Predictors of Fracture Risk in Patients with Probable Hypophosphatasia Identified within an Electronic Health Record Database

Presentation Number: MON 336
Date of Presentation: April 3rd, 2017

Joseph Biskupiak*1, Casey Tak1, Diana Brixner1, Ioannis C. Tomazos2, Gilbert L’Italien2, Bonnie Donato2 and Andrew E. Denker2
1University of Utah, Salt Lake City, UT, 2Alexion Pharmaceuticals, Inc., New Haven, CT


Hypophosphatasia (HPP) is a rare, systemic, metabolic disease caused by loss-of-function mutation(s) in the ALPLgene, leading to low tissue-nonspecific alkaline phosphatase (TNSALP) activity. Consequently, subnormal TNSALP activity leads to chronic increase of substrates, namely inorganic pyrophosphate (PPi) and pyridoxal-5′-phosphate (PLP). PPi is a known inhibitor of bone mineralization that negatively affects bone growth and homeostasis; PLP can negatively affect brain function and homeostasis. HPP manifestations are broad and systemic, ranging from stillbirth to developmental delays, failure to thrive, poor bone health/deficient mineralization, premature tooth loss, multiple fractures, severe pain, impaired mobility, and disability. HPP is often misdiagnosed, namely due to low disease awareness and absence of age-/sex-adjusted alkaline phosphatase (ALP) reference ranges. We sought to determine the predictors of fracture risk in HPP.

We queried the University of Utah Health Care electronic health record (EHR) database housing data from 1.6 million individuals (Jan 1990–Dec 2014) for disorders of phosphorous metabolism (ICD9 275.3) or low age-/sex-adjusted ALP (CPT 84705). Patients (pts) with persistently low ALP activity (≥2 measurements) were screened for clinical, biochemical, and radiographic evidence of HPP: seizures/respiratory failure in children <5 y old; elevated serum PLP; elevated urine phosphoethanolamine; HPP family history; radiographic evidence of hypomineralization/osteopenia/rickets; osteomalacia; history of nontraumatic and/or multiple fractures/premature tooth loss/craniosynostosis. Pts with ≥1 of these additional manifestations were considered pts with probable HPP. A Prentice-Williams-Peterson gap time regression model (extension of Cox proportional hazards model) for recurrent events was used to model overall effect of covariates on skeletal fractures. University of Utah IRB approved the protocol.

Eighty-three pts (4 infants [age 0–6 mo], 9 children [7 mo–5 y], 16 teens [6–17 y], 54 adults [≥18 y]) were identified as probable HPP pts by persistently low age-/sex-adjusted ALP and relevant symptomatology. Mean follow-up was 5.7 y; median number of serum ALP tests was 7 (range 2–85). Statistically significant predictors of fracture risk were pain (treatment/diagnosis; HR 1.820; P<0.0001), depression/anxiety (treatment/diagnosis; HR 1.332; P=0.0340), and both infant (HR 1.401; P=0.0347) and children (HR 1.617; P=0.0063) age groups (reference adults). No other covariates significantly predicted fracture risk.

EHR longitudinal data analysis suggests that a panel of symptomatic and clinical risk factors predicts fracture incidence in pts with probable HPP. Such analyses may prove useful in developing risk prediction algorithms for disease progression in this rare, debilitating disease.


Disclosure: JB: Research Funding, Alexion Pharmaceuticals, Inc.. CT: Research Funding, Alexion Pharmaceuticals, Inc.. DB: Research Funding, Alexion Pharmaceuticals, Inc.. ICT: Employee, Alexion Pharmaceuticals, Inc.. GL: Employee, Alexion Pharmaceuticals, Inc.. BD: Employee, Alexion Pharmaceuticals, Inc.. AED: Employee, Alexion Pharmaceuticals, Inc..