% of those given denosumab developed high calcium levels after treatment, and nearly half of those cases progressed to a severe condition called hypercalcemic crisis. Romosozumab and setrusumab block sclerostin, a protein that limits bone formation, with the aim of helping the body build more bone. However, evidence for these medicines in OI is still limited to adult case reports and early-stage trials. Fresolimumab blocks transforming growth factor-beta. In a phase I trial in adults, its effects on bone turnover depended on the dose, while changes in lower-spine bone density varied. No serious adverse events were linked to treatment in that trial. These medicines are still being studied for OI, and the available results do not prove that they work in children.

Experimental treatments aim to target the cells or genes responsible for OI. Mesenchymal stromal cells, which can come from bone marrow, umbilical cord blood or fetal liver, can develop into bone-building cells and influence nearby cells by releasing chemical signals. Clinical studies using repeated infusions of these cells from donors who shared one copy of some immune-system genes with the recipient have reported fewer fractures and better quality of life. In one prenatal transplant using cells from fetal liver, 7.4% of the transplanted cells became established in the recipient. Bone structure returned to a more normal appearance, and the child had a period after birth with very few fractures. However, the benefits faded as the donor cells disappeared. This highlights the challenge of keeping transplanted cells in the body for a long time. CRISPR-Cas9 gene editing has been used to correct COL1A1 and COL1A2 variants in induced pluripotent stem cells made from patients. After these cells were developed into other cell types, they showed improved collagen production, alkaline phosphatase activity and mineralization. In a mouse model of OI, researchers delivered Cas9 and a corrective DNA sequence throughout the body. They achieved a 12% rate of correction through a process called homology-directed repair in bone. The treatment was linked to improvements in bone mass, skeletal deformity, fracture healing and grip strength. These results come from laboratory experiments or animal studies, not established human treatments. Low editing efficiency, unintended changes in other parts of the DNA, delivery to the right cells and long-term effects remain major obstacles to clinical use.

Care tailored to each person also depends on surgery and support from several types of health professionals. Telescopic rods placed inside bones, including the Fassier-Duval system, have improved walking ability and reduced the need for repeat surgery compared with rods that do not lengthen. One study reported that 88% of telescopic rods remained in place and functional after four years, compared with 40% of non-telescopic rods. However, telescopic rods can still cause problems, including movement out of place, bending, failure to lengthen as intended and issues with fixing the lower end of the rod. The evidence therefore supports combining surgery with medicines that slow bone breakdown or treatments still under study, rehabilitation and support for development. Outcomes still depend on how severe the condition is, the shape and structure of the bones, and when treatment begins. As genetic classification improves, its most immediate benefit is to help identify the specific biological problem, guide checks for related complications and investigate whether people with a particular genetic subtype respond differently to treatment. This approach connects each person's bone problems to a measurable molecular cause while recognizing that care must also address fractures, deformities, breathing problems, hearing loss and imperfect tooth development.

Subject of Research: Genetic causes, differences in symptoms and personalized care for osteogenesis imperfecta

Article Title: Pathogenesis and precision management of osteogenesis imperfecta: from genetic mechanisms to novel therapies

Article References: Wang, X.-Q., Zhang, C., Li, X., Sun, Y., Chen, Q., Yang, B.-Y., Li, Y.-S., Jiang, X.-Y., Zhao, Z.-H., Wang, L., Sun, X.-H., Wang, H.-Y., Lu, W.-L., Wei, H.-Y., Su, Z., Ren, X.-Z., Cao, B.-Y., Wu, W., & Chen, T. (2026). Pathogenesis and precision management of osteogenesis imperfecta: from genetic mechanisms to novel therapies. World Journal of Pediatrics, 22(9), 982-1011. https://doi.org/10.1007/s12519-026-01086-7

Image Credits: AI Generated

DOI: 10.1007/s12519-026-01086-7

Keywords: osteogenesis imperfecta, fragile bones, genetic diagnosis, personalized medicine, collagen biology, bisphosphonates, gene editing, mesenchymal stromal cells, Asian populations, skeletal disorders, disease causes, precision treatment