| Characteristic | Numerical or molecular finding | Evidence type and inference limit | Concise source |
|---|---|---|---|
| Epidemiology and smoke exposure | Annual incidence is **0.2–19.1 per 100,000** children aged 0–14; males are affected approximately **3–5 times** as often as females. A Swedish case–control study (**852 cases; 4,432 controls**) found adjusted ORs of **1.4** for parental smoking under 10 cigarettes/day and **2.0** for over 10/day. | Population studies and reviews; incidence varies markedly by geography. Smoking evidence is observational, may retain socioeconomic confounding, and does not establish causation. | Joseph et al., 2023, DOI: [10.1177/18632521231203009](https://doi.org/10.1177/18632521231203009); Spasovski et al., 2023, DOI: [10.3390/diagnostics13030471](https://doi.org/10.3390/diagnostics13030471) (pqac-00000021, pqac-00000023, pqac-00000048) |
| Familial **COL2A1** disease | Heterozygous **COL2A1 c.1888G>A (p.Gly630Ser)** segregated with LCPD or avascular necrosis in a four-generation, **45-member** family; an asymptomatic seven-year-old carrier illustrated age-dependent or incomplete expression. Familial **c.3665G>A (p.Gly1170Ser)** has also segregated with LCPD, adult avascular necrosis, and osteoarthritis. | Human pedigrees support autosomal-dominant type II collagenopathy in rare families. Variant numbering can depend on transcript. These variants are **not established as the usual mechanism of sporadic LCPD**. | Li et al., 2014, DOI: [10.1371/journal.pone.0100505](https://doi.org/10.1371/journal.pone.0100505); Spasovski et al., 2023, DOI: [10.3390/diagnostics13030471](https://doi.org/10.3390/diagnostics13030471) (pqac-00000052, pqac-00000058, pqac-00000068, pqac-00000069) |
| Synovial inflammation | A research summary reports significant **IL-6 overexpression in synovial fluid from 13 patients** with active LCPD. Elevations of HMGB1, TNF-alpha, and IL-1beta and persistent synovitis have also been reported. | Small human biomarker studies support local inflammation but do not prove that IL-6 initiates ischemia. Mechanistic evidence from IL-6 knockout or receptor blockade is predominantly preclinical. No validated clinical IL-6 threshold exists. | Spasovski et al., 2023, DOI: [10.3390/diagnostics13030471](https://doi.org/10.3390/diagnostics13030471) (pqac-00000051, pqac-00000059, pqac-00000062) |
| Ischemia and staged evolution | Interruption of capital-femoral-epiphyseal blood supply leads to avascular necrosis. Modified Waldenström stages are **Ia/Ib** (necrosis), **IIa/IIb** (fragmentation), **IIIa/IIIb** (reconstitution), and **IV** (healed). Early loading deformation can be reversible; collapse becomes irreversible particularly in IIb–IIIa. Healing usually takes **2–4 years**. | Vascular imaging, sequential radiographs, pathology, and biomechanical observations strongly support ischemia and staged repair. The initiating cause of vascular interruption remains unknown. | Joseph et al., 2023, DOI: [10.1177/18632521231203009](https://doi.org/10.1177/18632521231203009) (pqac-00000018, pqac-00000022, pqac-00000023, pqac-00000032) |
| Vitamin D and stage duration | In a 2024 prospective cohort of **50** children, fragmentation lasted **15.5 ± 1.3 months** with 25-hydroxyvitamin D deficiency, **14.2 ± 1.9 months** with insufficiency, and **10.4 ± 3.1 months** with sufficient levels; **p=0.01**. | Single-center observational cohort restricted to unilateral stages 2–3A. The association may be prognostic but does not show that deficiency causes prolonged fragmentation or that supplementation changes outcomes. | Afaque et al., 2024, DOI: [10.7759/cureus.57274](https://doi.org/10.7759/cureus.57274) (pqac-00000037, pqac-00000038, pqac-00000039) |
| Rabbit transcriptomic model | RNA sequencing identified **77 differentially expressed lncRNAs, 239 miRNAs, and 1,027 mRNAs**. Angiogenesis- and platelet-activation genes were downregulated; a ceRNA network contained 29 lncRNAs, 28 miRNAs, and 76 mRNAs, including **HIF3A, ALOX12**, and **PTGER2**. | Induced rabbit osteonecrosis with computational network analysis. Findings are hypothesis-generating, not validated human biomarkers or proof that the nominated RNAs cause human LCPD. | Zhang et al., 2023, DOI: [10.3389/fgene.2023.1105893](https://doi.org/10.3389/fgene.2023.1105893) (pqac-00000055) |
| Natural canine disease and cell model | A 2024 study cultured MSC-like cells from **six naturally affected canine femoral heads** and compared them with **six canine bone-marrow MSC controls**. Cells showed MSC characteristics, trilineage differentiation, osteogenic potential, and measurable **HIF1A, VEGFA, VEGFB**, and **PDGFB** expression. | Natural canine LCPD provides comparative pathology, unlike induced piglet or rabbit models. Small samples, in-vitro conditions, and no healthy femoral-head MSC control prevent attribution of these properties specifically to LCPD. | Eto et al., 2024, DOI: [10.5455/ovj.2024.v14.i5.12](https://doi.org/10.5455/ovj.2024.v14.i5.12) (pqac-00000075, pqac-00000076, pqac-00000077, pqac-00000080) |
| Treatment evidence | Containment procedures, range-of-motion therapy, and selective weight relief are widely used. One nonrandomized comparison reported spherical healing in **75%** of hips protected from weight bearing until stage IIIb versus **49%** after six months. Prospective studies conflict on containment benefit, and no operation is clearly superior. | Evidence consists mainly of retrospective series and nonrandomized cohorts; the 2023 review found **no Level I evidence** for preventive intervention. Bisphosphonates, RANKL or IL-6 inhibition, stem cells, and molecular therapies remain experimental. | Joseph et al., 2023, DOI: [10.1177/18632521231203009](https://doi.org/10.1177/18632521231203009); IPSG cohort NCT02040714 (pqac-00000019, pqac-00000028, pqac-00000035, pqac-00000044) |


*Table: Key epidemiologic, genetic, mechanistic, comparative, and treatment findings for Legg-Calvé-Perthes disease, with quantitative results and explicit inference limits. Human observations are distinguished from animal and in-vitro evidence.*